Microbial water quality before and after the repair of a failing onsite wastewater treatment system adjacent to coastal waters


Rachel T. Noble, Institute of Marine Sciences, University of North Carolina at Chapel Hill, 3431 Arendell Street, Morehead City, North Carolina 28557, USA. E-mail: rtnoble@email.unc.edu


Aims:  The objective was to assess the impacts of repairing a failing onsite wastewater treatment system (OWTS, i.e., septic system) as related to coastal microbial water quality.

Methods and Results:  Wastewater, groundwater and surface water were monitored for environmental parameters, faecal indicator bacteria (total coliforms, Escherichia coli, enterococci) and the viral tracer MS2 before and after repairing a failing OWTS. MS2 results using plaque enumeration and quantitative reverse transcriptase polymerase chain reaction (qRT-PCR) often agreed, but inhibition limited the qRT-PCR assay sensitivity. Prerepair, MS2 persisted in groundwater and was detected in the nearby creek; postrepair, it was not detected. In groundwater, total coliform concentrations were lower and E. coli was not detected, while enterococci concentrations were similar to prerepair levels. E. coli and enterococci surface water concentrations were elevated both before and after the repair.

Conclusions:  Repairing the failing OWTS improved groundwater microbial water quality, although persistence of bacteria in surface water suggests that the OWTS was not the singular faecal contributor to adjacent coastal waters. A suite of tracers is needed to fully assess OWTS performance in treating microbial contaminants and related impacts on receiving waters. Molecular methods like qRT-PCR have potential but require optimization.

Significance and Impact of Study:  This is the first before and after study of a failing OWTS and provides guidance on selection of microbial tracers and methods.


Contamination of coastal and estuarine systems by faecal contamination, including bacterial and viral pathogens, represents a substantial public health risk, particularly in waters used for recreation or shellfish harvesting. Faecal contamination to coastal waters may originate from a variety of point and nonpoint sources. Point sources such as the effluent pipe of a wastewater treatment plant or ‘any discernible, confined and discrete conveyance…from which pollutants are or may be discharged’ [Code of Federal Regulations, CFR 40 502(14)] are regulated by the United States Environmental Protection Agency (USEPA) through the National Pollutant Discharge Elimination System permit program. Nonpoint sources encompass all other sources of surface water pollution, such as agricultural or stormwater runoff, leaking underground storage tanks, and decentralized or onsite wastewater treatment systems (OWTS, historically referred to as septic systems). Faecal contamination of coastal waters from nonpoint sources has been documented previously in eastern North Carolina (NC) (Kirby-Smith and White 2006; Line et al. 2008; Parker et al. 2010). However, the diffuse nature of nonpoint-source pollution causes difficulty in determining relative contributions from various sources, which is required for remediation and management.

Onsite wastewater treatment systems can be a nonpoint source of faecal contamination to coastal and estuarine systems (Paul et al. 2000; Lipp et al. 2001; Carroll et al. 2005; Cahoon et al. 2006). In a conventional OWTS, wastewater from a single-family home is treated onsite by a septic tank through flow equalization, settling and anaerobic biodegradation prior to discharge of the septic-tank effluent to a subsurface soil treatment unit. Septic-tank effluent either flows by gravity or is pumped to drainlines typically located in gravel trenches within 0·6 m of the ground surface. Effluent infiltrates the underlying native soil and percolates through the unsaturated vadose zone, where additional treatment of contaminants can occur via filtration, sorption, biodegradation and die off prior to groundwater recharge (USEPA 2002; Van Cuyk et al. 2004). Risk of contamination of groundwater or surface waters from OWTS may be higher in coastal areas with shallow groundwater and sandy soils. This is especially true during periods of high water tables combined with heavy rainfall events. In eastern NC, this risk is exacerbated because of the high density of OWTS (>50% of residences) and episodic heavy precipitation events including hurricanes and tropical storms, which leads to temporary or long-term failure of OWTS as evidenced by surfacing effluent, patterns of grass growth, and contamination of nearby ditches and streams.

Faecal indicator bacteria (FIB) such as Escherichia coli (E. coli) and enterococci are utilized as proxies of faecal contamination, and their occurrence in recreational waters have been associated with human health risk (Prüss 1998; Wade et al. 2006, 2008). Approximately 19% or 429 000 acres of NC’s potential shellfishing waters are permanently closed to harvesting because of the faecal coliform concentrations that exceed regulatory thresholds [NC Shellfish Sanitation and Recreational Water Quality Section (NCSSRWQS), personal communication]. In 2010, at least one swimming advisory was in effect along the NC coast for 177 days of the entire 214-day swimming season as a result of enterococci concentrations exceeding regulatory thresholds (NCSSRWQS 2010). However, E. coli and enterococci are found in the guts of all warm-blooded animals and therefore, are not detected only during a contamination event from a human-specific source such as faecal material from OWTS, which confounds source-tracking efforts.

The overall goal of the current investigation was to conduct before and after studies of a failing OWTS for treatment of microbial pathogens using a suite of parameters and tracers, and to investigate the potentially positive impact of a functioning system on the receiving water quality. Wastewater, groundwater, and surface water quality were monitored over time using environmental water quality parameters, FIB, dyes and the viral tracer MS2. The nonpathogenic, male-specific coliphage MS2 has been successfully employed as a tracer of enteric viral contamination in OWTS (DeBorde et al. 1998; Van Cuyk et al. 2004) and has typically been measured using plaque enumeration techniques (USEPA 2000). We quantified coliphage MS2 using both a plaque enumeration approach and quantitative reverse transcriptase polymerase chain reaction (qRT-PCR) and discuss the benefits of each method. The information will aid in understanding the impact of OWTS repair on microbial fate and transport and provide guidance into the selection of microbial tracers and their methodology in future faecal contamination studies.

Materials and Methods

Site characterization

The study was conducted near the Newport River Estuary in coastal Carteret County, NC (Fig. 1a). A high proportion (approximately two-thirds) of residences in the county utilizes OWTS for wastewater treatment and discharge. The region is characterized by sandy soils and shallow water tables that fluctuate with heavy rainfall and tidal exchange (NRCS 1987). In addition, the area is impacted by episodic weather events including hurricanes and tropical storms, with an average rainfall of approximately 130 cm year−1. Much of the Newport River Estuary has restricted use for shellfish harvesting because of the high concentrations of faecal coliforms (Fig. 1a) and cited nonpoint sources include OWTS (NCDENR 2005).

Figure 1.

 (a) Shellfish growing area classifications in the Newport River estuary in eastern North Carolina and (b) the location of the onsite wastewater treatment system monitored during the current study and its proximity to those waters. (Shellfish map courtesy of B. Pogue, North Carolina Department of Natural Resources. Aerial view courtesy of GoogleEarth).

An OWTS located adjacent to Core Creek which feeds into the Newport River Estuary was identified for study (Fig. 1b). Site and system details are described in Data S1. The system had been in operation for 13 years and was currently serving a four-person single-family home. In the original system, septic-tank effluent flowed to a distribution box (d-box) and was distributed by gravity flow into four drainlines. At a depth of 0·3–0·6 m below ground surface, the soil was characterized as a sandy loam soil, with a texture of 71·2% sand, 16·0% silt, and 12·8% clay, pH = 7·6, 0·36 mmhos cm−1 of conductivity, and 1·7% organic matter (Harris Laboratory, Lincoln, NE, USA). The groundwater table was <1·2 m below ground surface and could fluctuate near ground surface during periods of heavy precipitation.

The OWTS was determined to be failing based on wastewater odours above the soil treatment unit and surfacing effluent after heavy rainfall. In a preliminary dye study, fluorescein dye (20% w/v; Norlab Inc., Amherst, OH, USA) was added to the d-box 24 h before a rain event. Fluorescein was visually detected on the ground surface above the soil treatment unit and in a nearby ditch within 24 h after the rain event (Fig. S2).

A network of 11 groundwater-monitoring wells was installed to a depth of 1·2 m below the ground surface around the OWTS (Fig. S1A). Monitoring of wastewater from the d-box, groundwater from the 11 monitoring wells, and surface water from five nearby property line ditches and the adjacent coastal receiving waters for environmental parameters (pH, water temperature, dissolved oxygen, specific conductance and salinity), FIB (total coliforms, E. coli and enterococci) and the coliphage MS2 tracer was conducted from March through December 2008 (see Tracer tests and sample collection).

In August 2009, the failing OWTS was repaired (see Data S1). In the repaired system, septic-tank effluent from a pump tank was pumped through a pressure manifold system to drainlines located in a mounded system above a field of new, clean sand. Five monitoring wells were installed around the gravel bed (Fig. S1B). Monitoring of wastewater from the pump tank, groundwater from the five monitoring wells, and surface water from six nearby ditches and the adjacent coastal receiving waters for environmental parameters (pH, water temperature, dissolved oxygen, specific conductance, and salinity), FIB (total coliforms, E. coli, and enterococci), and the coliphage MS2 tracer was conducted from September through December 2009 (see Tracer tests and sample collection).

Tracer tests and sample collection

A viral tracer study using the male-specific coliphage MS2 was conducted in August 2008 (prerepair) and November 2009 (postrepair). A stock solution of MS2 (American Type Culture Collection, ATCC, number 15597-B1) was prepared by inoculating the ampicillin/streptomycin-resistant E. coli host (E. coli Famp, ATCC no. 700891) according to ATCC protocol. The final MS2 concentration in the stock solution was determined using the double agar layer procedure (USEPA 2000). MS2 was added to the d-box of the prerepaired system (2·75 × 1010 plaque-forming units, PFU, solution) and to the pressure manifold pipe of the postrepaired system (1·36 × 1014 PFU solution). In the postrepair system, MS2 was added to the pipe rather than the pump tank (which replaced the d-box) because of the large amount of tracer required to add a comparable dose (d-box volume ∼ 10 l vs pump tank volume = 5678 l) and the inability to add a single pulse of tracer (a fraction of the total pump tank volume was dosed to the field every 2 h, and the pump tank contents were regularly diluted with incoming septic tank effluent). The d-box wastewater was immediately sampled after addition to determine the starting concentration of MS2 in the prerepair test. In the postrepair test, it was assumed that all of the MS2 added to the pipe was distributed to the six drain lines during the first few pressurized dosing events because this addition method had been successfully tested at the site 3 months earlier.

Water samples were collected on a daily basis at the start of the experiments with decreasing temporal frequency as the tests continued. Fourteen sampling events occurred over 3 months during the prerepair test, and seven sampling events occurred over 3 weeks during the postrepair test. Grab samples of wastewater from the d-box or pump tank and surface water from ditches (when wet) and the coastal receiving waters were collected. Groundwater samples were collected via the monitoring wells using a peristaltic pump (Durham Geo Slope Indicator TR-200, Stone Mountain, GA, USA). The Tygon® tubing was disinfected with 5% bleach and rinsed with 5% sodium thiosulfate and then with deionized water between samples. All samples were collected in sterile nalgene bottles, stored at 4°C, and laboratory analyses were conducted within 8 h of sample collection.

Analytical methodology

Water temperature, dissolved oxygen and specific conductance were measured in the field using an Orion 5-Star portable meter (Thermo Scientific, Beverly, MA, USA). In the laboratory, pH was measured using a KCl electrode and benchtop meter (UltraBasic UB-5 and electrode model 300729·1, Denver Instrument, Bohemia, NY, USA) and salinity was measured using a refractometer (VISTA Model A366ATC). Samples were analyzed in laboratory replicates for total coliforms and E. coli using Colilert-18® and enterococci using EnterolertTM (IDEXX Laboratories, Inc., Westbrook, ME, USA) according to the manufacturer’s specifications. Each 100-ml sample (often diluted 1 : 10 or more with deionized water) was mixed with the reagent and poured into a 97-well tray. After incubation, the number of positive wells per tray was converted to a corresponding most probable number (MPN) per 100-ml sample based on the statistical method of Hurley and Roscoe (1983).

MS2 concentrations were determined by two methods: a plaque enumeration technique and a qRT-PCR technique modified from the technique followed by Gregory et al. (2006). In the single agar layer procedure (USEPA 2000), 100 ml of sample (after appropriate dilution) in laboratory replicates was added to an equal volume of 48°C tryptic soy agar containing ampicillin and streptomycin (100 μg each ml−1). The mixture was supplemented with MgCl2 and the host E. coli Famp, allowed to harden and followed by incubation at 36°C for 16–24 h. Plaques (clear zones of lysis) on the plates were counted and expressed as PFU per 100 ml.

For qRT-PCR, aliquots of 10–100 ml of sample were filtered through a 0·45-μm mixed cellulose ester type HA filter (Millipore, Billerica, MA, USA) in duplicate, and the filters were stored at −80°C. RNA was extracted using a modified version (see Data S2) of the Qiagen RNeasy® Mini kit (Valencia, CA, USA). Each sample was amended with 100 ng of mouse lung RNA coding for β-actin (AM7818, Applied Biosystems, Foster City, CA, USA), which served as an inhibition control to assess matrix inhibition during qRT-PCR. An MS2 standard (107 PFU MS2 per filter), an MS2 blank (HA filter), and a β-actin blank (HA filter with no β-actin added to the extraction buffer) were included in each set of 30 extractions. Eluted RNA was analyzed by qRT-PCR using the Qiagen OneStep RT-PCR Kit first for RNA β-actin to assess inhibition followed by analysis for MS2 (see Data S2). A standard curve was created by serially diluting the extracted RNA from the 107 PFU MS2 standard. A nuclease-free water negative template control was included in each qRT-PCR run. All β-actin sample analyses were run in singlet, and all MS2 analyses were run in duplicate. Reactions were performed on a SmartCycler® II (Cepheid, Sunnydale, CA, USA). The cycle at which sample fluorescence exceeded background fluorescence (cycle threshold, CT) was recorded. The concentration of MS2 (as PFU per 100 ml) was determined using the ΔCT method (Pfaffl 2001) by calculating the amplification efficiency (E) from the slope of the software-generated standard curve based on the equation: = 10(−1/slope)−1.

The optimal amount of added internal control was determined by comparing the amplification efficiency and sensitivity of qRT-PCR assays of five pre-extraction serial dilutions of a 108 PFU per filter MS2 standard containing 0, 10, 50 and 100 ng RNA β-actin per reaction. Once the optimal amount of internal control was determined, the consistency of MS2 standard curves containing the internal control was quantified by calculating the slope and efficiency statistics of a master linear regression containing seven MS2 standard curves. The statistics were compared with those from the individual standard curves. Similar statistics were calculated from five β-actin internal control standard curves containing MS2.

Quality assurance, quality control and data analysis

For comparison purposes, groundwater results were pooled into four groups: wells located less than 5 m from the wastewater source and within the soil treatment unit (GW <5), 5–15 m from the wastewater source (GW 5–15), greater than 15 m from the wastewater source (GW >15) and groundwater near the property line ditches (GW ditch). The mean concentrations of FIB and MS2 were log-transformed to normalize the data. Water quality was compared before and after the repair under dry weather conditions (7-day antecedent rainfall <2·5 cm). Impacts of rainfall (7-day antecedent rainfall >8 cm) was also assessed. Statistical differences were calculated using the two-tailed Mann–Whitney U-test. MS2 qRT-PCR nondetects and samples with CT values greater than 40 were assigned a concentration of 1 PFU per 100 ml for calculation and graphing purposes.


MS2 qRT-PCR method performance

The qRT-PCR assay consistently amplified MS2 over a 6-log concentration range regardless of the added amount of internal control, RNA β-actin (Fig. 2a). Therefore, 100 ng of the internal control was added to all samples prior to extraction. All negative extraction and template controls were negative throughout the study. The standard curves included in each qRT-PCR assay consistently quantified MS2 over 4 logs ranging from 8 × 101 to 8 × 105 PFU per reaction (103–107 PFU per filter). The amplification efficiency of the MS2 assay (Fig. 2b) was 93·7% (slope = −3·4837, R2 = 0·912, = 68 from seven standard curves). The average amplification efficiency of the individual MS2 standard curves was 92·0 ± 4·0%. The amplification efficiency of the β-actin assay (data not shown) was 95·1% (slope = −3·4445, R= 0·966, = 40 from five standard curves). The average amplification efficiency of the individual internal control standard curves was 94·8 ± 2·4%. The lowest quantifiable standard (∼80 PFU per reaction) accounted for losses during the extraction procedure. Extrapolating the standard curve linear regression to the typical instrument limit of CT = 40 produced a theoretical MS2 limit of detection of 18 PFU per sample (1·5 PFU per reaction). Using a range of 50–100 MS2 nucleic acid copies per PFU (Dreier et al. 2005; Williams 2009), the theoretical MS2 limit of detection was in the range of 900–1800 nucleic acid copies per sample or 75–150 copies per qRT-PCR.

Figure 2.

 Performance of the MS2 qRT-PCR assay: (a) MS2 standard curve performance with varying amounts of added internal control RNA β-actin, (inline image) 100 ng internal control; (inline image) 50 ng internal control; (inline image) 10 ng internal control; (inline image) 0 ng internal control and (inline image) MS2 standard and (b) MS2 standard curve compiled from seven standard curves amplified with the internal control. The amplification efficiency was calculated from the slope of the linear regression.

Samples with β-actin CT values ≥1 log unit (∼3·3 CT cycles) higher than that of the β-actin calibrator (107 PFU MS2 standard + 100 ng β-actin) were considered to be inhibited. Of the 95 environmental samples that were analyzed by qRT-PCR, 40 were inhibited. All 40 inhibited samples were prerepair groundwater samples. Of the remaining seven prerepair groundwater samples, five were partially inhibited (>1 but <3·3 CT cycles higher than the calibrator) and two were uninhibited. In contrast, there was no inhibition in the wastewater, surface water or postrepair groundwater samples (with the exception of one slightly inhibited wastewater sample). Dilution of the inhibited samples 1 : 10 and 1 : 100 with nuclease-free water failed to resolve the inhibition, so results are reported from the 1 : 1 dilution of all samples.

Transport of the viral tracer MS2

During the prerepair tracer test, MS2 levels in the d-box wastewater declined over time, with approximately a 4 log decrease (from ∼108 to ∼104 PFU per 100 ml) within the first 10 days after tracer addition (Fig. 3a). Much of the decrease in concentration (∼3 log) occurred within the first 3 days, as entering septic-tank effluent (∼750 l day−1) pushed d-box wastewater to the drainlines and diluted d-box concentrations. MS2 was detected in groundwater beneath the soil treatment unit within 3 days of tracer addition. MS2 continued to be measured in all sampled wells at low levels (1–70 PFU per 100 ml) 25–35 days after addition and in some wells more than 50 days after tracer addition (It should be noted that ‘GW ditch’ wells were not sampled). MS2 was also detected in the receiving waters. It was first detected 25 days after tracer addition and was measured at low levels (3–20 PFU per 100 ml) for longer than 50 days. The detections in the receiving waters may be due to MS2-impacted groundwater recharging the ditch system and receiving waters, or it may be due to overland transport of surfacing effluent from the d-box and drainlines under wet conditions (20 cm of rain fell during the 51 days of monitoring).

Figure 3.

 Comparison of the plaque enumeration results from the (a) prerepair (inline image) GW (< 5m), (inline image) GW (5–15m), (inline image) GW (> 15m) and (b) postrepair MS2 tracer test to the quantitative reverse transcriptase-polymerase chain reaction (qRT-PCR) results from the (c) prerepair and (d) postrepair MS2 tracer test. (Nondetects visually shown as 0 log PFU/100 ml; Error bar = 1 SD). (inline image) Wastewater; (inline image) Groundwater and (inline image) Receiving Waters.

During the postrepair tracer test, the amount of MS2 added was ∼4 log greater than that was added during the prerepair tracer test to ensure that similar amounts of tracer were discharged to the soil, given the many system design changes. In contrast to the prerepair tracer test when MS2 was detected in groundwater within 3 days, after the repair, MS2 was not detected in any of the wells for 10 days after tracer addition, even in samples collected within 24 h after Tropical Storm Ida delivered 20 cm of rainfall (Fig. 3b). In addition, MS2 was not detected in the receiving waters.

In general, MS2 results based on plaque enumeration (Fig. 3a,b) were confirmed by results using the qRT-PCR assays (Fig. 3c,d). For example, the initial MS2 concentration in the d-box wastewater was very comparable using plaque-based methods (7·5 × 107 PFU per 100 ml) or qRT-PCR (8·4 × 107 PFU per 100 ml). Also, after the repair, MS2 was not detected in any groundwater samples using either method (though positive controls were positive in both methods). However, there were some differences in results between the two methods. For example, 51 days after tracer addition, MS2 was still measured in d-box wastewater using qRT-PCR while it was not detected using culture methods. This may be because only infective phages are detected using plaque enumeration methods while qRT-PCR quantifies the RNA within the viruses regardless of infectivity. Also, concentrations below the lowest quantifiable qRT-PCR standard in this study (∼80 PFU per reaction, which incorporates extraction losses) were not differentiated, while culture-based results are reported down to 1 PFU per 100 ml.

Water quality indicated by environmental parameters and faecal indicator bacteria

Wastewater composition regarding environmental parameters and FIB was within the typical range for single-family home septic-tank effluent (Lowe et al. 2009). Throughout the study, pH values ranged from 7·14 to 8·02, and conductivity ranged from 940 to 1380 μS cm−1 (see Data S3). Faecal indicator bacteria were always present in wastewater samples. Total coliform concentrations ranged from 3 × 105 to >1 × 106 MPN per 100 ml (Fig. 4a), and E. coli ranged from 3 × 104 to 6 × 105 MPN per 100 ml (Fig. 4b). Levels of enterococci were lower (< 0·01, paired t-test) than levels of E. coli in wastewater and ranged from 3 × 101 to 6 × 103 MPN per 100 ml (Fig. 4c), typical of previous findings relating enterococci and E. coli concentrations in human waste (Blanch et al. 2006).

Figure 4.

 Comparison of average concentrations of (a) total coliforms, (b) Escherichia coli, and (c) enterococci in wastewater (WW), groundwater <5 m (GW <5), between 5 and 15 m (GW 5–15), and >15 m (GW >15) from the wastewater discharge point, groundwater near the property ditch (GW ditch), ditches, and the coastal receiving water (RecWat) before and after the repair of a failing onsite wastewater treatment system. (Averages calculated assuming values <10 = 5; Error bar = 1 SD). (inline image) Pre-repair, dry; (inline image) Pre-repair, wet; (inline image) post-repair, dry and (inline image) post-repair, wet.

Prior to the repair, conductivity values were lower (< 0·01) in groundwater collected from wells located outside of the soil treatment unit (‘GW ditch’ average = 870 μS cm−1, Data S3) as compared to groundwater from wells located within the soil treatment unit (‘GW <5’, ‘GW 5–15’, ‘GW >15’ = 1250, 1200, 1260 μS cm−1, respectively). Average total coliform concentrations in groundwater ranged from 2 × 103 to 1 × 104 MPN per 100 ml (Fig. 4a‘Pre-repair, Dry’) and were similar regardless of the distance from the d-box. Average E. coli concentrations in groundwater samples ranged from <10 to 20 MPN per 100 ml. Eighty per cent of groundwater samples had E. coli values <10 MPN per 100 ml, and the majority of detections were in well six (maximum concentration = 400 MPN per 100 ml) that was located 5–15 m from the wastewater source (Fig. 4b‘Pre-repair, Dry’). Interestingly, E. coli was detected only once in well seven that was located adjacent to the same drainline as well six but closer to the d-box (<5 m). While there was a greater abundance of E. coli than enterococci in wastewater, enterococci in groundwater were detected more frequently than E. coli (65 vs 20%) and were detected at least once in groundwater from every well at an average concentration of 100 MPN per 100 ml (Fig. 4c‘Pre-repair, Dry’). Levels of all three FIB were higher (< 0·001) in ditches than in groundwater (Fig. 4a–c‘Pre-repair, Dry’). Levels were higher in receiving waters than in groundwater for E. coli (< 0·001) and enterococci (< 0·05).

Total coliform concentrations were lower (< 0·01) after the repair as compared to before the repair for pooled groundwater data and for each groundwater group except ‘GW 5–15’ (Fig. 4a‘Postrepair, Dry’vs‘Pre-repair, Dry’). E. coli concentrations were less than 10 MPN per 100 ml in all groundwater samples (Fig. 4b‘Postrepair, Dry’). Enterococci (Fig. 4c‘Postrepair, Dry’) were detected in groundwater from three of the five wells, and average groundwater concentrations were similar after the repair (50 MPN per 100 ml) as compared to before the repair (100 MPN per 100 ml).

Despite improved OWTS function as shown by reduced concentrations of MS2, total coliforms and E. coli in groundwater, all three FIB were detected in the receiving waters after the repair (Fig. 4a–c‘Postrepair, Dry’). Before the repair, the E. coli geometric mean in the coastal receiving waters was 75 MPN per 100 ml under dry conditions (= 7), which exceeds NC’s faecal coliform limits for shellfish harvesting areas of 14 MPN per 100 ml (NCAC 15A.2B.0200). After the repair, E. coli concentrations were still elevated during dry conditions (58 MPN per 100 ml) and increased ∼2·5 log during wet conditions (6·8 × 103 MPN per 100 ml, Fig. 4b‘Postrepair, Wet’). In addition, a Tier III (low use) recreational marine beach is located across the waterway from the receiving waters. Enterococci concentrations under dry conditions before and after the repair (geometric mean = 100 MPN per 100 ml, = 6; single sample = 31 MPN per 100 ml, respectively) were below NC’s limit of two consecutive samples not to exceed 500 MPN per 100 ml (NCAC 15A.18A.3400). Under wet conditions, enterococci concentrations were below the limit before the repair (116 MPN per 100 ml) and well above the limit after the repair (4·3 × 104 MPN per 100 ml, sampled just after Tropical Storm Ida, Fig. 4c‘Postrepair, Wet’). In contrast, both before and after the repair, groundwater concentrations of FIB during wet conditions were not statistically different from dry conditions (Fig. 4).


Impact of onsite wastewater treatment system repair on water quality

Levels of MS2 (Fig. 3), FIB (Fig. 4) and environmental parameters (Data S3) suggest that groundwater beneath the soil treatment unit was being impacted by infiltrating effluent prior to the repair of the OWTS. The widespread and long-term detection of low levels of MS2 in groundwater suggest incomplete treatment and little directional flow away from the soil treatment unit. The inconsistent detections of E. coli (i.e., frequent detection in well six but infrequent detection in well seven which was closer to the d-box) suggests that groundwater impacts were not necessarily proportional to proximity to the d-box, possibly because of heterogeneities in subsurface characteristics such as soil texture and groundwater flow paths. The hydraulic loading rate and fluctuating water table were system features that likely contributed to the groundwater impacts. The daily hydraulic loading rate of septic-tank effluent to the soil treatment unit ranged from 0·91 to 4·5 cm day−1 (0·22–1·1 gal ft−2 day−1), based on measured water use ranging from 500 to 2500 l day−1 and a trench area of 55 m2. This ranged from less than half to almost double the designed hydraulic loading rate of 2·5 cm day−1 (0·6 gal ft−2 day−1). The water table could fluctuate from ∼0·5 m below the drainlines under dry conditions to the level of the drainlines under wet conditions. It is likely that, during periods of high water use and wet conditions, contaminants present in the wastewater effluent received little additional treatment in the soil before reaching the underlying groundwater.

Repairing the OWTS improved groundwater quality as measured by the coliphage MS2 (Fig. 3b), total coliforms (Fig. 4a) and E. coli (Fig. 4b). Changes in the hydraulic loading rate, effluent delivery method and extent of vadose zone likely contributed to the improved treatment of microbial indicators and tracers. After the repair, the daily hydraulic loading rate of septic-tank effluent to the soil treatment unit was limited to 1·7 cm day−1 (0·4 gal ft−2 day−1), based on a maximum dosing volume of 1100 l day−1 and a bed area of 65 m2. The pump system evenly distributed the effluent to the soil treatment unit both spatially and temporally, as the drainlines were pressurized during each dosing event that delivered approx. 96 l every 2 h. Dosing enhances alternating saturated and unsaturated conditions at the soil infiltrative surface, which may improve treatment via filtration, sorption and predation (USEPA 2002). By raising the drainlines in a mound above the ground surface, the extent of vadose zone between the drainlines and the groundwater increased, and the soil moisture content likely decreased. A reduction in soil moisture content in the unsaturated zone has been shown to enhance bacteria and virus sorption (Lance and Gerba 1984; Powelson and Mills 2001), and the increased retention time in the unsaturated zone may provide additional opportunities for cell die-off and predation. In addition, the property boundary ditches that had often been wet even under dry conditions prior to the repair were dry after the repair, lending evidence to the improved hydraulic performance of the system.

In addition to hydraulic changes, geochemical changes to the system likely also contributed to improved performance. The replacement of the native sandy loam soil in and 1·2 m below the soil treatment unit with sand that had been dredged from saline estuarine areas decreased the groundwater pH from neutral (6·99–8·72) to acidic levels (3·29–6·76), while conductivity increased from ∼1000 μS cm−1 to up to 8440 μS cm−1 (Data S3). Groundwater salinity was 0‰ in all prerepair samples and was measured up to 14‰ after the repair (Data S3). This increase in ionic strength and acidity in the unsaturated zone and groundwater likely enhanced the removal of virus and, to a lesser extent, bacteria by sorption to soil (Bales et al. 1995; Jewett et al. 1995; Walshe et al. 2010). As ionic strength increases, the double layer of ions surrounding soil grains decreases, allowing negatively charged particles such as E. coli, enterococci and MS2 to approach like-charged soil particles and be held there by van der Waals forces. Knappett et al. (2008) reported no MS2 breakthrough from saturated columns containing fine or medium sand under high-ionic-strength conditions (34 mmol l−1), while MS2 was detected in effluents from both column types under low-ionic-strength-conditions (5 mmol l−1). In addition, the electrostatic repulsion becomes weaker as pH decreases (the isoelectric point of MS2 is pH = 3·9), allowing for greater attachment of MS2 to the porous media.

The consistently high FIB concentrations, especially during wet weather, in surface waters adjacent to the OWTS both while it was failing and after it was repaired suggest that there are other sources of faecal material in the surrounding environment that are contributing to the elevated surface water levels. The highest ditch E. coli concentration (2 × 104 MPN per100 ml) was measured after the system was repaired from the ditch transporting water from a different part of the neighbourhood to the receiving water (Ditch 6, Fig. S1B). The highest enterococci concentration (4 × 104 MPN per 100 ml) was also measured after the repair in the ditch transporting water from the upstream neighbourhood to the receiving water (Ditch 1, Fig. S1B). Possible sources include flushing of untreated wastewater or naturalized soil bacteria (Brennan et al. 2010) from other OWTS in the neighbourhood, domestic pets such as dogs, wildlife such as deer and raccoons, and birds (Boehm et al. 2003), all of which are present in the study area.

Comparison of tracers

The traditional culture-based technique for MS2 enumeration requires more than 16 h after sample collection for results. In contrast, results were available within 5 h using the qRT-PCR method. However, less expense and training is required to conduct the culture-based methods as compared to the qRT-PCR technique. From a public health standpoint, it is likely that the actual number of infectious viruses in a sample may be overestimated by qRT-PCR (which quantifies RNA regardless of infectivity) and underestimated by plaque enumeration (as there are likely multiple viruses within a single plaque). Persistence of viable but noninfectious virus in the environment depends on a number of site conditions, including soil moisture, temperature, sunlight, ability to attach to soil or suspended particulates, and microbial activity (Tsai et al. 1995; Schijven and Hassanizadeh 2000). Plaque enumeration provides results down to 1 PFU per 100 ml, while the reporting level for the qRT-PCR method was higher (the lowest quantifiable standard was ∼80 PFU per reaction and the extrapolated reporting level was ∼1·5 PFU per reaction or ∼75–150 nucleic acid copies per reaction) because it accounted for losses because of extraction and inhibition. Indeed, inhibition is currently the major limitation of qRT-PCR.

This study identified possible factors affecting the level of inhibition that could guide future method optimization studies. All of the inhibited samples were prerepair groundwater samples. Similarly, of 97 environmental samples from three nearby OWTS that were analyzed for a related study (Habteselassie et al. 2011) the 25 that were inhibited were groundwater samples. This suggests that inhibitory compounds such as humic and fulvic acids are more likely in wastewater-impacted subsurface samples (i.e., underlying groundwater) than in other types of environmental samples (i.e., wastewater and surface water), perhaps because of the development of organic-rich biofilms in the subsurface. In addition, all seven prerepair groundwater samples that were uninhibited or only slightly inhibited were collected at the end of the tracer test in October, which suggests a need to test for seasonal differences in inhibition. We feel that the presented qRT-PCR approach has promise for future OWTS and faecal contamination studies. For example, the method has been employed to identify human enteroviruses (Gregory et al. 2006) to differentiate human and nonhuman sources of faecal contamination. However, thoughtful optimization of ultraconcentration approaches and attention to removal of inhibitory compounds is required.

The results of this investigation confirm the need to utilize a suite of tracers in future OWTS studies. Monitoring of environmental parameters such as pH and conductivity identified repair-induced geochemical changes in the subsurface that affected the fate and transport of micro-organisms. The differences in abundance and survival of FIB emphasize the limitations of using a single indicator of bacterial contamination in complex environmental systems such as OWTS. In particular, enterococci can survive longer than E. coli in faecal-impacted sediment and soil (Howell et al. 1996; Cools et al. 2001) and under extreme environments including variations in pH, salinity and moisture. In this study, the survival and transport of the viral tracer MS2 differed from that of FIB, which supports previous studies that have shown the occurrence of these FIB to be poorly correlated with the occurrence of viral pathogens in coastal waters (e.g., Noble and Fuhrman 2001; Fong et al. 2005). In future OWTS-monitoring studies, a multi-tiered approach could be employed utilizing simple cost-effective methods to monitor a suite of environmental parameters and conventional microbial tracers including FIB and phage. Molecular methods, such as qRT-PCR, need optimization and further testing but are promising additions as a second tier to the ‘toolbox’ of microbial tracers because of their pathogen specificity and rapidity. The findings are a first step of developing a systematic approach towards selecting the type, number and locations for pre/postmonitoring of OWTS and in the selection of specific microbial tracers to be monitored.


This research was funded by the North Carolina Division of Water Quality (Contract no. EW06032) through the US Environmental Protection Agency Section 319 grant program. The authors thank Troy Dees, Ben Kane, and Daniel Allen from the Carteret County Environmental Health Department; and Patti Fowler, J.D. Potts, Shannon Jenkins, Brad Pogue, Erin Bryan-Millush, and Ginger Kelly from the North Carolina Shellfish Sanitation and Recreational Water Quality Section. We are grateful to Marek Kirs, Raul Gonzalez, and Brianna Young for assistance with sampling and analysis, and Reagan Converse, Curtis Stumpf, and the anonymous reviewers for technical review. We thank IDEXX for materials support. We also thank the homeowners for access on their property and use of their onsite wastewater treatment system throughout the research project.