Field Evaluation of Indicator Bacteria Removal by Stormwater BMPs in North Carolina

In the United States Environmental Protection Agency’s National Water Quality Inventory in 2000, 13% of the river and stream miles that were surveyed were impaired by pathogen indicator bacteria (USEPA 2002). Stormwater runoff is a transport mechanism for indicator bacteria to receiving waters, resulting in an increased risk to public health through consumption of contaminated shellfish or ingestion by swimmers. Urban stormwater is commonly treated by stormwater Best Management Practices (BMPs), each of which provides some combination of natural treatment mechanisms and fosters certain environmental conditions. Although BMPs have been studied in detail for many pollutants, little peer-reviewed literature is available which documents their ability to remove or inactivate indicator bacteria. The North Carolina State University Department of Biological and Agricultural Engineering evaluated 10 stormwater BMPs in Charlotte and Wilmington, NC, to evaluate their efficiency with respect to indicator bacteria removal. The study practices included two bioretention cells, four stormwater wetlands, two wet ponds, and two dry detention areas. Data collected from these studies indicates that positive removal of indicator bacteria is possible in many types of BMPs; however, removal can be highly variable from practice to practice. Further, stormwater BMPs may foster environments where indicator bacteria can persist, becoming sources of indicator bacteria. Finally, even if positive reductions in indicator bacteria are noted, research indicates that achieving effluent concentrations of indicator bacteria consistent with USEPA standards may be difficult with many types of BMPs. Introduction In the United States Environmental Protection Agency’s (USEPA 2008) National Water Quality Inventory in 2006, approximately 12% of the river and stream miles that were surveyed were impaired by indicator bacteria. Of the stream and river miles designated as impaired, either unable or partially unable to meet their designated use, more were impacted by this pollutant than by any other. Indicator bacteria were also the number one source of impairment in bays and estuaries, the number two source of impairment in oceans and near coastal areas, and the number three source of impairment along coastal shorelines (USEPA 2008). In light of the negative impact that indicator bacteria have on surface waters in the United States (indicating the 1123 World Environmental and Water Resources Congress 2009: Great Rivers © 2009 ASCE Downloaded 15 Jun 2010 to 165.91.185.81. Redistribution subject to ASCE license or copyright. Visithttp://www.ascelibrary.org possible presence of pathogens), TMDLs have been established for impaired water bodies. Municipalities across the country are exploring options to reduce indicator bacteria inputs from point and non-point sources. Numerous studies have indicated that development in watersheds leads to increased export of indicator bacteria. In a study of 18 mixed land use watersheds in West Georgia, Schoonover and Lockacy (2006) indicated that watersheds consisting of greater than 24% imperviousness exhibit higher fecal coliform concentrations than watersheds with impervious percentages less than 5% during both base and storm flow. Studies by Line et al. (2008) and Mallin et al. (2000) conclude similarly that urbanization in watersheds leads to increases in indicator bacteria export. To test for the presence of harmful pathogens in surface waters, indicator species are used. Various indicator species have been used to assess water quality degradation due to pathogens including: total coliform, fecal coliform, Escherichia coli (E. coli), and enterococci. In 1986, the EPA’s Ambient Water Quality Criteria for Bacteria report (USEPA, 1986) discussed the merits of these various indicator species, and set a criteria whereby E. coli and enterococci are suggested as indicators in freshwater environments and enterococci is suggested as an indicator in marine environments. This criteria states that for fresh waters designated for use as full body contact recreational waters, the geometric mean over a 30 day period should not exceed 126 col/100 ml for E. coli and should not exceed 33 col/100 ml for enterococci. For marine waters designated for use as full body contact recreational waters, the geometric mean over a 30 day period should not exceed 35 col/100 ml for enterococci. The recommendation for fecal coliform, set in 1976 by the USEPA, is that the log mean over a 30-day period should not exceed 200 CFU/100ml (colony forming units per 100 ml) and no more than 10 percent of the samples should exceed 400 CFU/100ml (USEPA 1976). Indicator bacteria can be removed from surface waters and stormwater through a number of natural processes, such as ultraviolet light (from sunlight), sedimentation, filtration, and various environmental factors. These environmental factors can include temperature, moisture conditions, and salinity (USEPA, 2001; Schueler, 2000; Arnone, 2007; Davies-Colley et al., 1994). Urban stormwater is commonly treated by way of stormwater Best Management Practices (BMPs), each of which provides some combination of natural treatment mechanisms. These BMPs include wet ponds, dry detention basins, wetlands, bioretention areas, and proprietary devices. Although BMPs have been studied in detail for many pollutants, little peer-reviewed literature is available which documents their ability to remove or inactivate indicator bacteria. Six sites in Charlotte, NC, and 4 sites in Wilmington, NC, were monitored to determine indicator bacteria removal efficiency. Site Descriptions The stormwater BMPs evaluated in this project were monitored as part of the Charlotte – Mecklenburg Stormwater Services (CMSS) Pilot BMP Program and the 1124 World Environmental and Water Resources Congress 2009: Great Rivers © 2009 ASCE Downloaded 15 Jun 2010 to 165.91.185.81. Redistribution subject to ASCE license or copyright. Visithttp://www.ascelibrary.org Burnt Mill Creek Watershed Restoration program in Wilmington, NC. As part of these studies, grab samples were taken and analyzed for both fecal coliform and E. coli from 6 stormwater BMPs in Charlotte, NC, and 4 stormwater BMPs in Wilmington, NC. In Charlotte, data were gathered from two dry detention basins, one wet pond, two stormwater wetlands, and one bioretention area. In Wilmington, data were gathered from one wet pond, two stormwater wetlands, and one bioretention area. Although data collection has been completed in Charlotte, the study in Wilmington was ongoing at the time of this publication. The characteristics of the BMPs from each city are given in Table 1. Table 1: BMP and Watershed Characteristics Site (Charlotte) Watershed Size (ha) Description Estimated Curve Number Dry Detention 1 2.4 Office Park Buildings and Parking 85 Dry Detention 2 1.5 Office Park Buildings and Parking 94 Wet Pond 48.6 Residential 75 Wetland 1 21 Residential 80 Wetland 2 6.4 Residential and School 83 Bioretention 0.4 Municipal Parking Lot 98 Site (Wilmington) Watershed Size (ha) Description Estimated Curve Number Bioretention 2 0.14 Parking Lot 98 Wetland 3 12.7 School Parking Lot and Fields 73 Wetland 4 2 Mulit-Family Residential 80 Wet Pond 2 4.7 Commercial 81 Dry detention basins fill with runoff during storm events and provide temporary detention while slowly draining over a span of approximately 48 hours. The primary pollutant removal mechanism in these systems is sedimentation. Charlotte, dry detention 1 was an extended dry detention basin which received runoff from a 2.4 ha watershed comprised of an office park and its associated parking areas, landscape features and buildings. The dry detention facility was well vegetated with grass and had good sun exposure. There was some evidence of erosion and sedimentation within the facility. Charlotte dry detention 2 was sited in a similarly sized watershed, 1.5 ha, also comprised of an office park. Like dry detention 1, this facility had good sun exposure, was well vegetated with grass, and had evidence of some erosion and sedimentation. Both facilities appeared to be mowed frequently. CMSS staff noted the frequent presence of birds around the basins, with bird droppings noted on the boxes which housed flow and water quality sampling equipment. Wet ponds work on the principle of plug flow whereby influent runoff enters the pond and theoretically replaces the runoff that has been stored since then last storm event. Sedimentation in the basin is the primary pollutant removal mechanism as the stormwater slows, but some treatment is also provided via other mechanisms such as plant uptake, oxidation-reduction reactions, and adsorption as contact is made 1125 World Environmental and Water Resources Congress 2009: Great Rivers © 2009 ASCE Downloaded 15 Jun 2010 to 165.91.185.81. Redistribution subject to ASCE license or copyright. Visithttp://www.ascelibrary.org between the soils and plants in the pond and the captured stormwater. The Charlotte wet pond was fed by a small, perennial stream and received stormwater runoff from a 48.6 ha watershed that was primarily residential. This pond was likely not originally created for stormwater management, and was constructed with no detention component. The estimated age of the pond was between 50 and 70 years old. Waterfowl were frequently observed at the pond during site visits. The pond was retrofit in the late 1990’s to include a littoral shelf; however, the shelf was not planted and exhibited little vegetation during the study period. Despite the presence of trees around the BMP, there was good exposure on the pond. Wet pond 2 in Wilmington, NC, received runoff from a 4.7 ha commercial property which included a highly impervious 2 ha parking lot. Wetlands are commonly installed as water quality devices, whereby they are sized to treat small (2.5 cm) storm events. These BMPs promote sedimentation much like wet ponds, but provide more intense contact between the captured stormwater and wetland soils and plants in shallow a system. Charlotte wetland 1 received stormwater from approximately 21 ha of residential area. This wetland exhibited common wetland topography, and consisted predominantly of shallow water depths. During the course of the study, however, there was very little vegetation in the wetland, likely due to poor soil conditions, prolonged periods of high water levels due to slow drainage, and the impact of waterfowl grazing. This lack of vegetation resulted in a larger amount of full sun exposure to water in the wetland than would typically be expected. Waterfowl were commonly observed at this site. Charlotte wetland 2 was constructed with similar topography, but exhibited exceptional plant growth. Charlotte wetland 2 received stormwater from 6.4 ha of residential area and a school. This wetland had two inlets, thus, weighted average influent fecal coliform and E. coli concentrations were calculated by weighting the grab samples at each inlet based on the total flow they contributed to the system. Wildlife was observed at Charlotte wetland 2 during the study. Wilmington wetland 3 treated runoff from 12.7 ha of a school, which included parking lots and practice fields. Wilmington wetland 4 is located in a multi-family residential complex. This wetland exhibits high amounts of infiltration in between storm events resulting in a normal pool where water is held mostly inside the deep pools. Bioretention areas are filtration and infiltration BMPs. Stormwater enters the system and passes through a permeable soil media where pollutants are filtered, functioning similarly to sand filter systems. The BMP may pond water as much as 6 to 12 inches; however, it is drained within 12 to 24 hours. The system is intended to dry out inbetween storm events. The Charlotte bioretention site received stormwater from a highly impervious 0.4 ha parking lot. This bioretention cell was studied and described in detail by Hunt et al. (2008). On at least one occasion, a diaper was observed in the parking lot, providing a potential source of bacteria to the BMP. Additionally, trees in the parking lot attract birds, and evidence of bird droppings have been observed by CMSS staff. Sun exposure in the BMP was fair, as it was limited by fairly dense vegetation. Wilmington bioretention area 2 was a sodded bioretention that treated the 1126 World Environmental and Water Resources Congress 2009: Great Rivers © 2009 ASCE Downloaded 15 Jun 2010 to 165.91.185.81. Redistribution subject to ASCE license or copyright. Visithttp://www.ascelibrary.org runoff from a 0.14 ha parking lot at the headquarters of a coffee shop chain. The Wilmington bioretention area is broken into two paired cells, one with a 30 cm (1foot) soil depth and one with a 60 cm (2-foot) soil depth. Observations have been made that the 30 cm deep cell receives more stormwater due to parking lot grading. This cell characteristically appears to stay more moist, likely due to the shallow soil depth and greater watershed area. Detailed surveys of the watershed area for the Wilmington bioretention area are planned to determine the exact sub-watersheds for each cell. Sun exposure at the Wilmington bioretention area is high. With such highly variable uses, design specifications for the BMPs varied. Thus, there are some inherent differences in the function of the BMPs selected for this study with respect to both hydrology and water quality, making normalization problematic. However, these BMPs were selected because they are representative of the types of BMPs common to the City of Charlotte, NC, the City of Wilmington, NC, and elsewhere. Monitoring Methods Charlotte As part of the Pilot BMP Program, grab samples were taken due to the small sample hold times required of bacteriological samples (USEPA, 2002). Grab samples were tested for fecal coliform and E. coli. Samples were collected at the various sites in Charlotte between March 2004 and October 2006. The monitoring period and number of samples collected at each site varied (Table 2). Wilmington Grab samples were also collected at each Wilmington, NC, site beginning in August of 2007. The samples from Wilmington were analyzed for enterococci and E. coli. Enterococci has proven to be a more reliable indicator species in environments with higher salinity (USEPA, 1986). Table 3 shows the number of samples collected at each site. Table 2: Monitoring Period and Number of Samples Taken at Each Study Location Site Start End Number of Sample Tested For Fecal Coliform Number of Samples Tested For E. coli Dry Detention 1 Feb-05 Jul-06 9 9 Dry Detention 2 Jan-05 Dec-05 12 12 Wet Pond Aug-04 Apr-06 14 10 Wetland 1 Mar-04 Jun-05 9 6 Wetland 2 Sep-04 Dec-05 15 10 Bioretention Aug-04 Mar-06 19 14 Table 3 Number of samples taken at each Wilmington Study Site Site (Wilmington) Number of enterococci Number of E. coli Samples Bioretention 2 9 9 Wetland 3 9 8 1127 World Environmental and Water Resources Congress 2009: Great Rivers © 2009 ASCE Downloaded 15 Jun 2010 to 165.91.185.81. Redistribution subject to ASCE license or copyright. Visithttp://www.ascelibrary.org Wetland 4 8 8 Wet Pond 2 8 8 Data Analysis To adequately describe the bacteria sequestration and removal performance of each BMP, various analyses were performed. This included a calculation of concentration reduction efficiency and an analysis of effluent concentrations. The concentration reduction efficiency (CR) was determined by calculating the geometric mean of the influent and effluent indicator bacteria concentrations and using them in equation 1 below: Equation 1: CR = 1 – (geometric mean outlet concentration / geometric mean inlet concentration) Lastly, the geometric mean effluent concentrations of fecal coliform and E. coli leaving each site were compared to the maximum 30-day geometric mean for each indicator as established by the USEPA for full body contact (EPA, 1986; EPA 1976). This will aid in evaluating not only the efficiency of indicator bacteria removal for each system, but also the practicality of using stormwater BMPs to improve runoff from urban watersheds to indicator bacteria concentrations equal to or below targeted concentrations. Results and Discussion Table 3 presents the results for fecal coliform and Table 4 presents the results for E. coli for the BMPs studied in Charlotte, NC. Table 4 presents the results for E. coli and table 5 presents the results for enterococcus for the BMPs studied in Wilmington, NC. It should be noted that not all BMPs exhibit similar performance for both indicator bacteria for which they were tested. This indicates that BMP removal percentages generated for one indicator bacteria may not be applicable to other types of indicator bacteria data. For the Charlotte BMPs, the wet pond, wetland 1, wetland 2, and bioretention area, exhibited greater than 50% removal of fecal coliform. The high fecal coliform removal determined for wetland 1 and wetland 2, 99% and 70%, is similar to that found by Birch et al. (2004) and Davies and Bavor (2000). For E. coli, only wetland 1 and the bioretention area provided high (> 50%) concentration reductions. Overall, wetland 1 and the bioretention proved most proficient at reducing influent concentrations of both kinds of bacteria. Wetland 1 had good sun exposure, likely leading to higher die off rates. Stormwater wetlands and bioretention areas also facilitate sediment removal through sedimentation and, in the case of bioretention, filtration and drying. All of these factors likely have some impact on indicator bacteria removal in stormwater BMPs. The poorest performing BMPs were the two dry detention basins. These systems had good sun exposure but remained moist for a substantial period of time after each rain event (per CMSS staff observation). It is possible that the wet soil provided an environment where the indicator bacteria could 1128 World Environmental and Water Resources Congress 2009: Great Rivers © 2009 ASCE Downloaded 15 Jun 2010 to 165.91.185.81. Redistribution subject to ASCE license or copyright. Visithttp://www.ascelibrary.org persist for an extended period of time. Bird droppings were also noted by staff, likely leading to additional bacteria inputs to the BMP. Table 3: Fecal Concentration Efficiency for BMPs in Charlotte, NC. Geometric Mean Influent Geometric Mean Effluent Concentration