Recreational water sites such as beaches along lakes, rivers, or oceans, are one of the most popular activities in many parts of the world. Recently rainfall and runoff due to rainfall events has been associated with increasing microbial levels in recreational water. This runoff can lead to beach closures and potentially unsanitary conditions at popular swimming beaches. The impact of stormwater on beach water quality has led to a myriad of option for controlling stormwater. Some of these include grass buffer partitions, stormwater detention basins, media filters, catch basin inserts, and infiltration units. Biofilters, or infiltration units are gaining popularity as a treatment option for stormwater around the Great Lakes basin, but we are aware of no studies that have looked at the indicator organism (i.e. Escherichia coli, or E.coli) removal potential of these infiltration units and the media used in them. The overall objective of this study was to evaluate the performance of a stormwater biofilter medium in removing the indicator organism E.coli in a laboratory system. When several laboratory biofilter system were challenged with E.coli concentrations of 2.82E3 and 2.85E5 E.coli/100mL of simulated stormwater in a 1.25 cm rain event, the systems were able to remove between 83 and 100% of the E.coli in this influent. During a subsequent 1.25 cm rain event with E.coli-free water, the biofilter was able to retain 68%-100% of the E.coli originally inoculated into the system. The results of this study indicate that these systems hold promise for mitigation of E.coli from storm water near recreational beaches. These findings will assist beach managers, engineers, and municipal stake holders evaluate the usefulness of biofilter infiltration as a storm water management tool in order to decrease E.coli input into beach areas.
Introduction
Recreational water sites such as beaches along lakes, rivers, or oceans, are one of the most popular activities in many parts of the world. Fecal material from swimmers, domestic animals (dogs, cattle, and horses), as well as waterfowl (geese, gulls, and ducks), all lead to increases in microbial loading at beaches (Winfield and Groisman, 2003; Kleinheinz et al. 2003). Recently rainfall and runoff due to rainfall events has been associated with increasing microbial levels in recreational waters along numerous beaches including several coastal areas in Wisconsin (Ackerman and Weisberg, 2003; Kinzelman, Racine County, WI Health Department, personal communication, 2003). This runoff can lead to beach closures and potentially unsanitary conditions at popular swimming beaches (Sampson et al. 2006; Kleinheinz et al. 2006).
Heavy rainfall was implicated in increasing bacterial contamination at beaches in several areas of the country (Ackerman and Weisberg, 2003; Haack et al. 2003). Some beaches are automatically closed or restricted after a large rainfall event, even without microbiological testing of water (Ackerman and Weisberg, 2003; Kinzelman, Racine County, WI Health Department, personal communication, 2003). On the Santa Monica Bay beaches in southern California, health departments typically issue warnings for the public to avoid recreational water contact for 3 days following a rainfall event (Ackerman and Weisberg, 2003). This public health warning is based on 5 years of fecal coliform data taken daily and after rainfall events. Although storms are fairly infrequent in southern California, rainfall events did precipitate microbial contamination exceedences due to storm water runoff (Schiff et al. 2003). In Milwaukee, WI automatic beach closures occur following heavy rainfall events due to finding a positive correlation between rainfall and E. coli concentrations over many years (Kinzelman, Racine County, WI Health Department, personal communication, 2003).
The impact of stormwater on beach water quality has led to a myriad of option for controlling stormwater. Some of these include grass buffer partitions (Guber et al. 2007), stormwater detention basins (Hogan and Walbridge, 2007), media filters (Barrett, 2005, catch basin inserts (Morgan et al. 2005), and infiltration units (Birch et al. 2005) to name just a few. In general, these systems attempt to limit input of both chemicals, nutrients, and metals into water systems. The studies we are aware of have evaluated the success of the stormwater mitigation strategies in terms of the removal of total P, COD, Nitrogen, TSS, Pb, water volume, turbidity, and heavy metals (Hogan and Walbrudge, 2007; Guber et al. 2007; Furumai et al. 2005, Brezonik and Stadelmann, 2002; Bechet et al. 2006). In a few non-biofilter studies the transport of indicator organisms such as Escherichia coli (E.coli) and Enterococcus have been evaluated. There are a few evaluations of biofilter (infiltration media units) performance using metals, solids, nutrients, and organochlorine pesticides, hydraulic properties, (Birch et al. 2005; Hatt et al. 2006; Furumai et al. 2005; Bechet et al. 2006). However, we are aware of no studies that have looked at the indicator organism (i.e. Escherichia coli, or E.coli) removal potential of infiltration units and the media used in them. While there is certainly a multitude of evidence to support the effectiveness of these systems in removing many environmental contaminants, it is clear that more work is needed to evaluate their usefulness in removing biological contaminants associated with stormwater.
The overall objective of this study was to evaluate the performance of a stormwater biofilter medium in removing the indicator organism E.coli in a laboratory system. The results of this study will be used to assess the feasibility and design of full-scale stormwater management systems with the overarching goal of increasing beach water quality in Door County, WI (Lake Michigan).
Materials and Methods
Laboratory Biofilter Set-up
The laboratory biofilter column (Fig. 1) was borosilicate glass, 5 cm in diameter, contained a Teflon valve at the bottom to regulate flow, and was filled with varying amounts of biofilter media (Table 1) and/or sand (United States Department of Agriculture Classification – Medium to very coarse sand with a permeability of 0.006 cm/hr). These simulated biofilter columns of media were set-up at a depth of 15 or 30 cm. The biofilter media to sand mix was either 1:1 (Biofilter Media:Sand), or 1.5:1 (Biofilter Media:Sand) depending on which trial and as indicated in the results. The ratio of biofilter media to sand was recommended by the manufacturer of the biofilter media (Miller Engineers and Scientists, Sheboygan, Wisconsin, U.S.A.).
Table 1.
Characteristics of the biofilter media.
The water flow rates and E.coli concentrations were chosen to simulate real-world values found during several years of E.coli-stormwater relationships in Door County Wisconsin (Kleinheinz, 2007). The mixtures of Biofilter media were selected due to their impending use in re-engineered beaches along Wisconsin's Lake Michigan shoreline.
Preparation of E.coli inoculum
The E. coli inoculum water was prepared using the LS 232 strain of E.coli from the University of Wisconsin—Oshkosh's Environmental Microbiology Laboratory. This strain of E.coli was recovered from a swimming beach in 2005. The aforementioned E.coli was grown for 18-24 hours in nutrient broth. The resulting culture was then centrifuged at 12,000 × g for 10 minutes, the supernatant discarded, and the resulting pellet of E.coli resuspended in 0.85% NaCl. This centrifugation and washing procedure was repeated three times to yield an essentially nutrient-free inoculum. The starting inoculum was either 2,820 E.coli/100mL of water for the first trial or, 285,000 E.coli/100mL of water for the second trial. These concentrations of E.coli were chosen and typical and ‘worst-case’ E.coli loadings from stormwater in Door County, WI (Kleinheinz, 2005). E.coli analysis using the Colilert® test (described below) was used for this enumeration.
Water Sample Collection after Passing through column
In order to determine the effect of the biofilter medium on the removal of E.coli, an all-in-one, 1.25 cm of water rain event was simulated using 250 mL of E.coli inoculum. As stated previously, this was representative of rain events in Door County, Wisconsin, U.S.A. (Kleinheinz, 2007). This inoculum was allowed to wash through the laboratory system using only gravity flow. The first effluent from the column, indicated as First 1% on the graphs, was the first 2mLs of sample collected out of the column. The next approximately 75 mLs was collected (25%-50% samples), the next 75 mLs was collected (50%-75% samples), and the final 2 mLs of effluent (Final 1% samples). This initial inoculum was meant to simulate the ‘first-flush’ from a rain event. After the E.coli inoculum had washed completely through the column, and no further effluent was observed, 250 mL of E.coli-free water was washed through the column in an identical manner to the E.coli inoculum. This was intended to represent additional rainfall and determine how much, if any, of the trapped E.coli could be washed from the column.
All water samples were collected into sterile, polystyrene collection bottles (IDEXX Corp., Portland, ME). Upon collection, samples were immediately placed in a cooler at 4 °C until E. coli analyses of samples were conducted. Samples were analyzed within 4 hours of collection. The University of Wisconsin – Oshkosh Environmental Microbiology lab was utilized for all analysis; it is a Wisconsin State Certified Laboratory with a Quality Assurance plan on file with the WI Department of Agriculture, Trade and Consumer Protection (DATCP).
Amount of E.coli removed from the column was calculated by:
E.coli Removal = ((Influent) – (Effluent)) × Dilution
Sample analysis
The defined substrate test, Colisure® (IDEXX Corp., Portland, ME), was used to analyze all samples for E. coli (American Public Health Association. 1999). Incubation and microbial enumeration from samples were conducted following the manufacturer's recommendations. All results were reported as most probable number (MPN) of E. coli per 100 mL of water. Positive and negative controls were prepared in accordance with the laboratory's quality assurance plan.
Results and Discussion
Overall, E. coli was substantially reduced under all operational conditions using the proposed biofilter media. The 2.82E3/100mL loading of E.coli coupled with a biofilter column height of 15 cm (1:1 ratio of biofilter media to sand) showed 93%-100% removal of E.coli from a 2 cm of inoculum added to the column (Fig. 2). Subsequent washing of the column with another 2 cm of E.coli-free water still showed 89%-100% of the E.coli being removed in the column (Fig. 2). When the ratio of biofilter media to sand was increased to 1.5:1, the removal from the initial 2 cm inoculum increased to 99%-100% (Fig. 3, Table 2). Subsequent washing of the column with another 2 cm of E.coli-free water still showed 76%-100% of the E.coli being removed in the column. Two and four days after completing this study another 2 cm wash of biofilter column with E.coli-free water revealed 99.9% removal of E.coli, and no detectable E.coli, respectively.
Figure 2.
E.coli MPN/100mL from various portion of the effluent a15 cm test column and 1:1 (Biofilter mix:sand) ratio. Starting E.coli inoculum for this trial is represented by the dashed line and was 2.83E3 E.coli/100mL.
Figure 3.
E. coli MPN/100mL from various portion of the effluent a 15 cm test column and 1.5:1 (Biofilter mix:sand) ratio. Starting E.coli inoculum for this trial is represented by the dashed line and was 2.83E3 E.coli/100mL.
Table 2.
Percent removal of E.coli from laboratory microcosm during low E.coli loading. The initial incolumn concentration was 2.82E3 E.coli /100mL.
Furthermore, the 2.82E3/100mL loading of E.coli coupled with a biofilter column height of 30 cm (1:1 ratio of biofilter media to sand) demonstrated 93%-100% removal of E.coli from a 2 cm of inoculum added to the column (Fig. 4). Subsequent washing of the column with another 2 cm of E.coli-free water still demonstrated 77%-100% of the E.coli being removed in the column (Fig. 4). When the ratio of biofilter media to sand was increased to 1.5:1, the removal from the initial 2 cm inoculum increased to 83%-100% (Fig. 5, Table 3). Again, subsequent washing of the column with another 2 cm of E.coli-free water still showed 68%-100% of the E.coli being removed in the column. Two and four days after completing this study another 2 cm wash of biofilter column with E.coli-free water revealed 99.9% removal of E.coli, and no detectable E.coli, respectively.
Figure 4.
E.coli MPN/100mL from various portion of the effluent a 30 cm test column and 1:1 (Biofilter mix:sand) ratio. Starting E.coli inoculum for this trial is represented by the dashed line and was 2.83E3 E.coli/100mL.
Figure 5.
E.coli MPN/100mL from various portion of the effluent a 30 cm test column and 1.5:1 (Biofilter mix:sand) ratio. Starting E.coli inoculum for this trial is represented by the dashed line and was 2.83E3 E.coli/100mL.
Table 3.
Percent removal of E.coli from laboratory microcosm during high E.coli loading. The initial incolumn concentration was 2.85E5 E.coli/100mL.
In another trial with a loading of 2.85E5/100mL E.coli coupled with a biofilter column height of 15 cm (1:1 ratio of biofilter media to sand) showed 68%-100% removal of E.coli from a 2 cm of inoculum added to the column (Fig. 6). Subsequent washing of the column with another 2 cm of E.coli-free water still showed 80-100% of the E.coli being removed in the column (Fig. 6). When the ratio of biofilter media to sand was increased to 1.5:1, the removal from the initial 2 cm inoculum increased to 94%-100% (Fig. 7, Table 3). Subsequent washing of the column with another 2 cm of E.coli-free water still showed 78%-100% of the E.coli being removed in the column. Two and four days after completing this study another 2 cm wash of biofilter column with E.coli-free water revealed 99.4% removal of E.coli, and no detectable E.coli, respectively.
Figure 6.
E.coli MPN/100mL from various portion of the effluent a 15 cm test column and 1:1 (Biofilter mix:sand) ratio. Starting E.coli inoculum for this trial is represented by the dashed line and was 2.85E5 E.coli/100mL.
Figure 7.
E.coli MPN/100mL from various portion of the effluent a 15 cm test column and 1.5:1 (Biofilter mix:sand) ratio. Starting E.coli inoculum for this trial is represented by the dashed line and was 2.85E5 E.coli/100mL.
The 2.85E5/100mL loading of E.coli coupled with a biofilter column height of 30 cm (1:1 ratio of biofilter media to sand) showed 83%-100% removal of E.coli from a 2 cm of inoculum added to the column (Fig. 8). Subsequent washing of the column with another 2 cm of E.coli-free water still showed 73%-100% of the E.coli being removed in the column (Fig. 8). When the ratio of biofilter media to sand was increased to 1.5:1, the removal from the initial 2 cm inoculum increased to 81%-100% (Fig. 9, Table 3). Subsequent washing of the column with another 2 cm of E.coli-free water still showed 70%-100% of the E.coli being removed in the column. Two and four days after completing this study another 2 cm wash of biofilter column with E.coli-free water revealed 99.4% removal of E.coli, and no detectable E.coli, respectively.
Figure 8.
E.coli MPN/100mL from various portion of the effluent a 30 cm test column and 1:1 (Biofilter mix:sand) ratio. Starting E.coli inoculum for this trial is represented by the dashed line and was 2.85E5 E.coli/100mL.
Figure 9.
E.coli MPN/100mL from various portion of the effluent a 30 cm test column and 1.5:1 (Biofilter mix:sand) ratio. Starting E.coli inoculum for this trial is represented by the dashed line and was 2.85E5 E.coli/100mL.
At the lower E.coli loading, which is typical of many storm water discharge sites (Kleinheinz, 2007), the increased column depth did not appreciably increase E.coli removal indicating that the 15 cm bed depth was sufficient for the majority of E.coli removal at this loading. Increased bed depth and changing of the ratio of biofilter media to sand did not greatly reduce or increase the E.coli removal. At the higher E.coli loadings, which is typical of severe storm water discharge (Kleinheinz, 2007), the increased column depth did increase E.coli removal at the 1:1 ratio of biofilter mix to sand, but not at the 1.5:1 ratio. Additionally, the higher loadings did lead to increased breakthrough of E.coli during the initial inoculum washing, however, the system was still able to remove the majority of E.coli during even these elevated loadings. Remarkably, the system was very resilient when challenged with additional E.coli-free wash water and demonstrated continued filtration efficiency. Overall, the two ratio's of biofilter mix to sand—either 1:1 or 1.5:1, appeared to function in a similar manner in terms of E.coli removal. The overall efficiency of the system at the two loadings provides a good indication of the efficiency of these systems under very different loading scenarios. The fact that very low E.coli was able to be washed from the column two days after the studies (all concentrations and rations of media), and no detectable E.coli was recovered four days after the study appears to indicate that E.coli can not survive or multiply in the media. One-gram samples of the media analyzed four days after the study also indicated that no detectable E.coli was present.
Overall, this study provides valuable insight into the potential E.coli removal capabilities of a biofilter unit designed for E.coli mitigation from stormwater discharged in proximity to swimming beaches. To our knowledge this is the first attempt to investigate the E.coli removal capabilities of infiltration beds (i.e. biofilters) intended for storm water E.coli mitigation. There is obviously much more research that should be conducted into the effectiveness of these units, different media mixes, and the field evaluation of these units that can not be replicated in this simple laboratory unit. The best evaluation of this technology would be the detailed evaluation of full-scale units installed in a location that receives high concentrations of E.coli from stormwater. The evaluation of E.coli concentrations on a seasonal and year-to-year basis along with an evaluation of media condition, removal efficiency, and the impact on adjacent beach water quality would provide a very complete assessment of this technology. Certainly, there are additional engineering concerns such as filter maintenance and colmation effects that need to be elucidated, however, the results of this study indicate that these systems hold promise for mitigation of E.coli from storm water near recreational beaches. These findings will assist beach managers, engineers, and municipal stake holders evaluate the usefulness of biofilter infiltration as a storm water management tool in order to decrease E.coli input into beach areas.
Acknowledgements
The funding for this project was provided by the University of Wisconsin – Oshkosh Environmental Microbiology Research Laboratory. Special thanks to Miller Engineers & Scientists of Sheboygan, Wisconsin for providing the biofilter media and advice on mixture ratios.
