Reflections on Using Indicator Bacteria for Water Quality Impairment: Results from Two Case Studies in Georgia
This article is featured in The Georgia Operator Winter 2016 edition, page 63, and can be read online here.
Elevated fecal coliform bacteria account for the highest number of water quality impaired stream miles in Georgia (4,637 miles); this is over twice as many miles as the second highest, fish impairment (2,208 miles) and over 3.5 times that of dissolved oxygen impairment (1,291 miles) (EPA, 2015). Since fecal coliforms themselves are generally not harmful, they serve as cost effective indicator organisms (sample and identification costs) for those organisms that are known to cause diseases, such as other bacteria, viruses, and protozoans (EPA, 2015). The United States Environmental Protection Agency (USEPA) requires each state to protect surface waters from sewage contamination; currently the federal standard is either based upon fecal coliform or E. coli concentrations. Georgia Environmental Protection Division’s (GAEPD) adoption of that standard for streams designated as “fishable/swimmable” is a geometric mean of fecal coliform bacteria below 200 cfu/100mL from four samples taken within a 30-day period from May through October and a geometric mean below 1,000 cfu/100 mL from four samples taken within a 30 day period with no single sample greater than 4,000 cfu/100mL from November through April. Since most impaired stream miles in Georgia are associated with fecal coliforms, does this mean that our streams are full of raw sewage from broken sewer pipes, pet waste, and failing septic systems? In this article we will briefly discuss results of studies that we conducted over the past several years to address fecal impairment in streams and we will provide thoughts on how to improve the science, technology, monitoring, and regulation of this complex issue.
Case study 1: Augusta, Georgia
In collaboration with Augusta-Richmond County’s Engineering Department (AED), we were awarded a GAEPD 319(h) grant to address fecal coliform concentrations on sections of Rocky and Butler Creeks in Augusta, GA; these creeks were placed on the 303(d) list in 1998. We designed a sampling procedure to identify sources of bacteria, increased the number and frequency of sampling locations that AED were currently sampling, and developed materials to educate the public about fecal pollution in streams.
For Butler Creek, additional sampling showed that this creek was meeting the state’s standard for fecal coliform concentrations; it is currently being considered for delisting. For Rocky Creek, we found that concentrations of fecal coliform (and USEPA’s recommendation for E. coli of 126 cfu/ 100mL) were still not meeting the standard. However, from our increased sampling efforts there were no clear “point sources” that were causing the elevated bacteria levels. Our investigations then focused on potential diffuse, nonpoint source loadings by sampling according to travel time (Lagrangian sampling) using Rhodamine WT as a tracer, sampling stream sediment pore water for fecal coliform/E. coli concentrations, and sampling tributaries and other potential sources during storm and non-storm conditions. Our findings showed that the sediment pore water concentrations of E. coli were higher than the overlying water in nearly all cases (Figure 1), that E. coli loading increased nearly linearly with stream mile during one sampling event during the fall of 2014 (Figure 2), and E.coli concentrations exceeded 2,500 cfu/100mL in road runoff.
Case study 2: Thomson, Georgia
In collaboration with the City of Thomson, we were awarded a GAEPD 319(h) grant to develop a Watershed Management Plan to address elevated fecal coliform concentrations in Whites Creek, a tributary to Briar Creek; that creek was placed on the 303(d) list in 2002. For this project, we worked with the City of Thomson to identify an advisory council to guide the Watershed Management Plan process, increase the number and frequency of stream samplings to identify problem areas, and develop a Watershed Plan.
Since this grant was mostly focused on developing a Watershed Plan, additional sampling of this system was limited compared to the Augusta grant, so we could not perform a Lagrangian sampling scheme and were limited to two geometric mean sampling events, one in the November to April time frame, and one in the May through October time frame. Our findings were important nonetheless. The data showed that for all 88 samples analyzed during this study, not one geometric mean sampling event (16 in all) for any site (8 in all) was above the state standard for geometric mean fecal coliform concentrations. Of all 88 samples, three samples may have exceeded the single limit value of 4,000 cfu/100ml; these results were above the upper method limit for the IDEXX Colilert-18 analysis protocol used in this study (maximum 2,419 MPN/100mL) so we could not confirm the final value. Sediment pore water samples were analyzed for E. coli later in the study at 6 of the 8 sampling sites. When comparing the site-specific pore water results to the two geometric mean values for each site, the pore water had higher concentrations of E. coli 33% of the time. The ratio of sediment E. coli to water E. coli concentration for each of the six pore water sample sites ranged from 0.17 to 5.7. We also found that the wastewater treatment plant, which was suspected of causing high loading in the development of the criteria for listing the creek, was actually helping to decrease the total load to Whites Creek because the effluent had lower fecal coliform concentrations than the creek above the discharge.
What does all of this mean?
These two studies show that using indicator bacteria like fecal coliform or E. coli as a standard for protection of surface water is not straightforward or simple. We should also point out that we did not find concentrations of fecal coliform or E. coli concentrations in the tens or hundreds of thousands of bacteria per hundred milliliters; this would be indicative of a point source issue, so we were dealing with non-point or diffuse sources in these studies.
One of the most important findings was that fecal coliform, and specifically E. coli, concentrations in the sediment often exceeded concentrations in the overlying water. This suggests that bacteria associated with stream sediments may be contributing to fecal coliform loads in surface waters in some cases. The fact that fecal coliform and E. coli survive outside the gut is not a new finding; in fact, scientific literature has shown for decades that these organisms can survive in stream sediments for up to 85 days (Davies et al., 1995). If we are using these organisms as indicators of recent introduction of fecal material to water bodies, because they are assumed to only exist in the gut of warm-blooded animals, then regulating based upon their abundance in surface waters is problematic. While the use of indicator bacteria for protection of surface water quality is complex, it seems that we can do better science, monitoring, measurement, and regulation to decrease the seeming contradiction of using indicator bacteria.
From our studies, it appears that understanding the occurrence and persistence of fecal bacteria, including E. coli, in streams needs additional research. We present two areas of potential research that we are interested in. Firstly, we found that indicator bacteria existed in the sediment. Our research in Augusta showed that E. coli loading increased linearly with increased distance downstream, which indicated a constant, low level loading along the study reach. This would have required all sources of loading (e.g. pet waste, broken sewer pipes, malfunctioning septic systems, etc.) to be contributing uniformly throughout the nearly 4 mile stream study reach! Alternatively, this particular sampling event occurred in late fall when vegetation within the stream buffer was mostly dormant for the winter. Decreased evapotranspiration of the stream buffer vegetation (corroborated with a water level logger that showed differences in diurnal water level fluctuations between leaf-on and leaf-off conditions) resulted in increased groundwater recharge to the stream. If the sediment had higher bacterial concentrations than the overlying water, then the increased groundwater pressure, through the sediment, would have flushed the sediment-laden bacteria into the stream, thereby increasing stream E.coli concentrations. This could have caused a nearly linear trend if sediment bacteria were somewhat uniformly distributed. The role of stream sediments as potential “sources” of bacteria needs to be understood for “low level” bacterial concentrations in streams. The second area of research has to do with the condition of our urban streams. The “urban stream syndrome”, characterized by high, fast storm flows and decreased stream bed stability, generally results in an altered food web (Paul and Meyer, 2001, Walsh et al., 2005). Our hypothesis is that urban streams may lack sufficient bacterial predators to keep pathogen abundance in check; in essence, bacteria may be the top of the food pyramid in urban streams. Our point here is that elevated fecal bacteria concentrations in streams may not be due to broken sewer pipes but could be due to a lack of scientific understanding.
While better science is needed, we could also apply better methods to monitoring protocols for identification of non-point source fecal contamination. Infrared thermal imaging for identification of “warm” discharges and chemical fingerprinting using caffeine and artificial sweeteners (Tran et al., 2014) are some recent advances; but what about simply adding a dye tracer such as Rhodamine WT to sewer pipes and septic systems in reaches of streams suspected of leaking? Analytical probes specific for Rhodamine and Fluorescein dyes are readily available. If failing septic tanks and broken sewer pipes are suspected of contaminating the adjacent stream, dye introduced to the suspect sewer/septic system should readily show up in the stream. Such an approach would, at least, rule out suspected raw sewage-related sources. In addition, adding tracers directly to the stream and applying a Lagrangian sampling scheme has proven useful for understanding of fecal loading to streams in our studies as well.
One of the reasons we use fecal coliform and E. coli as indicators for other pathogens is that the test for those organisms is rapid and inexpensive relative to technologies needed to sample for all potential pathogens. This indicates a need for a new technology, one that is cost-effective and rapidly identifies all potential pathogens in a given sample. Next-generation or high throughput sequencing is a relatively new technology that allows for the analysis of millions of DNA sequences at the same time. In essence, this technology allows for the ability to sample for all pathogens in a given sample at the same time; this eliminates the need for indicator bacteria monitoring and the associated validity of the results. Unfortunately, the cost of this equipment (nearly $100,000), including the contract laboratory “per sample costs” (hundreds to thousands of dollars), are not currently feasible for conventional stream monitoring programs. The technology has been developed; we await the miniaturization and cost reduction of this important technology!
All life and economic sustainability is reliant upon access to clean water, so the importance of protecting our surface waters cannot be understated. However, if the parameter being regulated is a poor indicator of the actual condition, then change can lead to a cost savings or a reallocation of local, state, and federal funds for other important projects. Since it does not seem that regulating fecal coliform and E. coli is straightforward, we may need to rethink the federal regulation; here are several observations from our studies that confound regulation of fecal indicator bacteria.
Rocky Creek in Augusta, GA is an urban stream. As we continue to build cities, it seems that some of the best remaining wildlife habitat in urban areas are stream buffer zones; does this lead to higher wildlife fecal contributions to the stream? Who is responsible for that? Currently, municipalities are required to maintain water quality according to state standards, but the default position is usually that impairments are due to failing septic systems and broken sewer pipes, not density of wildlife in the urban greenspace.
From the Rocky Creek case, we sampled during a storm event and found high concentrations of bacteria in road runoff. An interesting sample, collected from water falling through a PVC pipe bridge drain (Figure 3), exceeded the 2,419 E. coli MPN/100 mL threshold. We suspected this area, which was a unique intact urban stream forest habitat, was a common bird flyway which may have resulted in high E. coli loads in the road runoff.
Finally, on multiple occasions during the Thomson project, we found deer entrails and deer and dog carcasses in the creek. Knowing that fecal coliform and E. coli can survive in stream sediments, incidences such as a gut pile in a stream or a broken sewer pipe that was repaired 2 years ago could be viewed as a one-time “inoculation” event; these events could have lasting effects on indicator bacteria concentrations in the surface water. It is likely that the entrails go unnoticed or the repaired pipe was forgotten when regulatory sampling occurs downstream of those sites. If the results return a geometric mean of 250 cfu/100mL in the May through October timeframe, there will likely be a presumption of a failing septic system somewhere upstream.
GAEPD is required to implement a bacteria-based standard by the USEPA. The current bacteria-based approach is not conclusive enough for a municipality or industry to determine whether or not a broken pipe or failing sewer system is causing the impairment. If our approach to protecting surface waters from fecal contamination remains the same, money and effort spent in source identification and development of Total Maximum Daily Load Plans to fix those problems will fall short and the solution will remain elusive. Over time, another cost of that elusiveness will be unnecessary conflict between the regulator/regulated communities.
At this point, it seems the short-term answer might be to consider each impaired section on a case-by-case basis and be comfortable that the problem might not be a broken pipe. The long term answer is that better science will lead to better technology which will lead to better monitoring which will lead to better regulation.
Davies, C. M., Long, J. A., Donald, M., & Ashbolt, N. J. (1995). Survival of fecal microorganisms in marine and freshwater sediments. Applied and Environmental Microbiology, 61(5), 1888-1896.
EPA. (2015). http://water.epa.gov/type/rsl/monitoring/vms511.cfm
Paul, M. J., & Meyer, J. L. (2001). Streams in the urban landscape. Annual Review of Ecology and Systematics, 333-365.
Walsh, C. J., Roy, A. H., Feminella, J. W., Cottingham, P. D., Groffman, P. M., & Morgan, R. P. (2005). The urban stream syndrome: current knowledge and the search for a cure. Journal of the North American Benthological Society,24(3), 706-723.
Tran, N. H., Hu, J., Li, J., & Ong, S. L. (2014). Suitability of artificial sweeteners as indicators of raw wastewater contamination in surface water and groundwater. water research, 48, 443-456.