Published online 27 October 2006
Published in J Environ Qual 35:2244-2249 (2006)
DOI: 10.2134/jeq2006.0243
© 2006 American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
TECHNICAL REPORTS
Surface Water Quality
Escherichia coli Loading at or Near Base Flow in a Mixed-Use Watershed
Randall W. Gentry*,
John McCarthy,
Alice Layton,
Larry D. McKay,
Dan Williams,
Shesh R. Koirala and
Gary S. Sayler
Center for Environmental Biotechnology, The University of Tennessee, Knoxville, TN 37996
* Corresponding author (rgentry{at}utk.edu)
Received for publication June 26, 2006.
 |
ABSTRACT
|
|---|
This study analyzed the occurrence of Escherichia coli in a mixed land-use watershed with human, cattle, and wildlife fecal inputs located in a karstic geologic region using synoptic monitoring (samples taken throughout the watershed system) during base-flow conditions. The objective of the study was to evaluate the occurrence of E. coli during base-flow conditions for several months at seven different main channel and nine different tributary sampling sites in the Stock Creek watershed, a 49.3-km2 basin located in Knoxville, TN. Escherichia coli densities were measured using the Colilert (Defined Substrate Technology) method. The instantaneous loads for E. coli were determined from measured flow rates and E. coli densities, with the highest loading rates observed in the late fall. The study indicated a strong correlation between E. coli load rate (colony-forming units [CFU]/d), 7-d antecedent precipitation, and turbidity. Water quality data, however, also exhibited a spatial dependency; for example, the E. coli load rate was better correlated with turbidity in the slower draining basin tailwater sampling sites than in the faster draining upstream headwater sampling sites. In the headwater sites, the E. coli load rate was better correlated with 7-d antecedent precipitation than turbidity.
Abbreviations: CFU, colony-forming units
|
INTRODUCTION
|
|---|
THE PROCESSES by which fecal waste are delivered to stream systems is an area of intense research interest (National Research Council, 2001; USEPA, 2002). Pressure is also growing for local and state organizations to better manage watershed systems to ensure that public health is not compromised by impacted waterways. A better understanding of the random nature of hydrologic influences, random land loadings, and complicated hydrogeology are crucial to defining the occurrence and persistence of fecal bacteria, such as Escherichia coli, and were explored in the field study of a mixed-use watershed presented here.
Recent research has sought to more robustly relate the influence of watershed-scale processes, such as flow, sediment transport, and precipitation, to fecal indicators in streams (Kistemann et al., 2002; Byappanahalli et al., 2003; George et al., 2004; Mallin et al., 2001; Reeves et al., 2004; Tyrrel and Quinton 2003). Some researchers have sought to develop better modeling methodologies for forecasting E. coli densities or loads (Olyphant et al., 2003; Reeves et al., 2004). Jamieson et al. (2004) reviewed the traditional methodologies for modeling microbial pollution at the watershed scale.
Total coliforms and E. coli have been used as indicators of potential fecal contamination for almost 100 yr (Feng et al., 2002), the primary assumption being that if bacteria commonly associated with mammalian intestinal tracts are present in a water system, then this water may contain pathogenic bacteria or viruses. Epidemiologic evidence correlating the presence of total coliforms and E. coli in recreational waters with higher incidence of gastrointestinal disease formed the basis for regulatory limits on permissible pathogen levels to reduce disease risk (Dufour, 1984).
The spatial and temporal variability of fecal indicator bacteria have been related to hydrologic processes such as the presence of seeps or springs, as well as precipitation events that can mobilize both pathogens and sediment (Dussart-Baptista et al., 2003). For example, Mallin et al. (2001) alluded to the possible importance of sediment transport of fecal indicator bacteria and found that both turbidity and rainfall in the previous 24 h strongly correlated with fecal coliforms. George et al. (2004) found that fecal coliform bacteria were linked to particles in small streams and that the fraction increased with suspended sediment content. Dussart-Baptista et al. (2003) correlated turbidity with sessile (attached) and planktonic (unattached) bacteria and also concluded that intrakarstic storage and resuspension played an important role, in agreement with Massei et al. (2002, 2003), who noted that groundwater transfer through a chalk karstic aquifer induced a large decrease in the concentration of planktonic bacteria but no reduction in the sessile population. Further information is needed on the spatial and temporal relationships of E. coli behavior in karstic systems, particularly in mixed land-use areas.
The goal of our study was to evaluate the processes contributing to variations in the spatial and temporal concentrations of E. coli in a mixed land-use watershed that includes some karstic features. The overall hypothesis of this study was that direct or indirect hydrologic variables control the occurrence and persistance of E. coli in karst stream networks at base flow or near base-flow conditions. The primary objective of this study was to test the hypothesis by systematically evaluating the relative statistical relationship (using covariance and correlation coefficients) of E. coli density and loading rate with other direct (flow and precipitation) and indirect (turbidity, temperature, pH, and conductivity) hydrologic variables at base flow or near base-flow conditions within a karst watershed study site.
|
MATERIALS AND METHODS
|
|---|
Study Site Description
The site chosen for the field investigations is the Stock Creek watershed located a few kilometers south of Knoxville in eastern Tennessee. The watershed is a 49.3-km2 basin that drains into the Tennessee River (Fig. 1). The watershed is located in the Valley and Ridge physiographic region, which is characterized by alternating northeast–southwest-trending ridges of Paleozoic sedimentary rocks. The approximate elevation of the outlet region of the watershed is 250 m National Geodetic Vertical Datum (NGVD) and the headwater ridges have an elevation of approximately 394 m NGVD (Fig. 2). Average annual precipitation in Knox County is 122.43 cm, with an average of 128 wet days during the year. The Stock Creek watershed is underlain mainly by karstic carbonate rock, which provides conditions favorable to rapid transport of pathogen-contaminated groundwater either to drinking water wells or back to surface water through seeps, springs, and fractures.
From April 2003 to February 2004, 84 individual grab samples from the water surface were collected at seven locations along the main channel of Stock Creek and at nine sites on tributaries of Stock Creek. Stream flow was measured at many main channel locations during each sampling event. The main channel sites were designated as SC-1 through SC-7, and the representative subbasins are identified in Fig. 2. The subbasins were defined based on the specific area of drainage represented by each sampling location along the main channel. Also, the nine tributary sites are named according to local features (e.g., GH-1 is named for Gun Hollow) and inflow locations to the main channel are shown on Fig. 2.
Synoptic evaluation of E. coli and water quality parameters throughout the Stock Creek watershed was performed throughout the year, which included dry and wet seasons. Seven sites (SC-1–SC-7) were monitored at base flow or near base-flow conditions. Site SC-1 was only monitored for E. coli density, with no flow measurements due to backflow or embayment conditions that exist near the watershed outlet, which discharges to a river system whose level is managed by the Tennessee Valley Authority. The reach length, drainage area and identified number of sinkholes associated with each sampling site are summarized in Table 1. In general, the reach lengths associated with the sample sites ranged from 1.00 to 2.60 km (mean = 1.81 km, 
= 0.65 km). The drainage area associated with each sampling site is more variable (mean = 7.05 km2, = 0.65 km2), with values ranging from 2.40 to 21.14 km2; however, the ratio of reach length to drainage area is less variable (mean = 0.34 km/km2, = 0.13 km/km2), with values ranging from 0.11 to 0.47. The largest drainage area was associated with Site SC-2, located toward the discharge end of the watershed (Fig. 2). The number of sinkholes per basin ranged from 1 to 23 (Tennessee Valley Authority, unpublished data, 2004), thus the sinkhole densities (count/km2 of drainage area) ranged from 0.42 to 2.26 per km2 (mean = 1.32 per km2, = 0.66 per km2).
Flow Measurements and Field Parameters
Flow and stage measurements were taken along the main channel. Flow measurements were performed at Sites SC-2 through SC-7 using the velocity-area method. Stream velocity measurements were collected using a Marsh-McBirney (Frederick, MD) Model 2000 Flow Mate current meter. The flow meter is calibrated for zero velocity by submerging the device in a stagnant pool of water. In June 2004, Solinst (Georgetown, ON) Model 3001 Levelogger pressure transducers were installed at Sites SC-2 to SC-6 to record detailed stage hydrographs throughout the watershed during a storm event. Site conditions at SC-7 would not allow reliable transducer measurements, so it was not included.
Individual grab samples for turbidity were collected from the water surface and delivered, within 3 h, to the Tennessee Department of Health's Knoxville Regional Lab. Turbidity analyses were performed using a nephelometric turbidimeter, and the sample results were reported as nephelometric turbidity units (NTU). Temperature, pH, and specific conductance were measured in the field during sample collection using a YSI (Yellow Springs, OH) 610 MPS multiprobe meter. The pH probe on the YSI 610 MPS was calibrated using two standard pH buffer solutions before each sampling event.
Escherichia coli Analyses
Eighty-four water samples were collected for E. coli analysis from the main channel sites. Samples were placed on ice and delivered to the Tennessee Department of Health's Knoxville Regional Lab within 3 h of collection. In general, the samples were collected at approximately the same time (mid to late morning) and delivered to the lab by early afternoon to meet the 6-h holding times required for the bacterial analysis. The samples were collected by submerging the sample bottles upstream, taking care to avoid the incorporation of disturbed bed sediments in the sample. The Tennessee Department of Health's Knoxville Regional Lab used the Colilert method to analyze samples, in accordance with Method 9223 (Eaton et al., 1998). The method uses a defined substrate technology, in which E. coli uses ß-glucuronidase to metabolize the nutrient-indicator 4-methylumbelliferyl-ß-D-glucuronide to create fluorescence. Sample densities were reported as colony forming units per 100 mL of samples (CFU/100 mL). Instantaneous load rates in the stream were calculated by multiplying the flow rate by the E. coli density and are reported as CFU per day.
|
RESULTS AND DISCUSSION
|
|---|
The water quality and flow data have been summarized by using the geometric mean for each sampling station. A summary of these statistics are provided in Table 2 for all of the sampling dates. In addition to the geometric means, the range of values measured are presented below.
Stock Creek generally gained flow from SC-7 to SC-2, within the range of sensitivity of the flow measurement, with flow rates ranging from 1.48 to 12.40 m3/s at SC-2. For comparison purposes, flow rates at SC-7 in the headwater portion of the watershed ranged from 0.21 to 3.18 m3/s. The geometric means for flows at each of the sites ranged from 0.60 to 4.84 m3/s. Two National Weather Service (NWS) gauge stations (40-4946 and 40-4950) were used to calculate the average 7-d antecedent precipitation associated with each sampling event. The 7-d antecedent precipitation ranged from 0.00 to 66.29 mm, in October and July 2003, respectively.
Densities of E. coli ranged from 10 to 3500 CFU/100 mL. The calculated E. coli loading rates ranged from 0.04 x 1010 to 77.41 x 1010 CFU/d. The highest densities and loading rates were observed toward the outlet of the watershed at Site SC-2, which had geometric means of 348 CFU/100 mL and 14.58 x 1010 CFU/d, respectively.
A scatter plot matrix with corresponding correlation coefficients for each of the bivariate combinations on the whole watershed data set is shown in Fig. 3. Correlation coefficients near –1 exhibit a perfect inverse relationship, whereas coefficients near +1 exhibit a perfect corresponding correlation. Bivariate 95% density ellipses, the area encompassing 95% of the data set, are also shown in Fig. 3 to demonstrate of the spread of the data. The more rounded ellipses indicate poor correlation, whereas elongated ellipses are indicative of stronger correlation. The analyses demonstrate that the E. coli load rate is most strongly correlated (
xy = 0.77) with turbidity and with antecedent precipitation (xy = 0.54) throughout the watershed. Density of E. coli alone correlated poorly with most hydrologic parameters.
Spatial dependencies were evaluated by calculating the degree of correlation for all bivariate combinations at individual main channel sites, as was done with the whole data set. The analysis indicates an interesting pattern throughout the watershed for E. coli load rates with turbidity and antecedent precipitation, which are summarized in Table 3 for each main channel site. The correlation coefficient for antecedent rainfall and turbidity decreased downstream to upstream (0.13 
xy 0.56). Antecedent precipitation's correlation to E. coli load was highest in the upstream reaches, SC-5 to SC-7 (0.80 xy 0.88), and decreased in the downstream reaches, SC-2 to SC-4 (0.54 xy 0.70). Load rates of E. coli correlated to turbidity best in the downstream reaches, SC-2 to SC-4 (0.81 xy 0.90), and correlated less in the upstream reaches, SC-5 to SC-7 (0.60 xy 0.66). These findings may indicate the same pattern noted by Dussart-Baptista et al. (2003), where high E. coli load rates were associated with slower draining, sediment-rich conditions in the lower elevations of the main channel associated with the runoff event, and low sediment persistent E. coli load rates where possible sessile bacteria exist (either by karst delivery or resuspension of bed material). A similar relationship between parasite, bacterial load, and turbidity has been identified by Kistemann et al. (2002). In this study, however, we see spatial dependencies in the relationship between turbidity and E. coli loads. The subbasins SC-1 and SC-3 had the highest sinkhole densities determined for the watershed, and the karst features could play a role in E. coli persistence in the tailwater portion of the watershed; however, the role of mixed-use land surface inputs could also play a role in the E. coli persistence in the tailwater portion of the watershed. Further studies should be performed to elucidate the combined role of karst and mixed land use.
The spatial dependencies observed throughout the watershed are probably connected to the drainage–storage relationships (surface and subsurface) of each subbasin. The total precipitation recorded from 16 to 17 June 2004 for NWS 40-4946 and 40-4950 were 26 and 1 mm, respectively. The resulting normalized stage hydrographs, h/hmax, from the storm systems is shown in Fig. 4. From the associated peak stages, differential times of hydrologic concentration were calculated between the respective subbasins and ranged from 0.5 to 0.95 h. More importantly, we observe differences between the subbasin hydrograph recession limbs. As seen in Fig. 4, SC-5 and SC-6 tended to have more rapid hydrologic responses than SC-2 to SC-4. The shape of the recession limbs and less defined peaks indicate that SC-2 to SC-4 are indicative of hydrologic storage with probable longer residence times associated with those systems. The expectation would be to find higher relative E. coli load rates within the slower draining systems after a storm event, which was observed at SC-2 and only occasionally at SC-4. The presence of E. coli in the faster response basins after the storm event is indicative of persistance due to a non-runoff supply to the stream, which was observed in the headwater basins of Stock Creek.
View larger version (15K):
[in this window]
[in a new window]
|
Fig. 4. Transducer stage hydrographs for select subbasins ( tc is the difference in time of concentration based on stage hydrograph peaks).
|
|
|
CONCLUSIONS
|
|---|
Our study of base-flow conditions indicates that the turbidity in the Stock Creek watershed correlates most strongly with E. coli load rates and not density. Moreover, there is a spatial dependency of the E. coli load rate on turbidity and 7-d antecedent precipitation. Due to the drainage–storage characteristics associated with individual subbasins at base flow, the E. coli load rate (i) correlated best with antecedent precipitation in headwater subbasins, and (ii) correlated better with turbidity in the higher order streams at the tailwater end of the basin. The slow shedding of storm water in the tailwater end of the watershed may have resulted in higher sediment and E. coli mass loadings. This suggests a strong tie to E. coli persistence in the slow-draining subbasins due to possible sediment attachment, although this hypothesis needs to be tested in future studies. The persistence of E. coli in faster draining subbasins may be tied more directly to the antecedent storm events and elevated base-flow response.
|
ACKNOWLEDGMENTS
|
|---|
This research was funded by the Tennessee Department of Environment and Conservation (TDEC) under the direction of Dr. Sherry Wang and Mr. Jonathon Burr, with cofunding from the Water Resources Research Institute Program (Project no. 2003TN7b) and the University of Tennessee Center for Environmental Biotechnology. We would like to thank Mr. Burr for additional collaboration in the selection of field sampling sites and for coordination of field equipment and sample analysis by the Tennessee Department of Health's Knoxville Regional Laboratory.
|
REFERENCES
|
|---|
- Byappanahalli, M., M. Fowler, D. Shively, and R. Whitman. 2003. Ubiquity and persistence of Escherichia coli in a midwestern coastal stream. Appl. Environ. Microbiol. 69:4549–4555.[Abstract/Free Full Text]
- Dufour, A.P. 1984. Health effects criteria for fresh recreational waters. EPA-600/1-84-004. USEPA, Cincinnati, OH.
- Dussart-Baptista, L., N. Massei, J.P. Dupont, and T. Jouenne. 2003. Transfer of bacteria-contaminated particles in a karst aquifer: Evolution of contaminated materials from a sinkhole to a spring. J. Hydrol. 284:285–295.[CrossRef]
- Eaton, A.D., L.S. Clesceri, A.E. Greenberg, and M.A.H. Franson. 1998. Standard methods for the examination of water and wastewater. Am. Public Health Assoc., Washington, DC.
- Feng, P., S.D. Weagant, and M.A. Grant. 2002. Enumeration of Escherichia coli and the coliform bacteria. Ch. 4. In Bacteriological analytical manual. Available at www.cfsan.fda.gov/
ebam/bam-4.html (verified 1 Aug. 2006). U.S. Food and Drug Admin., Washington, DC. - George, I., A. Anzil, and P. Servais. 2004. Quantification of fecal coliform inputs to aquatic systems through soil leaching. Water Res. 38:611–618.
- Jamieson, R., R. Gordon, D. Joy, and H. Lee. 2004. Assessing microbial pollution of rural surface waters: A review of current watershed scale modeling approaches. Agric. Water Manage. 70:1–17.[CrossRef]
- Kistemann, T., T. Classen, C. Koch, F. Dangendorf, R. Fischeder, J. Gebel, V. Vacata, and M. Exner. 2002. Microbial load of drinking water reservoir tributaries during extreme rainfall and runoff. Appl. Environ. Microbiol. 68:2188–2197.[Abstract/Free Full Text]
- Mallin, M.A., S.H. Ensign, M.R. McIver, G.C. Shank, and P.K. Fowler. 2001. Demographic, landscape, and meteorological factors controlling the microbial pollution of coastal waters. Hydrobiologia 460:185–193.[CrossRef]
- Massei, N., M. Lacroix, H.Q. Wang, B.J. Mahler, and J. Dupont. 2002. Transport of suspended solids from a karstic to an alluvial aquifer: The role of the karst/alluvium interface. J. Hydrol. 260:88–101.[CrossRef]
- Massei, N., H.Q. Wang, J.P. Dupont, J. Rodet, and B. Laignel. 2003. Assessment of direct transfer and resuspension of particles during turbid floods at a karstic spring. J. Hydrol. 275:109–121.[CrossRef]
- National Research Council. 2001. Assessing the TMDL approach to water quality management. Natl. Acad. Press, Washington, DC.
- Olyphant, G.A., J. Thomas, R.L. Whitman, and D. Harper. 2003. Characterization and statistical modeling of bacterial (Escherichia coli) outflows from watersheds that discharge into southern Lake Michigan. Environ. Monit. Assess. 81:289–300.[CrossRef][ISI][Medline]
- Reeves, R.L., S.B. Grant, R.D. Mrse, C.M.C. Oancea, B.F. Sanders, and A.B. Boehm. 2004. Scaling and management of fecal indicator bacteria in runoff from a coastal urban watershed in southern California. Environ. Sci. Technol. 38:2637–2648.[Medline]
- Tyrrel, S.F., and J.N. Quinton. 2003. Overland flow transport of pathogens from agricultural land receiving faecal wastes. J. Appl. Microbiol. 94(Suppl. S):87S–93S.
- USEPA. 2002. The twenty needs report: How research can improve the TMDL program. EPA-841-B-02-002. USEPA, Washington, DC.
This article has been cited by other articles:
|
|
|
|
 
P. Vidon, L. P. Tedesco, J. Wilson, M. A. Campbell, L. R. Casey, and M. Gray
Direct and Indirect Hydrological Controls on E. coli Concentration and Loading in Midwestern Streams
J. Environ. Qual.,
August 8, 2008;
37(5):
1761 - 1768.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
S. R. Koirala, R. W. Gentry, E. Perfect, J. S. Schwartz, and G. S. Sayler
Temporal Variation and Persistence of Bacteria in Streams
J. Environ. Qual.,
June 23, 2008;
37(4):
1559 - 1566.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
O. Vigiak, O. Ribolzi, A. Pierret, O. Sengtaheuanghoung, and C. Valentin
Trapping Efficiencies of Cultivated and Natural Riparian Vegetation of Northern Laos
J. Environ. Qual.,
May 1, 2008;
37(3):
889 - 897.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
R. W. Gentry, A. C. Layton, L. D. McKay, J. F. McCarthy, D. E. Williams, S. R. Koirala, and G. S. Sayler
Efficacy of Bacteroides Measurements for Reducing the Statistical Uncertainty Associated with Hydrologic Flow and Fecal Loads in a Mixed Use Watershed
J. Environ. Qual.,
July 17, 2007;
36(5):
1324 - 1330.
[Abstract]
[Full Text]
[PDF]
|
|
|
TOP
ABSTRACT
INTRODUCTION
for E. coli, turbidity, and 7-d antecedent rainfall (P) throughout the watershed.