Alexandria K. Graves, Doctoral Candidate and Charles Hagedorn, Professor; CSES Department
Alex Graves Sampling Mill Creek in Winter
Stream samples from Mill Creek, Montgomery County, VA, were collected monthly for one year, plus weekly for four consecutive weeks during seasonal high flows (March), and seasonal low flows (September-October), plus daily for seven consecutive days within each of the weekly schedules (30 stream samples per site for each of two sites, 60 total). Forty-eight isolates of E. coli per sample (total of 1,440 stream isolates) were classified by source using antibiotic resistance analysis (ARA) and comparing the resulting fingerprints against a known-source E. coli library (1,158 isolates). The same sampling and analyses were also performed on Enterococcus sp., but this report deals only with E. coli. The 12-month averages for bacterial source tracking (BST) were not different from the daily and weekly averages for high flow and low flow, both sites, and indicated that monthly sampling was adequate. There was a seasonality effect in that the human signature trended higher during high flow (if it was present at all) while the livestock signature dominated all samplings and the wildlife signature was slightly higher during low flow. Restoration of Mill Creek will have to primarily address cattle operations. There were essentially no horse or waterfowl signatures at any time and deer constituted the majority of the wildlife signature. Our results indicated that sampling should be done over a time period that includes both seasonal wettest and driest periods (at least eight months). Quarterly and every-other-month sampling was not adequate.
Nearly 65% of the impaired stream segments in Virginia are contaminated by fecal pollution (6, iii). Fecal bacteria are the major cause of impairments in Virginia's waterways, according to the Virginia Department of Environmental Quality (iv), with agriculture listed as the primary source of contamination (http://www.deq.state.va.us/water/98-305b.html). This fecal contamination results in increased health risks to persons exposed to the water, degradation of recreational and drinking water quality and shellfish bed closures. Based on recent developments in microbiology and molecular biology, it is now possible to determine with reasonable accuracy the sources of fecal bacteria in water (these new developments are referred to as bacterial source tracking (BST) (see i: http://soils1.cses.vt.edu/ch/biol_4684/bst/BST.html). Identification of the sources of fecal pollution in impaired waters will aid in the restoration of water quality, and reduce the danger of infectious diseases from exposure to contaminated recreational and shellfish-harvest waters (5, 7). In Virginia, state regulatory agencies face a mandate from the U.S. Environmental Protection Agency (EPA) to perform a total maximum daily load (TMDL) on some 600+ impaired waterways within the next 10 to 12 years (iii, iv). To perform TMDLs in Virginia, or anywhere else, it becomes critical no know how "representative" the data for such TMDLs actually are (6). To evaluate the "representativeness" of BST results, stream samples were collected monthly for one year, then weekly for four consecutive weeks during seasonal high flows (March), and seasonal low flows (September-October), and daily for seven consecutive days within both weekly schedules. The results from this frequency of sampling study will allow those responsible for TMDLs to obtain some idea regarding the number of samples that need to be collected, over what time period, and how often sampling should occur.
While many of Virginia's waters test positive for fecal coliforms, no useful watershed restoration plans or accurate TMDLs for bacteria can be developed until the source(s) of the fecal contamination can be identified (3, 6). By reliably and accurately identifying different fecal sources (e.g. human, cattle, poultry, wildlife), BST will provide an essential tool to those who are responsible for public health and environmental quality and are charged with reducing water pollution, protecting public health, and improving water quality (1, 2, 4). The results of this project will provide a mechanism for non-point source (NPS) problem identification in fecal contaminated waters and a means for determining TMDLs for fecal bacteria based on specific source(s) of the bacteria (ii). The possibility of establishing TMDLs for fecal bacteria by specific source(s) is both novel and unique at present. Until sources of pollution are identified, risk to communities cannot be accurately assessed, and water quality improvements will remain a hit-or-miss affair.
The goal of this study was to determine sources of fecal pollution in Mill Creek, Montgomery County, Va., as a function of sampling frequency under low and high stream flow conditions. The project objectives were (1) to build a small library of Escherichia coli (E. coli) isolates from known sources in the watershed and compare it to a larger, regional known-source library, and (2) compare E. coli isolates from the stream (unknown sources) against the library to determine their sources.
|Mill Creek Watershed at a Glance|
|Study area||3,767ha (9,308 ac)|
|Predominant Use||Pasture and cattle (3,800 animals, beef and dairy)|
|Length of stream||9.2 km (5.7 mi)|
|Number of impaired segments||1 (entire stream)|
|Land use||60% pasture, 25% crops, 10% residential, 5% forest|
|Other facts of interest||Eighty-eight percent of homes (567) are on septic tank/drainfield systems|
Mill Creek is a typical "ridge and valley stream," wide and shallow. The watershed is small (4,000 ha) and is dominated by pasture and cattle production. Two sampling sites were used for this study. Site MC-1 was located just above the confluence of Mill Creek with Meadow Creek, and Site MC-2 was located upstream approximately half the length of the stream. Site MC-1 averaged 11 ft wide, a high flow depth of 4.9 in. and a low flow depth of 4.0 in., while site MC-2 averaged 3.5 ft wide, a high flow depth of 3.4 in. and a low flow depth of 2.1 in. (see pictures below).
Mill Creek, Sample Site MC-1
Mill Creek, Sample Site MC-2
A watershed-specific library was constructed from known sources within the Mill Creek watershed. The Mill Creek library contained 384 E. coli isolates, and the regional library (384 Mill Creek known source isolates plus 774 others from the Blacksburg area, Holston River and Blackwater River watersheds) contained 1,158 E. coli isolates. Isolates for the Mill Creek library were collected from six sources within the Mill Creek watershed. These sources included dairy and beef cow 34 samples), horse (10 samples), deer (36 samples), wildlife (raccoon, muskrat, 5 samples), goose and duck (waterfowl, 5 samples), and human (30 samples). Three categories (human, cow, deer) were selected as the predominant potential sources of fecal pollution based on a sanitary/agriculture/wildlife survey provided by state regulatory officials and comments from citizens at public meetings.
Each known source E. coli isolate was tested for growth on 28 concentrations of seven different antibiotics and either "growth" or "no growth" was recorded for each concentration. This pattern of growth/no growth formed the antibiotic resistance profile, much like a bar code, for every E. coli isolate. Profiles of the E. coli isolates were combined to form the libraries (watershed-specific and regional), and these were analyzed by both 3-way (human vs. livestock vs. wildlife) and species-specific known (6-way) source classifications using discriminant analysis (3, 5, 8). Livestock samples were collected from local farms, wildlife samples from locations where wildlife had been observed and the scats could be identified, and human sources were collected from local septic tank pump-out companies. Individual fecal samples were placed in sterile whirlpac bags and liquid manure and septic samples were collected in sterile polystyrene bottles. No more than 10 isolates were taken from any one fecal sample (3, 8).
Thirty stream samples were collected from each of two sites (60 total) in the Mill Creek watershed (MC-1 and MC-2), and 48 isolates of E. coli per sample were classified by source using antibiotic resistance analysis (ARA) obtain a growth profile, for a total of 2,880 stream isolates. This profile of each E. coli isolate was compared to the known-source library to determine its source. Library construction and source classification of stream isolates was performed with logistic regression using JMP statistical software (v. 5.0, SAS Institute, Cary, NC). Stream samples were collected monthly for one year, then weekly for four consecutive weeks during seasonal high flows (March), and seasonal low flows (September-October), and daily for seven consecutive days within the weekly schedules. Both daily sampling schedules included at least one storm event. Fecal coliform counts were determined for every water sample and results were recorded as colony forming units (CFU) per 100 ml. Flow rates (Global Water flow meter model FP201, Global Water, Inc., Gold River, CA), turbidity, water temperature, and pH were determined on all water samples. Samples were collected in sterile polystyrene bottles, placed on ice in coolers, transported to the laboratory and assayed within 6 hours of collection.
Sampling at Site MC-1
Sampling at Site MC-2
Fecal Coliforms and Water Quality: Based on fecal coliform counts, the recreational standard (1,000 colony forming units (CFUs)/100ml) was exceeded, for monthly sampling (12 samples per site), a total of eight times, 4 each for sites MC-1 and MC-2, for a 25% violation of the standard for each site. Water temperature, dissolved oxygen, and turbidity were generally within acceptable ranges and were not problematic. Average stream volume was 22 times greater at high flow for MC-1 than at low flow (12 month average flow was 2,081 gal/min.), and 9 times higher for MC-2 at high flow than at low flow (12 month average flow was 339 gal/min.). The low flows must be regarded as unusually low due to the severe drought that occurred August - November 2001.
The Known-Source Library, 3-Way Classification: The average rate of correct classification (ARCC) for the known source E. coli library, including the Mill Creek isolates, was 87.0% for a 3-way human vs. livestock vs. wildlife classification (Table 1). The ARCC of the 384 Mill Creek isolates alone was 98.4% for a 3-way classification and was 94.0% for the 6-way classification (data not shown). Most importantly, the ARCCs of the larger regional library, while slightly lower, were comparable to the ARCC of the smaller Mill Creek-specific library. This can be taken as evidence that regional libraries, at least for some geographical areas, can be developed and used in more than one watershed. Based on this evidence it appears that new libraries will not have to be developed for every new watershed, but it is still important to obtain a few hundred known-source isolates for each new watershed to be sure these are correctly classified by the regional library (when treated as unknowns). This holds true for any library-based source tracking method, not just for ARA.
Table 1. Human, livestock, and wildlife isolate numbers and source classification averages (%) for Mill Creek E. coli isolates.
The Known-Source Library, 6-Way Classification: The average rate of correct classification (ARCC) for the known source E. coli library, including the Mill Creek isolates, was 88.9% for the species-specific (6-way) classification (Table 2). The rates of correct classification (RCC) ranged from 100% for wildlife to 76.5% for deer. The high RCCs for wildlife, goose, and horse are typical of library sources with small numbers of isolates. Results of the 6-way classification provided acceptable RCCs for the major sources of concern: human, cow, and deer; and were very similar to the RCCs for the 3-way classification (Table 1).
Table 2. Human, cow, horse, deer, waterfowl, and miscellaneous wildlife isolate numbers and source classification averages (%) for Mill Creek E. coli isolates.
The advantage of the 6-way classification (Table 2) is that it allows cattle and deer to be separated from livestock and wildlife, respectively, and this was an important separation due to beef and dairy operations being the major land use in the watershed.
Source Tracking on Stream Isolates: The results of source tracking (comparing stream isolates against the library for both 3-way and 6-way classifications) demonstrated that cattle (livestock) comprised an overwhelming signature that dominated both the sampling scheme and seasonality in Mill Creek (Tables 3 and 4).
Table 3. Monthly, daily, and weekly human, livestock, and wildlife source classification averages (%) for Mill Creek E. coli isolates. Daily (7 days) Weekly (4 weeks)
|Daily (7 days)||Weekly (4 weeks)|
|Montly (12 months)||High Flow||Low Flow||High Flow||Low Flow|
Table 4. Monthly, daily, and weekly human, cow, horse, deer, waterfowl, and miscellaneous wildlife source classification averages (%) for Mill Creek E. coli isolates.
|Daily (7 days)||Weekly (4 weeks)|
|Montly (12 months)||High Flow||Low Flow||High Flow||Low Flow|
Frequency of Sampling: For E. coli, the 12-month averages for source tracking were statistically equivalent to the daily and weekly averages for high flow and low flow, both sites, and indicated that monthly sampling was adequate (Tables 3 and 4). The averages were compared by Pearson's test, Student's t test, and the chi square test. There was a seasonality effect in that the human signature trended higher during high flow while the livestock signature tended to be highest during low flow (but predominated over all seasons, Table 3). The wildlife signature also tended to be slightly higher during low flow, especially with the weekly average. The two storm events provided higher numbers of fecal coliforms in the water, but did not change the proportions of the sources of E. coli that were present before and after the storms. Quarterly sampling and every-other-month sampling were not adequate based on comparison against the monthly results.
Classification of Specific Sources: The human signature was higher in the 3-way classification (1.9% to 16.2%, 8.3% average, Table 3) than in the 6-way classification (2.0% to 7.8%, 4.9% average, Table 4). None of the human signatures were different in the 6-way classification (Table 4) while three signatures (MC-1 daily high flow and weekly high flow and low flow) were different from the rest in the 3-way classification (Table 3). The human signatures in the 3- way classification tended to be higher under high flow conditions, but this same trend was not seen with the 6-way classification.
There was no detectable seasonality or differences between the livestock signature (Table 3) and the cattle signature (Table 4). The cattle signature tended to be higher under low flow conditions and was higher for site MC-2, both classifications. There were essentially no horse or waterfowl signatures at any time (Table 4) and deer constituted the majority of the wildlife signature. These results were in accordance with the watershed survey conducted prior to this project (e.g., lots of cattle and deer, less than 25 horses, very few waterfowl, a few other types of wildlife such as raccoon and muskrat). Both classifications provided essentially the same results and demonstrated that restoration of Mill Creek will have to primarily address cattle operations.
i. Bacterial source tracking, see Dr. Charles Hagedorn's website: http://soils1.cses.vt.edu/ch/biol_4684/bst/BST.html.
ii. Bacterial Source Tracking Methodology (BST): March 2001 Update. Crop and Soil Environmental News: http://www.ext.vt.edu/news/periodicals/cses/2001-03/bst.html.
iii. EPA's Total maximum daily load (TMDL) program: http://www.epa.gov/owow/tmdl/.
iv. VA-DEQ's state TMDL and water quality assessment program: http://www.deq.state.va.us/water/98-305b.html.
1. Booth, A.M., C. Hagedorn, A.K. Graves, S.C. Hagedorn, and K.H. Mentz. 2002. Sources of Fecal Pollution in Virginia's Blackwater River. J. Environ. Engineering. Accepted for Publication.
2. Hagedorn, C., J.B. Crozier, K.A. Mentz, A.M. Booth, A.K. Graves, N.J. Nelson, and R.B. Reneau, Jr. 2002. Carbon Source Utilization Profiles as a Method to Identify Sources of Fecal Pollution in Water. J. Applied Microbiology. Accepted for Publication.
3. Graves, A.K., C. Hagedorn, A. Teetor, M.Mahal, A.M. Bowman, and R.B. Reneau, Jr. 2002. Antibiotic Resistance Profiles to Determine Sources of Fecal Contamination in a Rural Virginia Watershed. J. Environ. Qual.31:1300-1308.
4. Hagedorn, C., S.L. Robinson, J.R. Filtz, S.M. Grubbs, T.A. Angier, and R.B. Reneau, Jr. 1999. Using antibiotic resistance patterns in the fecal streptococci to determine sources of fecal pollution in a rural Virginia watershed. Appl. Environ. Microbiol. 65:5522-5531.
5. Harwood, V.J., J. Whitlock, and V.H. Withington. 2000. Classification of the antibiotic resistance patterns of indicator bacteria by discriminant analysis: Use in predicting the source of fecal contamination in subtropical Florida waters. Appl. Environ. Microbiol. 66:3698-3704.
6. McClellan, P., V.O. Shanholtz, B. Petrauskas, and J. Kern. 2000. Bacterial source tracking: a tool for total maximum daily load development. p. 17. In T. Younos and J. Poff (ed.), Abstracts, Virginia Water Research Symposium 2000, VWRRC Special Report SR-19-2000, Blacksburg, VA.
7. Wiggins, B.A. 1996. Discriminant analysis of antibiotic resistance patterns in fecal streptococci, a method to differentiate human and animal sources of fecal pollution in natural waters. Appl. Environ. Microbiol. 62:3997-4002.
8. Wiggins, B.A., R.W. Andrews, R.A. Conway, C.L. Corr, E.J. Dobratz, D.P. Dougherty, J.R. Eppard, S.R. Knupp, M.C. Limjoco, J.M. Mettenburg, J.M. Rinehardt, J. Sonsino, R.L. Torrijos, and M.E. Zimmerman. 1999. Identification of sources of fecal pollution using discriminant analysis: Supporting evidence from large datasets. Appl. Environ. Microbiol. 65:3483-3486.
Visit Virginia Cooperative Extension