Chattahoochee BacteriALERT

By Stephen J. Lawrence (slawrence@usgs.go), U.S. Geological Survey, Atlanta, Georgia

Monitoring coliform bacteria in a Piedmont river arising from the
Appalachian region of Northern Georgia - Introduction

Because of historically high levels of indicator bacteria in the Chattahoochee River, the concept of a bacteria alert network was proposed as a means to inform people when bacteria levels in the river exceed U.S. Environmental Protection Agency (USEPA) criteria. Thus, a program of bacteria monitoring called BacteriALERT was initiated on the Chattahoochee River within the Chattahoochee River National Recreation Area (CRNRA, National Park Service) in Fall 2000. The CRNRA contains about three-fourths of all public green space in a 10-county area of Metropolitan Atlanta, Georgia. In 1999, the recreation area attracted about 2.9 million visitors with nearly 30 percent of those participating in water-based recreation. Within the area of the bacteria alert network, drinking water and recreation are the designated uses for the Chattahoochee River.

BacteriALERT is a partnership between State and Federal agencies and non-government organizations. This partnership includes the Georgia Environmental Protection Division, the National Park Service, and the U.S. Geological Survey (USGS) and non-governmental organizations such as the Upper Chattahoochee RiverKeeper, Georgia Conservancy, and Trust for Public Lands.

The main objective of this network is to collect and analyze water samples for total coliform and Escherichia coli (E. coli) bacteria and compute the percent probability that E.coli bacteria present a high-health risk (E.coli count over 235 colonies per 100 milliliters of water) at three sites on the Chattahoochee River. The sites are, from north to south, the Chattahoochee River at Medlock Bridge Road, the Chattahoochee River at Johnson's Ferry Road, and the Chattahoochee River at Paces Ferry. Results are posted on a publicly-accessible web site within 24 hours of data collection. A second objective is the statistical analysis and interpretation of these data under a wide range of seasonal, weather, and river conditions.

Sample Collection

[Picture of a USGS scientist lowing the sampling mechanism into the Chattahoochee River.] Water samples are collected daily using U.S. Geological Survey (USGS)-approved methods. The method uses a weighted-yoke to hold a sterile, narrow-mouth, 1-liter polypropylene bottle. A single, vertically integrated sample is collected at the center of flow. Turbidity and specific conductance in collected samples are measured in the laboratory. All analyses are completed in the bacteria laboratory at the USGS office in Atlanta.

Sample Analysis

The bacteria analysis for total coliform and E. coli is an enzyme substrate method called Colilert that is analogous to the commonly used multiple tube method. Bacteria counts are expressed as a most probable number (MPN) per 100 milliliters (mL). Three or four different dilutions are prepared for each site by adding an aliquot of sample to sterile, deionized/distilled (DI) water to produce 100 mL of liquid. A powdered reagent is added to each dilution bottle and the mixture added to a sterile, plastic tray containing 97 wells and incubated for 20 hours at 35 degrees Celsius. The wells produce a yellow color when total coliform bacteria are present and fluoresce under UV light when E. coli are present.

[Picture showing plastic tray with some cells indicating that total coliform is present (cells are yellow, rather than clear).]

Quality Control

Quality control is maintained by using sterile technique, collecting duplicate samples, and analyzing split samples using membrane filtration methods. Total coliform, E. coli, and fecal coliform bacteria are intermittently analyzed using the standard membrane filtration methods. Total coliform and E. coli filters are incubated in HACH's m-Coliblue24 broth.

Outreach

Three methods are used to inform the public, visitors to the park, and interested citizens of the previous day's E. coli bacteria counts: (1) via a web page where E. coli counts from samples collected the previous day are posted; (2) Electronic mail notification; and (3) signage at the entrances to each unit of the park.  
[Example of the BacteriALERT home page]
[Example of the daily email that goes out when someone signs up for the email service.]
[Picture of National Park Service sign telling of high bacteria levels] 3. Photo courtesy of National Park Service, 2001

Results: E.coli

One goal of the BacteriALERT program is the real-time estimation of E. coli bacteria concentration using real-time monitoring of stream turbidity. To that end, total coliform and E. coli bacteria levels in Panola Creek, the background site, and two sites on the Chattahoochee River were analyzed in relation to changes in streamflow and stream turbidity.
[Median concentrations of total coliform and E. coli bacteria at Chattahoochee sites.]

Median concentrations of total coliform and E. coli bacteria were higher at the Atlanta site on the Chattahoochee River (6,000 and 120 MPN/100 mL, respectively) than at the minimally impacted Panola Creek site (3,500 and 120 MPN/100 mL) or the Norcross site on the Chattahoochee River (1,200 and 65 MPN/100 mL). However, maximum concentrations of both bacteria were highest at the two Chattahoochee sites.
[Fifteen percent of the 214 E. coli bacteria samples at the Norcross site on the Chattahoochee River exceeded 235 MPN/100 mL]

Fifteen percent of the 214 E. coli bacteria samples at the Norcross site on the Chattahoochee River exceeded 235 MPN/100 mL, the USEPA criteria for streams with recreational uses; while at the Atlanta site, the USEPA criteria was exceeded in 32 percent of the 217 E. coli samples.

Results: Turbidity

Rain storms generally produce increases in streamflow and river turbidity within the Chattahoochee River. Statistical analysis showed that river turbidity was strongly related to streamflow at both Chattahoochee River sites. Furthermore, both total coliform and E. coli bacteria are strongly related to streamflow-adjusted river turbidity concentration at the Norcross and Atlanta sites. In contrast, E. coli concentrations were not related to river turbidity at the Panola Creek site. Although streamflow data were not available for Panola Creek, streamflow may govern bacteria transport in Panola Creek; whereas, rainfall-generated turbidity may govern bacteria movement in the Chattahoochee River.

[Statistical analysis showed that river turbidity was strongly related to streamflow at both Chattahoochee River sites. Furthermore, both total coliform and E. coli bacteria are strongly related to streamflow-adjusted river turbidity concentration at the Norcross.]
[Statistical analysis showed that river turbidity was strongly related to streamflow at both Chattahoochee River sites. Furthermore, both total coliform and E. coli bacteria are strongly related to streamflow-adjusted river turbidity concentration at the Atlanta sites. E. coli concentrations were not related to river turbidity at the Panola Creek site. Although streamflow data were not available for Panola Creek, streamflow may govern bacteria transport in Panola Creek. ]

Results: Logistic regression models

Logistic regression models of turbidity versus categorical data relating E. coli levels above and below the U.S. Environmental Protection Agency E. coli standard show that at the Norcross site all samples exceeded the E. coli standards when turbidity was above 50 nephelometic turbidity units (NTU). At the Chattahoochee River at Atlanta site, all samples exceeded the E. coli standard when turbidity was above 60 NTU.

Summary

The BacteriALERT program has proven to be successful in alerting the public to E. coli bacteria concentrations in the Chattahoochee River. The web site is visited hundreds of times per day and about 100 people routinely receive daily email giving bacteria counts from this program. Data analysis to date shows that total coliform and E. coli bacteria levels fluctuate greatly, even in a stream within a minimally impacted watershed. Although the bacteria levels at the background site, Panola Creek, do not appear related to turbidity, bacteria levels in the Chattahoochee are strongly related to turbidity. Using these relations, a logistic model will be developed to predict the probability that E. coli bacteria levels will exceed the US Environmental Protection Agency standard at a given turbidity. With the logistic model and continuous turbidity sondes deployed at each site, web-accessible real-time monitoring of E. coli concentration and potential health risk will be possible.