Escherichia coli at Ohio bathing beaches—Distribution, sources, wastewater indicators, and predictive modeling

Results of studies during the recreational seasons of 2000 and 2001 strengthen the science that supports monitoring of our Nation's beaches. Water and sediment samples were collected and analyzed for concentrations of Escherichia coli (E. coli).  Ancillary water-quality and environmental data were collected or compiled to determine their relation to E. coli concentrations. Data were collected at three Lake Erie urban beaches (Edgewater, Villa Angela, and Huntington), two Lake Erie beaches in a less populated area (Mentor Headlands and Fairport Harbor), and one inland-lake beach (Mosquito Lake).

The distribution of E. coli in water and sediments within the bathing area, outside the bathing area, and near the swash zone was investigated at the three Lake Erie urban beaches and at Mosquito Lake. (The swash zone is the zone that is alternately covered and exposed by waves.) Lake-bottom sediments from outside the bathing area were not significant deposition areas for E. coli. In contrast, interstitial water and subsurface sediments from near the swash zone were enriched with E. coli. For example, E. coli concentrations were as high as 100,000 colonies per 100 milliliters in some interstitial waters. Although there are no standards for E. coli in swash-zone materials, the high concentrations found at some locations warrant concern for public health.

Studies were done at Mosquito Lake to identify sources of fecal contamination to the lake and bathing beach. Escherichia coli concentrations decreased with distance from a suspected source of fecal contamination that is north of the beach but increased at the bathing beach. This evidence indicated that elevated E. coli concentrations at the bathing beach are of local origin rather than from transport of bacteria from sites to the north.

Samples collected from the three Lake Erie urban beaches and Mosquito Lake were analyzed to determine whether wastewater indicators could be used as surrogates for E. coli at bathing beaches. None of the concentrations of wastewater indicators of fecal contamination, including 3b-coprostanol and cholesterol, were significantly correlated (a=0.05) to concentrations of E. coli. Concentrations of the two compounds that were significantly correlated to E. coli were components of coal tar and asphalt, which are not necessarily indicative of fecal contamination.

Data were collected to build on an earlier 1997 study to develop and test multiple-linear-regression models to predict E. coli concentrations using water-quality and environmental variables as explanatory variables. The probability of exceeding the single-sample bathing-water standard for E. coli (235 colonies per 100 milliliters) was used as the model output variable. Threshold probabilities for each model were established. Computed probabilities that are less than a threshold probability indicate that bacterial water quality is most likely acceptable. Computed probabilities equal to or above the threshold probability indicate that the water quality is most likely not acceptable and that a water-quality advisory may be needed.

Models were developed at each beach, whenever possible, using combinations of 1997, 2000, and (or) 2001 data. The models developed and tested in this study were shown to be beach specific; that is, different explanatory variables were used to predict the probability of exceeding the standard at each beach. At Mentor Headlands and Fairport Harbor, models were not developed because water quality was generally good. At the three Lake Erie urban beaches, models were developed with variable lists that included the number of birds on the beach at the time of sampling, lake-current direction, wave height, turbidity, streamflow of a nearby river, and rainfall. The models for Huntington explained a larger percentage of the variability in E. coli concentrations than the models for Edgewater and Villa Angela. At Mosquito Lake, a model based on 2000 and 2001 data contained the explanatory variables rainfall, number of dry days preceding a rainfall, date, wind direction, wind speed, and turbidity. Additional research could include testing the threshold probabilities assigned for these models in subsequent years and comparing the models’ ability to predict recreational water quality to results from the current method—using antecedent E. coli concentrations. Each year the model is tested, new data can be added and model variables can be recalculated to determine whether the predictive ability improves with additional data.