Smarter models for predicting E. coli in Jacksonville rivers

By Megan Sam

A new machine learning model from the University of Florida could transform how Jacksonville monitors water quality in the St. Johns River, making it faster, more accurate and better able to protect public health. The model estimates E. coli levels using environmental and infrastructure data, reducing the need for costly, time-consuming water sampling. It also reveals which human activities are most strongly linked to elevated bacteria levels, giving communities a powerful tool for prevention and response.

Aerial view of a suspension bridge spanning a wide river under a clear blue sky. — A new machine learning model is helping scientists trace and predict bacterial pollution in Jacksonville’s St. Johns River (pictured above), offering faster, smarter tools to protect public health. What’s being learned in Jacksonville could help other communities across the country combat hidden bacterial threats in freshwater systems. (Photo credit: Wirestock – stock.adobe.com)

While most strains of E. coli, or Escherichia coli, are harmless, its presence in rivers signals fecal contamination, posing risks for swimmers, boaters and seafood consumers as some strains can cause serious illness if ingested. That makes E. coli a critical measure of waterborne health risks and a priority for water managers.

At the UF’s Center for Coastal Solutions (CCS) monthly forum on September 18, Ron Fick, Ph.D., introduced the model, developed with Kalindi Larios, Ph.D., a postdoctoral researcher in UF’s Agricultural and Biological Engineering Department.

Analyzing data from 1997 to 2023, the pair evaluated 89 different chemical substances. Nutrient levels such as nitrogen and phosphorus generally remained within healthy limits, whereas bacteria, including E. coli, enterococci and fecal coliform, frequently exceeded state standards, often by wide margins. The likely culprits? Leaking septic tanks, aging sewer systems, and increased urban development.

The model incorporates variables such as rainfall, temperature, population, pH and salinity of river water, number of septic tanks, and age of nearby homes. It quantifies how much each factor contributes to E. coli levels. Since Florida discontinued fecal coliform monitoring in 2017, and enterococci are primarily used in saltwater settings, E. coli currently serves as the most practical and reliable indicator of freshwater contamination. Paved surfaces, declining upland forests, and concentrations of septic tanks and older homes emerged as the strongest predictors of high E. coli levels, underscoring the influence of development on water quality.

“We can’t change things like rainfall,” Fick said. “But if the drivers are related to infrastructure, humans can change that. The goal is to predict bacteria levels and understand which factors drive them, so we can act where it matters most.”

A man in a blue Florida Gators polo shirt stands beside a display about AI combating red tide. A screen shows a coughing man. — Ron Fick, Ph.D., a research scientist at UF’s Center for Coastal Solutions, has worked with the city of Jacksonville on several projects to improve its wastewater infrastructure. This includes a collaboration with CCS researcher Tricia Kyzar, Ph.D., to study how climate-related impacts may affect the vulnerability of the city’s septic systems. (Photo credit: Megan Sam)

The findings suggest that bacterial pollution may be more widespread than current monitoring captures, underscoring the value of broader and more proactive management efforts. The model has already demonstrated strong predictive performance. Fick and Larios have reported results to representatives from the Jacksonville Environmental Protection Board and the Jacksonville Office of Resilience and are now finalizing a paper to share their findings.

For Fick, communicating the science effectively is important when it comes to building public understanding and support.

“Wastewater is a very personal topic; digging up yards to put in sewer lines isn’t something people always welcome,” he said. “But if you can show them the river is full of bacteria and explain why, that might convince them. Across Florida, and especially in Jacksonville, there’s momentum for wastewater improvements. With better science, we can give communities the evidence they need to drive change.”

A man stands in front of a projection screen showing a chart of the New Moon's position over months, with images of the horizon for each month. He points to the char — Arnoldo Valle-Levinson, Ph.D., a professor in UF’s Department of Civil and Coastal Engineering and an affiliate faculty of the CCS, demonstrates how the moon’s position aligns with decadal changes in sea levels. While human activity also affects sea level, his research focuses on astronomic influences. (Photo credit: Megan Sam)

The forum also offered a broader perspective on how gravitational forces from the moon and radiational forces from the sun can influence local water hazards, such as sunny-day flooding, saltwater intrusion and coastal erosion. Arnoldo Valle-Levinson, Ph.D., a professor in UF’s Department of Civil and Coastal Engineering, presented research linking the lunar nodal cycle to year-to-year and decade-to-decade shifts in sea levels and water quality. His work explores impacts of the moon’s decadal cycle on coastal dynamics, including the daily maxima in water level and the associated inland movement of saltwater into freshwater systems.

Together these presentations underscored the value of the CCS monthly forum as a space where students, researchers, staff and stakeholders can connect cutting-edge science, from the cosmic to the concrete, with real world challenges.