Modeling and Forecasting Vibrio Parahaemolyticus Concentrations in Oysters

Area of Science:

• Marine microbiology
• Environmental science
• Predictive modeling

Background:

• Vibrio parahaemolyticus (V.p) is a significant pathogen in coastal waters, impacting oyster safety and the shellfish industry.
• Effective forecasting models are crucial for public health interventions and economic risk mitigation.

Purpose of the Study:

• To develop and validate predictive models for V.p levels in oysters with varying lead times (1-4 days).
• To identify key environmental predictors and their time lags influencing V.p abundance.

Main Methods:

• Utilized the Random Forest algorithm with 227 datasets from two distinct geographic locations.
• Identified critical environmental predictors and optimal time lags for V.p forecasting.
• Developed four models (RF-1Day to RF-4Day) and quantified prediction uncertainty using bootstrapping.

Main Results:

• V.p abundance in oysters is significantly influenced by antecedent environmental conditions (Sea Surface Temperature and salinity) 1-11 days prior.
• Forecasting models demonstrated reliable prediction of V.p levels 1-4 days in advance.
• Model performance decreased with increasing lead time, with RF-3Day and RF-4Day suitable for preparedness and RF-1Day/RF-2Day for management interventions.

Conclusions:

• Time-lagged environmental data, particularly SST and salinity, are effective predictors for V.p in oysters.
• The developed Random Forest models provide valuable tools for predicting V.p, supporting both emergency preparedness and management strategies in the shellfish industry.