Forecasting Prorocentrum minimum blooms in the Chesapeake Bay using empirical habitat models

Aquaculturists, local beach managers, and other stakeholders require forecasts of harmful biotic events, so they can assess respond to health threats when algal blooms (HABs) are present. Based on this need, we developing empirical habitat suitability models for a variety Chesapeake Bay HABs forecast their occurrence based set physical-biogeochemical environmental conditions, start with the dinoflagellate Prorocentrum minimum (also known as P. cordatum ).To identify an optimal variables blooms, first assumed linear relationship between inverse logistic function used likelihood bloom presence, repeated method using more than 16,000 combinations variables. By comparing goodness-of-fit, found water temperature, salinity, pH, solar irradiance, total organic nitrogen represented most suitable The resulting algorithm forecasted overall accuracy 78%, though significant variability ~ 30-90% depending region season. To understand improve model performance, incorporated nonlinear effects into by implementing generalized additive model. Even without considering interactions five train model, yielded increase in (~ 81%) due model’s ability refine regions which occurred. Including increased even further 85%) accounting seasonality interaction irradiance temperature. Our findings suggest that influence predictors these change time space, complexity impacts performance our interpretation driving factors causing blooms. Apart from forecasting potential, results may be particularly useful constructing explicit relationships conditions presence mechanistic models.