Prediction of present and future spatial occurrence of cyanobacteria and the toxin nodularin in the Baltic Sea

Abstract. Blooms of filamentous cyanobacteria are recurrent phenomena in the brackish Baltic Sea. These blooms often include toxin producing species, however, predicting and modeling the toxins spatial distribution poses great challenges. In addition, projected rising temperature due to climate change is expected to increase the occurrence of cyanobacterial blooms, making it vital to understand the distribution of the blooms and the associated cyanotoxins across ecosystems. Herein, we integrated measured concentration of the cyanotoxin nodularin, abundance of the toxin producer Nodularia spumigena, and environmental variables using Empirical Bayesian Kriging (EBK) regression prediction, ensemble learning, and stacked species distribution modeling (SSDM). This setup was used to predict and interpret the current and future area distribution of N. spumigena and nodularin across the Baltic Sea. Predictions were based on results from biogeochemical models describing current and projected future concentrations of near surface chlorophyll, nitrate, phosphate, salinity, and temperature along with nitrate-to-phosphate ratio and a geographical variable of distance to shore. Prediction for the future distribution was performed using projected climate change scenarios in the year 2100. Findings show that the predicted area distribution of nodularin is determined by concentrations and interaction effects of salinity, temperature, phosphate, nitrate to phosphate ratio, and distance to shore, and is associated with the predicted area distribution of N. spumigena. Predicted site distribution shows increased nodularin occurrences in the Eastern and Western Gotland Sea, the Northern Baltic Proper, southern parts of the Bothnian Sea, and in the Arkona basin. By the year 2100, area distribution of nodularin is predicted to increase in the northern part of the Eastern Gotland Sea, Northern Baltic Proper, Åland Sea, southern parts of the Bothnian Sea, Arkona Basin, and slightly into the Bothnia Bay in response to projected climate change scenarios. Our developed modeling approach is useful for risk assessment and management of cyanotoxins where toxicological data are insufficient.