Bayesian inference for predicting ecological water quality under different climate change scenarios

The aim of this paper is to assess the separate and interactive effects of eutrophication and climate variables on the sea water quality in Parnu Bay (the Gulf of Riga, Baltic Sea) using multivariate statistical analyses and the Bayesian Belief Network (BBN) methodology. The assessment was based on the following biological quality elements: phytoplankton, submerged aquatic vegetation and benthic invertebrates. The multivariate statistical analyses suggest that zoobenthos communities are largely driven by weather conditions (i.e. climate variables), phytoplankton by nutrient loads while the dynamics of macrophyte communities was due to the combined effect of weather and nutrient loads. The BBN constructed for this study represents uncertainty in ecological water quality assessment. Probabilistic modeling shows that phytoplankton and zoobenthos are not sensitive to climate change impacts while phytobenthos would suffer from decrease in sea water salinity. Under climate change, therefore, phytobenthos is one of the key variables in determining the water quality in the study area.