Continuous Bayesian network for studying the causal links between phosphorus loading and plankton patterns in Lake Simcoe, Ontario, Canada.

An ecosystem perspective to restoring beneficial uses in Areas of Concern can be interpreted as a shift from the traditional elucidation of simple cause-effect relationships to a multicausal way of thinking that more effectively accommodates ecosystem complexity. This holistic management paradigm has also pervaded the contemporary ecological modeling practice, making compelling the adoption of more sophisticated ecosystem modeling tools. In this study, our primary objective is to develop a Bayesian hierarchical network of simple ecological models for Lake Simcoe, Ontario, Canada, aiming to establish a realistic representation of the causal connections among exogenous nutrient loading, ambient nutrient conditions, and epilimnetic plankton dynamics. In particular, we used a spatially explicit simple mass-balance model forced with idealized sinusoidal loading to predict total phosphorus concentrations. A structural equation model was then used to delineate the interplay among nutrients, ambient light conditions, phytoplankton, and herbivorous biomass. Our analysis highlights the strength of the causal linkages between total phosphorus and water clarity with phytoplankton as well as the capacity of zooplankton grazing to modulate the algal standing crop. Our Bayesian network is also used to examine the exceedance frequency of threshold values for total phosphorus (15 μg/L) and chlorophyll a (4 μg/L) concentrations under scenarios of phosphorus loading reduction. Our study suggests that a 15% phosphorus loading decrease will still result in >25% violations of the 4 μg chla/L value in the two embayments of Lake Simcoe (Cook's Bay and Kempenfelt Bay). The TP levels will decrease in response to the exogenous loading reductions and this improvement will be primarily manifested in the northcentral segments of the system.