A North Atlantic Marine Ecosystem Model: Model Calibration and Climate impact evaluation

The North Atlantic Ocean is a region marked by high value commercial fisheries as well as substantial uncertainties in present-day climate simulations and future climate modelling experiments, posing challenges to effective policymaking. To better explore these uncertainties and aid in policy development for fisheries management, a North Atlantic marine ecosystem model of intermediate complexity was created by repurposing a regional subsample of EcoOcean (a global Ecospace model) and calibrating it to historical catch and environmental data. The new model is intended for short term mass balance and temporally-dynamic modeling in the Ecopath with Ecosim environment, as well as spatio-temporal modeling using the Ecospace module, for scenario generation exploring fishing pressure and/or environmental change. The model was created by dividing the North Atlantic into coherent subregions based on common marine geochemical and ecological profiles. Historical catch and environmental data were used to calibrate the biomass of functional groups throughout each subregion, which were iteratively adjusted to withstand historical fishing pressure and forcing of primary productivity and sea surface temperature. Model parameters were varied during a Stock Reduction Analysis, which employed Monte Carlo simulations, in order to achieve best fits to historical landings data. The resulting calibrated subregional populations were aggregated back into a 204 functional group member North Atlantic-wide model. Subregional models achieved good fits to the landings data outside of the calibration period. The Sum of Squares Error (SSE) for most functional groups was less than 10, with many SSEs of less than 1. This is comparable to or better than the goodness of fit obtained by smaller regional models fit explicitly to landings of their immediate environments. We conclude that the calibrated model is suitable for short and longer term scenario generation, with external (using low (RCP2.6) and high (RCP8.5) greenhouse gas emissions trajectories) and internal (Altantic Multidecadal Variability) modes of climate variability as examples.