Towards end-to-end (E2E) modelling in a consistent NPZD-F modelling framework (ECOSMO E2E_v1.0): application to the North Sea and Baltic Sea

Abstract

Coupled physical-biological models usually resolve only parts of the trophic food chain and hence, run the risk of neglecting relevant ecosystem processes. Additionally, this imposes a closure term problem at the respective “ends” of the considered trophic levels. Here we propose a consistent NPZD-Fish modelling approach (ECOSMO E2E) to address the above-mentioned problem in lower trophic ecosystem modelling, and to understand how the implementation of higher trophic levels in a NPZD model affects the simulated response of the combined North Sea and Baltic Sea ecosystem. On the basis of the coupled ecosystem model ECOSMO II we implemented one functional group that represents fish and one group representing macrobenthos in the 3d model formulation. Both groups are linked to the lower trophic levels and to each other via predator-prey relationships. The model allows investigating bottom-up impacts on primary and secondary production and cumulative fish biomass dynamics, but also top-down mechanisms on the lower trophic level production. Model results for a ten-year long simulation period (1980–1989) were analysed and discussed with respect to the observed pattern. To address the relevance of the newly implemented trophic levels for the simulated model response, we compare the performance of the ECOSMO E2E to a respective truncated NPZD model (ECOSMO II), which simulated the same time period. Additionally, we performed scenario tests to analyse the new role of the zooplankton mortality closure term in the truncated NPZD and the fish mortality term in the end-to-end model, which summarizes pressure imposed on the system by fisheries and mortality imposed by apex predators. We found that the model-simulated macrobenthos and fish spatial and seasonal pattern agree well with current system understanding. Considering a dynamic fish component in the ecosystem model resulted in slightly improved model performance with respect to representation of spatial and temporal variations in nutrients, changes in modelled plankton seasonality and nutrient profiles. Model sensitivity scenarios showed that changes in the zooplankton mortality parameter are transferred up and down the trophic chain with little attenuation of the signal, while major changes in fish mortality and in fish biomass cascade down the food chain.