Modelling cyanobacteria in shallow coastal seas

There is an increasing need to describe cyanobacteria bloom dynamics using ecosystem models. We consider two fundamentally different ways how cyanobacteria are currently implemented: a simple approach without explicit consideration of the life cycle which assumes that cyanobacteria grow due to nitrogen fixation alone and an advanced approach that computes the succession of four different stages of the cyanobacteria life cycle based on internal quotas of energy and nitrogen. To qualitatively and quantitatively intercompare these different approaches and with observations, we use the Baltic Sea ecosystem model ERGOM coupled to the one-dimensional water column model GOTM. Four experiments are carried out: three, using the simple approach with either (a) a prescribed constant minimum production, (b) no minimum value or (c) a prescribed constant minimum concentration, and one with (d) the full predictive life cycle. The model data of 35 years (1970-2005) are analyzed for the timing of the bloom, the interannual variability, the annual mean nitrogen fixation rates and the effect of cyanobacteria on eukaryotic phytoplankton. The results show significant differences. In the climatological seasonal mean, only the advanced approach which resolves the life cycle produces a realistic bloom onset and duration. The interannual variability of blooms is unrealistically small in the experiments with a prescribed minimum value. Annual mean nitrogen fixation rates diverge by up to 30% between the four model solutions. Finally, the representation of the cyanobacteria also influences the seasonal cycle of eukaryotic phytoplankton, i.e., flagellates. This study demonstrates that the way how cyanobacteria are implemented in coupled biological-physical models strongly determines the fluxes into the system and between the individual compartments.