The use of simple models in the design and calibration of a dynamic 2D model of a semi-enclosed Australian bay

Like many ecosystems that of Port Phillip Bay (Australia) shows spatial and temporal variability. As such, we have constructed a spatially structured dynamic model (the full model) to describe it. However, such complex models are very difficult to design and analyse. We have derived earlier a simple model of a semi-enclosed marine ecosystem, whose analytical solution has allowed the determination of the dependence of major variables and fluxes on model formulation and parameter values. Here we apply this simplified model to the analysis of responses to changes in nutrient loads and zooplankton mortality formulation (trophic closure). Output of the simple model compares well with output from the full model (averaged in space and time) at current and moderately elevated nutrient loads; but at highly elevated loads, there is a breakdown in the linkage between the predictions of the two models due to the role of spatial variation in moderating ecosystem responses. Observations obtained from Port Phillip Bay are then used to select the most realistic zooplankton mortality formulation of this bay. Observed chlorophyll concentration supports a quadratic over a linear zooplankton mortality model. Observation of phytoplankton size fractions indicate that in productive waters of the bay picophytoplankton are grazer controlled, while microphytoplankton escape grazer control; analytical solutions show how rather subtle differences in parameter values allow this to occur. This analysis tool allows the prediction of the effects of changes to model formulation, parameter values and external forcing on the full model. However, the simple model is not intended to be, or used as, a realistic model in itself; rather it is a design tool that has been used for the development of more detailed models.