Influence of grazing formulations on the emergent properties of a complex ecosystem model in a global ocean general circulation model

Sensitivity to nonlinear equations may be a characteristic feature of biological models, particularly those that are complex. A complex marine ecosystem model (PlankTOM5.2) that incorporates multiple plankton functional types (PFTs) was embedded in a global ocean general circulation model (OGCM) and its performance assessed for four different formulations of multiple-prey zooplankton functional response: Michaelis-Menten (MM: Holling Type II), Sigmoidal (S: Holling Type III), Blackman (B) and Ivlev (Iv). Predictions of the four simulations were compared for the North Atlantic and North Pacific oceans. Remarkable differences were seen in both spatial extent and magnitude of predicted distributions of PFTs, as well as bulk properties, highlighting how the choice of functional response has a major impact on the resulting ecosystem structure. The range of average concentration of diatoms in surface waters was particularly marked, varying between 0.04 mg m(-3) (B and MM) and 0.13 mg m(-3) (S) in spring and between 0.01 mg m(-3) (B) and 0.07 mg M-3 (S) in autumn. Differences in ecosystem structure affected predicted export flux, which varied by more than 25% among the simulations. Overall, our work highlights that accuracy is required in ecosystem formulation if reliable predictions are to be made when using complex marine ecosystem models embedded in OGCMs and therefore the need for further studies, with appropriate validation, that address structural sensitivity. (C) 2010 Elsevier Ltd. All rights reserved.