Sensitivity of a pelagic ecosystem model to variations of process parameters within a realistic range

A one-dimensional physical-biological coupled model is constructed with reference to an oceanic area with a seasonal thermocline and to a pelagic trophic network dominated by a herbivorous pathway. The sensitivity of the model is then tested for the effect of parameter variations within the range actually encountered in the literature. In the reference simulation, the physical component of the model reproduces reasonably well the annual cycle of the hydrographic structure. The biological component simulates a cycle where the spring bloom is dependent on the critical depth criterion and yields, for the state variables and the fluxes, values which are consistent with available data. A comparison of the sensitivity tests within a realistic range (absolute sensitivity) is effected with the more commonly performed tests over a fixed proportion of variation in parameter values (relative sensitivity). The latter appear to essentially reflect the internal behaviour of the model and overestimate the sensitivity of the system to a number of parameters, especially those involved in complex relationships. The absolute sensitivity tests identify a restricted set of parameters with a major influence. Phytoplankton biomass and production, as well as the nitrate stock, are highly dependent on a small number of parameters, among which the most important are the coefficients of the photosynthetic curve and the sinking velocity of one class of particles. Zooplankton biomass and production, as well as the ammonium stock, show a more complex sensitivity to a larger number of parameters, among which the same as above, plus grazing parameters and excretion and remineralization rates. Overall, the fluxes within and from the system appear mostly dependent on the first step in the food web, and the latter, in turn, is mainly dependent on the physical forcing. (C) 1999 Elsevier Science B.V. All rights reserved.