Marine ecosystem models have grown increasingly complex over the past decade in response to the growing appreciation of the complexity of marine ecosystems, and also to satisfy the need for flexible models that can be used for global biogeochemical simulations and for coastal and estuarine management applications. This trend has been facilitated by rapid increases in computational power. The question naturally arises as to whether or not there will be a point of diminishing returns in the development of these models, that is, will these increasingly complex models give rise to increased predictive skill or will increased variability and indeterminancy be the result? Are there inherent limitations in present ecological modeling approaches that will ultimately limit the level of model complexity that can be usefully employed? The desire to address the former question has given rise to quantitative model intercomparison studies which aim to quantify the impact of increasing model complexity on predictive skill and model portability, where the latter refers to the ability of a model to 'adapt' and function properly under different environmental and forcing conditions. Results from these comparisons, which have focused on open ocean systems so far, indicate that more complex model formulations can give rise to increased predictive skill and greater portability as long as the models are properly constrained/tuned with the available data. In particular, it is important that complex model formulations should not be given too many degrees of freedom that allow the model to fit noise in the data, which decreases predictive skill and portability.
Whether or not there are inherent limitations in present ecological modeling approaches which will ultimately limit the level of complexity that can be usefully employed in marine ecosystem models falls into the realm of theoretical ecology. To our knowledge, this is still an open question.
Was this article helpful?