Ecosystem Model in Data-Poor Situations

Ecosystem assessment is one of the most interesting topics in ecology and fisheries science; modeling is an essential and indispensable part of ecosystem assessment. We briefly review existing ecosystem models that are employed around the world and present a new ecosystem model that can be applied in data-poor situations, e.g., when diet-composition data are unavailable. The new model is based on a multivariate state-space model with an allometric relationship between the biological parameters and body mass. The model generally does not require unrealistic assumptions, such as equilibrium prior to fishing and mass balance during a certain period. The simulation study demonstrated that the model outperformed a single-species assessment in terms of the inference of biological reference points. As an illustration, we applied the model to environmental index data and three species in the western North Pacific, which are known to show conspicuous species replacement (chub mackerel, sardine, and anchovy). The model can be extensively applied to various multispecies data in data-poor situations.