Comparative Dynamics of Ecosystem Components from the Northeast Atlantic Shelf and Pacific Coast: New Approaches to Forecasting and Understanding Variability and Structure in Marine Ecosystems

Principal Investigators: A. A. Rosenberg (UNH), George Sugihara (SIO/UCSD), M. J. Fogarty (NOAA), Les Kaufman (BU),Chi-Hao Hsieh (National Taiwan University)

Short Summary

Ecosystems can be investigated by reducing the dimensionality of their complex webs to a subset of interacting variables.  In this hypothetical example, three species (anchovy, squid, and copepods) appear to be uncorrelated in time even though they are functionally coupled.  However, by combining their time series (each as an axis to reconstruct an attractor) their dynamic functional relationship (attractor) is revealed. Such state space (attractor) reconstruction can be exploited for short-term predictability which, coupled with medium-term scenario exploration models, can feedback into forecasting the future states of marine ecosystems.

Play Audio Clip

Click on the link below to play the audio clip from the principal investigators

Project Summary

This pilot study will produce a general template for understanding the comparative dynamics of marine ecosystems. The main aims in developing the template are: 1) to develop new methods to provide ecology and fishery science with fundamental predictive understanding of ecosystem dynamics and, 2) to provide forecasting and structural tools (e.g., objective identification of dynamically coupled ecosystem units and their relationship to trophically defined subunits) for developing advice for policy makers. Two case studies are used to demonstrate proof of concept; the California Current large marine ecosystem and the Northeast Shelf large marine ecosystem with a focus on parallel ecosystem components. These two data rich, ecologically and economically important systems present different challenges to the methodology of developing predictive models because of differences in the data and understanding of the systems. On the other hand, they are representative of U.S. marine ecosystems and enable demonstration of the generality of the methods and tools. Using a combination of novel near-term time series prediction models (quarterly to 1 year ahead), with medium-term scenario exploration models (1-10 years ahead), the scientific groundwork will be laid for forecasting the future states of marine ecosystems and the interrelationships of ecosystem components in generating that future state. The methods employed in the study will enable objective identification of dynamically coupled ecosystem sub-units (hereafter called functional coupled units or FCUs) to reduce the dimensionality of the predictive models. Such FCUs may be statistically uncorrelated but dynamically dependent. FCUs will be compared with analysis of trophic groupings to assess their relative importance in determining dynamics. This predictive and fundamental organizational information combined with subsequent scenario exploration models will serve as a basis for scientific advice for policymakers. In particular, this study will create the basis for developing a comprehensive view of the scientific problem of nonequilibrium ecosystem changes on multiple time and space scales and present a completely new and rigorous predictive framework for ecosystem-based management.


Scheffer, M., Bascompte, J., Brock, W.A., Brovkin, V., Carpenter, S.R., Dakos, V., Held, H., van Ness, E.H., Rietkerk, M., and Sugihara, G. 2009. Early-warning signals for critical transitions. Nature 461(7620): 53-59.

For further information, contact or call 301-713-2239.