1. Scripps Institution of Oceanography, University of California, San Diego, La Jolla, California 92093-0202, USA

Correspondence to: George Sugihara¹ Correspondence and requests for materials should be addressed to G.S. (Email: gsugihara@ucsd.edu).

Abstract

The prospect of rapid dynamic changes in the environment is a pressing concern that has profound management and public policy implications^{1, 2}. Worries over sudden climate change and irreversible changes in ecosystems are rooted in the potential that nonlinear systems have for complex and 'pathological' behaviours^{1, 2}. Nonlinear behaviours have been shown in model systems³ and in some natural systems^{1, 4, 5, 6, 7, 8}, but their occurrence in large-scale marine environments remains controversial^{9, 10}. Here we show that time series observations of key physical variables^{11, 12, 13, 14} for the North Pacific Ocean that seem to show these behaviours are not deterministically nonlinear, and are best described as linear stochastic. In contrast, we find that time series for biological variables^{5, 15, 16, 17} having similar properties exhibit a low-dimensional nonlinear signature. To our knowledge, this is the first direct test for nonlinearity in large-scale physical and biological data for the marine environment. These results address a continuing debate over the origin of rapid shifts in certain key marine observations as coming from essentially stochastic processes or from dominant nonlinear mechanisms^{1, 9, 10, 18, 19, 20}. Our measurements suggest that large-scale marine ecosystems are dynamically nonlinear, and as such have the capacity for dramatic change in response to stochastic fluctuations in basin-scale physical states.

MORE ARTICLES LIKE THIS

These links to content published by NPG are automatically generated.