Detecting regime shifts in the ocean: data considerations

We review observational data sets that have been used to detect regime shifts in the ocean. Through exploration of data time series we develop a definition of a regime shift from a pragmatic perspective, in which a shift is considered as an abrupt change from a quantifiable ecosystem state. We conclude that such changes represent a restructuring of the ecosystem state in some substantial sense that persists for long enough that a new quasi-equilibrium state can be observed. The abruptness of the shift is relative to the life-scale or the reproductive time-scale of the higher predators that are influenced by the shift. In general, the event-forcing is external to the biological ecosystem, usually the physical climate system, but we also identify shifts that can be ascribed to anthropogenic forcing, in our examples fishing. This pragmatic definition allows for several different types of regime shift ranging from simple biogeographic shifts to nonlinear state changes. In practice it is quite difficult to determine whether observed changes in an oceanic ecosystem are primarily spatial or temporally regulated. The determination of ecosystem state remains an unresolved, and imprecise, oceanographic problem. We review observations and interpretation from several different oceanic regions as examples to illustrate this pragmatic definition of a regime shift: the Northeast Pacific, the Northwest and Northeast Atlantic, and Eastern Boundary Currents. For each region, different types of data (biological and physical) are available for differing periods of time, and we conclude, with varying degrees of certainty, whether a regime shift is in fact detectable in the data. (C) 2004 Elsevier Ltd. All rights reserved.