Detecting regime shifts in marine systems with limited biological data: An example from southeast Australia

• We attempted to detect regime shifts in a data-poor region, southeast Australia.
• Linear trends, rather than regime shifts, dominated regional ecological change.
• A minimum of ∼30 biological time series may be needed to detect regime shifts.
• Available data may be insufficient to detect regime shifts in southeast Australia.
• More long-term observations are needed to understand regional ecological change.

The ability to detect ecological regime shifts in a data-limited setting was investigated, using southeast Australian ecosystems as a model. Community variability was summarized for 1968–2008 with the first two principal components (PCs) of recruitment estimates for six fish stocks and reproductive parameters for four seabird species; regional climate was summarized for 1953–2008 with the first two PCs for three parameters (sea surface temperature [SST], sea surface salinity, surface nitrate) measured at two stations; and basin-scale climate variability was summarized for 1950–2012 with mean South Pacific SST and the first two PCs of detrended South Pacific SST. The first two biology PCs explained 45% of total community variability. The first two PCs of basin-scale SST showed abrupt shifts similar to “regime” behavior observed in other ocean basins, and the first PC of basin-scale SST showed significant covariation with the first PC of regional climate. Together, these results are consistent with the strong community variability and decadal-scale red noise climatic variability associated with Northern Hemisphere regime shifts. However, statistical model selection showed that the first two PCs of regional climate and the first PC of biology time series all exhibited linear change, rather than abrupt shifts. This result is consistent with previous studies documenting rapid linear change in the climate and biology of southeast Australian shelf ecosystems, and we conclude that there is no evidence for regime shift behavior in the region’s ecology. However, analysis of a large set of previously-published biological time series from the North Pacific (n = 64) suggests that studies using fewer than ∼30 biological time series, such as this one, may be unable to detect regime shifts. Thus we conclude that the nature of ecological variability in the region cannot be determined with available data. The development of additional long-term biological observations is needed for understanding change in southeast Australia and in many other marine ecosystems globally.