Prediction, characterisation, and detection of regime shifts in ecological systems

Regime shifts are persistent changes in ecosystem structure and function that can result in loss of ecosystem services upon which people depend. Thus, there is a clear need for methods capable of assessing ecosystems for vulnerability to, and occurrence of, regime shifts. Methods available for predicting and detecting ecological regime shifts generally require extensive data and understanding of ecosystem processes. However, for many ecosystems data availability is patchy, limiting the application of available methods to assess for regime shifts. This thesis uses qualitative and quantitative methods to improve capability for assessing ecosystems for vulnerability to regime shifts, predicting the nature of future regime shifts, and detecting regime shifts after they have occurred. I first undertake a review of the prediction, characterisation, and detection of regime shifts and in doing so, develop a framework for assessing ecosystems for regime shifts. This framework outlines ecosystem attributes that are linked to stability (and loss of stability), and describes methods that ecosystem practitioners can use to assess and incorporate understanding of these attributes for improved ecosystem management. The framework also explores a broader set of questions that ecosystem practitioners may want to answer about the nature and timing of regime shifts. A key question in this context relates to the consequences of not acting i.e. the likely future state of a given system and how urgently action must be taken i.e. how imminent a regime shift is, or when it is likely to occur. Suggested approaches for answering each question are provided by the framework. Finally, in this chapter I provide an overview of methods available for ascertaining whether a regime shift has occurred. Southern Ocean ecosystems are used as a case study to illustrate how this framework can help identify current capacity to assess for regime shifts, and to guide future targeted data collection where gaps in capacity exist. A key finding from the review is that ecosystems prone to regime shifts are often characterised by the presence of destabilising positive feedbacks. However, identifying positive feedbacks is particularly challenging for pelagic or remote ecosystems, because such feedbacks often involve non-trophic interactions which are difficult to observe (and quantify). An interesting case study is the presence of potentially important positive feedbacks in Southern Ocean ecosystems, which involve the release of a chemical cue (dimethyl sulfide) by phytoplankton that attracts the predators of the phytoplankton-grazers. A qualitative modelling approach is applied to explore the potential importance of this feedback in driving regime shifts in the Southern Ocean, and to identify key interactions determining the stability of the community. Thus, in this chapter I provide the first demonstration of the use of qualitative models with simulation to assess the importance of positive feedbacks for ecosystem stability, an approach that could usefully be applied to other systems. This thesis also improves capability for detecting regime shifts once they have occurred, and to distinguish regime shifts from an ecological fluctuation without a regime shift. In some ecosystems it is not always evident that a regime shift has occurred, and data constraints prevent the application of existing methods for detecting regime shifts. This thesis develops a new approach for detecting regime shifts from transect (1-dimensional spatial) data. This method (1D Characteristic Length Scale estimation) involves reconstruction of the system attractor and nearest neighbour prediction to identify the emergent spatial scale of the system, with a shift in this scale indicating regime shift. The method is tested successfully on model systems and then applied to a coral reef data set, which shows evidence of past regime shifts. This methodological advance broadens the range of ecosystems which can be tested for evidence of regime shifts, and removes some ambiguity in characterising the nature of community change. The final element of the thesis is a discussion of the management of ecological regime shifts in the context of global tipping points in climate and human systems, and some thoughts on how this thesis might help progress the study and management of regime shifts. Together the tools provided in this thesis help deepen our understanding of regime shifts, direct future research, and assist ecosystem managers adapt to a future in which these complex changes occur more frequently.