This project will develop and assess formal statistical methods for the retrospective identification of a regime shift in time series data of the abundances of interacting populations in an ecosystem. It will further develop and assess formal statistical methods for determining whether such an ecosystem is reactive (i.e., whether the dynamics of the system would amplify a perturbation). Well-established principles of statistics will be applied and based on a general model of the dynamics of a system of interacting populations. The performance of these methods will be assessed through a series of simulation experiments and with their application to previously published data sets. The possibility that a perturbation can cause a natural ecosystem to undergo a regime shift (i.e., an abrupt shift from one locally stable steady state to another) has important implications for management. To develop an understanding of this phenomenon in real systems, it is necessary to be able to distinguish between regime shifts and other types of variability. The first objective of this project is to provide such methods, without which scientific progress in support of management in this area is difficult. The second objective of the project is to develop methods for determining whether a system is reactive (i.e., capable of amplifying a perturbation). This is important because reactivity represents one possible "route to regime shift." Knowing whether a system is reactive is therefore important to managers seeking to avoid a regime shift. This work will help inform decision-making in fisheries management and conservation. The results of this project will be published in the peer-reviewed scientific literature and MATLAB routines for applying the methods will be made freely available.
