OOS 44-5 - Predicting ecological regime shifts - and post-transition dynamics - from time series

Many ecosystems can exist in alternate regimes, and tipping points occur when changes in an environmental driver cause an ecosystem to transition from one regime to another. Since alternate regimes can have unique economic and ecological consequences, there has been considerable interest in anticipating transitions, which has led to the development of early warning signals (EWS) to predict impending tipping points. While EWS are generic, they are limited in their ability to predict the qualitative nature of the system in the new regime, making management decisions challenging. Here we show that if we have knowledge about the driving parameter, Gaussian process - empirical dynamic modeling (GP-EDM) has the potential to predict tipping points and the dynamics in the new regime. As a proof of concept, we demonstrate GP-EDM's ability to estimate the qualitative dynamics of simulated and experimental data at unobserved values of the driving parameter. Thus demonstrating that this approach can help us anticipate an upcoming regime shift and make rational decisions about preventing or preparing for it.