Background/Question/Methods: Ecosystem thresholds are often identified by observing or inducing slow changes in different driver variables and investigating changes in the asymptotic state of the system, such as the response of lakes to nutrient loading or biome responses to climate change. Yet many anthropogenic pressures on ecosystems change rapidly or in discrete steps. Moreover, once an ecosystem crosses a bi-furcation point in the equilibrium states of the system, the transition is often assumed to be irreversible, but such irreversibility is rarely assessed. We tested these assumptions, investigating the temporal dimension of critical thresholds and the reversibility of slowly evolving dynamical systems responding to abrupt changes in external drivers. This study asks if and how long ecosystems can operate beyond a threshold –a bi-furcation point—before they they become trapped in an alternative state. We use two approaches: 1) simulation models; and 2) long-term experiments that pushed semi-arid grasslands and mesic grasslands past critical thresholds and then relaxed these pressures in some plots to test for reversibility.

Results/Conclusions: The simulated ecosystems, ecosystems or patches ecosystems remain trapped a new regime when the duration of forcing exceeds a threshold. However, some patches can operate beyond a bi-furcation point for years to decades before relaxing global change pressures fails to restore the original ecosystem state. Weaker exogenous forcing and lower endogenous resilience led to longer transient dynamics. Long-term experiments were consistent with this framework. After experimental cessation of N-addition, grazing, or fire suppression, many grasslands patches returned to their initial state, but the most altered patches of have yet to recover 5 to 20 years later after relaxing exogenous forcing. These responses support the argument that some real-world ecosystems can be restored even if they were pushed beyond thresholds that would eventually lead to irreversible shifts. We propose that ecosystem thresholds in slower systems have two dimensions: regime shifts occur when global change crosses a threshold defined by both magnitude and duration of exogenous forcing. An important implication of these results is that ecosystems that have yet to cross temporal thresholds may still be restored by appropriate and targeted management actions. Leveraging this capacity requires new tools to indicate when ecosystems are in an unstable state following press perturbations.  Abstract # COS 75-8.