Warning signals of regime shifts as intrinsic properties of endogenous dynamics

Ecosystem dynamics can exhibit large, nonlinear changes after small changes in an environmental parameter that passes a critical threshold. These regime shifts are often associated with loss of biodiversity and ecosystem services. Because critical thresholds for regime shifts are hard to determine with precision, some recent studies have focused on deriving signals from dynamics leading up to the thresholds. Models in these studies depend on using noise terms independent of system parameters and variables to add stochasticity. However, demographic stochasticity, an important source of random variability, arises directly from system dynamics. In this study, a framework is developed for modeling demographic stochasticity in a mechanistic way, incorporating system variables and parameters. This framework is applied to a deterministic, dynamic model of a coral reef benthos. The resulting stochastic model indicates that increasing variance-but not skewness-is consistently found in system dynamics approaching a critical threshold of grazing pressure. Even if the threshold is breached, attraction of transient dynamics by a saddle point provides an opportunity for regime shift reversal by management intervention. These results suggest that early warning signals of regime shifts can arise intrinsically in endogenous dynamics and can be detected without reliance on random environmental forcings.