Using State-and-Transition Simulation Models (STSMs) to Explore Dynamic Population Responses to Drought Cycles in Freshwater Ecosystems

Abstract

Climate variability and change pose significant threats to aquatic biodiversity, particularly in areas with low and variable streamflow. Quantifying the magnitude of risk from these threats is made more difficult by the variable responses of individual species to hydrologic stress. Patterns of population decline and recovery in response to drought cycles will depend on both the resistance traits (e.g., tolerance to harsh environmental conditions) and resilience traits (e.g., fecundity, age at maturity), both of which vary considerably among species. Collectively these traits can give rise to varied, and lagged patterns of decline and recovery in response to hydrologic variability, which ultimately can affect population viability in drought prone environments and in response to a changing climate. Such population cycles are typically modelled based on demographic rates (mortality and recruitment) under different climate conditions. However, such models are relatively data intensive, limiting their widespread development. A less precise but more tractable approach is to adopt state-and-transition approaches based on semi-quantitative population states (or population size estimates), and modelled transitions between states under different hydrologic conditions. Here we demonstrate the application of such models to a suite of diverse taxa, based on an expert elicitation of expected state-changes across those different taxa under a range of different flow conditions. The model results broadly conform with population changes observed in response to a major drought in the case-study system, mimicking the observed lags in recovery of species with different life-histories. Stochastic simulations of population cycles under scenarios of more protracted drought provide a semi-quantitative measure of the potential risk to different species under each scenario, as well as highlighting the large uncertainties that can arise when taking into account stochastic (rather than deterministic) state-transitions.