It is widely accepted that global warming will lead to the destruction of habitats and ecosystems. For example, global warming is predicted to increase both the frequency and severity of cyclones, which damage coral reefs. Predicting the likely future impacts of such events on communities requires realistic stochastic models of community dynamics. Here we describe a Bayesian model of cyclone effects at a single, well-studied location on the Great Barrier Reef. A 30-year dataset of permanent quadrats on Heron Island was used to develop a model of the effect of cyclones on community dynamics. Our aim was to produce a relatively simple, testable model, almost all of whose parameters can be estimated from field data. Our model is intermediate in complexity between simple linear Markov models and complex simulation models. An existing continuous time Lotka-Volterra competition model was used as the basis, to which we added a cyclone effect parameter and the dates of known cyclones. The duration of detrimental effects of cyclones was estimated from the literature. Bayesian estimates of the intensity of cyclone effects on coral cover and the interactions between states were obtained by Markov Chain Monte Carlo.
Results/Conclusions
The estimated rate of coral mortality was four orders of magnitude higher following a cyclone (95% credible interval for the cyclone effect: 2.48 x 104 – 2.90 x 104). However estimated coral mortality in the absence of cyclones was generally low (95% credible interval: 0.0406 - 0.1084 y-1). Thus, the overall effect of cyclones on these quadrats was modest, perhaps because of their relatively sheltered locations. How the model can be used to make predictions about future coral reef dynamics for given cyclone scenarios is discussed.
