Coral cover has been declining on the Great Barrier Reef (GBR), with reports as high as a ~50% reef-wide loss since long term monitoring began in the 1980’s. What we do not know however, is what the future will look like for corals and the communities they support, and whether there are bright spots of coral resilience (i.e. where decline is lesser than expected based on disturbance severity). Here we present a temporally (1986-2015) and spatially (1km x 1km) explicit dynamic coral cover model for the GBR. Gompertz-based growth curves applied to % coral cover were estimated in a hierarchical Bayesian framework for 46 reefs from the AIMS Long Term Monitoring Program (LTMP). These growth-curves were then interpolated over the entire GBR (16035 1km x 1km grid cells) using Multivariate Regression Trees trained on 34 topographic, environmental and spatial variables to determine the level of affinity between the unmonitored and the 46 monitored reefs. Stochasticity was built into the model using estimated frequencies and empirical data for four major disturbances on the GBR: 1) Crown-of-thorn Starfish (CoTS); 2) tropical cyclones; 3) coral bleaching; and 4) coral disease, providing the most comprehensive model of coral growth, disturbance and recovery on the GBR to date.
Results/Conclusions
Our model was able to reconstruct trajectories of coral cover for both our calibration and validation datasets (6.7% mean prediction error), and most importantly provides the first credible information for the unmonitored portions of the GBR (>99%). Our results show that whilst there has been a significant decline in reef-wide coral cover since the 1990’s (-0.56% yr-1), coral cover has increased on 10% of reefs, with 5% of reefs identified as bright spots of coral resilience. Furthermore our results indicate that despite the cumulative impacts of continued bleaching, cyclone activity and coral disease, coral cover in the central and northern regions of the GBR would increase if CoTS outbreaks were reduced by 10-20%. This project provides unprecedented insight into the history and future of the GBR. Being in principle applicable to any reef ecosystem worldwide, our model provides the most advanced platform developed to date for predicting coral cover.