Outbreaks of Crown-of-Thorns Starfish (CoTS), Acanthaster planci, are a major cause of coral decline on the Great Barrier Reef, second only to cyclones. Although various models have been developed in the past to assist management decision making, most of these models were cohort-based deterministic descriptions with little inclusion of parameter and individual uncertainties, or they were structured around a generic ecological modelling framework, or they were not calibrated with observational data. A major challenge of statistical modelling occurs when estimation of the likelihood is computationally expensive or even intractable. However, approximate Bayesian computation (ABC) can be a convenient solution to this problem. In this study, we developed a semi-individual agent-based model for CoTS. Unlike previous models, this model amalgamated a species-specific individual-based process model with statistical methods of parameter estimation, namely replenishment ABC (RABC) and ABC-rejection. This Bayesian modelling framework facilitates quantification of the uncertainty in parameter estimation while the individual-based aspect of the model enables it to take into account individual variation in life histories. In this relatively complex setting and in the absence of an informative prior, ABC-rejection was unable to efficiently estimate the posterior distribution of model parameters within a reasonable time frame. In contrast, RABC demonstrated promising results in estimating model parameters. The results of posterior predictive checks showed that most observations were within the 95% predicted intervals. Therefore, the semi-individual agent-based model developed in this study showed promising ability to accurately predict the general population trends at the three locations studied. The results of modelling identified a strong link between observed sharp declines in CoTS populations and depletion of their food (i.e. coral cover). As outbreaks of CoTS have caused substantial decline in the coral cover of the Great Barrier Reef, this model should eventually be combined with operation models to predict the outcome of different interventions, and the design of optimal control strategies.
- Approximate Bayesian computation
- Crown-of-Thorns Starfish
- Individual agent
- Individual uncertainty
- Model calibration
- Parameters uncertainty
- Replenishment sequential Monte-Carlo: ABC