Progress in being able to predict changes in ecosystems due to environmental stressors has been limited by our poor understanding of the mechanisms and interactions between physical, chemical and biological factors in dynamic ecosystems. Currently there are few predictive tools that are able to model the impacts of contaminants in complex environments, while explicitly acknowledging the associated uncertainties resulting from the inherent variability and lack of knowledge that characterises ecological systems. Bayesian network (BN) models have recently been recognised as a modelling tool that can: represent complex and poorly known cause-effect relationships; characterise the impacts of changes in ecological systems; and investigate the effectiveness of various management options.
The focus of the presentation will be on the recent progress made in developing BN models, within the context of an ecological risk assessment. Two case studies will be presented (a) the development and application of a BDN for the management of endangered species (Yarra Catchment, Victoria), and (b) the development of a BDN to aid management of land-based activities for the protection of a coastal habitat that supports threatened species (Great Barrier Reef, Queensland).