Using expert elicitation to identify effective combinations of management actions for koala conservation in different regional landscapes

Context: Choosing the most effective combinations of actions to manage threatened species is difficult. Aims: This study aimed to identify the most effective combinations of six management actions for reversing population declines of koala populations in three regional landscapes (Coastal, Riverine and Hinterland) in north-east New South Wales, Australia, based on expert-elicited knowledge. Methods: A Bayesian network was used to combine continuous probability distributions representing the impact of specific management actions and combinations of actions. Data were derived from expert elicitation. Variation within and among expert judgements was captured by incorporating submodels of individual responses. We evaluated alternative management solutions to address four mortality factors (disease and injury, vehicle strikes, domestic dogs and wild dogs/dingoes (Canis familiaris), and two habitat-related management actions (restoration and protection). Key results: We show that there are marked differences in the expected response of the koala populations to the various management options in the three regional landscapes over a 20-year period (2019-2039), and that multiple management actions are required to arrest and reverse the decline in koala populations of north-east NSW. Conclusions: Management actions for koala conservation should be based on regional context. Our model, in conjunction with robust expert-elicitation procedures, allows decision makers to distinguish effective from ineffective combinations of management actions for threatened species management, and has been structured so that new data can be incorporated into the model. Implications: Model design could be easily adapted to different species or conservation contexts, and updated as new evidence becomes available, making it valuable in adaptive management for local to regional-scale conservation problems.