Oral presentation- Open Session 12th International Mammalogical Congress

On-site variables improve the accuracy of a GIS-based occupancy model for a Critically Endangered arboreal marsupial, Leadbeater’s possum. (#178)

Jenny Nelson 1 , Michael Scroggie 1 , Jemma Cripps 1 , Louise Durkin 1 , Lindy Lumsden 1
  1. Arthur Rylah Institute for Environmental Research, Department of Environment, Land, Water and Planning, Heidelberg, Victoria, Australia

Occupancy models are widely used to predict the occurrence of threatened species to inform conservation planning, but evaluating their accuracy is less common. We developed an occupancy model for the Critically Endangered Leadbeater’s possum (an arboreal marsupial endemic to south-eastern Australia), to help identify population strongholds after extensive bushfires caused significant declines in range and abundance. Timber harvesting also threatens this species, with approximately one third of its potential habitat available for harvesting. Our model was developed from surveys at 180 sites, together with mapped environmental variables. Additional site-based habitat variables of known ecological importance (hollow-bearing trees, Acacia and mid-storey density) were collected at each survey site. These variables could not be included in the model used for spatial prediction, as they are not mapped across the species’ range. The results from surveys at an additional 287 sites were used to evaluate the predictive accuracy of the GIS-based and on-site variable based models. The model based solely on mapped environmental data performed poorly, with Leadbeater’s possum often detected at sites where they were not predicted to occur. The model incorporating on-site habitat data had much greater predictive accuracy. Our results illustrate the importance of good spatial understanding of critical habitat features to accurately predict the distribution of threatened species. Remotely-sensed data such as LiDAR and infrared imagery may provide spatial representations of critical habitat features for Leadbeater’s possum allowing us to improve the predictive accuracy of our model, to better inform fire and timber harvesting management within the species’ range.