Arboreal marsupial incidence in eucalypt patches in south-eastern Australia: A test of Hanski's incidence function metapopulation model for patch occupancy

The results of field tests are described of a simple model for the occupancy of habitat patches developed by Hanski. Patch occupancy was predicted for several species of arboreal marsupials in a system of fragmented patches of Eucalyptus forest in the Tumut region of southern New South Wales (NSW) in south-eastern Australia. The species of arboreal marsupials for which the model was tested were: common brushtail possum (Trichosurus vulpecula), mountain brushtail possum (Trichosurus caninus), common ringtail possum (Pseudocheirus peregrinus), and greater glider (Petauroides volans). Predictions of the model were compared against actual data on patch occupancy gathered from extensive field surveys. Four sub-models were tested for each of the four species of arboreal marsupials: With and without rescue effects and with and without mainland effects (i.e. dispersal from large population source areas). These sub-models were tested by deriving model parameters from a subset of 40 patches and then applying the model to make predictions of patch occupancy in an array of 86 other patches of remnant Eucalyptus forest habitat. Comparisons were then made between the actual values for patch occupancy found during field surveys and the probabilities of patch occupancy predicted by the model. Our analyses revealed large differences in the results for different species. In the case of P. peregrinus, there were several models for which it was not possible to derive the parameter estimates required by the model (i.e. relationships for extinction and colonisation as a function of patch size and patch isolation). The ability of the species to persist at low density in the surrounding landscape matrix may have precluded the calculation of these estimates. There was reasonable congruence between predictions and actual field data for patch occupancy for P. volans and also for most sub-models for T. vulpecula. Conversely, none of the sub-models for T. caninus produced results that were consistent with the observed data on patch occupancy for this species. The reasons for the marked differences in results for T. caninus and T. vulpecula are not clear, but they indicate that knowledge of the response of one species to disturbance (e.g. habitat fragmentation) may not necessarily provide a useful guide to the possible response of other taxa, including those that are very closely related.