Modeling Species’ Distributions to Improve Conservation in Semiurban Landscapes: Koala Case Study
JONATHAN R. RHODES,∗†‡†† THORSTEN WIEGAND,‡ CLIVE A. MCALPINE,∗† JOHN CALLAGHAN,§DANIEL LUNNEY,∗∗ MICHIALA BOWEN,∗ AND HUGH P. POSSINGHAM†
∗School of Geography, Planning and Architecture, The University of Queensland, Brisbane, QLD 4072, Australia
†The Ecology Centre, The University of Queensland, Brisbane, QLD 4072, Australia
‡Department of Ecological Modelling, UFZ-Centre for Environmental Research, PF 500136, D-04301, Leipzig, Germany
§Australian Koala Foundation, GPO Box 2659, Brisbane, QLD 4001, Australia
∗∗New South Wales Department of Environment and Conservation, P.O. Box 1967, Hurstville, NSW 2220, Australia
Models of species’ distributions are commonly used to in form landscape and conservation planning. In urban and semiurban landscapes, the distributions of species are determined by a combination of natural habitat and anthropogenic impacts. Understanding the spatial influence of these two processes is crucial for making spatially explicit decisions about conservation actions. We present a logistic regression model for the distribution of koalas (Phascolarctos cinereus) in a semiurban landscape in eastern Australia that explicitly separates the effect of natural habitat quality and anthropogenic impacts on koala distributions. We achieved this by comparing the predicted distributions from the model with what the predicted distributions would have been if anthropogenic variables were at their mean values. Similar approaches have relied on making predictions assuming anthropogenic variables are zero, which will be unreliable if the training data set does not include anthropogenic variables close to zero. Our approach is novel because it can be applied to landscapes where anthropogenic variables are never close to zero. Our model showed that, averaged across the study area, natural habitat was the main determinant of koala presence. At a local scale, however, anthropogenic impacts could be more important, with consequent implications for conservation planning. We demonstrated that this modeling approach, combined with the visual presentation of predictions as a map, provides important information for making decisions on how different conservation actions should be spatially allocated. This method is particularly useful for areas where wildlife and human populations exist in close proximity.