Combining abundance data collected in designed field surveys with biophysical data derived from geographic information systems is a powerful way to investigate predictors of spatial variation in the abundance of wildlife. We used such an approach to evaluate hypotheses about factors influencing the abundance of sambar deer (Cervus unicolour Kerr, 1792), a large non-native herbivore, in south-eastern Australia. We developed a spatial model for the abundance of sambar deer faecal pellets in a 3650-ha area in the Upper Yarra Ranges, Victoria. We counted the number of sambar deer faecal pellets along 100 randomly located transects and used a geographic information system to estimate biophysical variables around each transect. We formulated our hypotheses about how those variables might affect the abundance of sambar deer pellets into 22 candidate models and used the deviance information criterion to identify the ‘best’ model(s). Because five models had strong support we used model averaging to generate a predictive model. The three variables included in the predictive model were aspect (abundance of pellets declined with increasing ‘northerliness’ and increased with increasing ‘easterliness’), distance to water and elevation; the latter two variables were positively correlated and had a negative effect on the abundance of pellets. In contrast to previous models of sambar deer abundance in south-eastern Australia, our spatial predictions of the abundance of faecal pellets can be easily tested and updated. Our approach would be useful for modelling the abundances of other wildlife species at a range of spatial scales.
|Author||David M. Forsyth, Steve R. McLeod, Michael P. Scroggie and Matthew D. White|
|Secondary title||Wildlife Research|
|Institution||Arthur Rylah Institute for Environmental Research|
|Department||Department of Sustainability and Environment|