Modelling landscape distributions of large forest owls as applied to managing forests in north-east Victoria, Australia

Abstract

A landscape approach was taken to modelling distributions of large forest owls and conserving habitat for them in 1.2 million ha of forest on the Great Dividing Range in north-east Victoria, south-eastern Australia. Owls were surveyed using call playback at 472 sites, selected by stratified random sampling from geographical information system (GIS) data. Habitat data were collected at each site and at four spatial scales from GIS data. Six owl species were recorded in or near the study region. Data on powerful owls ( Ninox strenua) and sooty owls ( Tyto tenebricosa) were modelled using logistic regression, and predicted probabilities of occurrence were mapped using GIS. Mapped variables explained more variation than habitat variables assessed at survey sites. Powerful owls were most likely to be observed at sites with mature dry forest, many live hollow-bearing trees, diverse habitats within 2 km, and not much pure regrowth within 5 km. Sooty owls were most likely to be observed at sites with wetter more senescent forest associated with tree-ferns ( Cyathea australis and Dicksonia antarctica), blanket-leaf ( Bedfordia arborescens) and silver wattles ( Acacia dealbata), diverse habitats within 500 m and much senescent forest within 5 km. The models were field-tested and found to discriminate well between high and low probability sites. Actual records and then models were used to help select 225 protected areas for large owls, each of approximately 500 ha.