Habitat and spatial requirements of the eastern barred bandicoot (Perameles gunnii) at Hamilton, Victoria

Abstract

Perameles gunnii was greatly affected by the introduction of European agriculture to the volcanic plains in Victoria. At Hamilton, agricultural areas possessed little structural complexity and supported a fairly homogeneous composition of pasture species that were generally shorter than 100 mm because of heavy stock grazing. No P. gurznii were caught in agricultural areas at Hamilton. At the Hamilton Municipal Tip, most captures and nest sites occurred where food resources and structural complexity were greatest. Descriptions of 16 diurnal nest sites indicated that a range of natural and artificial materials was used for shelter, including fallen branches, sawn timber, cement culverts, galvanised iron, and scrap metal. Earthworms were most frequently observed in faecal and stomach material, while beetles and crickets were also common. Optimal habitat for P. gunnii was defined by high structural complexity and habitat heterogeneity. Older P. gunnii may usurp optimal habitat and force subordinate adult females into sub-optimal habitat. Associated with their occupation of optimal habitat, older P. gunnii may utilise smaller nocturnal foraging areas. Analysis of the movement of P. gunnii within the Hamilton Municipal Tip indicated that males occupied significantly larger nocturnal foraging areas than females. Mean female and male home ranges (defined by Minimum Convex Polygon analysis) were 1.6 ha (n = 13, range 0.02-5.9 ha) and 4 ha (n = 18, range 0.8-9.0 ha), respectively. Also, nocturnal foraging areas were analysed on the basis of pattern of use of an area, and these were referred to as utilisation distributions. The mean utilisation distributions (defined by Minimum Area v. Probability [0.95] analysis) for females and males were 0.64 ha (n = 13, range 0.01-4.7 ha) and 4.0 ha (n = 18, range 0.01-19.6 ha), respectively. Mean female and male observed range lengths were 173 m (tz = 26, range 0-364 m) and 249 m (n = 34, range 50430 m), respectively. Mean ( +/-s.e ) observed range length (214+/-20, n = 60, range 50-430 m) was less than half the grid width (500 m), which suggests that grid size did not negatively bias the calculated home-range sizes.