Pseudomys novaehollandiae is 'Endangered' in Victoria, where it is presently considered to be extant at only three localities Loch Sport, Providence Ponds, and Wilsons Promontory. This study aimed to determine indicators of suitable habitat for the species that could assist in identifying potential habitat and sites for planned re-introductions as part of a recovery program. Vegetation and site data (soils, topography, rainfall, fire age-time since fire) were assessed at localities where P. novaehollandiae was recorded. The species occurred in five structural vegetation groups -open-forest, woodland, heathland, shrubland, grassland, with the most common being open-forest and woodland. Grassland and shrubland were restricted to coastal sand-dunes in south Gippsland. Understorey vegetation at most sites was dominated by sclerophyllous shrubs ranging in cover from 10 -70%. Classification of quadrats produced eight floristic groups in which the trend was for quadrats to cluster according to geographical location. Ordination confirmed the classification pattern and vector-fitting produced significant correlations between vector points and five variables: species richness, latitude, longitude, fire age and annual rainfall. The study identified a range of vegetation communities where P. novaehollandiae occurs and provided evidence that the species is not restricted to floristically rich and diverse heathlands. The findings can be used to determine further localities with suitable habitat. However, factors other than vegetation are also likely to be important in predicting suitable habitat.Key words: Pseudomys novaehollandiae, New Holland mouse, habitat, Victoria, vegetation structure. (Keith and Calaby 1968; Mahoney and Marlow 1968;Seebeck and Beste 1970;Posamentier and Recher 1974;Hocking 1980). Recently, however, it has also been found at sites up to 100 km inland at altitudes of 550 -900 m in Queensland and NSW (Read 1988; Townley 1993;Van Dyck and Lawrie 1997). P. novaehollandiae occurs in a variety of habitats including heathland and woodland (Posamentier and Recher 1974;Kemper 1977;Braithwaite and Gullan 1978;Hocking 1980;Pye 1991), dry sclerophyll forest with a dense shrub layer (Keith and Calaby 1968;Seebeck and Beste 1970;Fox and McKay 1981) and on vegetated sand dunes (Keith and Calaby 1968;Fox and Fox 1978;Cockburn 1980;Pye 1991).
B.A. Wilson, School of Ecology and EnvironmentalThe microhabitat preferences of P. novaehollandiae have been found to be influenced by vegetation floristics and structure, and soil characteristics. Studies undertaken in NSW in coastal heath and open-forest regenerating after sand-mining found that the biomass of the species was highly correlated with the proportion of heath species, vegetation cover below 50 cm and softer substrates Fox 1978, 1984). At Cranbourne in Victoria, Braithwaite and Gullan (1978) found that floristic variables explained more of the habitat preferences of P. novaehollandiae than structural variables. In the eastern Otways, Victoria, the species preferred...