The relationship between environmental variables and the occurrence of potential hollows and hollow-bearing trees in three dry forest types (dry Eucalyptus delegatensis forest, E. pulchella -E. globulus -E. viminalis grassy/shrubby forest and dry E. obliqua forest) in south-eastern Tasmania, was examined using generalised linear modelling. The number of trees with potential hollows and the number of potential hollows was significantly higher in dry E. obliqua forest, compared with the other two

more »

... forest types. The number of potential hollows per tree was best explained by tree species, tree form, degree of burn damage and the interaction between burn damage and tree species. There was no significant relationship between the number of trees with potential hollows per site and the environmental variables measured. However, the number of potential hollows per site was best explained by several environmental variables: vegetation type (highest in dry E. obliqua forest); topographic position; amount of dead trees on the ground; the age of the stand; the average total basal area of all trees; the height of the overstorey vegetation and various interactions between these variables and other variables, such as understorey cover. A model developed using a subset of the environmental variables measured was coupled with GIS data to develop a map of the predicted occurrence of trees with potential hollows throughout the study area. The use of such a predictive map for landscape level planning, in particular to assess the implications of land use scenarios on the hollow resource, is illustrated.