Fire patterning in savanna landscapes and implications for wildlife management

Abstract

Fire has long been recognised as a major factor moulding vegetation patterning across Australia. Large proportions of northern Australia's tropical savannas are burned, resulting in declines of wildlife, including granivorous birds. These declines have been attributed to the combined impacts of pastoral use and alteration of fire regimes.

In the Fergusson River Catchment, an area with a mix of land uses. remote sensing imagery data accurately interpreted through an operational and robust method show that (i) more than 31.9% of the landscape burns annually (ii) contrary to prevailing interpretation and stated management goals, most fire occurs in the mid-dry season (June to August) (iii) all vegetation types, including riparian are subject to frequent fires, although the proportion burned varies significantly among vegetation types (iv) large areas are also burned late in the dry season when widespread fires are more likely.

Interpreting same burned pixel in different raster-based fire mapping layers in a temporal series of layers with three maps each year in Early, Middle and Late dry season provides a method to extract pixel-based fire interval, which demonstrates differences in different vegetation. In connection with vegetation mapping results, the understanding of fire return times in different vegetation helps for wildlife management.

In the study area, early dry season burning does not appear to influence the number, size or total extent of fires in the late season. This has important implications for management of fire in the region, because it illustrates the ineffectiveness of present application of this fire management technique.

Statistical modelling showed that all landscape and anthropogenic variables derivable from GIS coverages were associated with variation in fire frequency patterns. Vegetation, elevation and land use type were predominant in fire frequency models in all of the three different seasons examined. However, models explained a relatively small proportion of variance in fire frequency, suggesting that much remains to be learned about the determinants of fire patterns of this landscape. This has important implications for the management of a region supporting populations of rare and threatened fauna and adjoining or including important national parks.

Date of Award	2004
Original language	English
Supervisor	Peter Whitehead (Supervisor) & Diane Pearson (Supervisor)