The management of fire in savannas has been informed by a strong tradition of fire experiments, especially in Africa. This research tradition is much shorter in the 2 million square kilometres of tropical savannas in northern Australia, but has yielded several natural experiments, and three designed, manipulative, controlled field experiments (hereafter 'manipulative' experiments) of international significance (at Munmarlary, Kapalga and Kidman Springs in the Northern Territory). Here we assess the contributions of experiments, in particular the manipulative experiments, to ecological understanding and biodiversity management in Australia's savannas. Running from 1973 to 1996, the Munmarlary experiment comprised hectare-scale experimental plots with four replicated dry season fire treatments, and was designed to examine interactions between fire, landscape and biodiversity. The Kapalga experiment ran from 1989 to 1995, with a range of fire treatments broadly similar to those at Munmarlary. However, experimental units were 10–20�km2 sub-catchments, making it one of the largest, replicated fire experiments ever conducted. The Kidman Springs experiment focused on grass-layer productivity and composition to meet the needs of the pastoral industry, but also provided an opportunity to examine biodiversity responses to different fire regimes. Methodologically, the experiments have generally focused on phenomena—the responses to different fire treatments of individual taxa—rather than on mechanisms that determine response syndromes. They have highlighted that a range of responses to differences in fire regime is possible, and that no single fire regime can optimise all biodiversity outcomes. For effective conservation of biodiversity in the face of such complexity, conservation goals will need to be made explicit. The existing portfolio of manipulative experiments is incomplete, lacking especially a consideration of some critical savanna taxa and environments, and providing little information on the significance of spatial and temporal variability in fire patterns, especially at small scales. An understanding of fire in Australian savanna landscapes remains inadequate, so there is a continuing need for close partnerships between scientists and conservation managers, with fire management treated as a series of landscape experiments in an adaptive management framework.
Residual herbicide treatments reduce
Andropogon gayanus
(Gamba Grass) recruitment for mine site restoration in northern Australia