Effects of fire and drought in a tropical eucalypt savanna colonized by rain forest

Aim: This study documents the effects of multiple fires and drought on the woody structure of a north Australian savanna never grazed by domestic stock. Location: The study was conducted in a 500 ha pocket of Eucalyptus-dominated savanna surrounded by a late Quaternary lava flow. The flow is known as the Great Basalt Wall, located c. 50 km northeast of Charters Towers in semi-arid north-eastern Australia. This region was exposed to the largest 5-year rainfall deficit on record between 1992 and 1996. Methods: All individual woody plants were tagged within a 1.56 ha plot. Species were segregated into their habitat affinities (rain forest, ecotone, savanna) and regeneration strategy (resprouter, seeder). The survivorship of plants within these categories was analysed in relation to fire intensity from the first fire, and to each of four fires lit between 1996 and 2001. Results: Before the first fire, the plot contained thirty-one tree species including twenty-one typical of the surrounding dry rain forest. These rain forest species were represented by small individuals and constituted <1% of the total basal area of woody plants. The basal area of savanna trees was 7.5 m2 ha-1 at the commencement of monitoring, although 31% had recently died and others had major crown damage. Further death of the drought debilitated savanna trees was substantial during the first year of monitoring and the basal area of live savanna trees declined to 1.1 m2 ha-1 after 5 years. Most species from both rain forest and savanna were classified as resprouters and are capable of regenerating from underground organs after fire. Species without this ability (rain forest seeders and ecotone seeders) were mostly eliminated after the first two consecutive fires. Among resprouters, survivorship declined as fire intensity increased and this was more pronounced for rain forest than for savanna species. Repeated burning produced a cumulative effect of decreasing survivorship for rain forest resprouters relative to savanna resprouters. Main conclusions: The study provides evidence that savanna and rain forest trees differ in fire susceptibility and that recurrent fire can explain the restricted distribution of rain forest in the seasonally arid Australian tropics. The time of death of the savanna trees is consistent with the regional pattern after severe drought, and highlights the importance of medium term climate cycles for the population dynamics of savanna tree species and structure of Australian savannas.