Facultative and Obligate Trees in a Mesic Savanna: Fire Effects on Savanna Structure Imply Contrasting Strategies of Eco-Taxonomic Groups

Abstract

Fire is a major determinant of savanna tree communities and, as such, manipulation of fire frequency is an important management tool. Resolving the effects of fire management on tree size class distributions can help managers predict and plan for short-term ecological and economic outcomes, reveal different strategies by which woody plants cope with frequent fire, and help us predict vegetation changes under future fire scenarios. Savanna structure and size class distribution are strongly influenced by the ability of suppressed tree resprouts to escape stem death by frequent fire. A widespread assumption is that resprouts have an imperative to escape fire to reach sexual maturity in the canopy and thereby ensure long-term species viability. We use a census of Australian mesic savanna tree communities subjected to annual, triennial, and fire exclusion (unburnt) fire treatments to ask how fire frequency affects size class distributions within and between eco-taxonomic groups of species. Total tree densities did not significantly differ, but were highest in the triennial (7,610 ± se 1,162 trees ha^-1) and unburnt fire treatments (7,051 ± se 578 trees ha^-1) and lowest in the annual fire treatment (6,168 ± se 523 trees ha^-1). This was caused by increased sapling densities in the triennial and unburnt fire treatments, predominantly of Acacia and pantropical genera. Eucalypts (Eucalyptus and Corymbia spp.) dominated the canopy across all fire treatments indicating relatively greater success in recruiting to larger sizes than other species groups. However, in the sub-canopy size classes eucalypts co-dominated with, and in some size classes were outnumbered by, pantropicals and Acacia, regardless of fire treatment. We hypothesize that such results are caused by fundamental differences in woody plant strategies, in particular sexual reproduction, that have not been widely recognized in Australian savannas.

Keywords: escape; fire frequency; fire trap; savanna; size class distribution; tree; woody plant strategies.

Figure 1

Map of the Tiwi Islands showing locations of the Tiwi Carbon Study fire treatment sites (Imalu, Shark Bay, Taracumbi, Pikertaramoor). Fire treatment sites were in natural savanna vegetation areas between plantation blocks. Mean annual rainfall is shown for each site, collected from manual rain gauges between 2005/06 and 2008/09.

Figure 2

Timeline of fire history at Tiwi Carbon Study (TCS) fire treatment sites. Unburnt: fire-excluded for 10 years, Triennial: burnt every 3 years in the early dry season, Annual: burnt annually in the early dry season. Data used in our study were collected in May 2015, immediately prior to implementation of that year's planned burn program and 6 years after the first experimental burns of the Tiwi Carbon Study in 2009. Prior to plantation establishment in 2005/06 the study sites were subjected to ambient, unquantified, fire regimes.

Figure 3

Annual and triennial fire treatment coefficients (raw coefficient values on the log scale) relative to the unburnt reference class (dashed line). Error bars are 95% confidence intervals. The multiplicative effect is the exponentiated coefficient value and indicates relative change in density of each size class in annual and triennial fire treatments compared to unburnt.

Figure 4

Site intercepts (raw coefficient values on the log scale), which indicate departures of tree density within each size class from the average site (dashed line). Error bars are 95% confidence intervals. Sites are shown in order of decreasing rainfall from top to bottom within each size class.

Figure 5

Tree size class distributions in annual, triennial, and unburnt fire treatments. Error bars are 95% confidence intervals.

Figure 6

Mean total density (± 95% confidence interval) of trees in each eco-taxonomic group by fire treatment.

Figure 7

Mean total canopy tree (>10 cm diameter at breast height) density (±95% confidence interval) by eco-taxonomic group, averaged across fire treatments.

Figure 8

Model predictions of sub-canopy tree size class distributions in annual, triennial, and unburnt fire treatments by eco-taxonomic group. Error bars are 95% confidence intervals.

Figure 9

Species in the pantropical eco-taxonomic group that are capable of fruiting and flowering at very small height within the flame zone and within 1 year of burning. (A) Brachychiton megaphyllus (Malvaceae), (B) Planchonia careya (Lecythidaceae), (C) Buchanania obovata (Anarcardiaceae).