Can Australian fire regimes be managed for carbon benefits?

Landscape fires affect ecosystem structure and function and mediate interactions between the biosphere and the atmosphere (Bond & Keeley, 2005; Bowman et al., 2009; Flannigan et al., 2009). There is emerging concern about the role of fires as a source of greenhouse gas emissions; future increases in fire activity (via global warming and land use) could exacerbate this problem. A meeting of the Working Group No. 53 (‘Fire and Carbon Budgets’; http://www.vegfunction.net/wg/53/53_Fire_Carbon_Budgets.htm), of the Australia/New Zealand Network for Vegetation Function, was convened in Canberra in November 2008 to explore approaches for modelling emission and mitigation scenarios for Australian ecosystems, which are ubiquitously fire-prone. ‘Potential for emission abatement in differing ecosystems will depend on a trade-off between “prescribed burning efficacy”, on the one hand, and intensity/fuel consumption (“burning efficiency”) on the other.’ To what extent will carbon-mitigation benefits accrue under differing regimes of management – particularly the use of prescribed fire to mitigate the impacts of unplanned fire? The potential for major emission abatement via mitigation of wildfire impacts has been discussed for forests in North America (Hurteau et al., 2008) and in Europe (Narayan et al., 2007). In Australia, prescribed fire has been used successfully in northern Australia to achieve greenhouse gas emission abatement (Russell-Smith et al., 2009). A similar prospect has been mooted for southern fire-prone temperate eucalypt forests (Bushfire Cooperative Research Centre, 2006) but systematic analysis to support this proposition is lacking. Dick Williams (Commonwealth Scientific and Industrial Research Organisation (CSIRO), Australia) outlined the simple logic governing the problem: prescribed burning may decrease the intensity and extent of subsequent unplanned fires (wildfires) by reducing fuel. This may reduce net emissions, but quantification of any effect requires a detailed understanding of vegetation dynamics, carbon stocks and flows, accounting rules, the efficacy of prescribed burning and the trade-offs between different fire regimes. Extension of the savanna model to forests requires a comparative appraisal of these attributes. Malcolm Gill (Australian National University (ANU), Canberra, Australia) highlighted the mechanisms of fire and vegetation dynamics that fundamentally reduce terrestrial carbon stocks, compared with no fire (see Bond & Keeley, 2005). Alternative fire regimes (which vary in intensity and frequency) will affect these mechanisms. Potentially, an increase in either frequency or intensity (or both) will decrease terrestrial carbon stocks and thus increase emissions. Emissions mitigation has received scrutiny within the tropical savanna woodlands of northern Australia (Williams et al., 2004; Beringer et al., 2007; Cook & Meyer, 2009). Mick Meyer (CSIRO, Australia) highlighted the contribution of savannas, with an average of 340 000 km2 burnt annually, which produce 2–4% of Australia's accountable greenhouse gas emissions (Russell-Smith et al., 2007; Cook & Meyer, 2009). Jeremy Russell-Smith (Bushfires Northern Territory, Australia) described the program of experimental fire management in Western Arnhem Land, Northern Territory Western Arnhem Land Fire Abatement Project (WALFA; http://www.savanna.org.au/all/walfa.html), which is targeted at reducing emissions by reducing the area burned and the intensity of late, dry-season savanna fires (these annually burn an average of 340 000 km2) through the use of early, dry-season prescribed burning (Russell-Smith et al., 2009). Additional biodiversity and social benefits can accrue from altered fire regimes. How different are carbon stocks, fluxes and fire regimes in northern Australian savannas compared with southern temperate eucalypt forests? The degree to which managed changes in fire regimes will affect potential changes in carbon will depend on the nature of stocks and their interaction with characteristic fire regimes in each biome. Heather Keith (ANU, Canberra, Australia) and Lindsay Hutley (Charles Darwin University (CDU), Australia) demonstrated that above-ground and below-ground stocks of carbon are significantly higher in temperate forests compared with savannas, based on a variety of measurements, including estimates of net ecosystem exchange (NEE) from flux towers (Fig. 1a). Both forest and savannas appear to function as carbon sinks. However, infrequent, high-intensity disturbances (cyclones in savannas, severe fires in forests) cause major episodic losses that counteract this trend (Beringer et al., 2007). (a) Stocks of above-ground (stippled bars) and below-ground (filled bars) carbon (C t ha−1), and area of temperate eucalypt open forests and savannas (open bars). Carbon estimates were based on the results of Mackey et al. (2008) for forests and on the results of Chen et al. (2003) for savannas. (b) Ranges of inter-fire interval (IFI) (stippled bars) and fire intensity (white bars), in forests and savannas. Arrows indicate the net direction of change with increasing prescribed burning. The dashed arrow for savanna indicates potential change in inter-fire interval under strategic burning (see the text for details). Damian Barrett (University of Queensland, Australia) demonstrated that land-use change and fire across northern Australia savannas can result in significant losses to terrestrial carbon stocks, using modeled estimates (VAST model, Barrett et al., 2007) of carbon stocks and fluxes. Steven Roxburgh (CSIRO, Australia) showed that potential losses from terrestrial stocks in temperate forests, under clearing and certain types of logging (at a local scale), were likely to be higher per unit area than in savannas. The national significance of losses from stocks in forests compared with savannas may, however, be outweighed by their considerably smaller area (Commonwealth of Australia, 2007; Fig. 1a). Fire intensity is higher, but inter-fire intervals (IFI) are longer in forests than in savannas (Gill & Catling, 2002; Williams et al., 2002; Fig. 1b). Routine mapping of the spatial variability of fires (e.g. perimeters and intensity) will be crucial in any effort to quantify fire-regime changes and carbon benefits, as illustrated for WALFA by Stefan Maier (CDU, Australia) using remote sensing. This capacity is less comprehensive for temperate forests, in part because of the increased difficulties imposed by more highly varied terrain and complex vegetation structure. Gary Cook (CSIRO, Australia) described how changes to savanna fire regimes can switch pathways for breakdown of biomass (decomposition vs fire). By decreasing the frequency and intensity of fire in savannas, which are currently close to their maximum limits (Fig. 1b), more biomass is decomposed, resulting in reduced emissions of methane and nitrous oxide (accountable gases). By contrast, the current fire frequency in forests is such that a shift could have little impact on emissions because litter fuels are at, or near, steady-state levels. The results presented by Geoff Cary (ANU, Canberra, Australia), linking landscape fire-regime simulation with a carbon-accounting model, may reflect this. These indicated that changes to carbon stocks may be more sensitive to direct effects of future climate than to changes in fire management in montane eucalypt forests. Owen Price (University of Wollongong, Australia) used analysis of historical fire-area mapping to show that for each hectare treated using prescribed fire, the average reduction in unplanned fire area in savanna (WALFA landscape) is about 1 ha (1:1 ratio), meaning that the overall area burned is not reduced, although the intensity will be reduced. This ratio may be further reduced as prescribed burning becomes more strategically targeted (Fig. 1b). By contrast, similar historical fire-area analyses and simulation modelling for dry sclerophyll forests in both south-eastern (Owen Price and Ross Bradstock, University of Wollongong, Australia) and south-western (Mathias Boer, University of Western Australia, Perth, Australia) Australia indicated a higher range of ratios than in savannas (c. 3:1–4:1). This range contrasts the results of Narayan et al. (2007), who postulated that relatively small levels of prescribed fire could eliminate large unplanned fires in varied European ecosystems. The Australian data suggest that prescribed fire reduces, rather than eliminates, unplanned fire area. Given ratios of > 1:1, prescribed burning in Australian forests will tend to increase the overall fire frequency (i.e. decrease the IFI, Fig. 1b). The other component of efficacy is the effect of prescribed fire in altering fire intensity through changes to fuel quantity and structure. In savannas, prescribed burning in the early dry season has the potential to abate emissions by reducing average fire intensity (Williams et al., 2002), even without a net reduction in area burned or in fire frequency. Estimates indicated that prescribed burning is effective in shifting savanna fire intensity and fuel consumption. Prescribed burning can reduce fire intensity in forests, although the range is considerably wider (Fig. 1b) and the effects in the field are less well quantified. The potential for emission abatement in differing ecosystems will depend on a trade-off between ‘prescribed burning efficacy’, on the one hand, and intensity/fuel consumption (‘burning efficiency’) on the other. Prescribed burning involves an ‘outlay’ of emissions that will be potentially offset by a ‘return on investment’ via alteration of the area and intensity of subsequent unplanned fires. If, as in savannas, the return is relatively high (i.e. a 1:1 ratio), then a net ‘benefit’ (i.e. emission reduction) will accrue via a reduction in fire intensity and fuel consumption. The size of the benefit will be a function of the difference in average burning efficiencies of prescribed vs unplanned fires. In temperate eucalypt forests the predicted return is lower, because the effect of prescribed burning on the area burned is lower (i.e. ratio > 1:1), resulting in a net increase in the overall area burned. A ‘net benefit’ is therefore unlikely unless there is a very wide contrast in burning efficiency between prescribed and unplanned fires. Further measurement of fire intensities and fuel consumption in forests is required to clarify this. By contrast, the scenario postulated by Narayan et al. (2007) would yield a major net emission ‘benefit’ through elimination of area burned by unplanned fire via a relatively low outlay (i.e. ratio << 1:1). While there is potential for such an effect in savannas, it appears less likely to occur in temperate eucalypt forests. The workshop produced an equivocal prognosis. The relatively large potential emission abatement and mitigation benefits postulated for European and North American forests (Narayan et al., 2007; Hurteau et al., 2008), via the use of prescribed fire, do not appear to be plausible in Australian sclerophyll forests. The situation in savannas seems to be more favourable. Ongoing measurements of variation in fire regimes in the WALFA project, and evaluation of the efficacy of prescribed burning in reducing total area burnt, are required to confirm this potential. The conceptual model provides a basis for assessment of potential benefits in other Australian ecosystems – particularly the vast woodlands and rangelands of the dry interior.