• Open Access

Resolving conflicts in fire management using decision theory: asset-protection versus biodiversity conservation


  • Editor
    Atte Moilanen

Dr. Don A. Driscoll, Fenner School of Environment and Society, WK Hancock Building 43, Australian National University, Canberra, ACT 0200, Australia. Tel: +61 2 6125 8130; fax: +61 2 6125 0757. E-mail: don.driscoll@anu.edu.au


Agencies charged with nature conservation and protecting built-assets from fire face a policy dilemma because management that protects assets can have adverse impacts on biodiversity. Although conservation is often a policy goal, protecting built-assets usually takes precedence in fire management implementation. To make decisions that can better achieve both objectives, existing trade-offs must first be recognized, and then policies implemented to manage multiple objectives explicitly. We briefly review fire management actions that can conflict with biodiversity conservation. Through this review, we find that common management practices might not appreciably reduce the threat to built-assets but could have a large negative impact on biodiversity. We develop a framework based on decision theory that could be applied to minimize these conflicts. Critical to this approach is (1) the identification of the full range of management options and (2) obtaining data for evaluating the effectiveness of those options for achieving asset protection and conservation goals. This information can be used to compare explicitly the effectiveness of different management choices for conserving species and for protecting assets, given budget constraints. The challenge now is to gather data to quantify these trade-offs so that fire policy and practices can be better aligned with multiple objectives.


Fire management is receiving increased public attention in many parts of the world associated with increased ignition rates (Syphard et al. 2007) and an expanding wildland–urban interface (Radeloff et al. 2005). Fire management often incites controversy because there are conflicting management objectives (Morrison et al. 1996). Fire management and policy can be driven entirely by built-asset protection, with little or even no consideration of biodiversity conservation (DellaSala et al. 2004). Failure to recognize and resolve these conflicting objectives could lead to substantial environmental degradation and species loss (Gill 1977).

Contributing to this conflict is a frequent mismatch between policy rhetoric, which claims to manage fire for biodiversity conservation, and its implementation and knowledge base (DellaSala et al. 2004; Clarke 2008). For example, in the USA, the Healthy Forests Restoration Act 2003 implied a goal of benefit to the forest. However, the policy has been widely criticized because its implementation has detrimental effects for forest biodiversity (Franklin & Agee 2003). In Australia, ecological sustainability is emphasized in the policies of management agencies (Clarke 2008). However, the evaluation of policy outcomes is limited by the available evidence, and there is little systematic research or monitoring to gather adequate data (Clarke 2008). In the absence of appropriate evaluation, management practices that may be injurious to biodiversity are routinely implemented, even when the effectiveness of such practices for protecting assets is poorly known (Fernandes & Botelho 2003; Backer et al. 2004; Bradstock et al. 2005; Cary et al. 2009).

Given mismatches between stated policy goals and on-ground fire management, in this article we first identify fire management practices that might conflict with a policy of ecological sustainability, and then develop a rational decision-making approach that can evaluate conflicting objectives. We emphasize that there are relatively simple approaches that can help to achieve both conservation and asset-protection objectives using the logic espoused in decision theory.

Competing management objectives

Below, we identify a range of fire management actions that are commonly implemented before, during or after an unplanned fire to protect assets (nonbiodiversity assets; infrastructure, houses, human life). We discuss published evidence of their asset-protecting qualities, and evidence that illustrates ways in which fire management may compromise biodiversity objectives.

Competing management objectives before a fire

Planned burning within zones

Fire management zones define parcels of land with different fire management objectives. These can define protection zones adjacent to assets such as the wildland–urban interface (Gill & Stephens 2009), or have several zones representing different emphases on biodiversity and asset-protection (DellaSala et al. 2004). Burning zones within 100 m of infrastructure every 1–4 years can substantially reduce the risk of uncontrollable fire, except under extreme weather conditions (McArthur Forest Fire Danger Index (FFDI) > 100), and on steep slopes where any amount of burning will not reduce the risk (e.g., > 15 degrees and FFDI > 40, Bradstock et al. 1998). Such frequent burning is likely to have negative consequences for biodiversity in some ecosystems, such as Australian temperate forests (Morrison et al. 1995; Bradstock et al. 1998) and a range of North American forests (DellaSala et al. 2004). For example, Whelan et al. (2006) showed that frequently burnt buffer zones made 9–15% of potential habitat unsuitable for two threatened bird species in eastern Australia. This conflict may arise frequently. Housing development often preferentially occupies particular geographic landforms and therefore, habitat types (e.g., flat areas), leaving only small remnants of these habitats at the margins of development. The result is that the extent of housing development can be positively correlated with the number of threatened species (Underwood et al. 2009), with the threatened species occurring in the same areas that are zoned for frequent planned burning.

Large-scale, extensive planned burning

Planned burns aimed at protecting assets can involve burning large areas (e.g., Finney et al. 2005), throughout regions supporting predominantly native vegetation. Short periods between fires (1–4 years) are needed to minimize risk because fuels may rebuild within a few years (Bradstock et al. 1998). The effectiveness of such planned burning for protecting assets from unplanned fires is diminished under severe weather conditions (Fernandes & Botelho 2003; Cary et al. 2009, Keeley & Zedler 2009). The effectiveness of planned burning for reducing risk also depends on the distance of the burnt area from assets. Strategically locating burns in close proximity to assets may offer higher levels of risk reduction than burns that are more dispersed in the landscape (Ager et al. 2007; Bradstock et al. 2008; King et al. 2008; Cary et al. 2009). Widespread burning which leads to short inter-fire intervals will cause some species to decline and if conducted over a large region may threaten these species with extinction (Gill & Bradstock 1995; Morrison et al. 1996). Whether or not that risk becomes high will depend on the size of species’ distributions compared with the extent of burning. In addition to direct threats to biodiversity, planned burning and associated infrastructure and disturbance can facilitate weed invasion (Keeley 2006).

Although we emphasize the potential conflict between widespread planned burning and biodiversity conservation, in some environments, planned landscape burning can contribute to biodiversity conservation. For example, planned burns in tropical savannas can reduce the area burnt by intense late-dry-season fires (Andersen et al. 2005), and in North American Pinus ponderosa forest, planned burns can help to restore understorey characteristics (Noss et al. 2006). The same fire management action therefore has different effects on biodiversity and asset-protection in different environments. Evidence for appraising the effectiveness of management will thus need to consider local fire behavior and biotic responses (DellaSala et al. 2004).


Forest logging is frequently proposed as a means of reducing the impacts of unplanned fires on economic assets (Lindenmayer et al. 2009a). For example, in the USA, forest thinning, including removal of old-growth trees was promoted under the Healthy Forests Restoration Act 2003, with the objective of reducing the risk of unplanned fire (DellaSala et al. 2004). In some North American pine forests, forest thinning can reduce fire risk, particularly if most of the downed trees are removed (Stephens et al. 2009). In other forest types, logging can increase the risk of large fires because it creates areas with uniformly dense fuels (young trees) and increases ignition risk (DellaSala et al. 2004; Lindenmayer et al. 2009a). Certain logging regimes have well established negative effects on a broad range of species (Lindenmayer & Franklin 2002).

Competing management objectives during an unplanned fire

Roads and constructed “containment” lines

Cleared lines including “bare-earth” roads are commonly established during unplanned fires to facilitate access by fire crews and vehicles. To our knowledge, the effectiveness of building new cleared lines and roads during a fire for protecting assets, compared with using established roads, has not been examined, although if tracks are left open the risk of subsequent ignitions is increased (Syphard et al. 2007). The establishment of new roads and tracks has a range of environmental impacts (Gill 1977; Trombulak & Frissell 2000). Roads can cause soil compaction, erosion, and increased sedimentation (Backer et al. 2004). Road construction and roads can also facilitate the spread of exotic plants and animals and increase fragmentation of native biota (Forman & Alexander 1998).

Back-burning and burning out

Back-burning involves setting fires that burn back towards an unplanned fire to eliminate fuel and halt or slow the main fire front. These fires aim to be wide enough to prevent fire-fronts from starting spot fires downwind of the back-burnt area. It is possible that large proportions of the landscape are burned by back-burns compared with the main fire, although data are scant (Whelan 2002). Unburnt areas remaining after back-burning or after a fire front has passed may be deliberately burnt out. Although some back-burning is widely agreed to aid fire management, we have been unable to find published research documenting the effectiveness of different widths of back-burn, or extents of burning out, for protecting assets (but see: Gill & Stephens 2009). However, impacts on biodiversity are likely to arise by removing unburnt refuges. For some species, unburnt patches are important for survival and post-fire recolonization (Penman et al. 2007; Lindenmayer et al. 2009b). Patches of forest with low-flammability characteristics may normally remain unburnt during unplanned fires (Clarke 2002) and these can provide important refuges for species that otherwise would not survive in the more regularly burnt surrounding landscape (Gandhi et al. 2001).

Application of fire retarding chemicals

A broad range of fire-retarding and other suppressing chemicals is currently used in fire management. Although much is known about the fire limiting qualities of retarding chemicals, the benefits of retardants compared with water alone when fighting fires under field conditions is poorly studied (Gimenez et al. 2004; Àgueda et al. 2008). In one experiment, using water alone resulted in the fewest re-ignitions after initial fire suppression (Rawet et al. 1996). Many fire retardants have well-established negative impacts on aquatic biodiversity, and streams are now generally avoided during application (Gimenez et al. 2004). However, impacts on terrestrial biodiversity are poorly documented, although there is evidence that low-nutrient adapted plants and some birds are detrimentally affected (Buscemi et al. 2002; Bell et al. 2005). In addition, weed invasion can increase because fire retardants can add substantial quantities of phosphorus, nitrogen or other nutrients to the environment (Gill 1977; Gimenez et al. 2004; Bell et al. 2005). Given these risks to biodiversity, there is an urgent need to simultaneously evaluate the biodiversity and asset-protection value of using retardants compared with fresh-water alone for fire suppression.

Competing management objectives after an unplanned fire

Post-fire harvesting

Post-disturbance (salvage) logging operations are employed in many kinds of forests worldwide, particularly those in wet forest environments where stand-replacing unplanned fires predominate (Lindenmayer & Noss 2006). Post-disturbance logging can have economic benefits, and sometimes is thought to reduce fire risk, though evidence suggests there is an increased fire risk after post-disturbance logging (Donato et al. 2006). Such logging typically also has a range of ecological impacts (Lindenmayer et al. 2008). Post-disturbance logging can alter key ecological processes such as hydrological regimes (Donato et al. 2006), and can exacerbate soil erosion (Wilson 1999). In addition, the removal of biological legacies, including large living and dead fire-damaged trees, can have negative effects on a range of biota (Lindenmayer et al. 2008).

Post-fire rehabilitation

In attempts to minimize run off and subsequent damage to infrastructure after large fires, managers may introduce rapidly growing grasses, fell trees and spread straw or hay over burnt areas (Robichaud 2005; Wagenbrenner et al. 2006). The costs and benefits of these methods are being compared without consideration of impacts on biodiversity (Robichaud 2005). In reviewing these practices, Keeley (2006) has shown that stabilization can create a bigger problem than it solves, introducing new weeds into plant communities and suppressing natural germination. For example, after a large fire in eastern USA, mulching with straw and application of barley seeds suppressed pine regeneration and increased the prevalence of nonnative plant species (Kruse et al. 2004). Although mulching can substantially reduce the amount of sediment lost after fire (Wagenbrenner et al. 2006), these benefits can be outweighed by other impacts on the terrestrial environment.

In each of the practices discussed above, fire management is often implemented with one objective in mind (e.g., protecting buildings), without appraising the possible impacts on other objectives. In addition, there is often inadequate knowledge of how effectively the action achieves the first objective. Inattention to the effectiveness of management actions can mean that goals are not achieved, and at great expense (Possingham 2001). We contend that a repeatable and evidence-based approach to decision making is needed.

Decision theory framework

Decision theory offers a transparent method for choosing the most effective combination of management actions to achieve a clearly stated objective (Possingham et al. 2001). A decision theory approach to fire management can identify the most effective suite of management actions by (1) considering the relative costs and benefits of alternative management actions; (2) taking into account the uncertainty that pervades our knowledge of the system; and (3) acknowledging the constraints on the actions available to managers (Richards et al. 1999; McCarthy et al. 2001; Regan et al. 2009).

We present a conceptual model for the application of decision theory (Figure 1). After carefully defining the objectives and funding scenario (Steps 1 and 2, Figure 1) the critical next step is to catalogue the full range of potential management actions (Step 3, Figure 1). For asset-protection goals, the range of actions would encompass fuel-management options, ignition management, suppression effort, asset relocation, asset reconstruction, public education and engineering to protect assets in situ (e.g., Wakefield et al. 2009). Importantly, the metrics used to measure the effectiveness of these interventions (Step 4, Figure 1) must be aligned with the objectives. Surrogate measures of effectiveness, such as area burnt, or amount of fuel consumed, do not provide sufficient information about the actual objective (e.g. species persistence or assets protected). For example, fire mosaics may seem important for conserving biodiversity, but measuring implementation of the mosaic (Brockett et al. 2001) rather than the biodiversity response, could lead to perverse outcomes including unnecessary fire management expenditure or declines of biodiversity (Bradstock et al. 2005; Parr & Andersen 2006).

Figure 1.

Conceptual flow of logic for implementing decision theory. If none of the management options are satisfactory, there are three routes (A–C) to seek a resolution.

We suggest there are two pathways through the application of decision theory to identify management options that best achieve the objectives and make any trade-offs explicit (Steps 4–5, Figure 1). The first pathway is quantitative, using simulation models to predict outcomes of combinations of management actions (Richards et al. 1999). This approach describes the system using mathematical equations, and we give a simple example here. If one objective of fire management was to minimize the expected number of extinctions in the landscape over a given time period, an equation could be structured as


where EB(Ai) is the expected effectiveness of management suite Ai for ensuring the persistence of the S species in the management region; Ai one of i suites of actions that can be implemented given the total available resources. These management suites can be spatially explicit, can be applied heterogeneously across the management region, and can include combinations of approaches; fj(Ai) the probability that management option Ai will deliver one of J possible fire regimes. This formulation accommodates uncertainty in the capacity of actions Ai to influence the fire regime; bsj the expected benefit (i.e., probability of species persistence) that species s would derive if the management region is experiencing fire regime j; and ws the relative weighting (importance) of species s. The suite of actions Ai that maximizes effectiveness EB will represent the best allocation of fire management resources for achieving biodiversity outcomes.

Analogous to Equation (1), we can frame the asset-protection objective as


where EAP(Ai) is the expected effectiveness in protecting the X assets across the management region; qxj the probability that asset x will be protected under fire regime j; and vx the value of asset x. A manager whose objective is to maximize asset protection in the landscape would choose the management option Ai that maximizes EAP.

The straightforward nature of Equations (1) and (2) belies the considerable complexity behind the way management interventions filter through to influence asset-protection and biodiversity conservation. The quantitative approach therefore requires collaboration between managers, field ecologists, and modelers (Possingham 2001). To date, there are no models that consider how the full array of fire management choices might influence fire impacts on built assets and/or biodiversity, although steps have been made in this direction for the planned-burning subset of options (Bradstock et al. 2008). Studies that have modeled fire behavior, fire impacts and management interventions suggest that there are a number of regions where knowledge could be adequate to attempt this approach (Bradstock et al. 2005; Pinol et al. 2005; Ager et al. 2007; Bradstock et al. 2008).

Where resources or access to appropriate expertise is limited, a qualitative approach is also possible (Figure 1). This involves gathering evidence for likely outcomes of contrasting management approaches, then ranking or scoring the actions based on their expected effectiveness in achieving asset-protection and biodiversity goals.

As an example of how evidence might be used in a qualitative approach, we examined widespread planned burning compared with buffer burning at the wildland–urban interface. The spread of unplanned fires in temperate forests and shrublands can be influenced more by extreme weather than by fuel loads (Fernandes & Botelho 2003; Cary et al. 2009; Keeley & Zedler 2009). Due to the predominant effect of weather, widespread planned burning is predicted to have negligible asset-protection value (Cary et al. 2009) or only a small benefit within plausible funding scenarios (Bradstock et al. 2008). Using simulation models of randomly located planned fires, Bradstock et al. (2008) found that a large proportion of the landscape must be burnt annually to reduce the risk to assets (>10%: an order of magnitude greater than current levels, Bradstock et al. 2008). In contrast, Cary et al. (2009) found that the risk to assets is not substantially influenced by the proportion of the landscape that is burnt (up to 30%, Cary et al. 2009). However, these studies agree that strategic burning can be more effective than randomly placed planned burns (Bradstock et al. 2008; Cary et al. 2009). For example, Cary et al. (2009) found that annually burnt 150 m-wide buffers reduced the amount of landscape edge burnt by 89%, where edge is a surrogate for threat to assets. This effect was achieved by annually burning 0.4–1.5% of the simulated landscape (Cary et al. 2009). Widespread, frequent planned burns in temperate forests can cause native species to decline (Gill & Bradstock 1995; Morrison et al. 1996; Bradstock et al. 2008) and exotic species to increase (Keeley 2006). For example, in forest of eastern Australia, fire intervals of less than 7–8 years are likely to cause severe declines of woody shrubs (Morrison et al. 1996). Widespread planned burns that are not adjacent to assets are therefore likely to have low to moderate effectiveness for both the asset-protection and the biodiversity objectives. Alternatively, spatially targeted planned burns, taking into account the locations of threatened species, could reduce the threat to assets and biodiversity (Whelan et al. 2006; Ager et al. 2007; King et al. 2008; Cary et al. 2009).

Biodiversity and asset-protection objectives have different measures of success, and therefore multi-criteria optimization is needed to compare the effectiveness of these different objectives (Chankong & Haimes 1983; Drechsler 2004). Using either a qualitative or quantitative approach, each set of actions can be graphed based on the extent to which they achieve biodiversity and asset-protection goals (Step 5, Figure 1). The full range of management options could populate the biodiversity/asset-protection space as shown in Figure 2. From our example above, widespread planned burning may fall towards the bottom left hand corner of Figure 2, while strategic burning, with high values for both axes, would fall closer to the upper right. We do not know what actual shape this trade-off graph would take. In some systems, points further into the upper right hand region might be possible, where maximum conservation benefit coincides with maximum asset-protection (Andersen et al. 2005).

Figure 2.

A hypothetical trade-off graph where each marker represents the biodiversity conservation (EB) and asset-protection (EAP) effectiveness achieved by a particular suite of management interventions. The subset of strategies joined by the grey lines represents the optimal set of choices for a given budget. The budget required to implement each suite of actions is denoted by circle size. The biodiversity outcome can be substantially improved with only a small reduction in asset protection by choosing management suite 2 over suite 1.

From the full set of management strategies, an optimal subset can be identified for different management budgets (points joined by lines in Figure 2). Management options that do not belong to this optimal set (below and to the left of the grey lines, Figure 2) are less effective for both biodiversity and asset-protection, and better alternatives are therefore available. On the other hand, for options within the optimal subset, one objective may only be increased with some decrease in the other objective. However, trade-offs that are satisfactory for all stakeholders may be found. For the hypothetical example in Figure 2, changing from management action suite 1 to suite 2 (indicated by the numbered circles) substantially increases the biodiversity protection, but only slightly reduces asset protection. The definition of the optimal subset therefore not only indicates which of the available strategies are candidate solutions, but also quantifies any trade-offs required if managers wish to increase either objective.

This approach has been applied in a range of conservation contexts. For example, Venter et al. (2009) compared the global strategies for paying countries to keep their forests under the competing objectives of carbon sequestration and biodiversity conservation. They found that optimal allocation of resources for carbon sequestration was largely in South America, while optimal allocation for biodiversity protection was predominantly in Asia. However, using a trade-off curve that considered a range of investment strategies, they showed that biodiversity benefits could be doubled with only a 4–8% reduction in the amount of carbon retained.

Research that is closely linked with policy development is now needed to place different fire management strategies onto a trade-off graph, comparing quantitative and qualitative approaches (Figure 1). An iterative decision-making process might then be needed to find acceptable solutions. If none of the management options (selected in stage 5, Figure 1) achieve satisfactory effectiveness for each competing objective it may be necessary to identify and rank innovative management options (A, Figure 1), to obtain more funding (B, Figure 1), or at worst, to redefine the objectives (C, Figure 1). Finally, planned learning needs to be integrated with management implementation (Step 6, Figure 1; Possingham 2001; Wintle & Lindenmayer 2008). While knowledge is inadequate, management should be combined with appropriate monitoring and research to improve our ability to assess management effectiveness.


It might be possible to develop a qualitative trade-off graph based on existing knowledge, on which the relative positions of suites of management alternatives can be placed. This could be a rapid way to avoid the greatest failures to achieve multiple objectives by immediately separating clearly suboptimal fire management options from those that warrant more careful consideration. This alone would represent an advance beyond current fire management planning in many regions. Progress in modelling fire behavior, species’ responses and the effects of fuel-management suggests that a simulation-based application of decision theory is also practicable for well-studied regions. However, even in less studied regions, going through the process of building models and identifying uncertainties will highlight important knowledge gaps, enabling targeted research. Our review has shown that research is needed to relate management actions directly to asset protection and extinction risk. However, undertaking the decision-theory approach will identify the specific knowledge gaps that are most important to fill for a particular management region.

Finally, we think that there are three take-home messages for policy makers: (1) It is important to recognize that stakeholders place two conflicting demands on current fire management operations, and that the trade-offs between biodiversity conservation and asset protection need to be addressed explicitly and transparently; (2) Decision theory provides a framework for appraising conflicting objectives and so could be fruitfully applied to fire management; and (3) The successful application of decision theory requires two core sets of knowledge: the effectiveness of the full range of management actions in protecting assets; and the influence of those management options on biodiversity. Changing fire regimes pose a substantial, and increasing, threat to both biodiversity and built assets in many regions of the world. It is therefore imperative that a rational and effective response is developed quickly. This will include having evidence to address the competing objectives of biodiversity conservation and asset protection in fire management.


Thanks to Atte Moilanen and reviewers for suggestions that improved the article. This article developed from discussion held among the authors at a workshop funded by the Applied Environmental Decision Analysis Commonwealth Environment Research Facility. Matt Plucinski provided advice on fire suppressants.