A conceptual framework for predicting temperate ecosystem sensitivity to human impacts on fire regimes


  • Editor: Greg Jordan

Correspondence: Dave McWethy, Department of Earth Sciences, Montana State University, Bozeman, MT 59717, USA.

E-mail: dmcwethy@montana.edu



The increased incidence of large fires around much of the world in recent decades raises questions about human and non-human drivers of fire and the likelihood of increased fire activity in the future. The purpose of this paper is to outline a conceptual framework for examining where human-set fires and feedbacks are likely to be most pronounced in temperate forests world-wide and to establish and test a methodology for evaluating this framework using palaeoecological records.


Tasmania, north-western USA, southern South America and New Zealand.


We outline a conceptual framework for predicting the sensitivity of ecosystems to human impacts on fire regimes and then use a circum-Pacific comparison of existing historical reconstructions of fire, climate, human settlement and vegetation to evaluate this approach.


Previous research investigating important controls on fire activity shows that the sensitivity of temperate ecosystems to human-set fires is modulated by the frequency of natural fire occurrence, fuel moisture and fuel type and availability. Palaeoecological data from four temperate regions suggest that the effects of anthropogenic burning are greatest where fire is naturally rare, vegetation is poorly adapted to fire and fuel biomass is abundant and contiguous. Alternatively, where fire activity is naturally high and vegetation is well adapted to fire, evidence of human influence on fire and vegetation is less obvious.

Main conclusions

Palaeofire records suggest that the most dynamic and persistent ecosystem transitions occur where human activities increase landscape flammability through fire–vegetation feedbacks. Rapid forest transitions in biomass-rich ecosystems such as New Zealand and areas of Tasmania and southern South America illustrate how landscapes experiencing few fires can shift past tipping points to become fire-prone landscapes with new alternative stable state communities. Comparisons of palaeoecological data from different regions with similar biophysical gradients but different human settlement histories can provide new opportunities for understanding ecosystem vulnerability to fire–climate–human interactions.


Fire influences nearly all biomes of the world and plays a critical role in transferring carbon from the terrestrial biosphere to the atmosphere. Despite its importance, the drivers of fire and its ecosystem consequences over a range of temporal and spatial scales remain poorly understood. In recent decades, fire activity has increased in many temperate forests world-wide, even in ecosystems where natural ignitions are scarce, and this increase begs questions of whether climate change, human ignitions, land-use change and/or altered vegetation are responsible. Some of the largest fires occur in forests that are highly vulnerable to climate change, have little natural resilience to fire and are undergoing rapid land-use change (Meyn et al., 2007). The increased frequency of large fires has been attributed to drier than average summers and longer fire seasons (e.g. Westerling et al., 2006), and recent severe drought and increased plant mortality have exacerbated fire hazard, raising concerns about the trajectory of post-fire vegetation dynamics and future fire regimes (van Mantgem et al., 2009). Compounding the climatic drivers of fire are forest clearance, fire suppression and related fuel changes, invasion by exotic fire-prone plants and livestock grazing.

A number of recent studies have revisited alternative stable state theory to explain dynamic shifts in the distribution of forest and non-forest systems, many suggesting that abrupt changes in fire frequency often result in rapid transitions between seemingly stable vegetation states (Staver et al., 2011; Kitzberger et al., 2012). These studies illustrate the role of fire in maintaining multiple vegetation states through promoting and reducing landscape flammability depending on the starting point (Fig. 1). Palaeoecological studies from lake-sediment, tree-ring and anthropological records have revealed critical linkages among fire, vegetation, human activities and climate that cannot be inferred from modern observations alone. This research offers unique opportunities to consider climate–fuel–ignition feedbacks under different states of climate, vegetation and human influences across multiple temporal and spatial scales, and, coupled with conceptual models, it provides critical context for understanding dynamic vegetation transitions and vulnerabilities to future change.

Figure 1.

Conceptual model of vegetation dynamics (originally developed for south-west Tasmania) illustrated with the ‘ball and cup’ (or ‘stability landscapes’) commonly used to describe alternative stable states (ISLSCP2; Olson et al., 2001). In Jackson's ‘alternative stable states’ model, each vegetation community can occupy any possible underlying environmental setting. Each vegetation state is represented as resilient ‘basins of attraction’ that are modified by the effects of fire–vegetation–soil feedbacks. The community is initially resilient to disturbance by fire and will return to the original state, but if pushed beyond the threshold by repeated stand-replacing fires (or lack thereof), positive feedbacks from the alternative stable state will take over and the system will undergo transition to another stable state (Warman & Moles, 2009). Figure adapted from Wood & Bowman (2012).

A suite of biophysical conditions is required for fire to ignite and spread, and the significant driving variables change with spatial and temporal scale (Whitlock et al., 2010). At fine scales, fire requires a heat source, oxygen and fuel. At the scale of an individual fire event, weather, fuels and topography influence the rate of fire spread and intensity (Rothermel, 1972; Sullivan et al., 2012). A fire regime comprises the pattern of fire activity (e.g. fire frequency, intensity, seasonality and type) at landscape to regional spatial scales and decadal to centennial temporal scales (Gill, 1975); and more broadly, fire's role in an ecosystem (Krebs et al., 2010). At broader scales is the concept of a metafire regime, defined as the ensemble of fire regimes that characterize a biome during the duration of its existence (Whitlock et al., 2010).

Fires are frequent where flammable fuels are abundant, ignitions are frequent and weather conditions are conducive to fire spread (Krawchuk et al., 2009). Climate affects fuel characteristics (vegetation type, fuel moisture, continuity and abundance), rates of natural ignitions and fire weather (precipitation, humidity, air temperature and wind conditions). High-productivity/high-moisture (e.g. net primary productivity) fire regimes are supported by abundant, connected fuels, but fuel moisture is generally too high to allow for fire spread (Bradstock, 2010). In these systems, large fires occur during unusually dry years, when low fuel moisture and fire weather promote rapid fire spread (e.g. Bessie & Johnson, 1995). In less productive, drier, fuel-limited ecosystems, sparse or discontinuous fuels impede fire spread, even where low fuel moisture and weather conditions are suitable for burning. Hence, at extremes of productivity and environmental moisture gradients, fires are limited either by sparse, discontinuous plant cover (e.g. deserts) or by high fuel moisture (e.g. rain forests) (Krawchuk & Moritz, 2011).

Humans have altered fire regimes for millennia, although their ultimate impact relative to climate alone is still contested (Bowman et al., 2011). Human actions include igniting fires when and where they were naturally rare, modifying fuel composition and structure, synchronizing ignitions to match times of low fuel moisture and suppressing or eliminating fires (Fig. 2; Whitlock et al., 2010). Although it is widely known that dense, agricultural-based populations significantly altered past fire regimes and vegetation, evidence suggests that strong deviations from natural fire activity are also related to cultural practices and land-use histories linked to small, nomadic populations (e.g. Bush & Silman, 2007). Pre-European deforestation of island environments demonstrates that even small, mobile populations were able to transform large landscapes through the use of fire (e.g. McWethy et al., 2010). The role of humans in altering past and present fire regimes centres on three questions: (1) Which settings are most vulnerable to shifts in fire regimes resulting from human activities? (2) Where have humans promoted fire where it would not otherwise occur? (3) At what scales are human influences most relevant?

Figure 2.

Human influence on fire activity along a productivity/moisture gradient. The natural or background distribution of fire activity (i.e. frequency and area burned; solid black curve) can be altered by human impacts via increasing ignitions, actively suppressing fires, and altering vegetation by introducing new plants and land-use activities (e.g. promoting grazing, clearing forests) (modified from Whitlock et al., 2010; Krawchuk & Moritz, 2011). The dashed line indicates how human impacts can alter the natural distribution of fire activity, in some cases resulting in a persistent shift in biome type. For example, active suppression of fires in grassland biomes can result in a shift to woodland/savanna or scrub biomes depending on how this activity interacts with climatic controls and land use impacts on vegetation structure and composition.

Anthropogenic alteration of natural fire regimes has been greatest in high-productivity/moisture environments where fuels are abundant and natural ignitions are infrequent or asynchronous with times of low fuel moisture, and in low-productivity/moisture environments, where fuels are limiting (Fig. 3; Bowman et al., 2011). In high-productivity/moisture environments, spatial targeting of ignitions in more flammable, early-seral vegetation can initiate feedbacks that increase the amount of flammable vegetation and fire activity across the landscape promoting irreversible change in the vegetation (Kitzberger et al., 2012; Perry et al., 2012b). Ignitions deliberately timed to coincide with weather conducive to fire spread have amplified and extended the duration and intensity of the natural fire season (Holz & Veblen, 2011). In ecosystems where fire is fuel limited, introduced non-native species have increased fuel continuity and fire spread (Miller et al., 2010). In many intermediate-productivity grassland/savannas, anthropogenic ignitions have shifted the spatial distribution and timing of fires to reduce fire risk (i.e. spread), but deliberate burning has generally not altered natural fire frequency or area burned (van Wilgen et al., 2004). Human activities at the tails of the productivity/moisture gradient result in strong positive feedbacks that alter fuel conditions and landscape flammability. Our objective here is to outline a conceptual framework for understanding ecosystem sensitivity to human influence on fire regimes with particular attention to feedbacks driving widespread forest transitions in more productive forest ecosystems. To evaluate this framework we draw upon examples from four temperate forest regions in a circum-Pacific region: the north-western United States, southern South America, Tasmania and New Zealand.

Figure 3.

Hypothesized distribution of fire activity in relation to a productivity/moisture gradient (x-axis) and increased human activity (left y-axis). Probability of fire (right y-axis) is defined as the likelihood of ignition and fire spread. In the absence of human activity (y = 0), background fire activity is greatest (red = high fire activity, blue = low fire activity) where fuel is abundant and continuous and variation in climate provides seasonally dry fuels (centre of x-axis, e.g. grassland/savannas). Fire activity is low where fuel moisture is consistently high and/or ignitions are infrequent or asynchronous with conditions conducive to fire spread and where fuels are sparse or discontinuous (far left and right lower quadrants). Human activity has the greatest potential to increase fire activity in settings where fuels are abundant and fire activity is naturally low (upper left quadrant); here, targeted ignitions in time and space can initiate feedbacks that promote the development of fire-prone landscapes.


We propose that efforts to disentangle human versus climate drivers of past fire regimes can advance by examining the effects of climate and humans across a productivity/moisture gradient:

Drivers of fire activity

  • H1a: At millennial to centennial scales, fire activity is governed by both long-term variations in climate and decadal- to interannual-scale climate variability, which directly affect fire-conducive climate and weather, and indirectly shape fuel composition and structure.
  • H1b: Departures from the expectations imposed by climate can be ascribed to non-climatic variables, such as land-cover change, human-set ignitions, edaphic controls and interactions among these variables (Fig. 4).
Figure 4.

Hypothesized fire activity (probability of fire ignition and spread, right y-axis) at sites in New Zealand (NZ), Tasmania (Tas), north-western United States (NW. US) and southern South America (S. SA) along gradients in productivity/moisture (x-axis) and human presence (left y-axis) for pre-human (a), early-human (b) and European era (c) fire regimes. We identify a threshold at which human activity strongly alters natural fire regimes, whether through increased ignitions, fire suppression or shifts in fuel type, abundance and availability.

Sensitivity of vegetation to anthropogenic alteration of fire regimes

  • H2: The importance of non-climatic controls on fire regimes varies across gradients of productivity (e.g. net primary productivity) and moisture (Fig. 4; Bowman et al., 2011).
  • H2a: Fire regimes in high-productivity/moisture systems are most sensitive to human influences because human activities can create conditions that overcome or reinforce natural limits on fire. Deliberate fires in these settings may initiate positive feedbacks that maintain highly flammable, early-seral vegetation and prevent development of less flammable, late-seral vegetation (Kitzberger et al., 2012). In these settings, humans may also target ignitions to coincide with times that abundant, but normally wet biomass, is available for burning [e.g. periods of drought related to interannual variations in climate such as El Niño–Southern Oscillation (ENSO)]. Alternatively, human activity can limit fires in these settings through fire suppression and landscape fragmentation.
  • H2b: Fuel-limited systems are most sensitive to the introduction of non-native plants that increase fuel continuity and/or flammability, and to interannual variations in climate that can increase fuels in one year, and then promote fire through fuel-drying in the following year (e.g. ENSO). Human-set fires that are synchronous with periods of abundant flammable fuels can have dramatic impacts on fire activity (Bradstock, 2010).
  • H2c: Abundant and continuous fine fuels promote high fire activity in intermediate-productivity/moisture settings, such as in woodland or forest–steppe ecotones. Human-set fires alter the timing of fires but do little to increase fire frequency in these fire-prone settings. Fire suppression and landscape fragmentation may decrease fire activity.

Circum-Pacific Comparison

Changes in fire regimes following pre-European human and European arrival in temperate high-productivity regions (Fig. 4, Table 1) illustrate how our conceptual framework and hypotheses can be evaluated. The approach builds on and complements palaeofire reconstructions within particular regions and those under way at a continental and global scale (e.g. Marlon et al., 2009). We seek to identify where background fire activity deviates from a system driven by climate alone (H1a) by comparing biophysical gradients, flammability traits of native vegetation and records of past fire, vegetation and climate, and examine where humans have had the greatest influence in shaping past fire regimes (H1b). While palaeofire records themselves cannot identify the causal mechanisms of past fire activity, comparing fire–climate–human relationships across four temperate forest regions helps evaluate alternative hypotheses about the sensitivity of fire regimes to human activity. Ultimately, landscape-scale fire and ecosystem modelling, which can isolate and test alternative mechanisms linking climate, vegetation, humans and fire, are necessary to help infer potential mechanisms causing past changes.

Table 1. Human-settlement and biophysical gradients represented by palaeodata used in this study. Monthly temperature and precipitation data are from the Climate Research Unit (New et al., 1999), and net primary productivity (NPP) data are from the International Satellite Land-Surface Climatology Project II (ISLSCP2; Olson et al., 2001)
Gradient/parameterTasmaniaNorth-western United StatesSouthern South AmericaNew Zealand
Early human occupation (years)> 30,000> 12,00012,000800
European occupation (years)200200500200
Predominant early human economyHunter-gathererHunter-gatherer/localized intensive cultivationHunter-gatherer/localized intensive agricultureLocalized intensive cultivation (the North Island), hunter-gather (the South Island)
Total annual precip. (mm) min.-max.744–2226373–2974250–2508643–4076
Mean temp. of the warmest month (°C) min.-max.18–2217–3112–2718–25
NPP (gC m−2 year−1) min.-max.661–911177–857190–962356–1039
Historical fire regimes    
 Pre-humanInfrequent, small, and mostly ecologically insignificant fire activity in west and more frequent fires in mixed forests and non-forest in eastInfrequent, stand-replacing fires in subalpine and mixed conifer forests; frequent low-intensity surface fires in drier environmentsInfrequent, small and mostly ecologically insignificant fire activity in the west and more frequent fires in mixed forests and non-forest in the eastInfrequent, small, and mostly ecologically insignificant fire activity with slightly more frequent fires in the east
 Early humanFrequent, deliberate targeting of fires increased distribution of buttongrass moorlands in the west, forest/non-forest mosaic elsewhereLocalized burning targeting expansion of valley grasslands and forest margins – larger impact debatedFrequent but localized targeting of fires in wet forests increasing distribution of non-forested environments, scale of transitions still unclearDramatic increase in fire activity resulting in conversion of forest to non-forest
 European settlementExpansion of early-human deforestation followed by recent fire suppressionLarge and severe fires follow timber harvests c. early 1900s followed by recent fire suppressionExpansion of deforestation to increase pasture followed by recent fire suppressionExpansion of early-human deforestation followed by recent fire suppression c. 1970s

The mid-latitude forests of Tasmania, North and South America and New Zealand lend themselves well to such a comparison because they feature: (1) similar gradients in key fire controls such as productivity and climate; (2) different durations and intensities of human use; and (3) contrasting flammability traits in the native vegetation that shape the natural fire regime (Fig. 5, Table 1). In general, each region supports a west-to-east moisture gradient with wet forests experiencing low ignition frequencies and resource-focused pockets of human settlement in the west, and non-forest grassland/scrubland ecosystems with high natural ignition frequencies and more intense land use in the east.

Figure 5.

Location of paleofire records from four regions (black dots) in geographic space (top and middle rows) and climate space. Rectangles within maps identify each study region, and black dots indicate sediment charcoal records included in Version 2 of the Global Charcoal Database and additional sites targeted for future sampling. Climate space plots (Daniau et al., 2012) include each site (black dots), all land within each study region (dark grey squares) and all land across the globe (small, light grey squares), plotted using the biophysical variables most important for global fire activity. Monthly temperature and precipitation data are from the Climate Research Unit (bottom row; similar to Marlon et al., 2008) and have a spatial resolution of 0.5°.

The regions have different histories of human arrival, occupation and prehistoric land use, with people present first in Tasmania c. 35,000 cal. yr bp (Cosgrove, 1999), in the Americas and Patagonia by 12,000 cal. yr bp (Borrero, 1999; Meltzer, 2009) and in New Zealand only in the last 700–800 years (Wilmshurst et al., 2008). In all regions, the influence of Europeans on fire regimes was most profound in the 19th and 20th centuries.


Tasmania, to the south of mainland Australia, receives high annual rainfall on its west coast, has a somewhat drier climate in the interior lowlands due to a rain shadow effect and experiences a more seasonal and drier climate on its east coast. The geology of the island results in infertile soils (quartz-rich bedrock) on the rugged west coast to more fertile soils (e.g. dolerite bedrock) on the central and eastern half of the island. Both climate and edaphic controls create a vegetation mosaic of buttongrass moorland (sedge- and low-shrub-dominated vegetation of poorly drained sites), Nothofagus (southern beech) temperate rain forest and tall eucalypt forests in the west. The midlands and east coast support dry eucalypt forests, heathlands and fire-sensitive Callitris forests. Charcoal and pollen records suggest that fires were infrequent in west coast rain forests prior to human arrival, but midland and east coast forests experienced more frequent, low-severity fires (Fig. 4a; Fletcher & Thomas, 2010).

Pre- and early human fire regime

Along the west coast, the mosaic of tall forest to moorland vegetation has been ascribed to fire–vegetation–soil interactions and a long history of Aboriginal occupancy (> 30,000 years, Cosgrove, 1999). Jackson (1968) and later Bowman (1986) and Fletcher & Thomas (2010) proposed that the co-occurrence of forest and treeless vegetation in western Tasmania arose from interactions among fire-resistant and fire-prone vegetation related to anthropogenic burning. However, current assessments suggest that millennia without fire are required for buttongrass moorland to convert to forest (Wood & Bowman, 2012). It remains unclear if aboriginal burning created the current vegetation patterns in western Tasmania, or if aboriginal-set fires simply maintained vegetation patterns resulting from late Pleistocene and Holocene climatic changes (Wood & Bowman, 2012). Likewise, anomalous grasslands on fertile substrates may be the legacy of sustained aboriginal fire use, possibly maintaining Pleistocene plant communities (Gammage, 2008; Bowman et al., 2012).

European fire regime

European colonization of Tasmania and subsequent land clearance and settlement resulted in an initial increase in fire activity, potentially amplifying the influence of drought linked to interannual ENSO variability (Nichols & Lucas, 2007). Even with more intensive land use, Tasmania's wet forests experienced infrequent fires because of strong climatic controls on fire spread (H1a; Fig. 4c) but were more vulnerable to human-set fires than the fuel-limited regions of eastern Tasmania where fire closely tracks ENSO variability (H2b,c; Fig. 4c).

In western Tasmania, the prevalence of sharp boundaries between fire-prone moorlands and fire-sensitive forests suggest that early human activities may have played a large role in altering fire regimes and vegetation (H2a; Fig. 4b). Alternatively, in open grassland/savannas and drier eucalypt forests found in Tasmania's midlands and east coast, early human-set fires, while potentially frequent, appear less ecologically significant (H2b,c; Fig. 4b). With a long history of human presence, the challenge remains to clarify the scale at which human activities altered fire and vegetation from conditions governed by climate alone.

North-western United States

The north-western United States is a large, diverse region hosting multiple mountain systems. In general, climate varies widely from the Pacific slope, where mesic forests of Pseudotsuga, Tsuga, Picea and Thuja dominate, to the interior northern Rocky Mountains, where strong elevational gradients support wet-to-dry coniferous forests dominated by Pinus, Picea, Abies and Pseudotsuga. Prior to the arrival of humans (>12,000 yr bp), fire return intervals on the order of centuries, and large infrequent stand-replacing fires characterized subalpine forests, and frequent, low-severity surface fires shaped the driest low-elevation forests of ponderosa pine (Pinus ponderosa). In middle-elevation mixed-coniferous forests of ponderosa pine, Douglas-fir (Pseudotsuga menziesii), grand fir (Abies grandis) and western larch (Larix occidentalis), the pre-European fire regime was characterized by a mixture of surface and stand-replacing fires occurring at intervals from decades to centuries (Baker, 2009).

Pre- and early human fire regime

Palaeoecological records from wet forests of the west-side Cascades of Oregon and Washington suggest that late Holocene fire regimes have maintained much of the ‘old-growth’ forest in an intermediate successional state dominated by Pseudotsuga (Hemstrom & Franklin, 1982). Pre-European fire return intervals varied spatially along a climatic gradient with long fire intervals (i.e. several centuries) in wetter forests and shorter intervals (i.e. decades) in drier forests, and it is difficult to disentangle the relative influence of humans and climate, except to note that past fire activity has closely tracked changes in climate, suggesting a minor role for anthropogenic fires (Fig. 4b; Whitlock & Knox, 2002).

In the Rocky Mountains, hunter-gatherers were present for at least 12,000 years (Meltzer, 2009) but the extent and impacts of deliberate burning seem to have been spatially variable. High-elevation mesic conifer forests in Yellowstone, for example, where humans probably had minimal year-round presence, show variations in fire frequency that are consistent with climate change. Fires in these forests were more frequent during the early Holocene insolation maximum, when much of the region experienced warm and dry summer conditions (Millspaugh et al., 2000), and decreased in the last 7000 years as summers became cooler and wetter. There is little evidence that this climate-limited system was modified by human activities, except perhaps locally (Fig. 4b).

European fire regime

In general, across the north-western United States, EuroAmerican settlement resulted in increased fire activity in the late 19th century and early 20th century, followed by widespread suppression of fires and interruption of fuels by grazing through the mid 20th century, and then increased fire risk in recent decades, particularly in dry forests (Fig. 4c). Invasive grasses, forbs and shrubs have aggressively colonized recently burned forests at all elevations, and their life history (e.g. Bromus tectorum) has increased fire frequency in some locations. Logging and other land-cover changes, however, have fragmented the most flammable vegetation types and limited large fires to unusually dry years. Thus, the alteration of fuels and fire season with EuroAmerican land use has created feedbacks that perpetuate the presence of annual invasive plants.

Palaeoecological records from wet forests in the north-western United States indicate few natural fires in the last several millennia as a result of cooler wetter conditions in the late Holocene (Fig. 4a; Long et al., 2007). Prehistoric use of fire by hunter-gatherers may have been important in oak-dominated regions, but even there, their impact on the vegetation was local in scale (Walsh et al., 2010). Indigenous burning practices may have been more significant in shaping the fire regimes of mixed-conifer forests in the Rocky Mountains, but again examples point to local use of fire, not broad-scale burning. This localized use of fire by indigenous people stands in contrast to the large areas burned by EuroAmerican settlers (Fig. 4c).

Southern South America

Pre-human fire regime

In southern South America, westerly circulation and the Andes result in a steep rainfall gradient that supports wet forests and moorlands in the west and extensive grassland steppe ecosystems towards the east. Empirical research has documented that tall, closed Nothofagus forests inhibit fire spread, whereas post-fire successional communities dominated by resprouting shrubs (e.g. Nothofagus antarctica, Schinus spp.) are highly flammable (Mermoz et al., 2005; Veblen et al., 2011). Initial burning of tall Nothofagus forests that are dominated by obligate seeders is dependent on extreme drought, whereas tall shrublands are highly flammable under less severe drought. As in the north-western United States, palaeoecological data suggest shifts from fuel-limited fire regimes in the early Holocene to climate-limited fire regimes in the late Holocene, as a result of climate-driven vegetation changes (Fig. 4a; Whitlock et al., 2007). In the early Holocene (c. 12,000 cal. yr bp) nearly all sites south of latitude 40° S on both sides of the Andes experienced higher fire activity than today, and this period of regional burning has been attributed to a weakening and southward shift of the westerlies in response to the annual insolation maximum (Huber et al., 2004; Whitlock et al., 2007).

Early-human fire regime

Humans have been present in the region for > 12,000 years (Borrero, 1999). Heusser (1990) claims a broad impact of human-set fires on the landscape; however, Huber et al. (2004) argue that the human impact was probably local. It is difficult to imagine widespread fires in Tierra del Fuego where natural ignitions are rare today, and humans may have been the critical source of ignitions in the early Holocene (H2a; Fig. 4b); alternatively, climate conditions then may have included more convectional storms than today. In most regions, fire activity declined with the expansion of closed Nothofagus forest in the late Holocene; however, the pattern of fire occurrence has been spatially heterogeneous in the last few millennia, which may relate to the local use of fire as opposed to broad-scale burning. For most of southern South America few records show clear evidence of human influence on patterns of pre-European anthropogenic fire.

European era fire regime

Europeans arrived in southern South America about 500 years ago, and were accompanied by increased fires as well as the introduction of non-native plant species (Huber & Markgraf, 2003). European settlement in the late 19th to mid-20th centuries greatly increased fire activity (Veblen et al., 1999; Holz & Veblen, 2011), and replacement of burned Nothofagus forests by resprouting shrubs has been further exacerbated by introduced herbivores (livestock, exotic deer and European hares) which impede the regeneration of the obligate seeding tree species (Raffaele et al., 2011). As in other regions, xeric woodlands on the east side of the Andes have experienced increased fuel accumulation as a result of land-use practices and fire suppression (Veblen et al., 1999).

Palaeoecological reconstructions of fire activity over the past several centuries show the importance of interannual to interdecadal climate variability related to ENSO and the Southern Annual Mode (SAM) on both sides of the Andes (Kitzberger et al., 1997; Veblen et al., 1999; Holz & Veblen, 2012). European conversion of wet forests to shrublands, grazing and the spread of non-native plant species have increased landscape flammability, maintaining more fire-prone landscapes. In dry shrub ecosystems, the most evident human impact has been a reduction of fires due to pre-European depopulation and fire suppression (H2b; Fig. 4b & c). In recent decades, climate appears to be driving changes in fire activity as synchronous widespread fire activity is strongly linked to the intensification of interannual climate variability and associated drought conditions (Holz & Veblen, 2011). In addition, planting of introduced pines as well as invasion of the native vegetation by escaped pines is creating large areas of more homogeneous and highly flammable woody fuels over extensive areas of former steppe.

Examples from more productive and wet settings, such as Tierra del Fuego and western Patagonia, suggest that early human and European activities had a larger role in altering fire regimes and vegetation through increased fire activity than elsewhere in southern South America, where climate acted as the primary driver of changes in fire activity (H2a; Figure 4b). Land-use activities following European arrival have interacted with climate variability to alter landscape flammability, both promoting and inhibiting fire activity through fire suppression, grazing and the introduction of non-native plant species (Fig. 4c).

New Zealand

Pre-human fire regime

New Zealand has a maritime climate, with strong north–south temperature and east–west rainfall gradients as the primary broad-scale climatic controls on vegetation. Before initial human settlement in the 13th century ad (Wilmshurst et al., 2008), forest covered 85–90% of the New Zealand landscape (McGlone & Wilmshurst, 1999). The palaeoecological record suggests that while fires occurred in New Zealand prior to Māori settlement they were too infrequent to have had a widespread ecological effect (Ogden et al., 1998). With a few notable exceptions [e.g. Leptopsermum scoparium (mānuka) and Kunzea ericoides (kānuka)] New Zealand's plant species are poorly adapted to fire, with fire-stimulated resprouting, epicormic budding or germination being rare (Ogden et al., 1998). New Zealand's woody ecosystems were, however, highly susceptible to fire and were probably ignition-limited rather than inherently non-flammable (McGlone, 2001; Perry et al., 2012a). The rarity of pre-human fire in New Zealand was the consequence of the low rate of lightning and its association with high-rainfall frontal activity not conducive to ignition. Prior to human arrival, New Zealand's fire regimes were limited by infrequent ignitions that rarely occurred when and where fuels were sufficiently dry to facilitate fire spread (H1a; Fig. 4a).

Early-human fire regime

The arrival of humans dramatically altered New Zealand's forest landscapes, primarily via the introduction of fire. McWethy et al. (2009, 2009,2010) describe an ‘initial burning period’ (IBP) which occurred asynchronously across the South Island between 700 and 500 years ago, depending on location. In regions that prehistorically supported forest, the IBP was characterized by a short period of intense fire and loss of forest. Pollen and charcoal records from the North Island tell a similar story (Wilmshurst et al., 1997). The addition of fire to New Zealand's landscapes led to a dramatic transformation in vegetation, slope stability and limnology. In a period of 200 years or less 40–50% of New Zealand's forest cover was lost and converted to open shrubland and fernland, which in some regions has persisted until the present-day.

European fire regime

European settlement of New Zealand in the 19th century ad was accompanied by a second period of burning such that forest cover is now around 20% and rare in the lowland eastern areas. European fires and land use further expanded forest conversion to more flammable vegetation, although this trend was halted by active fire suppression in recent decades, with an average of < 6000 ha burned per year since 1991 (Anderson et al., 2008) (Fig. 4c). In the last century, invasive pyrophytic non-native plant species, such as gorse (Ulex europaeus) and Pinus spp., have altered landscape flammability setting the stage for increased fire activity. Feedbacks between land use, non-native species and fire regimes in recent times are thought to be critical for explaining the persistence of fire-prone landscapes in New Zealand (Perry et al., 2010) and fire risk is predicted to increase under future climate conditions (Pearce et al., 2005).

The abrupt and widespread increase in fire activity following Polynesian arrival indicates a dramatic deviation from a strongly climate- and ignition-limited fire regime (H1b) and suggests that New Zealand's native, fuel-rich forests, were highly sensitive to landscape flammability feedbacks responsible for the forest transitions that persist today (H2a; Fig. 4b). European fires and land use further expanded forest conversion to more flammable vegetation, even though recent fire suppression limits the extent of fire spread (Fig. 4c).


In circum-Pacific temperate forests, the greatest human impacts, both prehistoric and European, occur where human activities initiate self-reinforcing feedbacks that alter landscape flammability and underlie alternative stable state theory as described in seminal studies by Sauer (1944), Jackson (1968), Jones (1969), McGlone (1983), Denevan (1992) and Bowman & Prior (2004). Sauer (1944) proposed that the presence of grassland and savanna systems in climates that could support forests was largely a consequence of human-set fires in North America. In Australia, humans (‘fire-stick farmers’) are considered as the primary agents of fire that shape the distribution and structure of vegetation, especially where human activities and travel are pronounced (Jones, 1969; Bowman & Prior, 2004; Bird et al., 2008). Similarly, Jackson (1968) proposed that temperate rainforest would dominate most of western Tasmania were it not for human-set fires that promote flammable sclerophyll and buttongrass moorland vegetation. Jackson's model highlights how anthropogenic ignitions in areas with limited natural ignitions can lead to new stable states through altered landscape flammability. The ecological significance of human-set fires varies across biophysical gradients that shape fire regimes, and some of the most striking examples of human influence come from settings where human activities increase ignitions when and where they are naturally rare or alter fuel conditions in ways that overcome strong natural limits on fire.

Fuel-rich wet forests in New Zealand rarely experienced fires prior to human settlement, largely because few ignitions occurred when fuels were sufficiently dry to carry fire. The rapid anthropogenic transformation of the fire regime and extensive loss of forest when people arrived was a consequence of forests that were sensitive to fire but not inherently unburnable. The New Zealand story illustrates the vulnerability of fuel-rich ecosystems to fire–vegetation feedbacks and the fact that human ignitions can have dramatic impacts in settings where natural ignitions are rare. The loss of Nothofagus–Podocarpus forest represents an extreme case of vegetation modification by anthropogenic fire and one of the best examples where the influence of humans on fire risk and fire hazard can be isolated from those of climate change. Widespread forest transitions in New Zealand also show how ignition-limited ecosystems can be pushed past fire regime tipping points, shifting landscapes that rarely experience fire to fire-prone systems.

In contrast to New Zealand, where human-set fires initiated feedbacks that altered landscape flammability (Perry et al., 2012b), human impacts on past fire activity and vegetation in the north-western United States, southern South America and Tasmania, appear to have been limited in geographic extent and impact. Palaeofire records from wet forests of Tierra del Fuego and western Tasmania suggest that human-set fires may have been partially responsible for the creation of landscape mosaics of forest/non-forest vegetation with sharp boundaries between fire-prone and relatively fire-proof vegetation. Why much larger populations in the north-western United States had little impact on fire and vegetation remains unresolved. Wet forests here would have been vulnerable to early human-set fires but historical shifts in fire regimes closely track past climate variation. The lack of a strong human signal may be a consequence of different resource utilization strategies, the degree to which Douglas-fir forests are fire resistant, the frequency of natural spring–summer ignitions, the low flammability of early and middle-successional Douglas-fir forests or some combination of these factors. For example, the ecological impact of deliberate fires in the north-western United States may have been limited by a short period (years) of post-fire landscape flammability such that human-set fires resulted in little fire spread and, unlike New Zealand beech forests, mature Douglas-fir forests are resistant to widespread fire-caused mortality and are able to quickly recolonize extensive burned areas (Larson & Franklin, 2005). Thus, anthropogenic burning in the north-western United States appears to have had minimal ecological effect in widespread Douglas-fir forests, although early fire–human–climate interactions are still poorly resolved.

During the European era, all four regions experienced a similar sequence of conversion of native grasslands and adjacent forest margins to pasture, followed by grazing and forest fragmentation, and all show an expansion of forest cover in the 20th century as a result of fire suppression and elimination, although rates of expansion vary greatly (e.g. slow in button-grass moorlands of Tasmania). In recent decades, examination of changes in fire activity across regions suggests that even recent fire activity is strongly linked to interannual climate variability and its effects on fuel type and abundance, drought and fire-weather.


Our understanding of fire as a process in the Earth system has increased as a result of analyses of global fire-related datasets (e.g. the International Multiproxy Paleofire Database) yet our knowledge of how fire–climate–human interactions are contributing to persistent biome transitions is still limited (Bowman et al., 2011). Under future climate scenarios (without human influence), fire activity is predicted to increase in temperate forests and decrease in tropical broadleaf forests and savannas (Moritz et al., 2012), yet human activities may interact with climate in ways that reverse or amplify these projections. Hence, it is increasingly important to identify environmental settings that may be most vulnerable to these climate–fire–human interactions.

To address these key topics, fire scientists are calling for multi-proxy approaches across wide biophysical gradients, yet studies of this nature are rare, probably because they require an intensive sampling effort and replication, and collaboration across disciplines. Utilizing a comparison of palaeoenvironmental and climatological data across four temperate regions is an example of an approach that can provide new insights into the role of anthropogenic burning, climate change and their impacts on global distributions of forest and non-forest biomes. Integrating these palaeofire records into multiscale mechanistic models for scenario testing is a logical next step in further evaluating ecosystem vulnerabilities to fire–human–climate feedbacks and anticipating where these feedbacks are likely to be most pronounced.


This research was supported in part by National Science Foundation Grants OISE-0966472 (D.M., P.H., C.W., T.V., B.M.), BCS-1024413 and BCS-0956552 (D.M., C.W.), Australian Research Council grant DP110101950 (D.B., T.V., C.W.), New Zealand Foundation for Research Science and Technology funding (J.W., M.M.) and Royal Society of New Zealand Marsden funding (M.M., J.W., C.W.). We thank Patrick Bartlein for the suggestion of plotting sites in climate space and helping access the NPP dataset, and Mitch Power for providing access to Version 2 of the Global Charcoal Database. We also appreciate comments from Jed Kaplan and two anonymous referees.

  • Editor: Greg Jordan

David McWethy is currently an assistant research professor at Montana State University, Bozeman, USA. His research focuses on how past and present human and natural disturbances shape vegetation and influence the structure and function of ecosystems. McWethy and contributing authors are currently collaborating on research efforts aimed at better understanding the causes and consequences of fire in the past, present and future (http://www.wildfirepire.org/).