• broadleaf;
  • Canada;
  • charcoal;
  • drought;
  • forest fires;
  • multivariate adaptive regression splines;
  • needleleaf;
  • pollen


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results and Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information
  • Strategic introduction of less flammable broadleaf vegetation into landscapes was suggested as a management strategy for decreasing the risk of boreal wildfires projected under climatic change. However, the realization and strength of this offsetting effect in an actual environment remain to be demonstrated.
  • Here we combined paleoecological data, global climate models and wildfire modelling to assess regional fire frequency (RegFF, i.e. the number of fires through time) in boreal forests as it relates to tree species composition and climate over millennial time-scales.
  • Lacustrine charcoals from northern landscapes of eastern boreal Canada indicate that RegFF during the mid-Holocene (6000–3000 yr ago) was significantly higher than pre-industrial RegFF (ad c. 1750). In southern landscapes, RegFF was not significantly higher than the pre-industrial RegFF in spite of the declining drought severity. The modelling experiment indicates that the high fire risk brought about by a warmer and drier climate in the south during the mid-Holocene was offset by a higher broadleaf component.
  • Our data highlight an important function for broadleaf vegetation in determining boreal RegFF in a warmer climate. We estimate that its feedback may be large enough to offset the projected climate change impacts on drought conditions.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results and Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

The patterns of and controls on wildfire behaviour have interested ecologists and geophysical scientists for more than a century (Bell, 1889), and extensive bodies of work have been produced about the climatic controls on wildfire ignition, propagation and severity (Campbell & Flannigan, 2000; Turetsky et al., 2011; Zumbrunnen et al., 2011). The processes governing wildfire behaviour operate at several time-scales (e.g. days, seasons and decades) and are influenced by weather, climate and other environmental factors such as temperature, precipitation, wind, and the structure and composition of forests. The concentrated activity in wildfire science is worthy of its importance – the anticipated increase in global wildfire activity resulting from human-caused climatic change is a threat to communities living at wildland–urban interfaces world-wide and to the equilibrium of the global carbon cycle (Kurz et al., 2008; Flannigan et al., 2009; Westerling et al., 2011). In circumboreal forests, climatic change will probably act upon fuels through long-term increases in summer evapotranspiration and increased frequency of extreme drought years as a result of more persistent and frequent blocking high-pressure systems. Earlier arrival of spring, longer summer droughts and more frequent ignitions could also expose forests to higher wildfire activity (Wotton et al., 2010).

Manipulation of vegetation composition and stand structure has been proposed as a strategy for offsetting climatic change impacts on wildfires (Hirsch et al., 2004; Krawchuk & Cumming, 2011; Terrier et al., 2013). Broadleaf deciduous stands are characterized by higher leaf moisture loading and lower flammability and rate of wildfire ignition and initiation than needleleaf evergreen stands (Päätalo, 1998; Campbell & Flannigan, 2000; Hély et al., 2001). Therefore, their introduction into dense needleleaf evergreen landscapes as strategic barriers could decrease the intensity and rate of spread of wildfires, improving suppression effectiveness, and reducing wildfire impacts (Amiro et al., 2001; Hirsch et al., 2004). Considerable portions of boreal forests are currently being harvested and there may be opportunities for using planned manipulation of vegetation for management of future wildfire risks (Hirsch et al., 2004). The concept has a long history, and its potential effect has been demonstrated through model simulation experiments (Hirsch et al., 2004; Krawchuk & Cumming, 2011). Nevertheless, determining the efficiency of planned manipulation of vegetation with respect to wildfire behaviour at the landscape scale is a daunting task because ecological processes resulting from stand dynamics (e.g. canopy closure, mortality and species turnover) succeed one another over many decades, and the biotic feedback from these could be confounded by other factors that influence wildfire activity, namely increasing land uses and human ignition (Niklasson & Granström, 2000; Zumbrunnen et al., 2011), active wildfire suppression efforts (Woolford et al., 2010), and episodic shifts in drought regimes as a result of oceanic forcing (Shabbar et al., 2011). Analyses of ecological features and feedback processes (climate and vegetation) in paleoecological records may provide significant insights for future wildfire management policies, particularly in terms of the magnitude of treatments required for an effect on wildfires over time (Willis & Birks, 2006; Gavin et al., 2007; Higuera et al., 2009; Marlon et al., 2012).

Here we provide an assessment of the response of boreal wildfire activity to changes in vegetation as well as the strength of vegetation feedback to limit or amplify climatic change impacts on wildfires. This assessment was carried out by integrating into a wildfire modelling scheme information about millennial-scale changes in wildfire activity reconstructed from analysis of charred particles accumulated in lake sediments, climate simulated by general circulation models (GCMs), and vegetation changes inferred from pollen analysis. We tested the hypothesis that increasing wildfire risks in needleleaf boreal forests brought about by more wildfire-prone climatic conditions may be offset by an increasing broadleaf component in landscapes.

Materials and Methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results and Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Reconstruction of regional fire frequencies

Variations in charcoal accumulation rate or influx (sedimentary charcoal load per time unit; e.g. mm−2 cm−2 yr−1) provide a continuous record of local wildfire frequency in a point-based manner (i.e. average fire number per time unit at a given point) within the sampling resolution of the sediment record (Clark, 1990) that may be used in cross-comparison with, for instance, model simulations of past climate and pollen-based vegetation reconstructions. Here, fire events that occurred during the Holocene were reconstructed using charred particles extracted from the sediments of 11 small lakes from the transition zone of the boreal mixedwood and the dense needleleaf forests of eastern boreal Canada (Fig. 1, Table 1). At all sites it was possible to reconstruct wildfire frequency since the onset of sediments that followed the retreat of the Laurentide Ice Sheet in eastern North America. The investigated forests remained largely unaffected by humans until European settlement in the early 20th Century. Before that time, there is no record of the specific influence of Amerindian practices on fire activity for the region under study, but it is generally assumed that, in the boreal forest, Amerindians were using fire for clearing land around campsites and trails (Patterson & Sassaman, 1988). Fires were generally set during periods of low fire susceptibility and consequently were of low intensity and small in size (Lewis, 1982). All sampled lakes are located within the Central Canadian Shield Forest ecoregion (Olson et al., 2001).

Table 1. Main features of studied lakes
Lake nameLac PessièreLac aux CèdresLac aux GeaisLac ProfondLac RaynaldLac à la Loutre
Vegetation zoneNeedleleafNeedleleafNeedleleafNeedleleafNeedleleafNeedleleaf
DataCharcoal and pollenCharcoal and pollenCharcoalCharcoalCharcoalCharcoal
Elevation (m asl)283307278274279270
Lake surface (ha)
Water depth (m)10.151610.15> 2010.2810.63
Length of organic core (cm)603573603223472227
Median deposition time (yr cm−1)1413.
ReferenceCarcaillet et al. (2001)Carcaillet et al. (2006)Ali et al. (2009)Ali et al. (2009)Ali et al. (2009)Ali et al. (2009)
Lake nameLac Jack PineLac HuardLac ChristelleLac FrancisLac Pas-de-Fond
Vegetation zoneBoreal mixedwoodBoreal mixedwoodBoreal mixedwoodBoreal mixedwoodBoreal mixedwood
DataCharcoalCharcoal and pollenCharcoal and pollenCharcoal and pollenCharcoal and pollen
Elevation (m asl)341346265305290
Lake surface (ha)
Water depth (m)
Length of organic core (cm)338712427302368
Median deposition time (yr cm−1)
ReferenceA. A. Ali (unpublished)Genries et al. (2012)Genries et al. (2012)Carcaillet et al. (2001)Carcaillet et al. (2001)

Figure 1. Location of the 11 sampled lakes in Canada. Sampled lakes located north of the modern transition zone of the boreal mixedwood and dense needleleaf forests are: 1, Lac Pessière; 2, Lac aux Cèdres; 3, Lac aux Geais; 4, Lac Profond; 5, Lac Raynald; and 6, Lac à la Loutre; sampled lakes located south of the transition zone are: 7, Lake Jack Pine; 8, Lac Huard; 9, Lac Christelle; 10, Lac Francis; and 11, Lac Pas-de-Fond. Also shown are forest cover types obtained from Natural Resources Canada (2008) 250-m resolution land cover classes. The information relating to vegetation openness was discarded. The dimensionless scale ranges from needleleaf dominance (dark grey) to broadleaf dominance (light grey).

Download figure to PowerPoint

Standard methods were used for charred particle extraction, dating procedures, and reconstruction of fire events (Higuera et al., 2008). Briefly, charred areas (CHARa, in cm2) were interpolated to constant time steps (Cinterpolated), corresponding to the median temporal resolution of each record (Table 1). Low-frequency variations in CHARa, namely Cbackground, represent changes in charcoal production, transport, sedimentation, mixing, and sampling. We therefore decomposed CHARa into background (Cbackground) and peak (Cpeak) components using a locally defined threshold that identifies charcoal peaks probably related to the occurrence of one or more local fires (i.e. ‘fire events’ within c. 1 km). The locally weighted regression was applied with a 500-yr-wide window that maximized a signal-to-noise (peak-to-background) index and the goodness of fit between the empirical and modelled Cbackground distributions (Kelly et al., 2011). The residual series related to peaks was obtained by subtraction (i.e. Cpeak = CinterpolatedCbackground).

Consistent with theoretical evidence (Higuera et al., 2007) and empirical studies (Whitlock & Millspaugh, 1996; Carcaillet et al., 2001; Higuera et al., 2009), we assumed in a second step that Cpeak was composed of two subpopulations, namely Cnoise, representing variability in sediment mixing, sampling, and analytical and naturally occurring noise, and Cfire, representing significant peaks of charcoal input from local fires. For each peak, we used a Gaussian mixture model to identify the Cnoise distribution. We considered the 99th percentile of the Cnoise distribution as a possible threshold separating samples into ‘fire’ and ‘nonfire’ events; between-record differences were similar using other threshold criteria. We did not screen peaks based on the original charcoal counts of each peak, as in Higuera et al. (2009). All charcoal time-series analyses were performed using the program CharAnalysis (P.E. Higuera; freely available at

To document past millennial to centennial time-scale fluctuations in regional wildfire activity, sites were grouped into northern (hereafter ‘North’) and southern (‘South’) landscapes with respect to their location along the gradient of transition from boreal mixedwood to dense needleleaf vegetation zones (Fig. 1). This grouping was historically supported by pollen grain concentrations of major tree species and plant macroremain data (Terasmae & Anderson, 1970; Vincent, 1973; Richard, 1980; Liu, 1990; Gajewski et al., 1993; Ali et al., 2008; Genries et al., 2012). The vegetation composition in the northern landscapes did not change significantly over the last 6000 yr. In southern landscapes, a reduction in the proportion of broadleaf taxa since 1200 calibrated years before present (hereafter bp) tended to reduce the differences between the boreal mixedwood and needleleaf forests (Carcaillet et al., 2010). From fire event dates extracted from Cfire over past millennia, we computed regional fire frequencies (RegFFs) using a kernel-density function (Mudelsee, 2002) that allowed a detailed inspection of time-dependent event frequencies (Mudelsee et al., 2004). RegFF can be viewed as an arithmetic average of all fire frequencies determined in a designated area during a specified time period, and is herein expressed in n fires 1000 yr−1. We used a Gaussian kernel, K, to weigh observed fire event dates, T(i), i,…, N (where N is the total number of events), and calculated the regional frequency, RegFF, at each time t as:

  • display math(Eqn 1)

(n(t), the total number of sampled cores at time t.) Selection of the bandwidth (= 500 yr) was guided by cross-validation aimed at finding a compromise between large variance and small bias (which occurs under shorter h) and small variance and large bias (longer h). We assessed the significance of changes with the help of bootstrap confidence intervals (CIs) computed from confidence bands (90%) around RegFF (Mudelsee et al., 2004).

Drought severity

We used paleoclimatic simulations provided by the UK Universities Global Atmospheric Modelling Programme to develop a mechanistic understanding of the climatic variations associated with the reconstructed paleofire regime. These simulations were performed with the Hadley Centre climate model (HadCM3; Singarayer & Valdes, 2010), which is a state-of-the-art global climate model (GCM) used in both the third and fourth assessment reports of the Intergovernmental Panel on Climate Change (2001, 2007). The GCM is a three-dimensional time-dependent numerical representation of the atmosphere, oceans and sea ice and their phenomena over the entire Earth, using the equations of motion and including radiation, photochemistry, and the transfer of heat and water vapour. The HadCM3 GCM simulations used in the present study consist of climatic averages at 1000-yr intervals (i.e. maximum temporal resolution available) covering the last 120 000 yr at a spatial resolution of 2.5° in latitude by 3.75° in longitude. These simulations include forcing from a prolonged presence of the residual Laurentide Ice Sheet in eastern North America and an improved way of handling the isostatic rebound that was previously less effective (Singarayer & Valdes, 2010). For each millennium interval, anomalies for air temperature (the difference between a given millennia and the pre-industrial (ad c. 1750) period) and precipitation (the percentage of change between a given millennia and the pre-industrial period) were computed. A downscaling method was used in which means of HadCM3 GCM anomalies of temperature and precipitation were applied to Climate Research Unit spatial grids TS 3.1 (period ad 1901–2008; Mitchell & Jones, 2005) over an area compatible for comparison with our RegFF reconstructions (48.5°–51.5°N and 86.5°–78.0°W, for a total of 126 CRU pixels encompassing nine HadCM3 pixels). The produced time-series of monthly temperature and precipitation (108-yr monthly time-series for each millennium) were then used to compute the monthly Drought Code, which is a monthly adaptation of the daily Drought Code (DC) index of the Canadian Forest Fire Weather Index System (Girardin & Wotton, 2009). The DC is used in several countries by fire agencies to predict the risk of fire ignition based on weather conditions (de Groot et al., 2007). It represents the net effect of changes in evapotranspiration and precipitation on cumulative moisture depletion in the organic matter of the deep humus layer (18 cm thick, 25 kg m−2 dry weight, and 138.9 kg m−3 bulk density). The DC (and its monthly version) is significantly correlated with wildfire activity in our study area (Balshi et al., 2009; Girardin et al., 2009). We nonetheless feel the need to specify that the DC might not apply to locations where there is a distinctly thin or absent deep duff layer. Calculation started every simulated year in April and ended in October (Van Wagner, 1987; Terrier et al., 2013). An overwintering adjustment was included in the calculation, such that the starting values in spring depend on antecedent autumn drought severity and winter precipitation (Girardin & Wotton, 2009). Medians of April to October monthly Drought Code values were computed for each year, and then across the 108 yr and 126 CRU pixels, and at each millennium, to produce a multi-millennia seasonal Drought Code (SDC) severity time-series. Confidence intervals (90%) for uncertainty in the regional climate history were built by bootstrap resampling of single-HadCM3 pixel SDC anomaly time-series.

Monthly temperature and precipitation data collected from eight GCMs and four scenarios of greenhouse gas (GHG) emissions were used for projection of changes in SDC over the next century (Supporting Information Table S1). The objective was to assess whether the magnitude of past simulated drought conditions is analogue to plausible scenarios expected for the late 21st Century. GCM selections were made according to the availability of monthly means of daily maximum temperature outputs necessary for simulation of the SDC. For the present study, GCM data were collected from four to six cells, depending on model resolution, and for the interval 1961–2100. To account for differences between the CRU data and the GCM projections, the monthly simulations were adjusted relative to the absolute difference from the 1961–1999 monthly means of CRU data (Balshi et al., 2009). A correction was also applied to the interannual variability by changing the width of the distributions so that mean monthly GCM projections and CRU data had equal standard deviations over their common period 1961–1999 (details in Bergeron et al., 2010). These anomaly correction methods were intended to capture the future changes in the frequency of precipitation events that could cause the year-to-year variability in the SDC to also change significantly. These anomaly correction methods were not applied to the HadCM3 paleoclimatic projections because the necessary information was not available.


Data on modern forest composition were extracted from a database of temporary sample plots established by the ministère des Ressources naturelles et de la Faune for the province of Quebec (third and fourth forest inventory programs). Aboveground biomass was estimated for each stem within a plot using measured diameter at breast height and the species-specific tree biomass equations of Lambert et al. (2005). Values were summed to obtain estimates of species plot-level aboveground biomass and averaged across the vegetation zones defined by Terrier et al. (2013).

Past changes in forest composition were documented by summarizing the timing and magnitude of palynological changes at six sampled lakes (two from the northern landscapes and four from the southern ones; Table 1 and Fig. 1), and by examining the differences between the sites through time. Pollen percentages were calculated for 100-yr time windows corresponding to the median resolution of the 11 pollen records. Average pollen percentages per flammable needleleaf species (Pinus banksiana Lamb. and Picea mariana (Miller) BSP) were compared with average percentages of less flammable broadleaf species (Populus tremuloides Michx. and Betula sp.) and a needleleaf index was computed, which is equivalent here to the needleleaf percentage. Our analysis of the pollen data deals mainly with qualitative interpretation. No attempt was made to calibrate the needleleaf index on numerical data sets of modern forest attributes because of the low pollen-site replication in the forests under study.

Fire modelling framework

To develop projections of past and future wildfires that take into account regional climate and tree composition changes, we used empirical models (Terrier et al., 2013) describing the distribution of wildfire occurrences in eastern Canada as a function of sets of wildfire bioclimatic zones determined from modern fire weather (FW) and tree species composition (TreeComp). The wildfire occurrence models were formulated by Terrier et al. (2013) using piecewise regression models:

  • display math(Eqn 2)

In Eqn (Eqn 2), FireOcc is the number of lightning-caused fires above a specified size-threshold per year per 1000 km2 for a period j, c1 and c2 correspond to constants, and BF1 and BF2 are basis functions for nonlinear interactions between FireOcc and FW and TreeComp variables. Lightning is the primary source of wildfire ignition in boreal North America and usually results in fires that account for the majority of the area burned (Stocks et al., 2003). FW is defined using fuel moisture codes at different forest floor levels computed from April to October, and averaged over 10-yr periods. TreeComp takes the form of binary variables to indicate the presence of a given vegetation category. A parsimonious model for large fires (size > 200 ha; Stocks et al., 2003) with mean SDC as a predictor variable is shown in Fig. 2 and used in this study. Therein, FireOcc of size > 200 ha in boreal mixedwood landscapes (i.e. with vegetation attributes described by Fig. 2b) progressively increases as mean SDC increases above 125 units (Fig. 2d). The presence of a tree composition dominated by needleleaf species Pmariana (Fig. 2a) contributes significantly to increasing the FireOcc quantity; a compositional group dominated by nonboreal broadleaf species (e.g. sugar maple, Acer saccharum Marshall; Fig. 2c) contributes significantly to lowering it (zero slope model; Fig. 2). An application of the FireOcc > 200 ha model to a gridded climatology data set suggested that the model was adequate for projecting patterns of wildfire occurrences across the boreal mixedwood and needleleaf forests under study (Fig. S1). In this work, we project past FireOcc > 200 ha using HadCM3 GCM median SDC simulations and pollen-based vegetation information based on the needleleaf index as input data for the model. For the future, projections were made using the multiple GCM simulations and scenarios with and without vegetation changes. For this analysis, 10-yr SDCs computed from the GCM simulations were used as inputs into the FireOcc model. The change from needleleaf (vegetation attributes described by Fig. 2a) to boreal mixedwood forests (vegetation attributes described by Fig. 2b) was arbitrarily set at ad 2040.


Figure 2. Modern vegetation attributes in the studied forests and their modelled effect on wildfire activity. (a–c) Relative contribution of dominant tree species to total stand biomass in dense needleleaf, boreal mixedwood and nonboreal broadleaf forests in Canada; statistics were obtained from analysis of Quebec's temporary sample plots from the third and fourth inventories (= 4665, 10 047 and 20 937 plots, respectively). (d) Empirical model for the occurrence of large wildfires (FireOcc) as a function of mean seasonal Drought Code (SDC) severity and vegetation composition (refer to Terrier et al., 2013).

Download figure to PowerPoint

Statistical analyses

Relationships among RegFF, SDC and FireOcc were assessed using Pearson's r coefficient (von Storch & Zwiers, 1999; Systat Software Inc, 2004). For these analyses, RegFF reconstructions were downsampled to the time resolution of HadCM3 simulations (i.e. 1000-yr intervals). Linear relationships between two variables were visually inspected using scatter-plots. The statistical significance of correlations was determined using bootstrap resampling (von Storch & Zwiers, 1999). When the confidence interval contains zero, the hypothesis of ‘no correlation’ cannot be rejected at the 90% level.

Significant differences in needleleaf index between pollen series of northern and southern landscapes were analysed using a moving two-sample Student's t-test (two-sided and equal variance; von Storch & Zwiers, 1999). Cubic smoothing splines were fitted to the individual series before conducting the Student's t-test analysis; the smoothing was automatically determined using a cross-validation procedure (AutoSignal, 1999). We are seeking to disprove the null hypothesis of equal means of the needleleaf index when the P-value is lower than 0.05.

The significance of changes in FireOcc projected using the multiple GCM simulations, and under scenarios with and without vegetation changes, was tested using the two-sample Student's t-test (one-sided) comparing the 1961–1999 and 2041–2100 intervals (= 10 decades). A Holm–Bonferroni correction was applied to counteract the problem of multiple comparisons (Holm, 1979). One seeks to disprove the null hypothesis asserting that 2041–2100 FireOcc is not greater than 1961–1999 FireOcc when the P-value is lower than 0.05/(– 1), where m is the number of P-values being tested at a given iteration.

Results and Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results and Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Climate controls on boreal wildfires

Past millennial to centennial time-scale fluctuations in wildfire activity are herein documented by two composite reconstructions of RegFF (north and south) covering the mid-Holocene period to the late-Holocene pre-industrial period (Fig. 3a,b). Analysis of sensitivity to site selection and data treatment confirmed the robustness of the reconstructions for the period from 6000 bp to present (results not shown). Unstable RegFFs in both reconstructions before 6000 bp (particularly evident in the southern reconstruction) were mainly associated with the successive inclusion of sampling sites in association with the heterogeneous deglaciation in North America (Dyke, 2004, 2005). Therefore, the period before 6000 bp was discarded from further analyses. The two RegFF reconstructions were uncorrelated with one another during the period from 6000 bp to the pre-industrial era (= −0.15; 90% bootstrap confidence interval (90% CI) −0.78, 0.59; = 7 millennia), implying different temporal wildfire trajectories and probably different controls on wildfires (Bremond et al., 2010). In northern landscapes, wildfires were frequent c. 6000 and 2000 bp with a maximum of RegFF estimated at 7.2 wildfires per millennium at 2500 bp (Fig. 3a). The period around 2000 bp marked the onset of a gradual decline in North-RegFF towards a minimum value attained during the pre-industrial period at 3.5 wildfires per millennium (Fig. 3a). These changes between the mid-Holocene period and the pre-industrial period are significant according to the bootstrap resampling of wildfire event dates (with the exception of the period around 4000 bp which is not statistically different from the pre-industrial period). By contrast, the South-RegFF remained constant during the last 6000 yr, with values fluctuating at c. 4–6 wildfires per millennium (Fig. 3b). Estimates of pre-industrial RegFFs are equal in both reconstructions, as can be judged from the overlapping 90% CIs around RegFF (Fig. 3a,b). They are also in the range of plausible values in these forests, as documented by the stand-replacing fire history studies (Fig. S2).


Figure 3. Past fluctuations in wildfire activity in the transition zone of the dense needleleaf and boreal mixedwood forests of eastern Canada, and their associated forcings. (a, b) Northern and southern regions' fire frequencies. Shaded areas denote 90% bootstrap confidence intervals (CIs) for uncertainty in fire frequencies. Horizontal bars: the period covered by fire data for each individual sampled lake. (c) June–August solar insolation computed at 45°N (Berger & Loutre, 1991). (d) Median seasonal Drought Code (SDC) severity computed from simulated climate outputs of the Hadley Centre climate model (HadCM3) with 90% CI. A high value indicates a high seasonal fire danger. (e) Needleleaf index inferred from the mean proportions of total pollen counts of black spruce and jack pine. A high percentage indicates a dominance of needleleaf over broadleaf species. Significant differences (< 0.05) in the needleleaf index between northern and southern landscapes are indicated by the thick horizontal red line. By definition, the period of 0 calibrated years before present (cal yr bp) is equivalent to the pre-industrial period (ad c. 1750).

Download figure to PowerPoint

Changes in regional wildfire frequencies can reflect a millennial-scale climatic control on wildfire danger. In high latitudes of the Northern Hemisphere, the input of summer solar irradiance has declined over the last 6000 yr as a result of changes in the Earth's axial tilt (Berger & Loutre, 1991). The period c. 4000 bp was a millennial-scale transitional period between the mid-Holocene, characterized by very high positive anomalies in summer solar irradiance, and the late-Holocene, marked by an ongoing decrease in solar irradiance up until today (Fig. 3c). Model simulations and reconstructions of past temperature changes indicate a millennial-scale summer cooling over the last 2000 yr as a direct response to the solar forcing (Kaufman et al., 2009; Viau & Gajewski, 2009; Marcott et al., 2013). Recent studies suggested that declining incoming solar irradiance has had an impact on boreal wildfire danger and activity (Hély et al., 2010; de Lafontaine & Payette, 2011). Accordingly, the Holocene median SDC severity assessed from HadCM3 GCM simulations decreased through the last 2000 yr, falling from 152 units at 3000 bp to 139 units during the pre-industrial period (Fig. 3d). Low North-RegFFs recorded during the mid- and the late-Holocene correspond to low wildfire season severities and vice versa. Altogether, the North-RegFF reconstruction is correlated with simulated median SDC from 6000 to 0 bp (= 0.83 with 90% CI 0.74, 0.98; = 7). Such close similarities between climate controls and RegFF are not distinguished when analysing the South-RegFF (= −0.23 with 90% CI −0.86, 0.52; = 7), suggesting another controlling factor for wildfire activity in southern landscapes over recent millennia. Below, we provide an explanation for the diverging North- and South-RegFF trajectories from the mid- to the late-Holocene that involves an offsetting effect on the climate forcing brought on by regional vegetation changes.

Vegetation feedback

Important vegetation modifications in eastern North America marked the transition from the mid- to the late-Holocene. Noticeable through investigations of pollen records was a southerly displacement of the transition zone of the mixedwood and needleleaf forests in association with cooler climatic conditions (Liu, 1990; Dyke, 2005; Carcaillet et al., 2010). We examined the potential links between changes in RegFF and vegetation inferred from published sedimentary pollen data sets (Table 1). Pollen assemblages vary according to the vegetation composition and structure surrounding the study sites (Jackson & Lyford, 1999; Broström et al., 2005), and modification of these assemblages can occur with canopy disturbances and climatic changes (Richard, 1980; Koff et al., 2000). In eastern boreal North America, a post-disturbance transition from broadleaf to needleleaf species can occur under abundant needleleaf regeneration. The relative dominance of needleleaf species can be greater in stands under long fire-return intervals, or lower under dry climatic conditions (Bergeron et al., in press). Given that the 11 sites are within a transition zone of two forest types and that species' abundance at each site is dynamically related to the changing climate (Carcaillet et al., 2001), we expected some changes in sites belonging to forest types over past millennia. This was seemingly the case. Modern vegetation composition in northern landscapes is dominated by black spruce (P. mariana; Fig. 2). This dominance was already set some 6000 yr ago according to pollen analysis and persisted throughout the intervening millennia (Fig. 3e). By contrast, a gradual development towards flammable needleleaf species such as black spruce and jack pine (P. banksiana) was recorded in southern landscapes c. 2000 bp (Fig. 3e). This change came at the cost of a decrease in the abundance of broadleaf species, that is, birches (Betula papyrifera and B. alleghaniensis), grey alder (Alnus incana) and aspen (P. tremuloides; Fig. 3e). Therefore, the pre-industrial composition of the southern sites is much closer to the dense needleleaf forest type than it was some 6000–3000 yr ago (Fig. 3e; Carcaillet et al., 2010).

A potential explanation for the divergence between the two RegFF trajectories may lie in the changing vegetation. The declining risks brought about by less wildfire-prone climatic conditions (Fig. 3d) may well have been offset by an increasing needleleaf component in southern landscapes some 2000 yr ago (Fig. 3e). We tested this hypothesis by integrating the HadCM3 GCM median SDC simulations and pollen-based vegetation information into the fire model for projection of FireOcc of size > 200 ha. Model projections for northern landscapes suggest a decline in FireOcc > 200 ha, with the median FireOcc of the last two millennia being c. 20% lower than the median of 6000–2000 bp (Fig. 4). This difference closely matches the decline of c. 25% seen in North-RegFF over the same periods (Fig. 3a). The opposite pattern is found in southern landscapes. Therein model projections indicate higher levels of FireOcc during the last 2000 yr relative to the mid-Holocene period (Fig. 4b), which is coherent with the wildfire trajectory deduced from the South-RegFF observations (Fig. 3b; = 0.80 with 90% CI 0.50, 0.99; = 7). This stable state in FireOcc occurs because the induced shift in vegetation from boreal mixedwood to dense needleleaf landscapes at 2000 bp was sufficient for offsetting the FireOcc decline brought on by a lowering of the median SDC (Fig. 3d). If these landscapes had remained in a boreal mixedwood state as they were some 5000 yr ago, they would have undergone a significant decline in large wildfires towards levels approximating 0.07 fires per year per 1000 km (Fig. 4b). But, as seen with the South-RegFF (Fig. 3b), this was not observed. Hence, the modelling results suggest that biotic feedback arising from vegetation changes was strong enough to modulate past climatic change influences on large wildfire activity.


Figure 4. Projected changes in the occurrence of large wildfires (FireOcc) in (a) northern and (b) southern landscapes from 8000 to 0 bp (calibrated years before present) simulated using climatic data from the HadCM3 general circulation model (GCM) and vegetation changes deduced from pollen analyses (see Fig. 3d–e). The shaded area is the 90% CI. In (a), a fixed vegetation composition dominated by needleleaf forests was set throughout the entire period. In (b), vegetation was manipulated (Vegetation + climate) with a shift from a boreal mixedwood to a needleleaf-dominated forest at 2000 bp. The status quo scenario of no vegetation change (Climate) is also shown in (b).

Download figure to PowerPoint


Figure 5. Projected changes in the occurrence of large wildfires (FireOcc) in the modern northern needleleaf landscape over the 21st Century simulated from an ensemble of eight global climate models forced by various scenarios for greenhouse gas emissions. A 90% bootstrap confidence interval (CI) for the ensemble median is shown (red shading). In scenario (a), vegetation was set with a fixed needleleaf forest throughout all periods. In scenario (b), vegetation was manipulated with a shift from a needleleaf to a boreal mixedwood forest at ad 2040. In scenario (a), mean FireOcc calculated over the interval 2041–2100 is significantly greater than mean FireOcc calculated over the interval 1961–1999 (one-sided Student's t-test < 0.05). In scenario (b), 2041–2100 FireOcc is not significantly greater than 1961–1999 FireOcc.

Download figure to PowerPoint

Our results indicate that modification of vegetation composition has contributed to a lower wildfire probability in a warmer climate. Nonetheless, uncertainty remains about the efficiency of such effects in the 21st Century if GHG levels and climate conditions become significantly different from historical levels. Under excessive droughts, the biotic feedback might not be strong enough to limit future wildfire activity. To address this question, we used projected drought from an ensemble of eight global climate models forced by various scenarios of GHG emissions (Table S1) as input into the FireOcc model. For this experiment, we induced a vegetation shift from dense needleleaf to boreal mixedwood landscapes at ad 2040 and compared the results with a status quo scenario (Fig. 5). Results indicated that median SDC at the end of the 21st Century could reach levels similar to those simulated from the HadCM3 GCM during the mid-Holocene (i.e. c. 30 SDC units above the ad 1961–1999 baseline). Keeping vegetation composition in a needleleaf state produced a doubling in FireOcc for the late 21st Century compared with ad 1961–1999 levels (ensemble median; Fig. 5a). Increases in FireOcc were significant in seven out of 21 experiments (one-sided Student's t-test with correction for multiple comparisons). Inducing a vegetation change to boreal mixedwood landscapes (Fig. 5b) was effective in offsetting climatic change impacts on FireOcc in six out of these seven experiments (the effect failed for MIROC3.2 medres A1B). Altogether, the negative vegetation feedback was sufficient to limit the rise in FireOcc calculated over the interval 2041–2100 to 30% of the level projected for the baseline period (ensemble median; Fig. 5b) and well within the range of historical variations (i.e. < 0.20 fires 1000 km−2 yr−1; Fig. 4b).


With the urgent necessity for strategic decisions to cope with the increasing threat future wildfires pose, there is a requirement for sound assessments of the costs and benefits of planned manipulation of vegetation in wildland–urban interfaces of global boreal forests. The use of paleoecological data and GCM simulations in a wildfire model for testing the sensitivity of biotic feedback is a significant contribution to this objective. Manipulative vegetation treatments have been suggested as potential climate-change adaptation strategies in boreal forests, mostly on the basis of simulation experiments (Hirsch et al., 2004; Krawchuk & Cumming, 2011; Terrier et al., 2013). Our assessment of millennial-scale variations of seasonal wildfire danger, vegetation flammability, and fire activity suggests that feedback effects arising from vegetation changes are large enough to offset climatic change impacts on fire danger. Our quantitative results are subject to uncertainties, including those associated with the increasing impact of human ignitions and differences in fire seasonality (Wotton et al., 2010). However, our main finding is robust: in spite of the warm climate some 6000–3000 yr ago in eastern Canada, RegFF in southern landscapes was not significantly higher than the pre-industrial RegFF and this was a result of the lower landscape proportion of flammable needleleaf species. Future climate warming will lead to increases in the proportion of hardwood forests in both southern and northern boreal landscapes (McKenney et al., 2011; Terrier et al., 2013). However, this effect will spread over long periods because of low species migration and dispersal rates. If lower proportions of flammable needleleaf species in landscapes are a natural feature of a warmer climate (Carcaillet et al., 2010; Terrier et al., 2013), then in the short term forest management should gradually provide more space for approaches promoting broadleaf and boreal mixedwood forests. This would be a way to reduce wildfire risk during the transition to a new vegetation equilibrium. This consideration is important as it could make vegetation changes socially and environmentally acceptable. There are also many other benefits brought about by the increasing dominance of broadleaf species in landscapes. This may include the higher albedo and summer evapotranspiration from deciduous trees, which would cool and counteract regional warming (Rogers et al., 2013), and the increase in the resilience of forests to climatic changes (Drobyshev et al., 2013). Further studies should address questions dealing with the magnitude of vegetation composition changes needed to attain the wildfire management objectives (e.g. relative abundance of species and size of the managed areas), the existence of potential constraints on the success of species establishment (e.g. nutrient limitations), and how the treatments would interfere with other values and concerns of the forest sectors (e.g. forest conservation and timber supply).


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results and Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Financial support was provided by Canadian Forest Service Funds to M.P.G., the programme PALEO2-BOREOFIRE to A.A.A., the Natural Sciences and Engineering Research Council of Canada to Y.B. and A.A.A., and the contribution from the École Pratique des Hautes Etudes to C.C. The research was carried out within the framework of the International Associated Laboratory (LIA France-Canada). We thank X. J. Guo and D. Gervais for their assistance with the analysis of the HadCM3 data, M. D. Flannigan and two anonymous reviewers for comments on an earlier version of this manuscript, and W. Finsinger for contributing to ideas.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results and Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information
  • Ali AA, Asselin H, Larouche AC, Bergeron Y, Carcaillet C, Richard PJH. 2008. Changes in fire regime explain the Holocene rise and fall of Abies balsamea in the coniferous forests of western Québec, Canada. The Holocene 18: 693703.
  • Ali AA, Carcaillet C, Bergeron Y. 2009. Long-term fire frequency variability in the eastern Canadian boreal forest: the influences of climate vs. local factors. Global Change Biology 15: 12301241.
  • Amiro BD, Stocks BJ, Alexander ME, Flannigan MD, Wotton BM. 2001. Fire, climate change, carbon and fuel management in the Canadian Boreal forest. International Journal of Wildland Fire 10: 405413.
  • AutoSignal. 1999. AutoSignal, version 1.5 for Windows. Mapleton, OR, USA: AISN Software.
  • Balshi MS, McGuire AD, Duffy P, Flannigan M, Walsh J, Melillo J. 2009. Assessing the response of area burned to changing climate in western boreal North America using a Multivariate Adaptive Regression Splines (MARS) approach. Global Change Biology 15: 578600.
  • Bell R. 1889. Forest fires in northern Canada. American Forestry Congress, Atlanta meeting. Washington, DC, USA: Gibson Brothers.
  • Berger A, Loutre MF. 1991. Insolation values for the climate of the last 10 million years. Quaternary Science Reviews 10: 297317.
  • Bergeron Y, Chen HYH, Kenkel NC, Leduc AL, Macdonald E. in press. Boreal mixedwood stand dynamics: ecological processes underlying multiple pathways. The Forestry Chronicle.
  • Bergeron Y, Cyr D, Girardin MP, Carcaillet C. 2010. Will climate change drive 21st century burn rates in Canadian boreal forest outside of its natural variability: collating global climate model experiments with sedimentary charcoal data. International Journal of Wildland Fire 19: 11271139.
  • Bremond L, Carcaillet C, Favier C, Ali AA, Paitre C, Bégin Y, Bergeron Y, Richard PJH. 2010. Effects of vegetation zones and climatic changes on fire-induced atmospheric carbon emissions: a model based on paleodata. International Journal of Wildland Fire 19: 10151025.
  • Broström A, Sugita S, Gaillard M-J, Pilesjö P. 2005. Estimating the spatial scale of pollen dispersal in the cultural landscape of southern Sweden. The Holocene 15: 252262.
  • Campbell ID, Flannigan MD. 2000. Long-term perspectives on fire-climate-vegetation relationships in the North American boreal forest. In: Kasischke ES, Stocks BJ, eds. Fire, climate change, and carbon cycling in the boreal forests. New York, NY, USA: Springer-Verlag, 151172.
  • Carcaillet C, Bergeron Y, Richard PJH, Fréchette B, Gauthier S, Prairie YT. 2001. Change of fire frequency in the eastern Canadian boreal forests during the Holocene: does vegetation composition or climate trigger the fire regime? Journal of Ecology 89: 930946.
  • Carcaillet C, Richard PJH, Asnong H, Capece L, Bergeron Y. 2006. Fire and soil erosion history in East Canadian boreal and temperate forests. Quaternary Science Reviews 25: 14891500.
  • Carcaillet C, Richard PJH, Bergeron Y, Fréchette B, Ali AA. 2010. Resilience of the boreal forest in response to Holocene fire-frequency changes assessed by pollen diversity and population dynamics. International Journal of Wildland Fire 19: 10261039.
  • Clark JS. 1990. Fire and climate change during the last 750 yr in northwestern Minnesota. Ecological Monographs 60: 135159.
  • Drobyshev Y, Gewehr S, Berninger F, Bergeron Y. 2013. Species specific growth responses of black spruce and trembling aspen may enhance resilience of boreal forest to climate change. Journal of Ecology 101: 231242.
  • Dyke AS. 2004. An outline of North American deglaciation with emphasis on central and northern Canada. Developments in Quaternary Sciences 2: 373424.
  • Dyke AS. 2005. Late Quaternary vegetation history of northern North America based on pollen, macrofossil, and faunal remains. Géographie Physique et Quaternaire 29: 211262.
  • Flannigan MD, Krawchuk MA, de Groot WJ, Wotton BM, Gowman LM. 2009. Implications of changing climate for global wildland fire. International Journal of Wildland Fire 18: 483507.
  • Gajewski K, Payette S, Ritchie JC. 1993. Holocene vegetation history at the boreal-forest – shrub-tundra transition in north-western Québec. Journal of Ecology 81: 433443.
  • Gavin DG, Hallett DJ, Hu FS, Lertzman KP, Prichard SJ, Brown KJ, Lynch JA, Bartlein PJ, Peterson DL. 2007. Forest fire and climate change in western North America: insights from sediment charcoal records. Frontiers in Ecology and the Environment 5: 499506.
  • Genries A, Finsinger W, Asnong H, Bergeron Y, Carcaillet C, Garneau M, Hély C, Ali AA. 2012. Local versus regional processes: can soil characteristics overcome climate and fire regimes by modifying vegetation trajectories? Journal of Quaternary Science 27: 745756.
  • Girardin MP, Ali AA, Carcaillet C, Mudelsee M, Drobyshev I, Hély C, Bergeron Y. 2009. Heterogeneous response of circumboreal wildfire risk to climate change since the early 1900s. Global Change Biology 15: 27512769.
  • Girardin MP, Wotton BM. 2009. Summer moisture and wildfire risks across Canada. Journal of Applied Meteorology and Climatology 48: 517533.
  • de Groot WJ, Field RD, Brady MA, Roswintiarti O, Mohamad M. 2007. Development of the Indonesian and Malaysian fire danger rating systems. Mitigation and Adaptation Strategies for Global Change 12: 165180.
  • Hély C, Flannigan M, Bergeron Y, McRae D. 2001. Role of vegetation and weather on fire behavior in the Canadian mixedwood boreal forest using two fire behavior prediction systems. Canadian Journal of Forest Research 31: 430441.
  • Hély C, Girardin MP, Ali AA, Carcaillet C, Brewer S, Bergeron Y. 2010. Eastern boreal North American wildfire risk of the past 7000 years: a model-data comparison. Geophysical Research Letters 37: L14709.
  • Higuera PE, Brubaker LB, Anderson PM, Brown TA, Kennedy AT, Hu FS. 2008. Frequent fires in ancient shrub tundra: implications of paleorecords for arctic environmental change. PLoS ONE 3: e0001744.
  • Higuera PE, Brubaker LB, Anderson PM, Hu FS, Brown TA. 2009. Vegetation mediated the impacts of postglacial climate change on fire regimes in the south-central Brooks Range, Alaska. Ecological Monographs 79: 201219.
  • Higuera PE, Peters ME, Brubaker LB, Gavin D. 2007. Understanding the origin and analysis of sediment charcoal records with a simulation model. Quaternary Science Reviews 26: 17901809.
  • Hirsch K, Kafka V, Todd B. 2004. Using forest management techniques to alter forest fuels and reduce wildfire size: an exploratory analysis. In: Engstrom RT, Galley KEM, de Groot WJ, eds. Fire in temperate, boreal, and montane ecosystems. Tallahassee, FL, USA: Tall Timber Research Station, 175184.
  • Holm S. 1979. A simple sequentially rejective multiple test procedure. Scandinavian Journal of Statistics 6: 6570.
  • Intergovernmental Panel on Climate Change. 2001. Climate Change, 2001. Impacts, adaptation and vulnerability – contribution of Working Group II to the Third Assessment Report of IPCC. Cambridge, UK: Cambridge University Press.
  • Intergovernmental Panel on Climate Change. 2007. Climate Change, 2007. The physical science basis – contribution of Working Group I to the Fourth Assessment Report of the IPCC. Cambridge, UK: Cambridge University Press.
  • Jackson ST, Lyford ME. 1999. Pollen dispersal models in Quaternary plant ecology: assumptions, parameters, and prescriptions. The Botanical Review 65: 3975.
  • Kaufman DS, Schneider DP, McKay NP, Ammann CM, Bradley RS, Briffa KR, Miller GH, Otto-Bliesner BL, Overpeck JT, Vinther BM et al. 2009. Recent warming reverses long-term Arctic cooling. Science 325: 12361239.
  • Kelly RF, Higuera PE, Barrett CM, Hu F. 2011. A signal-to-noise index to quantify the potential for peak detection in sediment-charcoal records. Quaternary Research 75: 1117.
  • Koff T, Punning JM, Kangur M. 2000. Impact of forest disturbance on the pollen influx in lake sediments during the last century. Review of Palaeobotany and Palynology 111: 1929.
  • Krawchuk MA, Cumming SG. 2011. Effects of biotic feedback and harvest management on boreal forest fire activity under climate change. Ecological Applications 21: 122136.
  • Kurz WA, Stinson G, Rampley GJ, Dymond CC, Neilson ET. 2008. Risk of natural disturbances makes future contribution of Canada's forests to the global carbon cycle highly uncertain. Proceedings of the National Academy of Sciences, USA 105: 15511555.
  • de Lafontaine G, Payette S. 2011. Long-term fire and forest history of subalpine balsam fir (Abies balsamea) and white spruce (Picea glauca) stands in eastern Canada inferred from soil charcoal analysis. The Holocene 22: 191201.
  • Lambert MC, Ung CH, Raulier F. 2005. Canadian national tree aboveground biomass equations. Canadian Journal of Forest Research 35: 19962018.
  • Lewis HT. 1982. A time of burning. Occasional publication number 17. Edmonton, AB, Canada: Boreal Institute for Northern Studies, University of Alberta.
  • Liu KB. 1990. Holocene paleoecology of the boreal forest and Great Lakes-St. Lawrence forest in northern Ontario. Ecological Monographs 60: 179212.
  • Marcott SA, Shakun JD, Clark PU, Mix AC. 2013. Reconstruction of regional and global temperature for the past 11,300 years. Science 339: 11981201.
  • Marlon JR, Bartlein JP, Gavin DG, Long CJ, Anderson RS, Briles CE, Brown KJ, Colombaroli D, Hallett DJ, Power MJ. 2012. Long-term perspective on wildfires in the western USA. Proceedings of the National Academy of Sciences, USA 109: E535E543.
  • McKenney DW, Pedlar JH, Rood RB, Price D. 2011. Revisiting projected shifts in the climate envelopes of North American trees using updated general circulation models. Global Change Biology 17: 27202730.
  • Mitchell TD, Jones PD. 2005. An improved method of constructing a database of monthly climate observations and associated high-resolution grids. International Journal of Climatology 25: 693712.
  • Mudelsee M. 2002. XTREND, a computer program for estimating trends in the occurrence rate of extreme weather and climate events. In: Arnold RA, Arnold K, eds. Scientific reports. Leipzig, Germany: Institute of Meteorology, Institute for Tropospheric Research, 149195.
  • Mudelsee M, Börngen M, Tetzlaff G, Grünewald U. 2004. Extreme floods in central Europe over the past 500 years: role of cyclone pathway “Zugstrasse Vb”. Journal of Geophysical Research 109: D23101.
  • Natural Resources Canada. 2008. Land Cover Map of Canada 2005, Canada. Earth Sciences Sector Program, Ottawa. [WWW document] URL ftp:\\AD\EMS\Landcover2005 [accessed 20 February 2013].
  • Niklasson M, Granström A. 2000. Numbers and sizes of fires: long-term spatially explicit fire history in a Swedish boreal landscape. Ecology 81: 14841499.
  • Olson DM, Dinerstein E, Wikramanayake E, Burgess N, Powell G, Underwood EC, D'Amico J, Itoua I, Strand H, Morrison J et al. 2001. Terrestrial ecoregions of the world: a new map of life on Earth. BioScience 51: 933938.
  • Päätalo M-L. 1998. Factors influencing occurrence and impacts of fires in northern European forests. Silva Fennica 32: 185202.
  • Patterson WA III, Sassaman KE. 1988. Indian Fires in the Prehistory of New England. In: Nicholas GP, ed. Holocene human ecology in northeastern North America. New York, NY, USA: Plenum Press, 107135.
  • Richard PJH. 1980. Postglacial history of the vegetation, south of Lake Abitibi, Ontario and Québec. Géographie physique et Quaternaire 34: 7794.
  • Rogers BM, Randerson JT, Bonan GB. 2013. High-latitude cooling associated with landscape changes from North American boreal forest fires. Biogeosciences 10: 699718.
  • Shabbar A, Skinner W, Flannigan M. 2011. Prediction of seasonal forest fire severity in Canada from large-scale climate patterns. Journal of Applied Meteorology and Climatology 50: 785799.
  • Singarayer JS, Valdes PJ. 2010. High-latitude climate sensitivity to ice-sheet forcing over the last 120 kyr. Quaternary Science Reviews 29: 4355.
  • Stocks BJ, Mason JA, Todd JB, Bosch EM, Wotton BM, Amiro BD, Flannigan MD, Hirsch KG, Logan KA, Martell DL et al. 2003. Large forest fires in Canada, 1959–1997. Journal of Geophysical Research 108: 8149.
  • von Storch H, Zwiers FW. 1999. Statistical analysis in climate research. Cambridge, UK: Cambridge University Press.
  • Systat Software Inc. 2004. SYSTAT Version 11.0 Software. Chicago, IL, USA: SPSS Inc.
  • Terasmae J, Anderson TW. 1970. Hypsithermal range extension of white pine (Pinus strobus L.) in Quebec, Canada. Canadian Journal of Earth Sciences 7: 406413.
  • Terrier A, Girardin MP, Périé C, Legendre P, Bergeron Y. 2013. Potential changes in forest composition could reduce impacts of climate change on boreal wildfires. Ecological Applications 23: 2135.
  • Turetsky M, Kane ES, Harden JW, Ottmar RD, Manies KL, Hoy E, Kasischke ES. 2011. Recent acceleration of biomass burning and carbon losses in Alaskan forests and peatlands. Nature Geoscience 4: 2731.
  • Van Wagner CE 1987. Development and structure of the Canadian Forest Fire Weather Index System. Forestry Tech. Rep. 35. Ottawa, ON, Canada: Canadian Forest Service.
  • Viau AE, Gajewski K. 2009. Reconstructing millennial-scale, regional paleoclimates of boreal Canada during the Holocene. Journal of Climate 22: 316330.
  • Vincent JS. 1973. A palynological study for the Little Clay Belt, northwestern Québec. Le Naturaliste Canadien 100: 5970.
  • Westerling AL, Turner MG, Smithwick EH, Romme WH, Ryan MG. 2011. Continued warming could transform Greater Yellowstone fire regimes by mid-21st Century. Proceedings of the National Academy of Sciences, USA 108: 1316513170.
  • Whitlock C, Millspaugh SH. 1996. Testing the assumptions of fire history studies: an examination of modern charcoal accumulation in Yellowstone National Park, USA. The Holocene 6: 715.
  • Willis KJ, Birks HJB. 2006. What is natural? The need for a long-term perspective in biodiversity conservation. Nature 314: 12611265.
  • Woolford DG, Cao JG, Dean CB, Martell DL. 2010. Characterizing temporal changes in forest fire ignitions: looking for climate change signals in a region of the Canadian boreal forest. Environmetrics 21: 789800.
  • Wotton BM, Nock CA, Flannigan MD. 2010. Forest fire occurrence and climate change in Canada. International Journal of Wildland Fire 19: 253271.
  • Zumbrunnen T, Pezzatti GB, Menéndez P, Bugmann H, Bürgi M, Conedera M. 2011. Weather and human impacts on forest fires: 100 years of fire history in two climatic regions of Switzerland. Forest Ecology and Management 261: 21882199.

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results and Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Please note: Wiley Blackwell are not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office.

nph12322-sup-0001-FigureS1.docWord document632KFig. S1 Observed versus projected number of forest fires of size > 200 ha yr−1 per 1000 km2 in the province of Quebec (Canada).
nph12322-sup-0002-FigureS2.docWord document197KFig. S2 Verification of RegFF against independent fire history studies from needleleaf and boreal mixedwood landscapes.
nph12322-sup-0003-TableS1.docWord document31KTable S1 General circulation models and their greenhouse gas forcing scenarios