ArticlePDF Available

Climatic and Landscape Influences on Fire Regimes from 1984 to 2010 in the Western United States


Abstract and Figures

An improved understanding of the relative influences of climatic and landscape controls on multiple fire regime components is needed to enhance our understanding of modern fire regimes and how they will respond to future environmental change. To address this need, we analyzed the spatio-temporal patterns of fire occurrence, size, and severity of large fires (> 405 ha) in the western United States from 1984–2010. We assessed the associations of these fire regime components with environmental variables, including short-term climate anomalies, vegetation type, topography, and human influences, using boosted regression tree analysis. Results showed that large fire occurrence, size, and severity each exhibited distinctive spatial and spatio-temporal patterns, which were controlled by different sets of climate and landscape factors. Antecedent climate anomalies had the strongest influences on fire occurrence, resulting in the highest spatial synchrony. In contrast, climatic variability had weaker influences on fire size and severity and vegetation types were the most important environmental determinants of these fire regime components. Topography had moderately strong effects on both fire occurrence and severity, and human influence variables were most strongly associated with fire size. These results suggest a potential for the emergence of novel fire regimes due to the responses of fire regime components to multiple drivers at different spatial and temporal scales. Next-generation approaches for projecting future fire regimes should incorporate indirect climate effects on vegetation type changes as well as other landscape effects on multiple components of fire regimes.
Content may be subject to copyright.
Climatic and Landscape Influences on Fire
Regimes from 1984 to 2010 in the Western
United States
Zhihua Liu*, Michael C. Wimberly
Geospatial Sciences Center of Excellence, South Dakota State University, Brookings, South Dakota, United
States of America
An improved understanding of the relative influences of climatic and landscape controls on
multiple fire regime components is needed to enhance our understanding of modern fire
regimes and how they will respond to future environmental change. To address this need,
we analyzed the spatio-temporal patterns of fire occurrence, size, and severity of large fires
(>405 ha) in the western United States from 19842010. We assessed the associations of
these fire regime components with environmental variables, including short-term climate
anomalies, vegetation type, topography, and human influences, using boosted regression
tree analysis. Results showed that large fire occurrence, size, and severity each exhibited
distinctive spatial and spatio-temporal patterns, which were controlled by different sets of
climate and landscape factors. Antecedent climate anomalies had the strongest influences
on fire occurrence, resulting in the highest spatial synchrony. In contrast, climatic variability
had weaker influences on fire size and severity and vegetation types were the most impor-
tant environmental determinants of these fire regime components. Topography had moder-
ately strong effects on both fire occurrence and severity, and human influence variables
were most strongly associated with fire size. These results suggest a potential for the emer-
gence of novel fire regimes due to the responses of fire regime components to multiple driv-
ers at different spatial and temporal scales. Next-generation approaches for projecting
future fire regimes should incorporate indirect climate effects on vegetation type changes
as well as other landscape effects on multiple components of fire regimes.
Fire is an integral component of the earth system and plays a key role in regulating vegetation
structure and ecosystem function [13]. Understanding the relative influences of multiple con-
trolling factors on fire regimes is one of the fundamental objectives of fire ecology, and this
knowledge is critical for improving our ability to anticipate future fire regime changes. Climatic
variability is a major driver of fire in many terrestrial ecosystems, as reflected in Bradstocks
PLOS ONE | DOI:10.1371/journal.pone.0140839 October 14, 2015 1/20
Citation: Liu Z, Wimberly MC (2015) Climatic and
Landscape Influences on Fire Regimes from 1984 to
2010 in the Western United States. PLoSONE
10(10): e0140839. doi:10.1371/journal.pone.0140839
Editor: Lucas C.R. Silva, University of California
Received: May 16, 2015
Accepted: September 29, 2015
Published: October 14, 2015
Copyright: © 2015 Liu, Wimberly. This is an open
access article distributed under the terms of the
Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any
medium, provided the original author and source are
Data Availability Statement: Data and R code are
available at:
Funding: Financial support for this work was
provided through Research Work Order Number
G12AC20295 from the United States Geological
Survey ( and Grant Number
NNX11AB89G from National Aeronautics and Space
Administration ( The funders
had no role in study design, data collection and
analysis, decision to publish, or preparation of the
conceptual model of four climatic switchesthat influence fire regimes by controlling fuel
amount, fuel moisture, and fire weather at contrasting temporal scales [4]. However, fire
regimes are also affected by other controls such as landscape-scale patterns of vegetation,
topography, and human activities [5]. For example, recent analyses in boreal Canada found
that vegetation and fuels influenced the spatial and temporal patterns of fires, even in systems
where climate was considered the most limiting factor [6,7]. Topography also influences fire
regimes through its effects on fuel loads and fuel moisture via site productivity and microcli-
mate [8]. Humans can modify fire regimes by changing ignition patterns [9] and by altering
fuel amount and continuity [10]. Therefore, understanding how fire regimes respond to land-
scape controls in addition to climatic shifts is critical in this era of unprecedented global
change, and will require research that explores the effects of multiple, interacting drivers of fire
regimes [11].
Fire regimes are typically described by statistical distributions of frequency, size, severity,
and seasonality in a particular area during a given time period. Thus, the environmental deter-
minants of fire regimes can be assessed by exploring how environmental drivers operating over
a range of scales affect the spatial and temporal patterns of these fires (Fig 1). The behavior and
effects of an individual wildfire emerge over days to weeks as a result of weather interacting
with fine-grained spatial variability in fuels and vegetation. However, these interactions are
also constrained by biogeographic drivers that vary over broader spatial and temporal scales.
Climate, for example, is connected to fires at two distinct temporal scales [12]. Short-term cli-
matic anomalies (months to years) affect fires by modifying vegetation growth and fuel mois-
ture before the fire and by influencing weather during the period of fire spread. In addition,
climate has more indirect, long-term (decadal or longer) effects on the distributions of major
vegetation types, which in turn constrain the landscape-scale mosaics of fuels and vegetation.
Topography provides a relatively stable physical template that influences fire through direct
interaction with fire spread and indirect effects on vegetation, fuel amounts, and fuel moisture.
Humans can affect fire through a variety of pathways including ignition, suppression, and
alteration of fuels and vegetation [13]. These human impacts are in turn strongly influenced by
variability in human population density, land ownership, and the resulting patterns of land use
and natural resource management activities.
Because climate, vegetation, topography, and human activities interact with fire behavior
and effects at different spatial and temporal scales [14,15], they are likely to have distinctive
effects on fire occurrence, size, and severity. These multiple fire regime components interact
with climate along with other biophysical and human drivers to form characteristic fire regimes
in different geographic settings [1]. Studies conducted at a global scale have shown that fire fre-
quency and burned area tend to be highest in intermediate levels of productivity and moisture
[1620]. However, a more complete understanding of the other components of fire regimes,
including size distribution and severity, remains rudimentary for most biomes on earth [21].
Moreover, fire size and severity have strong influences on the ecosystem structure, function,
and landscape heterogeneity [2224], and have been central to debates on whether climate
change and fire management have altered fire regimes in many parts of world, including the
western US [2527]. To advance our understanding of environmental control on fire regimes,
there is currently a need to explore the determinants of multiple components of the fire regime
including fire frequency, size distribution, and severityusing a comprehensive analytical
framework [28].
Our overarching hypothesis in this research was that different fire regime components
(occurrence, size, and severity) would exhibit characteristic spatial and spatio-temporal pat-
terns that reflected differences in the relative importance of various environmental drivers. We
used the western US as a model system to test this hypothesis because of the diversity of fire
Controls of Fire Regimes in the Western US
PLOS ONE | DOI:10.1371/journal.pone.0140839 October 14, 2015 2/20
Competing Interests: The authors have declared
that no competing interests exist.
regimes spanning a broad range of vegetation and climate gradients and the availability of data
on multiple characteristics of fires along with high-resolution geospatial climatic and biophysi-
cal data. We utilized an ensemble-based decision tree model to analyze how distributions of
individual-fire characteristics varied continuously in space and time. We focused on large fires
(>= 405 ha) because they are consistently documented across space and time, account for the
majority of area burned, and have the most significant ecological and socioeconomic impacts.
Specific research questions included (1) Do the spatial patterns and spatial synchrony of differ-
ent fire regime components (fire occurrence, severity, and size) differ? And (2) How does the
relative importance of short-term climatic anomalies, vegetation types, topography, and geo-
graphic patterns of human activity differ among these fire regime components?
Materials and Methods
Study Area
The study area encompassed the area west of the continental divide, which covers 2 707 515
km2 (approximately 33.5% of the conterminous US, S1 Fig). Climatic conditions vary consid-
erably throughout the study area, ranging from desert and semi-arid regions to mountain
Fig 1. Conceptual model of major factors affecting fire occurrence, size, and severity. Red lines: human influences; green lines: direct climate
influences; black lines: indirect climate (vegetation) influences; blue lines: topographic influences. Bold text represents groups of variables included in the
analysis. Non-bold text represents implicit relationships that were not directly analyzed.
Controls of Fire Regimes in the Western US
PLOS ONE | DOI:10.1371/journal.pone.0140839 October 14, 2015 3/20
ranges with abundant precipitation and maritime climates along the Pacific Coast. There is a
high diversity of vegetation and fire regimes in the study area due to the complexity of geology,
landform, and climate. For example, the coastal Pacific Northwest is dominated by temperate
rainforests with infrequent, high-severity wildfires [29]. The Sierra Nevada are covered by a
variety of forest, shrubland, and grassland with a mixture of different fire regimes [30]. South-
ern California is characterized by Mediterranean climate, and chaparral vegetation that burns
at a relatively high frequency and severity [31]. The mountain ranges of the interior West are
covered by a variety of forest types and associated fire regimes that vary along elevation gradi-
ents [32]. Much of the low-elevation area in the intermountain West is dominated by drought-
adapted vegetation, such as shrub and grass, which historically supported a diversity of fire
Data Sources
Fire dataset. Perimeter and severity data for all large wildfires (>= 405 ha) from 1984
2010 (S1 Fig) were obtained from the Monitoring Trends in Burn Severity (MTBS) project
( A total of 6071 fires (85 for wildland fire use, and 5986 for wildfire)
were included in the analysis after removing duplicate records and prescribed fires. Wildland
fire use (WFU) are naturally ignited wildland fires which can be managed to accomplish spe-
cific resource objectives when people are not threatened. WFU constituted 1.4% of total fires,
and the inclusion or exclusion of these fires did not affect the results. The fires that are being
studied are mostly those that escape initial attack, and thus likely occurred under relatively
extreme fire weather conditions. Fire ignition and extinction dates were used to calculate cli-
matic conditions before and during the fires. Fire occurrence location, size, and percent of high
severity burning were used as response variables. For fire occurrence analysis, we used a case-
control approach in which cases were the observed fires and controls were artificially generated
fire perimeters with random locations, timing, and sizes (see Sampling Random Fire Polygons
for details). Size referred to individual fire extent (mean = 3918 ha, s.d. = 11277 ha). Unburned
islands within the MTBS fires are not always mapped, sometimes resulting in an overestima-
tion of the area burned [33]. Fire severity was measured by the degree of change in vegetation
(i.e., mortality, or biomass consumption) and soil (i.e., char, mineral soil, ash) one year post-
fire relative to pre-fire conditions, as measured by delta normalized burn ratio (dNBR) derived
from Landsat imagery [34]. Higher dNBR indicated an increase in vegetation mortality and
biomass consumption, and therefore higher fire severity. The dNBR has been shown to corre-
late well with field-based assessments of fire severity over a broad range of ecosystem types in
the western U.S. [35,36]. Percent of high severity burning referred to the proportion of area
burned by high severity fires, which corresponds to at least 80% vegetation mortality
(mean = 8.83%, s.d. = 12.73%). Methods for processing and extracting the wildfire data are pro-
vided in S1 File.
Independent Variable: Climate, Vegetation, Topography, and Human
Gridded climate variables at 4 km spatial resolution, including daily maximum temperature,
precipitation, wind speed, and minimum relative humidity, were downloaded from the Univer-
sity of Idaho ([37,38]. For each fire, daily climate vari-
ables were spatially aggregated over the grid cells whose centers were within the fire
perimeters. A variety of short-term antecedent climate variables known to influence fire, sum-
marized over 90 days preceding fire ignition [39], previous year growing seasons (May to Sep-
tember) [40], and previous winters (October to March) [41], were calculated as means and
Controls of Fire Regimes in the Western US
PLOS ONE | DOI:10.1371/journal.pone.0140839 October 14, 2015 4/20
anomalies from climatological normals over 19812010. To capture post-ignition weather con-
ditions, means and maxima were calculated for the fire spreading period, defined as the period
between fire ignition and extinction dates. The mean and median lengths of the fire spreading
period were 25 and 12 days respectively.
Ideally, vegetation composition and structure at the time of each fire should be used to eval-
uate vegetation effects on fire occurrence, size and severity. However, vegetation characteristics
have changed over our 27-year study period as a result of succession, forest management, and
natural disturbances, and spatially and temporally consistent vegetation information for the
western US from 19842010 cannot be obtained from currently available national datasets. For
this study, we used 30 m spatial resolution Biophysical Setting (BpS) data from LANDFIRE
project ( to characterize the broad vegetation types for each fire. BpS is
generated from the current biophysical environment (climate, soils and topography) and his-
torical fire regimes by a state-and-transition simulation approach, providing a characteristic
range of structures and fuels. BpS was aggregated to generate a set of 13 vegetation types with
distinctive species composition, vegetation structure, and fuels (S2 Fig &S1 Table). We over-
laid this vegetation type map with fire perimeters, and computed the percent of each vegetation
type within each fire perimeter.
Elevation (meter) and slope (percent) were derived from a 30 m spatial resolution digital
elevation model. Aspect-related variables (e.g., heat load index and terrain shape index) were
also considered, but preliminary analysis indicated that they did not exhibit significant variabil-
ity among fires, and they were not used in subsequent models. A gridded river density (km/
) dataset at 250 m spatial resolution was calculated from 1: 1 million scale river network
data (, using the line kernel density function with bandwidth = 1000
meters in ArcGIS 10.1. The gridded topographic indices were extracted for each fire perimeter
and summarized as means for each fire.
Major road network data were downloaded from the National Atlas at 1: 1 million scale
( This dataset included highways and other major transportation cor-
ridors, but not local road systems. Wildland-urban interface (WUI) data for 2000 (the approxi-
mate mid-point of our study period) were obtained from Radeloff et al [42]. Two grids at 250
m resolution were created using the Euclidean distance function in ArcGIS 10.1 to represent
distance to the nearest major road and the nearest WUI. These distance surfaces were overlaid
and averaged within each fire perimeter. Land ownership was obtained through Protected
Areas Database of the US (, and was
classified into private ownership, public (non-wilderness) ownership, and wilderness. We over-
laid the land ownership map with fire perimeters, and computed the percent of each ownership
type within each fire perimeter.
There were a large number of climatic and landscape variables, many of which were strongly
correlated with one another and therefore redundant. We selected a subset of these variables
with Pearson correlations <|0.65| according to our a priori hypotheses about the climatic driv-
ers of fire. For the antecedent climate variables, we preferentially selected climatic anomalies
for the 90 days preceding the fire and for the previous winter and growing seasons, based on
the rationale that these anomalies remove the underlying spatial variability in long-term cli-
mate and emphasize temporal deviations and are therefore effective indicators of regional
droughts and other meteorological fluctuations. For the period of fire spread we included
mean and maximum values of climatic variables to capture the extremes, focusing on variables
expected to have the strongest influences on fire behavior such as wind speed, temperature,
and humidity.
Controls of Fire Regimes in the Western US
PLOS ONE | DOI:10.1371/journal.pone.0140839 October 14, 2015 5/20
Sampling Random Fire Polygons
We used a case-control approach to evaluate the effects of landscape and climatic factors on
the probability of large fire occurrence. Random control fires were sampled based on the fol-
lowing criteria: (1) non-burnable land cover types, including urban areas, and agriculture, were
masked out based on the NLCD 2001 land cover dataset; (2) control fire sizes were sampled
from the empirical distribution of fires sizes for all observed fires, and fire perimeters were
assumed to be circular to remove the influence of spatial controls on fire spread; (3) the years
for the control fires were sampled from a uniform distribution from 19842010 to remove the
effect of interannual climate variability on fire occurrence; (4) the dates of control fires were
sampled from the empirical distribution for all observed fires to ensure that control fires
occurred within the fire season; and (5) the lengths of the periods of spread were sampled from
the empirical distribution for all observed fires. The landscape and climatic variables for each
random fire (controls) were processed in the same way for observed fires (cases) described
above. We sampled a number of control fires equal to the total number of observed fires in the
Analysis Methods
Research Question 1: Spatial Patterns and Spatial Synchrony of Fire
Regime Components
Spatial patterns of large fire occurrence were visualized by kernel smoothing [43]. We also
mapped kernel-weighted mean values of total fire size and high-severity fire size (the total area
of each fire that was classified as high severity). Smoothing was carried out using a 50 km
radius circular window and an isotropic Gaussian kernel. Edge effects were mitigated using the
border method in which points that were <50 km from the nearest edge were excluded. A
smoothed percent of high severity burning map was produced by dividing the smoothed high-
severity fire size map by smoothed total fire size map. These analyses were carried out using
the spatstat package [44] in R 3.0 [45]. Spearman's rank coefficient was used to assess the corre-
lation between the spatial patterns of different fire regime components derived from kernel
smoothed surfaces.
Spatial synchrony analysis was used to assess whether temporal fluctuations in fire occur-
rence and other fire characteristics were synchronized across large areas or more localized. We
used the spline correlogram, a modification of the nonparametric covariance function (NCF),
to characterize spatial synchrony. The NCF denotes how spatial covariance, an indicator of
strength of spatial synchrony, varies as a function of geographic distance between locations.
The spline correlogram is an adaptation of the NCF that (1) provides a direct estimate of the
spatial covariance function; (2) uses a bootstrap algorithm to provide a confidence envelope for
the function; and (3) can handle univariate and multivariate spatial data. We computed spline
correlograms for the number of fires, mean fire size, and mean percent of high severity burn-
ing. We employed a 100 km × 100 km grid to summarize the annual number of fires, mean fire
size, and mean percent of high severity burning within each grid cell for each year, and these
gridded time series datasets were used to conduct the spline correlogram analyses. This resolu-
tion was selected based on a tradeoff between sample size to calculate the correlograms and the
amount of data available to compute fire statistics within a grid cell. For example, larger grid
sizes reduced the sample size to calculate the correlograms and increased the minimum dis-
tance between samples, while smaller grid sizes resulted in highly skewed and zero-inflated dis-
tributions of the fire statistics. The mean annual number of fires in each 100 km grid cell was
0.81 (SD = 1.64, min = 0, max = 17, CI
= [0.77, 0.85]). We use 1000 bootstrap samples to
Controls of Fire Regimes in the Western US
PLOS ONE | DOI:10.1371/journal.pone.0140839 October 14, 2015 6/20
calculate confidence intervals and assumed that the NCF was isotropic. Detailed information
on these techniques can be found elsewhere [46]. Spline correlogram analyses was conducted
by using the ncf package [46] in R 3.0.
Research Question 2: Relative Importance of Landscape and Climatic
Controls on Fire Regime Components
We used boosted regression trees (BRT), a machine learning method, to examine the influences
of landscape and climatic controls on individual fire occurrence, fire size, and percent of high
severity burning. The BRT method combines the advantages of regression trees, which relate a
response to their predictors by recursive binary splits, and boosting algorithms, which combine
many simple models to give improved predictive performance. BRT is well-suited for ecological
analyses because of its ability to handle various types of relevant variables, automatically model
interactions among variables, and produce easily interpretable results [47,48]. In BRT, trees are
fitted iteratively to the residual of the existing collection of trees, which often results into higher
predictive accuracy than traditional model averaging methods [48]. The final BRT model can be
understood as an additive regression model in which the individual terms are simple trees.
To avoid overfitting, BRT uses learning rate (lr), tree complexity (tc), and number of trees
(nt) to balance model fit and predictive performance. The lr is used to shrink the contribution
of each tree as it is added to the model. The tc controls the number of nodes or variable interac-
tions in each tree. The best combination of parameters (lr, tc and nt) that achieves maximum
predictive accuracy, can be determined by 10-fold cross validation [48]. To reduce stochastic
errors that might be caused by random subsampling and bagging, a bag fraction of 0.75 was
used for all models; meaning that at each iteration, 75% of the data were drawn at random,
without replacement, from the full training set. BRT analyses were conducted using the dismo
package [48] in R 3.0.
In this analysis, we used a binomial model for large fire occurrence probability based on
binary case-control fire perimeter data (observed fire perimeters-cases; random fire perime-
ters-controls). The best fire occurrence model was selected by maximizing the area under the
receiver operating characteristic curve (0.97), with tc = 3, nt = 2600, and lr = 0.075. We used
Gaussian models for fire size, and percent of high severity burning based on only the observed
fires. The best fire size models were selected by maximizing the variance explained by the
model (0.76), with tc =3,nt = 1880, lr = 0.05. The best percent of high severity burning model
was also selected by maximizing the variance explained by the model (0.60), with tc =3,
nt = 360, lr = 0.1.
The BRT method automatically models the interactions among variables, because every suc-
cessive tree node constitutes a potential interaction. We plotted the first trees from the fire
occurrence model, fire size model, and percent of high severity burning model, because dis-
playing a single tree (usually the first tree) can provide a clear and easily interpretable depiction
of complex interactions [48,49]. We reported the variables that had a relative influence greater
than 5%, along with their marginal effects on fire regime components. The relative importance
of each explanatory variable was calculated by averaging over all trees in a model the number
of times a variable was selected for splitting weighted by the squared residual improvement
resulting from these splits. The marginal effect of a variable on fire regime components was
determined from partial dependency plots, which showed the effect of a variable on the
response after accounting for the average effects of all other variables [48].
We stratified the study area into forest and non-forest sub-regions based on major vegeta-
tion type of Omerniks level III ecoregions. Non-forest subregions included ecoregions domi-
nated by grasslands and shrublands as well as woodlands and chaparral. We repeated the
Controls of Fire Regimes in the Western US
PLOS ONE | DOI:10.1371/journal.pone.0140839 October 14, 2015 7/20
analysis for each sub-region to test whether relative importance of landscape and climatic con-
trols was consistent across the western US or differed between the two sub-regions (S3 Fig).
The 6071 large fires reported in the MTBS database burned 24 265 610 ha over 19842010
(10.6% of the total burnable area of the western US), of which 2 979 817 ha (12.3% of the area
burned and 1.3% of the total burnable area of the western US) was high severity. Fire rotations,
defined as the estimated time required to burn an area of interest, were 254 and 2077 years for
all fire and high severity fire, respectively, based on the total burnable area of the western US.
The mean fire occurrence density was 1.37 large fires per million ha per year, and the median
fire size was 1207 ha. Smoothed maps revealed that the three fire regime components were spa-
tially heterogeneous, and their spatial patterns were not entirely concordant, suggesting they
were potentially influenced by different sets of spatial controls (Fig 2a, 2b, and 2d). Spearman's
rank correlation of the spatial patterns of fire regime components showed fire occurrence and
size had a positive correlation (r= 0.34, p<0.001), whereas fire occurrence and percent high
severity had a negative correlation (r= -0.30, p<0.001). There was a weak positive correlation
between fire size and percent of high severity burning (r= 0.04, p<0.001).
The spline correlograms showed that fire occurrence had the strongest spatial synchrony,
followed by fire size, and percent of high severity burning (Fig 3). The strength of spatial
covariance at the shortest lags dropped from 0.35 for fire occurrence, to 0.2 for fire size, and 0.1
for percent high severity at the grid size of 100 km. The strength of spatial covariance was
stronger when larger grid sizes was used (S3 Fig), suggesting large fire tender to cluster at
broader spatial scales. The values of covariance for fire regime components decreased with
increasing distance, and reached zero at approximately 1500 km, which can be considered as
the scale of the spatial synchrony. There was no difference of the scale of spatial synchrony
among fire occurrence, size, and percent high severity. Local Indicators of Spatial Association
(LISA) analysis highlighted regions with spatially synchronous fire responses and also con-
firmed that temporal patterns of fire occurrence tend to have stronger local associations than
fire size and severity (S4 Fig).
BRT results indicated that precipitation anomaly 90 days before fire ignition had the stron-
gest influence on large fire occurrence, followed by elevation (Fig 4a). Generally, precipitation
had relative stronger influences on large fire occurrence than temperature and humidity, and
short-term antecedent climate conditions (90 days before fire) had a relatively stronger influ-
ence on fire occurence than long-term antecedent climate (previous winter and growing sea-
son) (S5a Fig). Elevation had a negative influence on large fire occurrence. Of the vegetation
types, California Chararral and Shrub-steppe vegetation were the most influential variables,
and both were positively associated with large fire occurrence (Fig 4a).
Vegetation types were the main variables influencing fire size in the western US (Fig 4b). Of
the vegetation types, Grass, Hardwood, Shrub-Steppe, Mixed Conifer, Desert scrub, Pinyon-
Juniper Woodland were selected as the most influential variables. These vegetation types were
positively correlated with fire size except for Mixed Conifer, Pinyon-Juniper Woodland and
Hardwood forest (Fig 4b). Percent of private land was negatively associated with fire size.
Topography and climate had weaker effects on fire size (S5b Fig). Human influences on fire
size were stronger than many topographic and climate variables. Climate variables generally
had a weak effect, and humidity during the fire spreading period and precipitation anomaly 90
days before fire ignition were the most influential climate factors (S5b Fig).
Vegetation and topographic variables had the strongest influences on percent of high sever-
ity burning (Fig 4c). Subalpine and Mixed Conifer were positively associated with percent of
Controls of Fire Regimes in the Western US
PLOS ONE | DOI:10.1371/journal.pone.0140839 October 14, 2015 8/20
high severity burning. Elevation and slope were also positively associated with percent of high
severity burning, likely because they were moderately (0.30.4) positively correlated with Sub-
alpine, Mixed Conifer, and Pinyon-Juniper Woodland (S6 Fig). Precipitation anomaly 90 days
before fire ignition was ranked as the most important climate variable influencing percent high
severity, and its effect was larger than any other climate or human influence variable (S5c Fig).
Fig 2. Spatial distributions of (a) smoothed density of large fires (fires per million ha per year), (b) smoothed mean fire size (ha), (c) smoothed
mean high severity fire size (ha), and (d) smoothed percent of high severity burning (%) in the western US. Non-burnable areas are displayed in white.
Controls of Fire Regimes in the Western US
PLOS ONE | DOI:10.1371/journal.pone.0140839 October 14, 2015 9/20
Fig 3. Spline correlograms illustrating the non-parametric spatial covariance function and 95%
confidence intervals (gray area) for (a) large fire occurrence; (b) fire size; and (c) percent of high
severity burning.
Controls of Fire Regimes in the Western US
PLOS ONE | DOI:10.1371/journal.pone.0140839 October 14, 2015 10 / 20
The first three nodes of the simplified BRT tree provide information about the dominant
effects of climate and landscape factors on fire regime components. The first node of the fire
occurrece tree showed that the precipitation anomaly 90 days before fire was the most impor-
tant factor, and that it interacted with vegetation type to influence probability of large fire
occurrence (Fig 5a). In contrast, the first three nodes for fire size and percent of high severity
burning trees were all vegetation factors, highlighting the dominant controls of vegetation on
fire size and severity (Fig 5b and 5c).
Overall, the results in forest and nonforest dominated sub-regions were similar to the results
for the entire western US. Although the ranks of some individual variables differed, the relative
importance of climate, vegetation, topography, and human influences were consistent for the
fire regime components that we examined (S7 Fig).
Spatial and spatio-temporal patterns of large-fire occurrence differed from patterns of fire size
and severity in the western US, and these differences can be explained by the distinctive effects
of key environmental drivers on various components of the fire regime. In particular, large fire
occurrence had higher spatial synchrony and was most strongly associated with short-term cli-
matic anomalies. This finding is supported by previous analyses in the western US, which
found that large fires were often preconditioned by frequent or more numerous consecutive
Fig 4. Relative influences of variables that explained greater than 5% of the variation and marginal effects (red trend lines within each bar) from
boosted regression tree models for (a) large fire occurrence, (b) fire size, and (c) percent of high severity burning. Values are specified for truncated
bars. Abbreviations of predictor variables and their corresponding full names are described in Table 1.
Controls of Fire Regimes in the Western US
PLOS ONE | DOI:10.1371/journal.pone.0140839 October 14, 2015 11 / 20
days of hot and dry climatic conditions which result in lower fuel moisture, particularly for live
fuels and larger dead fuels [39,50,51]. These droughts are often associated with climate pat-
terns such as the Pacific Decadal Oscillation and El Niño Southern Oscillation, and can affect
regional, inter-annual fluctuations in large fire occurrence of western North America [52,53].
Our results are also consistent with studies employing different analytical methods based on
the MTBS data in the western US. For example, Riley et al [39] found that precipitation during
the past 13 months was a strong predictor of large fire occurrence because dead fuel moisture
is strongly influenced by short-term antecedent climate conditions. Similarly, Abatzoglou and
Kolden [18] found that climatic indices of drought during the current fire season had stronger
relationships with area burned than antecedent climate variables from previous years, and that
these regional anomalies synchronized the area burned across both forested and non-forested
Fig 5. Simplified versions of the first tree of (a) fire occurrence model (b) fire size model, and (c) percent of high severity burning model computed
with the boosted regression tree algorithm. The first three splits of each tree are shown to illustrate the interactions between key variables. The splitting
variable and its corresponding splitting value are shown in oval above the node (variable abbreviation and units are provided in Table 1). The values in the
rectangles at the terminal nodes represent the mean prediction and number of the records in the terminal nodes (n). The total number of records in the
terminal nodes equals 0.75 (the bag fraction of the BRT model) of the total number of fires. Abbreviations: P: relative probability of fire occurrence; MFS:
mean fire size; PHS: percent of high severity burning.
Controls of Fire Regimes in the Western US
PLOS ONE | DOI:10.1371/journal.pone.0140839 October 14, 2015 12 / 20
sub-regions of the western United States. However, our results further suggested that the short-
term antecedent climatic anomalies interact with vegetation type to influence patterns of fire
occurrence and have stronger influences on fire occurrence than on fire size or fire severity
(Fig 5a).
Fire size and severity exhibited weaker spatial synchrony and were more strongly influenced
by relatively static driving variables characterizing vegetation and topography, consistent with
results from a previous study of fire severity patterns in the western US [54]. These differences
reflect the fact that fire is controlled by different sets of factors at different spatio-temporal
scales [55,56], incorporating cross-scale interactions and nonlinearities at different stages of
fire growth [57]. At the initiation stage, when and where a fire occurs is highly stochastic and is
influenced by ignition sources and fuel moisture; the latter being largely governed by short-
term antecedent climatic anomalies [58]. The initial spread of a fire involves within-patch
spread stage (see Fig 1 in [57]) and depends on local fuel characteristics within the initiation
patch, such as fuel type, amount, chemistry, and spatial distribution. The large fires that are
being studied here all likely reached the next stage of spread among multiple patches (see Fig 1
in [57]), when fire spread and fire effects are sensitive to the spatial pattern of topography and
the landscape connectivity of multiple fuel and vegetation patches. Much of the western US is
topographically complex and involve sharp gradients of vegetation and fuels related to eleva-
tion, topography, and climate. Given that the large fires that we studied already tended to
occur during time periods characterized by relatively dry conditions, differences in size and
severity of these fires were more closely associated with vegetation and fuels than with short-
term climatic variability.
Our findings of vegetation and topographic controls on fire severity are supported by a vari-
ety of other, more localized studies. For example, Steel et al [59] and Collins and Stephens [60]
showed that fir-dominated mixed conifer forest in the Sierra Nevada tends to be dominated by
higher fire severity than pine- or hardwood-dominated forest because of greater vertical conti-
nuity in fuels from the ground surface to the upper canopy strata. Sherriff et al [61] demon-
strated that elevation and slope steepness were both positively related to high-severity fires in
Colorado Front Range based on an analysis of forest structure and tree-ring fire history data.
Harvey et al [62] investigated the controls on severity of five recent fires in upper-montane and
subalpine forests in the U.S. Northern Rockies, and concluded that topography significantly
influenced fire severity. Our regional analysis also suggested that humans had a relatively
stronger influence on fire size than the other two fire regimes components, echoing previous
studies that have documented the complex influences of roads and the wildland-urban inter-
face on fire suppression and fire management policies [63,64].
The finding that different fire regime components are controlled by different sets of drivers
suggests that climate change will have varied effects on fire regimes over a range of spatial and
temporal scales. Large fire occurrence in the western US was mainly influenced by short-term
climate variability during the 90 days preceding the fire, which constrains fuel moisture and
the availability of fine fuel to burn. Therefore, short-term variability in the frequency and dura-
tion of drought events will likely have the strongest influence on fire occurrence in western US.
In contrast, fire size and severity were less affected by these climatic switches and more influ-
enced by the spatial distribution of broad vegetation types [59,6567], which are more con-
strained by longer-term climate influences on species composition and rates of biomass
accumulation and decomposition. As a result, longer-term climate-driven changes in vegeta-
tion types and associated fuel conditions may have particularly strong influences on the sizes
and ecological effects of the largest fires. Future fire regime changes will thus be driven by the
combined effects of both fast-changing climatic variability and slower-changing landscape fac-
tors. A subsequent study confirmed that changes in regional patterns of major vegetation
Controls of Fire Regimes in the Western US
PLOS ONE | DOI:10.1371/journal.pone.0140839 October 14, 2015 13 / 20
types, in addition to broad scale climate, had a significant influence on future regional fire
regime in the western US [68].
These results are supported by other recent studies that have documented distinctive
responses of different fire regime components to environmental variability. Another study of
fire regimes in the western US based on the MTBS dataset by Parks et al [69] found that burned
area increased monotonically with fuel amount and had a unimodal relationship with fuel
moisture, whereas fire severity increased monotonically with both fuel amount and fuel mois-
ture. A paleoecological study of Rocky Mountain subalpine forests over the past 6000 years
found that fire severity varied independently of fire frequency and was more influenced by
changes in vegetation and fuels than by direct effects of climatic variation [70]. A remote sens-
ing-based analysis found that change in vegetation and fuels due to cheatgrass (Bromus tec-
torum) invasion in recent decades has increased both fire frequency and size and has therefore
substantially altered the regional fire regime across the Great Basin of the western US [71].
Simlar to the plant community reassembly that resulted from individualistic responses of tree
species to historical climate change [72], the independent responses of different fire regime
components to environmental change have the potential to result in novel fire regimes and
produce ecological surprises in the future. These findings underscore the necessity of explicitly
incorporating indirect effects of climate change on vegetation, as well as the influences of other
mediating landscape variables, on multiple fire regime components in future fire projections at
large spatiotemporal scales [11,19].
Several limitations should be considered when interpreting these results. First, climate vari-
ables were measured at a much coarser spatial resolution than topography, vegetation, and
human influences. However, our analysis was conducted at the fire level rather than the pixel
level [54,73,74], and summarizing the landscape variables for each fire reduced these scale
inconsistences. Second, we did not analyze human and lightning-ignited fires separately. In the
future, the development of longer-term fire datasets with information on the cause of fires
would be useful for exploring potential differences in human versus lightning-caused fires [9,
75]. Third, despite the widespread use of dNBR to measure fire severity in the western US, they
have significant limitations including inadequate characterization of fire effects on the surface
and ground layers, reduced sensitivity at higher NBR values, and inconsistent relationships
with field-based measurements across different fires and vegetation types [35]. High burn
severity was classified by the MTBS project using a custom threshold for each fire using based
on the remote sensing indices, plot data, expert knowledge, and published literature [34].
Given the subjectivity of this procedure, we focused on the high severity class (>80% overstory
vegetation mortality), which causes the most distinctive change in spectral signatures and
therefore should be more consistently classified than lower burn severity classes. However, it
should be recognized that other aspects of fire severity are not be captured by this burn severity
metric and that the ecological consequences of high-severity fire can vary across different envi-
ronmental settings. Finally, although there was considerable interannual variability in fire
activity during this 27-year period, we expect that climatic variables would be more important
relative to landscape variables in analyses of longer time series that encompass a larger range of
climatic conditions.
In this analysis, we characterized climate and landscape effects on fire regimes across the west-
ern United States using a consistent analytical framework to produce general insights about the
environmental factors that control the spatial and temporal patterns of major fire regime com-
ponents. Our results showed that landscape and climatic factors had varied effects on fire
Controls of Fire Regimes in the Western US
PLOS ONE | DOI:10.1371/journal.pone.0140839 October 14, 2015 14 / 20
Table 1. Environmental variables summarized for each fire and used in the analysis.
Variable Description Mean+s.d
TMx (degree) Maximum temperature during re spread 32.36 +5.39
PMn (mm) Mean precipitation during re spread 0.799+1.203
HMn (%) Mean relative humidity during re spread 18.63+7.27
WMx (m/s) Maximum wind speed during re spread 3.94+1.39
WMn (m/s) Mean wind speed during re spread 3.24+0.80
TAo (degree) Maximum temperature anomaly during re spread 10.10 +5.16
PAo (mm) Precipitation anomaly during re spread -0.63 +1.13
HAo (%) Relative humidity anomaly during re spread -12.52+6.65
T90Ao (degree) Mean maximum temperature anomaly for 90 days preceding re start
6.99 +4.55
P90Ao (mm) Mean precipitation anomaly for 90 days preceding re start date -0.45 +0.90
H90Ao (%) Relative humidity anomaly for 90 days preceding re start date -8.35+5.34
TPAo (degree) Previous year growing season temperature anomaly -0.22+1.04
PPAo (mm) Previous year growing season precipitation anomaly 10.14+51.53
HPAo (%) Previous year growing season relative humidity anomaly for each re 0.19+2.86
TWAo (degree) Previous winter temperature anomaly 8.64+1.82
PWAo (mm) Previous winter precipitation anomaly -0.59+0.98
Growing season temperature anomaly 2 years prior -0.078+1.28
PP2GAo (mm) Growing season precipitation anomaly 2 years prior -0.085+0.34
PCDF (%) Percent Pacic coast Douglas-r forest 0.717+6.75
InDF (%) Percent Interior Douglas-r 5.36+13.76
PJW (%) Percent Pinyon-Juniper Woodland 4.81+10.87
PIPO (%) Percent Interior Ponderosa Pine 6.17+16.25
Salp (%) Percent Subalpine forest 6.89+19.62
Mixd (%) Percent Mixed conifer 4.73+15.40
Hdwd (%) Percent Hardwood 7.05+16.28
CaCr (%) Percent California Chararral 7.84+20.26
DeSc (%) Percent Desert scrub 10.70+24.41
MMsh (%) Percent Mesic Mountain Shrub 1.05+4.57
Sgbr (%) Percent Sagebrush 18.23+29.32
Shsp (%) Percent Shrub-steppe 12.62+26.61
Gras (%) Percent Grass 5.43+16.01
Slop (in percent) Mean slope 14.33+10.26
DEM (m) Mean elevation 1434+621
RiverD (km*km
Mean river density 0.11503
D2Rd (m) Mean distance to nearest road 11864+11240
D2WUI (m) Mean distance to Wildland Urban interface 10432+9045
PerPrv Percent private land 64+38
PerPub Percent public non-wilderness land 25+34
PerWdn Percent wilderness land 10+27
Mean and s.d. for environmental variables were calculated from all the res.
Controls of Fire Regimes in the Western US
PLOS ONE | DOI:10.1371/journal.pone.0140839 October 14, 2015 15 / 20
occurrence, size, and severity. In particular, the spatio-temporal patterns of fire size and sever-
ity exhibited weaker spatial synchrony than fire occurrence, and were more strongly con-
strained by spatial patterns of vegetation, topography, and human activities. In contrast, the
probability of fire occurrence was mainly influenced by recent climate anomalies, and showed
a stronger spatial synchrony. These findings underscore the value of studying individualistic
response of different fire regime components and ultimately incorporating multiple environ-
mental drivers of different fire regime components into projections of future fire regimes. In
particular, our results suggest the possibility that fire regimes with novel combinations of fre-
quency, severity, and size may emerge as a result of the interacting effects of changes in climate,
shifts in vegetation distributions, and continuing expansion of the human footprint. Ongoing
efforts to project the effects of future global change on regional fire regimes should therefore
incorporate the indirect effects of climate on vegetation type, as well as other types of landscape
controls on multiple fire regime components.
Supporting Information
S1 Fig. Study area boundary overlaid by large fire locations from the MTBS dataset
between 1984 and 2010 (red dots).
S2 Fig. Spatial distribution of vegetation types based on biophysical settings from the
LANDFIRE project (
S3 Fig. Sensitivity of spatial covariance to different grid size (columns) for different fire
regime components (rows).
S4 Fig. Local Indicators of Spatial Association (LISA) analysis for spatial synchrony of a)
fire occurrence, b) fire size, and c) percent of high severity burning in the western US.
Square and circle symbols indicate positive and negative associations, respectively. Sizes of the
symbols indicate strength of association. Filled symbols indicate significant (p <0.05) associa-
S5 Fig. Relative influences of predictor variables from boosted regression tree models for
(a) large fire occurrence, (b) fire size, and (c) percent of high severity burning in western
US. Values are specified for truncated bars. Abbreviations of variables and their corresponding
full names are described in Table 1.
S6 Fig. Correlations among Subalpine, Mixed Conifer, Pinyon-Juniper Woodland, eleva-
tion, and slope. Abbreviations of variables and their corresponding full names are described in
Table 1.
S7 Fig. Relative influences of variables from boosted regression tree models on (a) large fire
occurrence, (b) fire size, and (c) percent of high severity burning for forest dominated sub-
regions, and (d) large fire occurrence, (f) fire size, and (g) percent of high severity burning
for non-forest dominated sub-regions in the western US. Values are specified for truncated
bars. Abbreviations of variables and their corresponding full names are described in Table 1.
Controls of Fire Regimes in the Western US
PLOS ONE | DOI:10.1371/journal.pone.0140839 October 14, 2015 16 / 20
S1 File. Data source and processing procedure for fire variables.
S1 Table. Biophysical settings reclassification used in the analysis.
The contents of this publication are solely the responsibility of the authors and do not necessar-
ily represent the official views of the funding agencies. The authors thank Peter J Weisberg for
assisting with the definition of BpS vegetation types and Joshua Picotte (EROS USGS) for pro-
viding data on start and end dates for the MTBS fires. Jingfeng Xiao, Patrick Freeborn, and
Francis Dwomoh provided helpful comments on earlier drafts of this paper. All the data and R
code can be downloaded from
Author Contributions
Conceived and designed the experiments: ZL MW. Performed the experiments: ZL. Analyzed
the data: ZL. Contributed reagents/materials/analysis tools: ZL MW. Wrote the paper: ZL MW.
1. Bowman DMJS, Balch JK, Artaxo P, Bond WJ, Carlson JM, Cochrane MA, et al. Fire in the earth sys-
tem. Science. 2009; 324(5926):4814. doi: 10.1126/science.1163886 PMID: 19390038
2. Bond W, Woodward F, Midgley G. The global distribution of ecosystems in a world without fire. New
Phytol. 2005; 165(2):52538. PMID: 15720663
3. Pausas JG, Keeley JE. A burning story: the role of fire in the history of life. BioScience. 2009; 59
4. Bradstock RA. A biogeographic model of fire regimes in Australia: current and future implications.
Global Ecol Biogeogr. 2010; 19(2):14558.
5. Bowman D, Balch J, Artaxo P, Bond WJ, Cochrane MA, D'Antonio CM, et al. The human dimension of
fire regimes on Earth. J Biogeogr. 2011; 38(12):222336. PMID: 22279247
6. Parisien M-A, Parks SA, Krawchuk MA, Little JM, Flannigan MD, Gowman LM, et al. An analysis of con-
trols on fire activity in boreal Canada: comparing models built with different temporal resolutions. Ecol
Appl. 2014; 24(6):134156.
7. Heon J, Arseneault D, Parisien MA. Resistance of the boreal forest to high burn rates. Proc Natl Acad
Sci U S A. 2014; 111(38):1388893. doi: 10.1073/pnas.1409316111 PMID: 25201981
8. Rollins MG, Morgan P, Swetnam T. Landscape-scale controls over 20th century fire occurrence in two
large Rocky Mountain (USA) wilderness areas. Landscape Ecol. 2002; 17(6):53957.
9. Liu Z, Yang J, Chang Y, Weisberg PJ, He HS. Spatial patterns and drivers of fire occurrence and its
future trend under climate change in a boreal forest of Northeast China. Global Change Biol. 2012; 18
10. Pausas JG, Fernández-Muñoz S. Fire regime changes in the Western Mediterranean Basin: from fuel-
limited to drought-driven fire regime. Clim Change. 2012; 110(12):21526.
11. Bowman DMJS, Murphy BP, Williamson GJ, Cochrane MA. Pyrogeographic models, feedbacks and
the future of global fire regimes. Global Ecol Biogeogr. 2014; 23(7):8214.
12. Hessl AE. Pathways for climate change effects on fire: Models, data, and uncertainties. Prog Phys
Geogr. 2011; 35(3):393407.
13. Liu Z, Wimberly M, Lamsal A, Sohl T, Hawbaker T. Climate change and wildfire risk in an expanding
wildlandurban interface: a case study from the Colorado Front Range Corridor. Landscape Ecol.
2015: doi: 10.1007/s10980-015-0222-4
14. Falk DA, Heyerdahl EK, Brown PM, Farris C, Fulé PZ, McKenzie D, et al. Multi-scale controls of histori-
cal forest-fire regimes: new insights from fire-scar networks. Front Ecol Environ. 2011; 9(8):44654.
15. Liu Z, Yang J, He HS. Identifying the Threshold of Dominant Controls on Fire Spread in a Boreal Forest
Landscape of Northeast China. PLoS ONE. 2013; 8(1):e55618. doi: 10.1371/journal.pone.0055618
PMID: 23383247
Controls of Fire Regimes in the Western US
PLOS ONE | DOI:10.1371/journal.pone.0140839 October 14, 2015 17 / 20
16. Archibald S, Roy DP, van Wilgen BW, Scholes RJ. What limits fire? An examination of drivers of burnt
area in Southern Africa. Global Change Biol. 2009; 15(3):61330.
17. Krawchuk M, Moritz M. Constraints on global fire activity vary across a resource gradient. Ecology.
2011; 92(1):12132. PMID: 21560682
18. Abatzoglou JT, Kolden CA. Relationships between climate and macroscale area burned in the western
United States. Int J Wildland Fire. 2013; 22(7):100320.
19. Morton D, Collatz G, Wang D, Randerson J, Giglio L, Chen Y. Satellite-based assessment of climate
controls on US burned area. Biogeosciences. 2013; 10:24760.
20. Pausas JG, Ribeiro E. The global fireproductivity relationship. Global Ecol Biogeogr. 2013; 22
21. Bowman DM, O'Brien JA, Goldammer JG. Pyrogeography and the global quest for sustainable fire
management. Annual Review of Environment and Resources. 2013; 38(1):5780.
22. Yates CP, Edwards AC, Russell-Smith J. Big fires and their ecological impacts in Australian savannas:
size and frequency matters. Int J Wildland Fire. 2008; 17(6):76881.
23. Turner MG. Disturbance and landscape dynamics in a changing world. Ecology. 2010; 91(10):2833
49. PMID: 21058545
24. Liu Z, Yang J. Quantifying ecological drivers of ecosystem productivity of the early-successional boreal
Larix gmelinii forest. Ecosphere. 2014; 5(7):84.
25. Stephens S, Agee J, Fulé P, North M, Romme W, Swetnam T, et al. Managing forests and fire in chang-
ing climates. Science. 2013; 342(6154):412. doi: 10.1126/science.1240294 PMID: 24092714
26. Williams MA, Baker WL. Spatially extensive reconstructions show variable-severity fire and heteroge-
neous structure in historical western United States dry forests. Global Ecol Biogeogr. 2012; 21
27. Odion DC, Hanson CT, Arsenault A, Baker WL, DellaSala DA, Hutto RL, et al. Examining historical and
current mixed-severity fire regimes in ponderosa pine and mixed-conifer forests ofwestern North Amer-
ica. PloS one. 2014; 9(2):e87852. doi: 10.1371/journal.pone.0087852 PMID: 24498383
28. Krawchuk MA, Moritz MA. Burning issues: statistical analyses of global fire data to inform assessments
of environmental change. Environmetrics. 2014; 25(6):47281.
29. Wimberly MC, Liu Z. Interactions of climate, fire, and management in future forests of the Pacific North-
west. For Ecol Manage. 2014; 327(0):2709.
30. Miller J, Collins B, Lutz J, Stephens S, van Wagtendonk J, Yasuda D. Differences in wildfires among
ecoregions and land management agencies in the Sierra Nevada region, California, USA. Ecosphere
3 (9): 80. 2012.
31. Jin Y, Randerson JT, Faivre N, Capps S, Hall A, Goulden ML. Contrasting controls on wildland fires in
Southern California during periods with and without Santa Ana winds. Journal of Geophysical
Research: Biogeosciences. 2014:2013JG002541.
32. Noss RF, Franklin JF, Baker WL, Schoennagel T, Moyle PB. Managing fire-prone forests in the western
United States. Front Ecol Environ. 2006; 4(9):4817.
33. Sparks AM, Boschetti L, Smith AMS, Tinkham WT, Lannom KO, Newingham BA. An accuracy assess-
ment of the MTBS burned area product for shrubsteppe fires in the northern Great Basin, United
States. Int J Wildland Fire. 2015; 24(1):708.
34. Eidenshink J, Schwind B, Brewer K, Zhu Z-L, Quayle B, Howard S. A project for monitoring trends in
burn severity. Fire Ecol. 2007; 3(1):119.
35. French NH, Kasischke ES, Hall RJ, Murphy KA, Verbyla DL, Hoy EE, et al. Using Landsat data to
assess fire and burn severity in the North American boreal forest region: an overview and summary of
results. Int J Wildland Fire. 2008; 17(4):44362.
36. Miller JD, Thode AE. Quantifying burn severity in a heterogeneous landscape with a relative version of
the delta Normalized Burn Ratio (dNBR). Remote Sens Environ. 2007; 109(1):6680.
37. Abatzoglou JT. Development of gridded surface meteorological data for ecological applications and
modelling. Int J Climatology. 2013; 33(1):12131.
38. Abatzoglou JT, Brown TJ. A comparison of statistical downscaling methods suited for wildfire applica-
tions. Int J Climatology. 2012; 32(5):77280.
39. Riley KL, Abatzoglou JT, Grenfell IC, Klene AE, Heinsch FA. The relationship of large fire occurrence
with drought and fire danger indices in the western USA, 19842008: the role of temporal scale. Int J
Wildland Fire. 2013; 22(7):894909.
40. Littell JS, McKenzie D, Peterson DL, Westerling AL. Climate and wildfire area burned in westernUS
ecoprovinces, 19162003. Ecol Appl. 2009; 19(4):100321. PMID: 19544740
Controls of Fire Regimes in the Western US
PLOS ONE | DOI:10.1371/journal.pone.0140839 October 14, 2015 18 / 20
41. Westerling AL, Hidalgo HG, Cayan DR, Swetnam TW. Warming and earlier spring increase western
US forest wildfire activity. Science. 2006; 313(5789):9403. PMID: 16825536
42. Radeloff VC, Hammer RB, Stewart SI, Fried JS, Holcomb SS, McKeefry JF. The wildland-urban inter-
face in the United States. Ecol Appl. 2005; 15(3):799805.
43. Diggle PJ. Statistical analysis of spatial point patterns, 2nd edition: Oxford University Press, New
York; 2003.
44. Baddeley A, Turner R. Spatstat: an R package for analyzing spatial point patterns. J Stat Soft. 2005; 12
45. R Core Team. R: A language and environment for statistical computing. R Foundation for Statistical
Computing, Vienna, Austria. URL 2013.
46. Bjørnstad ON, Ims RA, Lambin X. Spatial population dynamics: analyzing patterns and processes of
population synchrony. Trends Ecol Evol. 1999; 14(11):42732. PMID: 10511718
47. De'Ath G. Boosted trees for ecological modeling and prediction. Ecology. 2007; 88(1):24351. PMID:
48. Elith J, Leathwick J, Hastie T. A working guide to boosted regression trees. J Anim Ecol. 2008; 77
(4):80213. doi: 10.1111/j.1365-2656.2008.01390.x PMID: 18397250
49. Parisien M-A, Moritz MA. Environmental controls on the distribution of wildfire at multiple spatial scales.
Ecol Monogr. 2009; 79(1):12754.
50. Meyn A, White PS, Buhk C, Jentsch A. Environmental drivers of large, infrequent wildfires: the emerg-
ing conceptual model. Prog Phys Geogr. 2007; 31(3):287312.
51. Schoennagel T, Veblen TT, Romme WH. The interaction of fire, fuels, and climate across Rocky Moun-
tain forests. Bioscience. 2004; 54(7):66176.
52. Kitzberger T, Brown PM, Heyerdahl EK, Swetnam TW, Veblen TT. Contingent PacificAtlantic Ocean
influence on multicentury wildfire synchrony over western North America. PNAS. 2007; 104(2):5438.
PMID: 17197425
53. Swetnam TW, Betancourt JL. Fire-southern oscillation relations in the southwestern United States. Sci-
ence. 1990; 249:101720. PMID: 17789609
54. Dillon GK, Holden ZA, Morgan P, Crimmins MA, Heyerdahl EK, Luce CH. Both topography and climate
affected forest and woodland burn severity in two regions of the western US, 1984 to 2006. Ecosphere.
2011; 2(12):130. doi: 10.1890/ES11-00271.1
55. Pyne SJ, Andrews PL, Laven RD. Introduction to wildland fire: John Wiley and Sons; 1996.
56. Moritz MA, Morais ME, Summerell LA, Carlson J, Doyle J. Wildfires, complexity, and highly optimized
tolerance. Proceedings of the National Academy of Sciences of the United States of America. 2005;
102(50):17912. PMID: 16332964
57. Peters DPC, Pielke RA Sr, Bestelmeyer BT, Allen CD, Munson-McGee S, Havstad KM. Cross-scale
interactions, nonlinearities, and forecasting catastrophic events. PNAS. 2004; 101(42):151305. PMID:
58. Pyke DA, Brooks ML, D'Antonio C. Fire as a restoration tool: a decision framework for predicting the
control or enhancement of plants using fire. Restor Ecol. 2010; 18(3):27484.
59. Steel ZL, Safford HD, Viers JH. The fire frequency-severity relationship and the legacy of fire suppres-
sion in California forests. Ecosphere. 2015; 6(1):art8.
60. Collins BM, Stephens SL. Stand-replacing patches within a mixed severityfire regime: quantitative
characterization using recent fires in a long-established natural fire area. Landscape Ecol. 2010; 25
61. Sherriff RL, Platt RV, Veblen TT, Schoennagel TL, Gartner MH. Historical, observed, and modeled wild-
fire severity in montane forests of the Colorado Front Range. PloS one. 2014; 9(9):e106971. doi: 10.
1371/journal.pone.0106971 PMID: 25251103
62. Harvey BJ, Donato DC, Turner MG. Recent mountain pine beetle outbreaks, wildfire severity, and post-
fire tree regeneration in the US Northern Rockies. PNAS. 2014; 111(42):151205. doi: 10.1073/pnas.
1411346111 PMID: 25267633
63. Syphard AD, Radeloff VC, Keeley JE, Hawbaker TJ, Clayton MK, Stewart SI, et al. Human influence on
california fire regimes. Ecol Appl. 2007; 17(5):1388402. PMID: 17708216
64. Haire SL, McGarigal K, Miller C. Wilderness shapes contemporary fire size distributions across land-
scapes of the western United States. Ecosphere. 2013; 4(1):art15.
65. Miller J, Skinner C, Safford H, Knapp EE, Ramirez C. Trends and causes of severity, size, and number
of fires in northwestern California, USA. Ecol Appl. 2012; 22(1):184203. PMID: 22471083
Controls of Fire Regimes in the Western US
PLOS ONE | DOI:10.1371/journal.pone.0140839 October 14, 2015 19 / 20
66. Miller J, Safford H, Crimmins M, Thode A. Quantitative evidence for increasing forest fire severity in the
Sierra Nevada and southern Cascade Mountains, California and Nevada, USA. Ecosystems. 2009; 12
67. Mallek C, Safford H, Viers J, Miller J. Modern departures in fire severity and area vary by forest type,
Sierra Nevada and southern Cascades, California, USA. Ecosphere. 2013; 4(12):art153.
68. Liu Z, Wimberly MC. Direct and indirect effects of climate change on projected future fire regimes in the
western United States. Sci Total Environ. In review(0).
69. Parks SA, Parisien M-A, Miller C, Dobrowski SZ. Fire activity and severity in the western US vary along
proxy gradients representing fuel amount and fuel moisture. PLoS ONE. 2014; 9(6):e99699. doi: 10.
1371/journal.pone.0099699 PMID: 24941290
70. Higuera PE, Briles CE, Whitlock C. Fire-regime complacency and sensitivity to centennial-through mil-
lennial-scale climate change in Rocky Mountain subalpine forests, Colorado, USA. J Ecol. 2014; 102
71. Balch JK, Bradley BA, D'Antonio CM, Gómez-Dans J. Introduced annual grass increases regional fire
activity across the arid western USA (19802009). Global Change Biol. 2013; 19(1):17383.
72. Davis MB, Shaw RG. Range shifts and adaptive responses to Quaternary climate change. Science.
2001; 292(5517):6739. PMID: 11326089
73. Collins BM, Kelly M, van Wagtendonk JW, Stephens SL. Spatial patterns of large natural fires in Sierra
Nevada wilderness areas. Landscape Ecol. 2007; 22(4):54557.
74. Holden ZA, Morgan P, Evans JS. A predictive model of burn severity based on 20-year satellite-inferred
burn severity data in a large southwestern US wilderness area. For Ecol Manage. 2009; 258(11):2399
75. Narayanaraj G, Wimberly MC. Influences of forest roads on the spatial patterns of human-and light-
ning-caused wildfire ignitions. Appl Geo. 2012; 32(2):87888.
Controls of Fire Regimes in the Western US
PLOS ONE | DOI:10.1371/journal.pone.0140839 October 14, 2015 20 / 20
... climatic, edaphic, and vegetation patterns) (Muzika et al. 2015;Guyette et al. 2016;Johnson and Kipfmueller 2016;Kipfmueller et al. 2017Kipfmueller et al. , 2021. Analysing anthropogenic and biophysical drivers of fire regimes across multiple scales is a fundamental objective of fire ecology (Lui and Wimberly 2015), and can reveal interactions among both drivers at regional scales. One example observed in fire-scar records across the upper Great Lakes Region is the substantial effect of fire suppression. ...
... Collins et al., 2007;Fernandes et al., 2010;Dillon et al., 2011;Viedma et al., 2015;Harvey et al., 2016;Fernández-García et al., 2022) in forest ecosystems worldwide. Only a few studies assessed how fire behavior drivers modulate the relationships between fire size and fire severity in North America (Cansler and McKenzie, 2014;Parks et al., 2014;Liu and Wimberly, 2015;Harvey et al., 2016) and Asia (Fang et al., 2015). However, linkages between contemporary fire attributes and their driving factors remain scarcely documented in southern Europe, namely in Portugal, the case-study territory of this paper. ...
Characterizing the fire regime in regions prone to extreme wildfire behavior is essential for providing comprehensive insights on potential ecosystem response to fire disturbance in the context of global change. We aimed to disentangle the linkage between contemporary damage-related attributes of wildfires as shaped by the environmental controls of fire behavior across mainland Portugal. We selected large wildfires (≥100 ha, n = 292) that occurred during the 2015-2018 period, covering the full spectrum of large fire-size variation. Ward's hierarchical clustering on principal components was used to identify homogeneous wildfire contexts at landscape scale on the basis of fire size, proportion of high fire severity, and fire severity variability, and their bottom-up (pre-fire fuel type fraction, topography) and top-down (fire weather) controls. Piecewise Structural Equation Modeling was used to disentangle the direct and indirect relationships between fire characteristics and fire behavior drivers. Cluster analysis evidenced severe and large wildfires in the central region of Portugal displaying consistent fire severity patterns. Thus, we found a positive relationship between fire size and proportion of high fire severity, which was mediated by distinct fire behavior drivers involving direct and indirect pathways. A high fraction of conifer forest within wildfire perimeters and extreme fire weather were primarily responsible for those interactions. In the context of global change, our results suggest that pre-fire fuel management should be targeted at expanding the fire weather settings in which fire control is feasible and promote less flammable and more resilient forest types.
... This result is not unexpected as urbanisation and infrastructure development alter the natural fire regime of a region [30]. Increases to anthropogenic activity may increase ignition sources across a region as fires are ignited to clear land for urbanisation, and also for industry such as agriculture [121]. The effects of fire on an ecosystem depend on the ecosystem's evolutionary background. ...
Full-text available
This study identifies the spatial and temporal trends, and explores the drivers of fire dynamics in the Bolivian Amazon basin. Bolivia ranks in the top ten countries in terms of total annual burnt, with fires affecting an estimated 2.3 million hectares of forest in 2020. However, in comparison to the Brazilian Amazon, there has been little research into the fire regime in Bolivia. The sparse research and the limited literature on the subject indicate that fire activity is higher in the Bolivian Amazon basin’s dry forests and flooded savanna zones, and that agriculture and drought are the primary causes of fire activity. In this study, trend analysis and emerging hotspot analysis are deployed to identify the spatial and temporal patterns of fire activity and boosted regression tree models to identify the drivers of forest fire within each ecoregion of the Bolivian Amazon basin. Comparable to most of the Brazilian literature, this study finds that fire activity and fire season length is higher in the flooded Beni Savanna, and Chiquitano seasonally dry tropical forests than in the Bolivian Amazon ecoregion. This study also identifies moisture stress and human activity as the main drivers of fire dynamics within the region. It is intended that this research will offer a foundation for future research and conservation activities aimed at better understanding the fire regime of the Bolivian Amazon basin.
... The results in Table 4 reveal that at the provincial (P) scale, the digital elevation model and the relief degree of the land surface were the key factors associated with the variations in the regression coefficients of the four driving forces in Sichuan Province. Topography, as a fundamental physical element, plays an essential role in determining the distribution of land use [8,73,74], human activities [26,75], vegetation [76][77][78], and water resources [79] across the world. In Sichuan Province, there are a plethora of landform types, including plateaus, mountains, hills, basins, and plains. ...
Full-text available
Driving forces are the factors that lead to the observed changes in the quantity and quality of ecosystem services (ESs). The relationship between driving forces and ESs involves considerable scale-related information. Place-based ecological management requires this information to support local sustainable development. Despite the importance of scale in ES research, most studies have only examined the association between ESs and their drivers at a single level, and few studies have examined this relationship at various scales or analyzed spatial heterogeneity. The purpose of this paper is to explore the significance of the scale-dependent effects of drivers on ESs for localized ecological management. The biophysical values of ESs were calculated using several ecological simulation models. The effects of driving forces on ESs were explored using the geographically weighted regression (GWR) model. Variations in the effects of driving forces on ESs were examined at three scales: provincial, ecoregional, and subecoregional scales. Finally, canonical correlation analysis was used to identify the major environmental factors associated with these variations in each ecoregion. Our results show that (1) the distribution of soil conservation and water yield is highly heterogeneous; (2) four driving forces have significant positive and negative impacts on soil conservation and water yield, and their effects on the two services vary spatially (p < 0.05); (3) the impacts of drivers on ESs vary across different spatial scales, with a corresponding shift in the related environmental factors; and (4) in the study area, at the provincial scale, physical, topographical, and biophysical factors were key factors associated with the variations in the relationship between ESs and drivers, and at the ecoregional and subecoregional scales, physical, socioeconomic, topographical, and biophysical factors all contributed to these changes. Our results suggest that significant differences in topographical conditions (e.g., altitude, slope) can be incorporated for exploring the relationship between drivers and ESs and optimizing ecological management at the provincial scale, whereas significant differences in physical and socioeconomic conditions (e.g., urbanization levels, human activity, vegetation coverage) are more meaningful for localized ecological management at the ecoregional and subecological scales. These findings provide a basis for understanding the relationship between drivers and ESs at multiple scales as well as guidelines for improving localized ecological management and achieving sustainable development.
... The magnitude of anthropogenic influence has been identified as an important fire driver in the Brazilian Atlantic forest [3]. To comprehend the spatial patterns, causes and ecological impacts of fire, various components of the fire regime must be examined [57]. This knowledge is essential when developing strategies for combating dynamic natural fire in the future. ...
Full-text available
The fire susceptibility of the Atlantic Forest has largely increased over the past two decades due to a combination of climate change and anthropogenic factors such as land cover change and human modification. High rates of forest fragmentation have contributed to escalating fires in this imperilled global biodiversity hotspot. Understanding fire patterns is essential to developing an effective forest fire management strategy. In this research, we utilized the Random Forest (RF) machine learning approach for identifying the role of climatic and anthropogenic factors in influencing fire occurrence probability and mapping the spatial distribution of fire risk. We found that the Normalized Difference Vegetation Index value and climate variables (i.e., temperature and solar radiation) were significant drivers of fire occurrence risk. Results also confirm that forest fragmentation increases with fire density in the region.
... These results highlight the indirect effect of rainfall on fire severity in more arid ecosystems, through its influence on biomass productivity. Severity patterns in response to rainfall dynamics can, however, be partially offset by the effect of vegetation composition and structure in response to land use/land cover changes and fire-suppression policies, fire history, and topography (Dillon et al., 2011;Estes, Knapp, Skinner, Miller, & Preisler, 2017;Fang, Yang, White, & Liu, 2018;García-Llamas et al., 2019;Liu & Wimberly, 2015). Fuel is the only factor that can be modified through land management aiming at reducing fuel accumulation, through silvicultural operations such as thinning and pruning and modifying fuel structure based on the creation open patches to reduce fuel continuity. ...
Active volcanoes continuously register morphological changes during their activity by the emplacement of lava flows and pyroclastic loose material on their flanks. Precipitation represents an external agent that reshapes the volcano edifice by remobilizing large volumes of ash and pyroclastic flow deposits producing lahars, sediment-water, or/and gravity-driven flows. Depending on the regional climatic conditions, lahars can be a very common phenomenon, highly frequent just after an eruption episode and they can persist for several years. The most common lahar initiation mechanism corresponds to the progressive water erosion of material from the catchment and the riverbed, up to bulk to a saturated mixture of sediment and water. Precipitation duration and intensity are the main factors in lahar generation. Orographic (or stationary) and regional (o migratory) rain events can both trigger lahars. In the first case, high intensity, short duration rains trigger low-volume (10*⁴–10*⁵ m³), low-duration (less than one hour) lahars; in contrast, extreme hydrometeorological events such hurricanes can trigger large volume (10*⁶ m³) and long-duration (several hours) lahars with catastrophic effects on the surrounding area. The hydrological response of the watershed feeding the ravine is a key factor in determining flow volume, flow discharge behavior, and morphological evolution of the drainage. On the other hand, during cataclysmic eruptions, with the formation of long-lasting, sustained eruptive columns, atmospheric conditions can rapidly change at the regional scale, inducing rainfall that can quickly trigger syneruptive lahars also promoted by the accumulation of fine ash layers on the volcano slopes with hydrophobic behavior. In addition, the large amounts of SO2 gas released during catastrophic explosive eruption can induce climate changes, with an important impact on the Global Monson rainfalls, changing their distribution for a few years, on a global scale. Studying the interaction between volcanic activity and precipitation will help to understand and model lahar initiation processes and their behavior and will allow to establish more accurately future hazard scenarios.
... These results highlight the indirect effect of rainfall on fire severity in more arid ecosystems, through its influence on biomass productivity. Severity patterns in response to rainfall dynamics can, however, be partially offset by the effect of vegetation composition and structure in response to land use/land cover changes and fire-suppression policies, fire history, and topography (Dillon et al., 2011;Estes, Knapp, Skinner, Miller, & Preisler, 2017;Fang, Yang, White, & Liu, 2018;García-Llamas et al., 2019;Liu & Wimberly, 2015). Fuel is the only factor that can be modified through land management aiming at reducing fuel accumulation, through silvicultural operations such as thinning and pruning and modifying fuel structure based on the creation open patches to reduce fuel continuity. ...
Climate tends toward drier conditions in most regions, particularly in the Mediterranean Basin. These changes may affect forest fire regime parameters, such as severity. Rainfall events also control postfire regeneration, mainly in the early stages, and water availability may lead to differential response of the vegetation and could induce changes in the proportion of plant functional types. Postfire rainfall type and intensity could affect the ecosystems in two different ways. On the one hand, it may favor the regeneration of vegetation (gentle rain), and on the other hand, it could lead to loss of nutrients and seeds due to surface runoff (heavy rains), thus limiting the regeneration of the ecosystems. Therefore the application of emergency postfire rehabilitation treatments is often recommended in severely burned areas. These treatments should adequately protect the soil while not impeding natural plant regeneration or enhancing the establishment of exotic species.
Orman yangınları dünya üzerindeki ekosistemleri etkileyen doğal bir sürecin parçasıdır. Yangın, mikro ölçekli fenomenden geniş peyzaj desenleri ve süreçlerine kadar birden fazla zamansal ve mekânsal ölçekte biyofiziksel süreçleri etkiler. Yöneticiler yangın sonrası karar verme süreçlerini gerçekleştirebilmek için yanan alanların karakterizasyonunu iyi yapabilmelidir. Bu ise hem bitki örtüsü hem de toprakta yangının neden olduğu ekolojik değişimin derecesi olarak kabul edilen ve yangın rejiminin en önemli bileşenlerinden olan yanma derinliğinin tespiti ile mümkündür. Yanma derinliği, yanıcı madde tüketimi derecesinin görsel olarak gözlemlenmesi, yangından sonra tüketilmemiş bitki ve toprak yüzeylerindeki kömürleşme miktarı temelinde sınıflandırılabilir. Bu sınıflandırmalar genellikle yanmamış, az yanmış, orta derecede yanmış ve çok yanmış alanlara ayrılarak yapılır. Bazen yangınlar çok büyük alanlarda meydana gelir ve bu alanlarda zarar tespitleri yapmak çok fazla emek, para ve zaman gerektirebilir. Bu yüzden büyük yangınlar sonrası yangının ekosisteme verdiği zararın derecesi belirlenirken uzaktan algılama yöntemleri sıklıkla kullanılır. Yanma derinliği sınıflandırmaları genellikle optik uzaktan algılama verilerinden türetilen spektral indeksler ile ya da SAR ve LiDAR gibi aktif uzaktan algılama yöntemlerinden elde edilen haritalar ile ifade edilir. Yanma derinliğini sınıflandıran haritalar, orman yangınların toprak, su, ekosistem florası ve faunası, atmosfer üzerindeki etkilerini tanımlayabilir ve yangınlar sonucu ortaya çıkan farklı derecelerde yanmış/ yanmamış alanların sürdürebilir ekolojik bir yaklaşım ile planlanmasında kullanılabilir. Bu çalışmada yanma derinliği kavramı ve yanma derinliği sınıflandırmalarının aşamaları hakkında bilgiler verilmiş, bu konu hakkında bundan sonra yapılacak çalışmalar için literatür özetlenerek konunun iyi ve eksik yönleri tartışılmıştır.
Bandwidth plays a crucial role in the Geographically Weighted Regression modelas it affects the model’s ability to describe spatial dependencies. If the bandwidth is too large, the model will be similar to a normal regression model. Conversely, if it is too small, the model will be too rough. Bandwidth can be selected in several ways, e.g. manually determined by experts or using Akaike Information Criteria, Cross-Validation, and Lagrange Multiplier methods. This study offers an alternative approach to choosing bandwidth based on the covariance function representing a linear combination between the Bessel and Gaussian-Type functions. We applied this function to analyze the land price in Manado with four infrastructure accessibility variables, such as accessibility to government offices, education facilities, shopping centers, and healthcare facilities. Therefore, the proposed method is different from the index methods (AIC and CV) which have been used by other researchers. The results showed that the non-parametric covariance function provides a smaller bandwidth than conventional methods, specifically Akaike Information Criteria and Cross-Validation. In addition, the value of R ² (adjusted) given by the covariance function is greater than the one given by the proportional method. This means that the optimal bandwidth obtained using the covariance function is more suitable to explain the land price in the city of Manado.
Full-text available
O Brasil tem vivenciado incêndios sem precedentes na última década. Imagens de imensas áreas queimadas ou animais mortos que não conseguiram escapar dos incêndios em 2020 chocaram o mundo. Para prevenir ou minimizar desastres similares no futuro nós devemos compreender os fatores que levaram a esses eventos catastróficos. As causas e consequências dos incêndios estão vinculadas a interações complexas entre as esferas biofísica e sociocultural, e decisões acertadas de gestão requerem uma sólida base científica. Apresentamos aqui o panorama recente do aumento dos focos de incêndio nos biomas brasileiros e discutimos as causas que têm contribuído para esses incêndios, seus impactos no meio ambiente e as consequências gerais para o bem-estar humano, com base numa extensa revisão da literatura sobre o assunto, no conhecimento especializado dos autores e informações fornecidas por gestores ambientais, pesquisadores e políticos durante uma oficina organizada para debater o tema dos incêndios no Brasil. Esta revisão atualizada é direcionada ao público acadêmico, gestores ambientais, tomadores de decisão e formuladores de políticas. Num primeiro momento, apresentamos evidências dos contrastantes efeitos do fogo nos diferentes ecossistemas. A seguir, traçamos, numa perspectiva histórica, a percepção e as principais políticas relacionadas ao uso e manejo do fogo no Brasil, desde a colonização até o presente. Depois, propomos meios de avançar na prevenção e desenvolvimento de estratégias de gestão do fogo. Finalmente, nós procuramos esclarecer e/ou desmistificar algumas questões nem sempre apropriadamente abordadas pela mídia
Full-text available
Context Wildfire is a particular concern in the wildland-urban interface (WUI) of the western United States where human development occurs close to flammable natural vegetation. Objectives (1) Assess the relative influences of WUI expansion versus climate-driven fire regime change on spatial and temporal patterns of burned WUI, and (2) determine whether WUI developed in the future will have higher or lower wildfire risk than existing WUI. Methods We projected the spatial pattern of the WUI and its associated wildfire risk from 2005 to 2050 at 90-m spatial resolution and 5-year intervals in Colorado Front Range using CHANGE, a landscape change model that simulates land cover and land use change, natural vegetation dynamics, and wildfire in a unified framework. A total of four scenarios from a factorial design with static versus changing WUI and static versus changing fire regimes were simulated to examine the effects of WUI expansion and climate-driven fire regime change on burned area in the WUI. Results Both WUI expansion and fire regime change contributed to the increase of burned WUI, but fire regime change had a stronger influence. The effects of WUI expansion and fire regime change had a combined influence greater than the sum of their individual effects. This interaction was a result of projected WUI expansion into regions of higher wildfire risk than existing WUI. Conclusions The human footprint will continue to expand into wildland areas and must be considered along with climate effects when assessing the impacts of changing fire regimes in future landscapes.
Full-text available
Fire is one of the most important natural disturbance processes in the western United States and ecosystems differ markedly with respect to their ecological and evolutionary relationships with fire. Reference fire regimes in forested ecosystems can be categorized along a gradient ranging from "fuellimited" to "climate-limited" where the former types are often characterized by frequent, lower-severity wildfires and the latter by infrequent, more severe wildfires. Using spatial data on fire severity from 1984- 2011 and metrics related to fire frequency, we tested how divergence from historic (pre-Euroamerican settlement) fire frequencies due to a century of fire suppression influences rates of high-severity fire in five forest types in California. With some variation among bioregions, our results suggest that fires in forest types characterized by fuel-limited fire regimes (e.g., yellow pine and mixed conifer forest) tend to burn with greater proportions of high-severity fire as either time since last fire or the mean modern fire return interval (FRI) increases. Two intermediate fire regime types (mixed evergreen and bigcone Douglas-fir) showed a similar relationship between fire frequency and fire severity. However, red fir and redwood forests, which are characterized by more climate-limited fire regimes, did not show significant positive relationships between FRI and fire severity. This analysis provides strong evidence that for fuel-limited fire regimes, lack of fire leads to increasing rates of high-severity burning. Our study also substantiates the general validity of "fuel-limited" vs. "climate-limited" explanations of differing patterns of fire effects and response in forest types of the western US.
Full-text available
Increased wildfire activity (e.g. number of starts, area burned, fire behaviour) across the western United States in recent decades has heightened interest in resolving climate-fire relationships. Macroscale climate-fire relationships were examined in forested and non-forested lands for eight Geographic Area Coordination Centers in the western United States, using area burned derived from the Monitoring Trends in Burn Severity dataset (1984-2010). Fire-specific biophysical variables including fire danger and water balance metrics were considered in addition to standard climate variables of monthly temperature, precipitation and drought indices to explicitly determine their optimal capacity to explain interannual variability in area burned. Biophysical variables tied to the depletion of fuel and soil moisture and prolonged periods of elevated fire-danger had stronger correlations to area burned than standard variables antecedent to or during the fire season, particularly in forested systems. Antecedent climate-fire relationships exhibited inter-region commonality with area burned in forested lands correlated with winter snow water equivalent and emergent drought in late spring. Area burned in non-forested lands correlated with moisture availability in the growing season preceding the fire year. Despite differences in the role of antecedent climate in preconditioning fuels, synchronous regional fire activity in forested and non-forested lands suggests that atmospheric conditions during the fire season unify fire activity and can compound or supersede antecedent climatic stressors. Collectively, climate-fire relationships viewed through the lens of biophysical variables provide a more direct link to fuel flammability and wildfire activity than standard climate variables, thereby narrowing the gap in incorporating top-down climatic factors between empirical and process-based fire models.
Full-text available
Wildfire area and severity are projected to increase in Eurasian larch (Larix spp.) forest, potentially altering ecosystem dynamics and carbon balance under warming climate. However, the ecological drivers that influence post-fire ecosystem productivity in larch forest are poorly understood. Here we assess the influence of several factors controlling plant aboveground net primary productivity (ANPP) on the early successional stage of Dahurian larch (Larix gmelinii) forests following fire. We quantified the relative importance of fire severity, competition, topography, and soil inorganic N [NH4+ and NO3-] availability on ecosystem productivity 11 years post-fire in a Dahurian larch forest of Northeast China using a boosted regression trees technique. Results showed that fire severity had the strongest negative influence on ANPP of saplings, followed by a positive influence of soil inorganic N availability and negative influence of elevation. Sapling density had the strongest negative influence on ANPP of understory (shrub and grass), followed by positive influence of aspect and inorganic N availability. N may limit ANPP of sapling, but not of understory vegetation. Fire severity had a dominant influence on the proportion of total ANPP contributed by the understory. Fire severity was positively correlated to understory productivity and proportion of total ANPP contributed by understory, but negatively correlated to sapling productivity. Our analysis suggests that increasing fire severity may increase understory productivity, while decreasing the sapling productivity, which may contribute to the potential changes of vegetation type and carbon cycling in Dahurian larch forests of Northeast China under warming climate.
Global pyrogeographic study is necessary to inform climate change impact assessments used for management and decision-making. Climate is a strong driver of spatial and temporal patterns of fire such that ongoing climate change is expected to alter global fire activity. A growing number of statistical-correlative analyses examine environmental drivers of current patterns of global fire occurrence or burned area, but few studies ask important "what if" questions about the potential future of fire under scenarios of a changing climate. Accordingly, our goal is to engage the broader statistical community in analysis of global fire data products to spur further understanding of fire regimes and the complex links they demonstrate between the biosphere and atmosphere. We provide an overview of constraints over fire regimes and the role of fire in the biosphere-atmosphere, describe general approaches being used for global fire-climate assessment, summarize opportunities and pitfalls in the public-access global fire datasets, and highlight thinking on next steps for analysis of global fire and fire regime data.