Translator Disclaimer
1 August 2015 Climate Contributors to Forest Mosaics: Ecological Persistence Following Wildfire
Author Affiliations +
Abstract

It is hypothesized that climate impacts forest mosaics through dynamic ecological processes such as wildfires. However, climate-fire research has primarily focused on understanding drivers of fire frequency and area burned, largely due to scale mismatches and limited data availability. Recent datasets, however, allow for the investigation of climate influences on ecological patch metrics across broad regions independent of area burned and at finer scale. One area of particular interest is the distribution of fire refugia within wildfire perimeters. Although much recent research emphasis has been placed on high-severity patches within wildfires, unburned and low-severity patches provide critical remnant habitat and serve as seed sources to initiate colonization and succession in recently burned landscapes. These patches of persistence also may yield insights into approaches for developing fire-resilient landscapes by forest managers and communities seeking to reduce wildfire hazard. Here, we present results showing no decline in proportion of persistent patches in three study areas surrounding National Parks in the western United States, even as research and anecdotal information suggests that fires have become larger and more severe. We also show climate linkages to metrics of persistence that echo previous findings in climate-fire research, and we introduce a framework for addressing global change impacts on forest pattern more broadly. Specifically, we discuss the interactions of multiple drivers at landscape scales and the need to disaggregate relative influences using mixed methods that can address both social and ecological phenomenon.

Introduction

There is widespread evidence that fire regimes are changing in response to a myriad of drivers, with negative impacts to human lives, infrastructure, and ecosystem goods and services (Butry et al. 2001, Rocca et al. 2014). Fire activity is increasing, as larger fires with a longer duration are becoming more frequent (Westerling et al. 2006, Bowman et al. 2011, Dennison et al. 2014). There have also been preliminary assessments that wildfires have become more severe (Miller and Safford 2012), although these suggestions are based on widelydebated definitions and proxies of fire severity (Key and Benson 2006, Lentile et al. 2006, French et al. 2008, Keeley 2009, Kolden and Rogan 2013), and on relatively short time series of data in only a limited regional scope.

Although definitive empirical evidence currently lags, numerous studies have advanced the idea of increasing fire severity by identifying likely drivers of such a trend. These drivers include climate change, fire meteorology, impact of fungal and insect outbreaks, and the biogeographic extents of plant species distributions (Dale et al. 2001, Allen et al. 2010, Crimmins et al. 2011); a century of fire exclusion producing increased tree density and higher fuel loading in forests where frequent fire maintained lower tree densities prior to Euro American settlement (Arno 1980, Covington and Moore 1994; Marlon et al., 2012); the introduction and spread of more flammable invasive species (Abatzoglou and Kolden 2011a, Balch et al. 2013); land use changes such as altered grazing-related reduction of fine fuel loading in rangeland ecosystems (Zimmerman and Neuenschwander 1984); and changes in fire management and suppression practices (Pyne 2001, Millar et al. 2007, Pyne 2010). The contributions of fire management are particularly difficult to identify because management impacts fire severity before, during, and after the fire; by changing fuel loadings and structure pre-fire (Stephens and Moghaddas 2005, Prichard and Kennedy 2014), suppressing wildfire using burnout techniques and backfires that can produce higher severity impacts than the main fire, and conducting post-fire mitigation efforts that increase post-fire vegetative cover at greater rates than would naturally occur, such as aerial seeding (Robichaud et al. 2000). Of the many potential drivers of fire severity, however, climate variability and change remain one of the most critical to explore due to the observed and projected trajectory of anthropogenic climate change and the associated fire effects (Westerling et al. 2006, Littell et al. 2009).

Climate-fire research utilizing contemporary area burned as the dependent variable has generally found two dominant modes: either fire activity is predicted by antecedent conditions prior to the fire season, or it is correlated to concurrent conditions during the peak of the fire season (Westerling et al. 2003, Littell et al. 2009, Abatzoglou and Kolden 2013). In the former mode, climatic relationships to fire are often explained as conditions conducive to either an increase in fine fuel abundance (e.g., wetter-than-normal winters in fine fuel-limited ecosystems [Swetnam and Betancourt 1990, Littell et al. 2009, Abatzoglou and Kolden 2013]) or conditions conducive to relatively low live and large dead fuel moistures (e.g., early snowmelt leading to mid-summer drought stress in living trees [Westerling et al., 2006]). In the latter mode, strong climate-fire relationships are attributed to fire meteorology conducive to large fire spread (e.g., a delayed onset of summer precipitation resulting in relatively high temperatures and low relative humidities that lower fine fuel moistures past critical thresholds [Abatzoglou and Kolden 2011b]).

Climate-fire research to-date has primarily utilized area burned as the fire variable, treating fire as simply present or absent and ignoring the role fire plays in controlling heterogeneous forest pattern at finer-scale resolution (Larson and Churchill 2012, Kane et al. 2013, Lutz et al. 2013). Relatively recent development of consistent, higher spatial resolution fire data sets has allowed for quantification and characterization of heterogeneous burn severity patterns and trends (Table 1), including a few studies investigating climate-fire severity relationships (Lutz et al. 2009, Dillon et al. 2011, Cansler and McKenzie 2014, Birch et al. 2015). The research seeking to characterize fire severity trends, in particular, has focused on the highest severity class, which is usually associated with stand-replacement in small to large patches (Turner et al. 1994, Holden et al. 2007, Miller et al. 2009, Dillon et al. 2011, Miller and Safford 2012). This focus results from both the magnitude of ecological change associated with severe fire and the associated hazards to resources-at-risk, such as loss of forest canopy and potential landslide events (Turner et al. 1994, Pierce et al. 2004).

The emphasis on high severity largely ignores the impacts of fire at the opposite end of the spectrum. Areas classified as ‘unchanged’ or ‘low severity’ within the larger fire perimeter that are either entirely unburned or experience minimal impacts of lower intensity fire. These relatively unchanged areas have been described and defined in several ways; most commonly as either “fire refugia” (Camp et al. 1997, Swengel and Swengel 2006, Keppel et al. 2012) or as unburned islands (Kolden and Weisberg 2007, Kolden et al. 2012).

We posit that another approach to defining these unburned and low severity fire effects is to link them to forest process and pattern through the utilization of State and Transition (S&T) models. S&T models identify succession states and transitional pathways between them and quantify the temporal and spatial probability of landscapes existing in a state or transitioning to a new state. For example, the Vegetation Dynamics Development Tool (VDDT) is a S&T model that was utilized during the development of the LANDFIRE program to characterize fire regimes (Rollins 2009; Figure 1). It describes the average residence time for a vegetation type in a given succession stage (e.g., early, mid, or late age; open or closed canopy) and the probability of transitioning to a new state based on disturbance frequency intervals. While the probabilities of change and state residency times have been portrayed as static, these probabilities themselves should change in response to longterm forcing mechanisms. For example, a model frequency probability of 0.002 for stand-replacing fire would double if trend analysis revealed such an increase across that ecotype over a sufficient temporal interval in response to an external driver like climate change.

TABLE 1.

Examples of peer-reviewed studies which have quantified the within-fire burn severity pattern using spatial statistics, and statistically analyzed or summarized these patterns in relation to one or more variables that may influence them. Metrics include mean patch size (MP), other metrics calculated at the scale of the patch (PA), measurements of patch interior, core area, or distance to the edge of a patch (IN), metrics calculated at the scale of the pixel (PI), measurements of the amount or density of edges between different patch types (ED), and metrics measuring the diversity or interspersion of different class types across the landscape (DI). For studies that included a complete census of fires down to a minimum fire size cutoff “years” refers to total years in study period; for studies that included only a sample of fires, years refers to the total individual years that during which sampled fires occurred.

t01_219.gif

Figure 1.

Conceptual model of succession showing how differential fire severity produces a landscape mosaic comprising of multiple successional stages. Each line color represents differential fire severity and effects, and the successional outcome of that severity.

f01_219.jpg

Within an S&T framework, sites that remain unburned or experience such low severity fire that they do not change successional state as a result of a fire event are said to ‘persist’ in the same state; we therefore define these areas for this study as ‘persistent patches.’ Although persistent patches are critical ecologically as post-fire seed sources, microclimatic havens for seedlings, wildlife habitat refugia, and buffers to downslope surface erosion (Turner et al. 1997, DeLong and Kessler 2000), the factors contributing to their development are poorly understood. In contrast to the strong area burned-severity relationships identified by Miller et al. (2009), Kolden et al. (2012) found that unburned proportion within wildfire perimeters corresponded only weakly to fire size, and found significantly different average size and density of unburned patches between a mixed-severity fire regime in California and two stand-replacing fire regimes in Montana and Alaska. This result could be explained by a variety of determinants across a range of spatial scales (e.g., management histories, forest composition, underlying geology, etc.), but one such potential larger-scale driver is well-documented changes in regional climate (Abatzoglou et al. 2014). Because large regional studies of burn severity have found that aggregated burn severity atlases tend to have a normal distribution for forested ecosystems (Kolden 2010, Thode et al., 2011), a trend towards higher severity could equally be represented as either a shift to the right or a change in the shape of the fire severity distribution, resulting in potential changes in the proportion of both high-severity and persistent patches (Figure 2).

Figure 2.

Conceptual model of the distribution of fire severity for an aggregated fire regime or region (i.e., not a single fire) for both the historical period and three potential future models: (a) one based on a shift in the normal distribution towards higher severity, (b) one based on a greater range of fire effects, and (c) one based on a change in shape to a bimodal distribution dominated by high-severity and persistent patches. Thresholds for areas of persistence and high severity show the respective increases and decrease in persistent proportion and high severity for each conceptual model.

f02_219.jpg

Here we examine whether climate is a significant determinant of the pattern of persistent patches within fire perimeters, and if climate-persistent patch relationships reflect climate-area burned relationships more broadly. Following prior studies (Table 1) we utilize both the Severity Metric (SM; Lutz et al. 2011) as a measure of overall fire severity and four measures of persistent patch pattern, including persistent proportion within the fire perimeter, median persistent patch size (ha), area-weighted mean persistent patch size (ha), and persistent patch density (patches/ha). We answer three specific questions: 1) is overall fire severity correlated to persistent patch metrics?, 2) are climate variables significant predictors of persistent patch metrics?, and 3) do temporal trends exist for persistent patch metrics?

Methods

Our study area included forested zones around and including three protected areas in the western US: Glacier National Park in the northern Rocky Mountains of western Montana, North Cascades National Park and adjacent wilderness areas in north-central Washington, and Yosemite National Park in the central Sierra Nevada of California (Figure 3). For Glacier and Yosemite, a buffer distance around the National Park boundaries was identified and applied, allowing us to include fires that either began in or ultimately burned into the respective study site; this buffer was drawn to exclude non-forested areas. The full extents are described in Kolden et al. (2012) for Glacier and Yosemite and Cansler and McKenzie (2014) for North Cascades. These three sites represent different dominant forest ecotypes and fire regimes in the western US: the Northern Rockies mountain forest (historically dominated by stand-replacing fire), the Cascade mixed conifer forest (mixedseverity fire), and the Sierra Nevada mixed conifer forest (low-to-mixed-severity fire) (Bailey 1998, Kolden et al. 2012, Cansler and McKenzie 2014). They are also characterized by minimal land use change over the past century, and have relatively lower active fire suppression compared to the surrounding national forests.

Figure 3.

Western US study area locating the three sites analyzed in the present effort: Yosemite National Park in California, Glacier National Park in Montana, and North Cascades National Park and surrounding wilderness in Washington.

f03_219.jpg

Fire Data

Regional-scale landscape assessment of fire pattern over time has only recently been made possible with the development of fire atlases from relatively fine resolution (30 m) data acquired by the Landsat Thematic Mapper (TM), Enhanced Thematic Mapper-plus (ETM+), and Operational Land Imager (OLI) sensors. These sensors are more commonly known by their numeric order as Landsat 4 and 5, Landsat 7, and Landsat 8, respectively. The entire Landsat scene archive was made available for free by USGS in 2007 ( http://landsat.usgs.gov). The Monitoring Trends in Bum Severity (MTBS) project utilizes these data beginning in 1984 to map fire effects and pattern on all large fires in the US (Eidenshink et al. 2007). We developed atlases for each study area independent of MTBS that included both wildland and prescribed fires down to much lower size thresholds: 20 ha for Glacier, 40 ha for North Cascades, and 40 ha for Yosemite, area thresholds chosen so that our fire atlases included > 95% of area burned in each study site. We utilized standard protocol followed by MTBS in developing these atlases from pre- and post-fire Landsat data transformed to the differenced Normalized Burn Ratio (dNBR), but with a focus on following best practices as outlined by Key (2006), which MTBS was not able to follow during the early part of the project due to high per-scene costs prior to 2007, and including normalization for phonological difference, which MTBS does not include in their dNBR product. Although some authors have advocated for the use of the relativized version of the dNBR (RdNBR) to study fire effects (Miller and Thode 2007), improved accuracy of RdNBR compared to dNBR occurs primarily in the high-severity range (Miller and Thode 2007), and RdNBR may have anomalously high or low values, causing misclassification of burn severity in areas where there is little pre-fire vegetation (Parks et al. 2014). Many fires in the study sites have areas of limited vegetation where snow and ice dominate the site more than half the year, and some of these locations are of interest as persistent islands. Therefore, we utilized dNBR, which has been validated as a proxy for burn severity in all three of the study areas (Thode 2005, Key and Benson 2006, Cansler and McKenzie 2012).

Our atlases include all fires above the size threshold from 1984 to 2009, including 145 fires in Glacier, 106 fires in North Cascades, and 151 fires in Yosemite. For each fire, we first classified persistent pixels as those with dNBR values between -100 and 100 (Kolden et al. 2012). We then identified patches of contiguous persistent pixels, and filtered out all persistent pixel patches with less than a minimum of four adjacent pixels to remove the noise inherent in spectral reflectance data associated with atmospheric scattering (Tanré et al. 1987) and scene registration mismatches. Only the remaining patches (> 3.6 ha) were used to calculate all metrics. Although the -100 to 100 dNBR thresholds have previously been identified as the thresholds for pixels ‘unchanged’ or ‘unburned’ by the fire in several studies (Key and Benson 2006, Cansler and McKenzie 2012, Kolden et al. 2012), Kolden et al. (2012) identified several conditions where fire impacts are seen, particularly with low to moderate severity, but a pixel would be spectrally unchanged.

Therefore, our consideration of persistent patches includes areas that are both unburned and those which have in fact been burned, but in such a manner as to be undistinguishable by remote sensing techniques classified from spectral data. First, we recognize that although discrete burn severity classes more easily translate to single metrics for science or management applications, fire effects are inherently continuous across the landscape, and usually only abruptly change where there is a physical boundary such as a river, road, or recently burned area. Second, we have found that in many closed-canopy forests, the effects of sub-canopy surface fire can go undetected by passive spectral reflectance data (Kane et al. 2015). Because the closed canopy is the defining feature of older forests, from an S&T modeling perspective, this low-to-moderate-severity fire effect results in persistent areas (Figure 1).

There have been numerous approaches taken by prior authors to translate continuous fire effects to individual, representative metrics of fire patterns (Table 1). These include statistical representations such as mean and proportions as well as landscape ecology metrics that describe fragmentation in classified data, with these measures being applied to varying spatial (e.g., pixel, polygon, fire, landscape) and temporal scales (e.g., fire duration, month, annual). Given the relatively coarse spatial scale of the climate data, we attempted to match it by calculating fire pattern metrics at the scale of the fire. From the original dNBR data we calculated the Severity Metric (SM; Lutz et al. 2011) of each fire to describe the overall fire severity but maintain the continuous nature of fire effects. From the classified data, we calculated four measures of the spatial arrangement of persistent areas within fires at the per-fire level: persistent proportion within the fire perimeter, median patch size (ha), area-weighted mean patch size (ha), and patch density (patches/ha). These metrics have been widely utilized in multiple studies of fire effects (Table 1) and are more easily interpretable in both ecological and management contexts.

Climate Data

In-situ meteorological observations of temperature, humidity, wind, precipitation and solar radiation are a limiting factor in broad-scale ecological analysis that incorporates atmospheric forcing. These problems are heightened in the western United States due to landscape heterogeneity and a sparse observational network. We overcome some of these limitations by using the daily surface meteorological dataset of Abatzoglou (2013) that provides high-resolution (2.5 arc minute ∼ 4 km) temperature, precipitation, 10 m wind velocity, solar radiation and humidity that have been cross-validated using a set of meteorological observations across the United States. We used these data to derive the Palmer Drought Severity Index (PDSI), fire danger indices from the National Fire Danger Rating System (NFDRS, Burgan 1988) and Canadian Forest Fire Danger Rating System (CFFDRS Van Wagner 1987), and calculated reference potential evapotranspiration (ETo) using the Penman-Montieth equation (Allen et al. 1998) with zero canopy stomatal resistance (e.g., Littell and Gwordz 2011). Finally, we ran a modified Thornthwaite water balance model (Willmott et al. 1985; implemented by Lutz et al. 2010 and Dobrowski et al. 2013) at monthly time steps with standard 150 mm soil water holding capacity to model monthly snow water equivalent (SWE), soil moisture, actual evapotranspiration and climatic water deficit (defined as the unmet atmospheric demand, or the difference between ETo and actual evapotranspiration).

For each fire, a suite of 18 climate variables was calculated as the average of the values of all of the 4 km voxels (volumetric pixels) covered by that fire, inclusive of pixels across which the fire perimeter fell. All values were calculated as anomalies referenced to the 1981–2010 normals for that location permitting comparison across the study sites (temperature in degrees C, precipitation and soil moisture as percent of normal, evapotranspiration in mm, vapor pressure deficit in kPa, and raw values for fire danger indices). We considered six variables that represent antecedent conditions for winter and spring months prior to the summer of the fire: winter (December through February) precipitation (PPT-DJF), winter temperature (T-DJF), spring (March through May) precipitation (PPT-MAM), spring temperature (T-MAM), late summer (July through September; JAS) Palmer Drought Severity Index (PDSI; a normalized measure of long-term drought), and March-April snow-water equivalent total (Snow-MA). The remaining variables correspond to concurrent conditions during the season in which the fire burned, and consist of a mix of variables describing synoptic conditions, fire danger, and vegetation-based moisture availability. These include late summer (JAS) vapor pressure deficit (VPD), early summer (June through August) precipitation (PPT-JJA), late summer precipitation (PPT-JAS), early summer temperature (T-JJA), late summer temperature (T-JAS); late summer Energy Release Component (ERC) and Burning Index (BI) from the NFDRS; late summer Duff Moisture Code (DMC) from the CFFDRS ; and late summer Potential Evapotranspiration (POTET), Actual Evapotranspiration (AET), Water Deficit (DEF), and soil moisture (Soil-JAS) from the water balance model.

Analysis

We first conducted cross-correlation analysis to assess relationships between overall fire severity and persistent patches. We then correlated climate predictor variables with persistent patch metrics. For all correlation analyses, we calculated Spearman's Rho rank correlation coefficients. Spearman's Rho rank correlation has the advantage of being more robust to outliers (which occur often in fire datasets) than a Pearson correlation, and is appropriate for data that are not normally distributed. Significance was defined as p < 0.05. We transformed non-linear variables (median patch size, patch density, and area-weighted mean patch size) using a log transform. We developed time series of each of our five fire metrics by calcu-lating the mean of all fires for a given year. One drawback of this averaging process is it provides equal weighting to each year, whether there are only a few fires or many. We suggest that because we are examining climate interannual variability that is expressed across broad spatial scales likely exceeding the study area extents, significant trends should be evident despite the averaging process. We then developed an ordinary least squares regression trend line and calculated the mean absolute percentage error (MAPE; a MAPE value of zero is a perfect fit with zero error).

Results

Fire severity (SM) was significantly negatively correlated with the proportion of persistent area within fires and significantly negatively correlated to median persistent patch size for all three study areas (Table 2). Greater fire severity (higher SM) was linked to significantly higher persistent patch density in Yosemite, lower persistent patch density for Glacier, and was not significant for North Cascades.

Of the original 18 climate variables, only two were not significantly correlated to any metric of persistent patches (T-DJF and P-DJF). The three strongest statistically significant correlation coefficients for each fire metric by park are reported in Table 3; where there are less than three coefficients reported, all significant relationships are presented. In Yosemite, lower spring snow and precipitation was correlated to higher areaweighted mean persistent patch size and lower persistent patch density, while lower summer precipitation was correlated to a greater severity metric, smaller persistent proportion, and smaller median persistent patch size. By contrast, the primary predictors of persistent patches in Glacier and North Cascades were summer conditions. In Glacier, summer precipitation and soil moisture were negatively correlated to fire severity (SM) and persistent patch size, and positively correlated to persistent patch density. Wind also played a factor in Glacier, as BI was positively correlated to both persistent patch size metrics. In North Cascades, drier summer conditions (lower precipitation, soil moisture and évapotranspiration metrics) were correlated to higher fire severity (SM), lower persistent proportion, smaller median persistent patch size, larger area-weighted mean persistent patch size, and lower persistent patch density. Although nearly all of the climate variables were significantly correlated to some metric of persistent pattern, the correlations were relatively weak, with the highest correlation coefficient only -0.415 (between area-weighted mean persistent patch size and PDSI in Glacier).

TABLE 2.

Spearman's Rho rank correlation coefficients relating the severity metric (SM) to the persistent proportion, median patch size, and patch density for each of the three parks. For all coefficients, p ≤ 0.001. *Denotes no significant relationship.

t02_219.gif

No significant trends were found for the time series. Trends in fire severity (SM), proportion persistent, median persistent patch size, areaweighted mean persistent patch size, and persistent patch density were interannually variable (Figure 4). The lowest MAPE values found were for trends in persistent patch density and fire severity (SM) (Table 4), but the lowest MAPE value was only 13.02 associated with a weak positive trend in Yosemite persistent patch density, indicating greater than 13 percent error in the trend.

Discussion

Climate metrics were weak but significant predictors of the spatial pattern of persistent patches, suggesting that large-scale climatic influences have an influence on the mosaic of fire effects across the three forested ecosystems, but that other drivers are also important. As expected, the nature of these relationships varied by study area, consistent with the broad literature identifying differential relationships between area burned and climatic anomalies across the study area. Overall, Glacier exhibited the strongest correlations between persistent patch metrics and climate, particularly between relatively drier and windier summers and larger persistent patch sizes. These results are consistent with findings from Abatzoglou and Kolden (2013) that the strongest linkages between climate and area burned were in the Northern Rockies region, and with concurrent climatic anomalies rather than antecedent conditions (Westerling et al. 2003, Littell et al. 2009). Because this study focused on persistent patch metrics instead of area burned, we interpret the agreement in fire-climate relationships to be consistent with the notion of regional climate controls on fire activity in general, potentially mediated through fuel availability to burn.

TABLE 3.

Spearman's Rho rank correlation coefficients and p-values for each of the three study areas between five wildfire metrics of persistence (Severity Metric [SM], proportion persistent [PERS], median patch size [MEDP], Area-weighted mean patch size [AWMP], patch density [DENS] and 15 climate variables. Climate variables calculated by seasons are denoted: March/April/May (MAM), June/July/August (JJA), July/August/September (JAS), and December/January/ February (DJF). The top three predictors significant at the p < 0.05 level are reported.

t03_219.gif

Figure 4.

Time series of annual means for Severity Metric (SM), proportion persistent (PERS), log of median patch size (MEDP), and log of persistent patch density (DENS) over the 1984–2009 study period for each of the three study areas. Individual years identified with markers, trend lines (none significant) superimposed for Yosemite (blue circles, short dash line), Glacier (red triangles, solid line), and North Cascades (black diamonds, long dash line).

f04_219.jpg

TABLE 4.

Trend line equations and mean absolute percentage error (MAPE) for severity metric (0–1), persistent proportion (proportion), patch density (patches/ha), median patch size (square meters), and area-weighted mean patch size (m2) for the three sites over the 26-year study period (1984–2009). A larger MAPE value indicates greater variation around the regression line.

t04_219.gif

Climate Predictors of Persistent Patches

In Glacier, increased fire severity and lower persistent proportion were only weakly associated with lower summer soil moisture. Stronger associations were found between persistent patch metrics and numerous metrics that represented warmer, drier summer conditions; high fire danger concurrent to the fire was associated with lower persistent patch density and larger area-weighted mean patch size. It is notable that the Burning Index, the only metric that incorporates wind, was only significantly positively correlated with persistent patch size and density in Glacier, where the fire regime is characterized by stand-replacing fire, but not the other two sites, which are characterized by more mixed-severity fire regimes. This contrasts studies identifying wind as significantly correlated to fire size in each of the three ecoregions (Abatzoglou and Kolden 2013), and suggests that wind may play a role in Glacier in producing fast-moving fire “runs” that skip over large patches, a hypothesis that is supported by Birch et al.'s (2014) findings that large fire “runs” in the Northern Rockies are not significantly higher severity than smaller runs. Because BI here was a summer average, we interpret the results as a windier summer being associated with fire behavior that produces fewer, but larger persistent patches. This is consistent with findings from Birch et al. (2015) that wind speed is a predictor of higher severity on large fires in the Northern Rockies.

In North Cascades, greater proportion persistent was linked primarily to higher summer moisture availability, through summer precipitation, evapotranspiration, and soil moisture. Drier summers were associated with lower density of persistent patches and smaller persistent patch sizes, although a positive relationship between summer precipitation and area-weighted mean persistent patch size suggests that outliers are potentially influencing the drought-patch size relationship. The emphasis on summer climatology is consistent with other findings for northwest forested areas that point to concurrent summer drought as a driver of area burned (Abatzoglou and Kolden 2013, Cansler and McKenzie 2014). Although Cansler and McKenzie (2014) found that some winter and spring conditions also significantly contributed to area burned for the North Cascades study area, they note that not every summer is dry enough to burn (in contrast to an area like Yosemite, where every summer is dry enough to burn). Thus, the summer variables dominate, and the winter and spring variables are secondary contributors to the initial requirement that the summer be dry enough to burn.

In Yosemite, spring and summer moisture both contributed to persistent patch patterns. Cooler, wetter springs were associated with smaller persistent patch sizes and higher patch densities, while a wetter summer produced lower fire severity and higher proportion persistent. More severe fires were linked to smaller, higher density persistent patches. Lutz et al. (2009) found that lower spring snowpack was a predictor of larger fires with greater proportions of high severity, so our results point to the likelihood that late season and remnant snow patches, as well as the patterns of summer precipitation (which is almost entirely convective) play a role in controlling landscape severity mosaics. Microclimates at a finer scale than this analysis likely modify both vegetation types and flammability thus influencing the fire heterogeneity.

Trends in Persistent Patches

None of the three study areas demonstrated a strong trend over the 26-year study period. Although MAPE values are simply a percentage error and not a measure of significance, they can be interpreted as the inverse of percent confidence in a result. The lowest MAPE value was 13 percent (corresponding to 87% confidence in the result), suggesting relatively low confidence (87%) in a real trend. This is in contrast to Miller et al. (2009), the primary study that has examined trends in fire severity, who found a strong positive trend in high-severity fire over a similar period for a larger study area in California. Although Miller et al. (2009) utilized an Autoregressive Integrated Moving Average (ARIMA) with an 11-year moving average to assess trends in high severity for the Sierra Nevada region, we felt that it was inappropriate to apply similar methodology here for two reasons. First, our three study areas are small enough that there is considerable interannual variability in the number of fires, and because our metrics are distilled on a per fire basis, some years are smoothed by the effect of averaging numerous fires across a wide range, while other years are skewed by the occurrence of only one or two fires (Figure 5). Second, given the considerable interannual variability in our small number of fires and a relatively short temporal period, we felt it was inappropriate to use a moving average that would shorten the number of values in the trend, thus giving greater weight to the years of fewer fires. This is particularly problematic for Glacier, as six of the 26 years in the study were years without fires > 20 ha (thus, they became ‘no data’ years in our time series, and were removed from the analysis).

Implications

Our results suggest two broad implications. First, climatic conditions and anomalies do influence patterns of persistent areas, including both the proportion of the fire area that is persistent and the ecological metrics that are often of greatest concern to wildlife biologists and ecologists: patch size and patch density. Because many of the relationships we found to be significant are consistent with prior studies of climate and burned area (e.g., spring snowpack in Yosemite and summer drought in the Northern Rockies), even though this study analyzed fire patterns, we suggest that the same mechanisms by which climate influences burned area are also at play with regard to persistent patches. Although fine scale environmental and management factors (e.g., topography, suppression actions) likely also contribute to variance in fire pattern across the landscape, the regional fire-climate signals are strong enough to be significant. Because antecedent conditions associated with fuel growth were not significant (e.g., pluvial conditions the winter prior to the fire season in fuel-limited environments [Abatzoglou and Kolden 2013]), concurrent conditions controlling fuel moisture levels likely determine persistent patches in the forested ecosystems analyzed here by delineating which fuels are available to burn.

Across all three study areas, we observed decreases in persistent patch density and increases in persistent patch size with warmer and drier conditions. The fire behavior interpretation of this is potentially larger fire runs occurring under more extreme conditions (e.g., active crown fire) that leave a few large pockets unburned, as opposed to less aggressive fire behavior occurring under less extreme conditions (e.g., surface fire with passive torching) that leaves more numerous smaller pockets of persistent vegetation. This suggests that as climate change produces more extreme summer drought conditions or reduced spring snowpack, as is already being observed in the western US (Mote et al. 2005), persistent patches will become more fragmented and isolated. This fundamental alteration of the landscape mosaic produced by fire may increase distances to seed sources and limit regeneration of some flora, and reduce the quantity and quality of habitat for species that prefer closed canopy forests.

Figure 5.

Example of how years of large and small distributions affect time series. Distribution of proportion unburned for individual fires per year for Glacier (top panel), North Cascades (middle panel) and Yosemite (bottom panel). Dots denote mean for that year, fences denote 95% confidence interval.

f05_219.jpg

Figure 6.

Conceptual framework for climate impacts on wildfire in a coupled human-environment system, where climate produces fire effects to both the ecological system and the human system. This occurs through both biophysical pathways (i.e., fire behavior) through weather, fuel type, fuel abundance and fuel flammability, and human factor pathways (i.e., fire vulnerability) based on how humans respond to the fire risk both through top-down, government mitigation and response and bottom-up, individual response.

f06_219.jpg

Second, our relatively weak (and sometimes conflicting) correlations between macroscale climate and persistent patch patterns further support the argument made in the introductory section that drivers of fire pattern are complex and multiscalar. Research on fire-climate relationships has primarily focused on macroclimatic drivers of area burned, primarily because these relationships, once quantified, can be applied to project future area burned utilizing global climate model outputs. There are, however, numerous feedbacks through which climate change has already and will likely continue to affect fire activity and severity, occurring across the coupled human-natural system spectrum (Figure 6). These include changes in forest management and thinning practices specifically designed to increase carbon stocks and reduce the risk of catastrophic fire (Stephens et al. 2013, Prichard and Kennedy 2014), changes to prescribed burning practices based on increased variability during shoulder seasons (Kolden and Brown 2010), changes to fire suppression and management tactics such as increased burnout operations that reduce remnant unburned islands or increasing use of climate information to make management decisions (Owen et al. 2012), and changes in the way human populations act in fire-prone environments (Ray et al. 2012) such as through regulation of further development in the wildland-urban interface (Moritz et al., 2014). Prescribed fires and fire suppression decisions, in particular, have an impact on this type of analysis because decisions about when to ignite prescribed fires and when and how to fully suppress wildfires often account for climatic conditions, leading to significantly different burn patterns (Kolden and Brown 2010). However, as noted above, the consistency of regional fire-climate signals across studies assessing two very different metrics of fire activity (i.e., area burned versus fire pattern) support the theory that coarse-scale climate controls fire activity broadly.

Limitations and Future Work

There are several key limitations of landscapescale fire effects studies that challenge our ability to try and predict how climate change will impact persistent patches in the future. As has been noted by previous efforts (Dillon et al. 2011, Birch et al., 2015), there is a scale mismatch between the Landsat-derived burn severity data (30 meter pixels) and the coarse-resolution climate data (4 km pixels). This scale mismatch can lead to finerresolution predictors (e.g., topography) being able to explain more variance in severity data simply because they capture the fine-scale spatial variance. Remotely-sensed indices, such as NBR and its derivatives, are more sensitive to higher-severity fire effects; this is a product of both a low signal-to-noise ratio (i.e., fire-induced change versus phenological and other change) and the inherent difficulties of quantifying sub-canopy change from a space borne sensor (Key and Benson 2006, Key 2006, Kolden et al. 2012, Kolden et al. 2015). Finally, there are numerous potential contributors to fire severity and persistent patch development, including fire exclusion-related increases in fuel loading, invasive species, human-induced land use and land cover change, fire management and suppression tactics, soils, high-resolution meteorology, etc., that are difficult to analyze empirically due to a lack of representative data across time and space. Until these types of datasets and analyses begin to be compiled, our attempts to predict burn severity and important ecological aspects of fire—like persistent islands—will continue to be limited in scope.

Conclusion

Understanding and predicting wildfire burn severity is of wide interest because of the key role burn severity plays in determining landscape mosaics. Particularly crucial are the persistent patches that serve as fire refugia but are potentially threatened by the larger, more severe fires associated with a changing climate. Although other studies have found significant trends in high-severity fire, we found no significant trends in persistent patches in Glacier, Yosemite, and North Cascades National Parks and surrounding areas, although the abundance of persistent patches was correlated to many of the same spring and summer climate variables as prior fire-climate studies have found for these regions.

This lack of trends points to a potential flaw in the underlying assumption that climate change will produce more frequent, larger, and more severe wildfires. Depending on the definitions and metrics by which severity is measured, understanding the future may not be so simple as an oft-hypothesized shift of the normal distribution of severity towards higher severity. Alternative scenarios may include decreases in low and moderate severity (i.e., a switch to a bimodal severity distribution), or multi-decadal climate cycles associated with periods of characteristic landscape fuel composition, where lower severity dominates for a period of years, then higher severity dominates. Only long-term studies of fire severity that better account for a variety of predictive factors can fully address potential severity patterns. Ultimately, we suggest that managers and scientists begin to explore alternative theoretical frameworks than the standard one of greater frequency, size, and severity currently begin emphasized in both the literature and the mass media. Fire pattern and the full range of fire severity, including persistent areas, provide equally important metrics of ecological sustainability.

Acknowledgments

This manuscript developed from a presentation given as part of the University of Montana Plum Creek Distinguished Lecture series; the authors are grateful for the opportunity to present and receive feedback on the ideas presented herein. Funding for this research was provided by the US Geological Survey Global Change research program (‘Climate change impacts on burn severity in three forest ecoregions of the US') and the USGS Northwest Climate Science Center through Grant/Cooperative Agreement Number G14AP00177. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the USGS. This manuscript was improved by valuable comments from two reviewers and the editor. Any use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the US Government.

Literature Cited

  1. J. T. Abatzoglou 2013. Development of g ridded surface meteorological data for ecological applications and modeling. International Journal of Climatology 33:121–131. Google Scholar

  2. J. T. Abatzoglou , and C. A. Kolden . 2011a. Climate change in western US deserts: potential for increased wildfire and invasive annual grasses. Rangeland Ecology & Management 64:471–478. Google Scholar

  3. J. T. Abatzoglou , and C. A. Kolden . 2011b. Relative importance of weather and climate on wildfire growth in interior Alaska. International Journal of Wildland Fire 20:479–486. Google Scholar

  4. J. T. Abatzoglou , and C. A. Kolden . 2013. Relationships between climate and macroscale area burned in the western United States. International Journal of Wildland Fire 22:1003–1020. Google Scholar

  5. J. T. Abatzoglou , D. E. Rupp , and P. W. Mote . 2014. Seasonal climate variability and change in the Pacific Northwest of the United States. Journal of Climate 27:2125–2142. Google Scholar

  6. R. G. Allen , L. S. Pereira , D. Raes , and M. Smith . 1998. Crop evapotranspiration, guidelines for computing crop water requirements. FAO Irrigation and Drainage Paper 56, Food and Agriculture Organization of the United Nations. Rome, Italy. Google Scholar

  7. C. D. Allen , A. K. Macalady , H. Chenchouni , D. Bachelet , N. McDowell , M. Vennetier , T. Kitzberger , A. Rigling , D. D. Breshears , E. H. Hogg , P. Gonzalez , R. Fensham , Z. Zhang , J. Castro , N. Demidova , J. Lim , G. Allard , S. W. Running , A. Semerci , and N. Cobb . 2010. A global overview of drought and heat-induced tree mortality reveals emerging climate change risks for forests. Forest Ecology and Management 259:660–684. Google Scholar

  8. S. F. Arno 1980. Forest fire history in the northern Rockies. Journal of Forestry 78:460–465. Google Scholar

  9. R. G. Bailey 1998. Ecoregions. Springer, New York. Google Scholar

  10. J. K. Balch , B. A. Bradley , C. M. D'Antonio , and J. Gómez-Dans . 2013. Introduced annual grass increases regional fire activity across the arid western USA (1980–2009). Global Change Biology 19:173–183. Google Scholar

  11. D. S. Birch , P. M. Morgan , C. A. Kolden , J. T. Abatzoglou , G. K. Dillon , A. T. Hudak , and A. M. S. Smith . 2015. Vegetation, topography and daily weather influenced burn severity in central Idaho and western Montana forests. Ecosphere 6:art17. Google Scholar

  12. D. S. Birch , P. M. Morgan , C. A. Kolden , A. T. Hudak , and A. M. S. Smith . 2014. Is proportion burned severely related to daily area burned? Environmental Research Letters 9:064011. Google Scholar

  13. D. M. Bowman , J. Balch , P. Artaxo , W. J. Bond , M. A. Cochrane , C. M. D'Antonio , R. DeFries , F. H. Johnston , J. E. Keeley , M. A. Krawchuk , C. A. Kull , M. Mack , M. A. Moritz , S. Pyne , C. I. Roos , A. C. Scott , S. I. Sodhi , and T. W. Swetnam . 2011. The human dimension of fire regimes on Earth. Journal of Biogeography 38:2223–2236. Google Scholar

  14. R. E. Burgan 1988. 1988 Revisions to the 1978 National Fire Danger Rating System. USDA Forest Service Research Paper SE-273, Southeastern Forest Experiment Station, Ashville, NC. Google Scholar

  15. D. T. Butry , E. D. Mercer , J. P. Prestemon , J. M. Pye , and T. P. Holmes . 2001. What is the price of catastrophic wildfire? Journal of Forestry 99:9–17. Google Scholar

  16. A. E. Camp , C. D. Oliver , P. F. Hessburg , and R. L. Everett . 1997. Predicting late-successional fire refugia predating European settlement in the Wenatchee Mountains. Forest Ecology and Management 95:63–77. Google Scholar

  17. C. A. Cansler , and D. McKenzie . 2012. How robust are burn severity indices when applied in a new region? Evaluation of alternate field-based and remotesensing methods. Remote Sensing 4:456–483. Google Scholar

  18. C. A. Cansler , and D. McKenzie . 2014. Climate, fire size, and biophysical setting control fire severity and spatial pattern in the northern Cascade Range, USA. Ecological Applications 24:1037–1056. Google Scholar

  19. B. M. Collins , and S. L. Stephens . 2010. Stand-replacing patches within a “mixed severity” fire regime: quantitative characterization using recent fires in a long-established natural fire area. Landscape Ecology 25:927–939. Google Scholar

  20. B. M. Collins , M. Kelly , J. W. Wagtendonk , and S. L. Stephens . 2007. Spatial patterns of large natural fires in Sierra Nevada wilderness areas. Landscape Ecology 22:545–557. Google Scholar

  21. W. W. Covington , and M. M. Moore . 1994. Postsettlement changes in natural fire regimes and forest structure: ecological restoration of old-growth ponderosa pine forests. Journal of Sustainable Forestry 2:153–181. Google Scholar

  22. S. M. Crimmins , S. Z. Dobrowski , J. A. Greenberg , J. T. Abatzoglou , and A. R. Mynsberge . 2011. Changes in climatic water balance drive downhill shifts in plant species' optimum elevations. Science 331:324–327. Google Scholar

  23. V. H. Dale , L. A. Joyce , S. McNulty , R. P. Neilson , M. P. Ayres , M. D. Flannigan , P. J. Hanson , L. C. Irland , A. E. Lugo , C. J. Peterson , D. Simberloff , F. J. Swanson , B. J. Stocks , and B. M. Wotton . 2001. Climate change and forest disturbances. BioScience 51:723–734. Google Scholar

  24. S. C. DeLong , and W. B. Kessler . 2000. Ecological characteristics of mature forest remnants left by wildfire. Forest Ecology and Management 131:93–106. Google Scholar

  25. P. E. Dennison , S. C. Brewer , J. D. Arnold , and M. A. Moritz . 2014. Large wildfire trends in the western United States. 1984–2011. Geophysical Research Letters 41:2928–2933. Google Scholar

  26. G. K. Dillon , Z. A. Holden , P. Morgan , M. A. Crimmins , E. K. Heyerdahl , and C. H. Luce . 2011. Both topography and climate affected forest and woodland burn severity in two regions of the western US, 1984 to 2006. Ecosphere 2:art130. Google Scholar

  27. S. Z. Dobrowski , J. T. Abatzoglou , A. K. Swanson , A. Mynsberge , J. A. Greenberg , Z. Holden , and M. K. Schwartz . 2013. The climate velocity of the contiguous United States during the 20th century. Global Change Biology 19:241–251. Google Scholar

  28. K. E. Eberhart , and P. M. Woodard . 1987. Distribution of residual vegetation associated with large fires in Alberta. Canadian Journal of Forest Research 17:1207–1212. Google Scholar

  29. J. Eidenshink , B. Schwind , K. Brewer , Z. Zhu , B. Quayle , and S. Howard . 2007. A project for monitoring trends in burn severity. Fire Ecology 3:3–21. Google Scholar

  30. N. H. F. French , E. S. Kasischke , R. J. Hall , K. A. Murphy , D. L. Verbyla , E. E. Hoy , and J. L. Allen . 2008. Using Landsat data to assess fire and burn severity in the North American boreal forest region: an overview and summary of results. International Journal of Wildland Fire 17:443–462. Google Scholar

  31. S. L. Haire , and K. McGarigal . 2009. Changes in fire severity across gradients of climate, fire size, and topography: A landscape ecological perspective. Fire Ecology 5:86–103. Google Scholar

  32. S. L. Haire , and K. McGarigal . 2010. Effects of landscape patterns of fire severity on regenerating ponderosa pine forests (Pinus ponderosa) in New Mexico and Arizona, USA. Landscape Ecology 25:1055–1069. Google Scholar

  33. J. J. Hayes , and S. M. Robeson . 2011. Relationships between fire severity and post-fire landscape pattern following a large mixed-severity fire in the Valle Vidal, New Mexico, USA. Forest Ecology and Management 261:1392–1400. Google Scholar

  34. Z. A. Holden , P. Morgan , M. A. Crimmins , R. K. Steinhorst , and A. Smith . 2007. Fire season precipitation variability influences fire extent and severity in a large southwestern wilderness area, United States. Geophysical Research Letters 34:L16708. Google Scholar

  35. V. R. Kane , J. A. Lutz , C. A. Cansler , N. A. Povak , D. J. Churchill , D. F. Smith , J. T. Kane , and M. P. North . 2015. Water balance and topography predict fire and forest structure patterns. Forest Ecology and Management 338:1–13. Google Scholar

  36. V. R. Kane , J. A. Lutz , S. L. Roberts , D. F. Smith , R. J. McGaughey , N. A. Povak , and M. L. Brooks . 2013. Landscape-scale effects of fire severity on mixedconifer and red fir forest structure in Yosemite National Park. Forest Ecology and Management 287:17–31. Google Scholar

  37. R. E. Keane , J. K. Agee , P. Fulé , J. E. Keeley , C. Key , S. G. Kitchen , R. Miller , and L. A. Schulte . 2008. Ecological effects of large fires on US landscapes: benefit or catastrophe? International Journal of Wildland Fire 17:696–712. Google Scholar

  38. J. E. Keeley 2009. Fire intensity, fire severity and burn severity: a brief review and suggested usage. International Journal of Wildland Fire 18:116–126. Google Scholar

  39. G. Keppel , K. P. Van Niel , G. W. Wardell-Johnson , C. J. Yates , M. Byrne , L. Mucina , A. G. T. Schut , S. D. Hopper , and S. E. Franklin . 2012. Refugia: identifying and understanding safe havens for biodiversity under climate change. Global Ecology and Biogeography 21:393–404. Google Scholar

  40. C. H. Key 2006. Ecological and sampling constraints on defining landscape fire severity. Fire Ecology 2:34–59. Google Scholar

  41. C. H. Key , and N. C. Benson . 2006. Landscape Assessment: Ground measure of severity, the Composite Burn Index: and remote sensing of severity, the Normalized Burn Ratio. In D.C. Lutes et al. (editors). FIREMON: Fire Effects Monitoring and Inventory System, USDA Forest Service General Technical Report RMRS-GTR-164-CD: LA 1-51. USDA Forest Service, Ogden, UT. Google Scholar

  42. C. A. Kolden 2010. Characterizing Alaskan wildfire regimes through remotely sensed data: assessments of large area pattern and trend. Ph.D. Dissertation, Clark University, Worcester, MA. Google Scholar

  43. C. A. Kolden , A. M. S. Smith , and J. T. Abatzoglou . 2015. Limitations and utilisation of Monitoring Trends in Burn Severity products for assessing wildfire severity in the USA. International Journal of Wildland Fire WF15082. Google Scholar

  44. C. A. Kolden , and J. Rogan . 2013. Mapping wildfire burn severity in the Arctic tundra: novel approaches for an extreme environment. Arctic, Antarctic and Alpine Research 45:64–76. Google Scholar

  45. C. A. Kolden , and P. W. Weisberg . 2007. Assessing accuracy of manually mapped wildfire perimeters in topographically dissected areas. Fire Ecology 3:22–31. Google Scholar

  46. C. A. Kolden , and T. J. Brown . 2010. Beyond wildfire: perspectives of climate, managed fire and policy in the USA. International Journal of Wildland Fire 19:364–373. Google Scholar

  47. C. A. Kolden , J. A. Lutz , C. H. Key , J. T. Kane , and J. W. Van Wagtendonk . 2012. Mapped versus actual burned area within wildfire perimeters: characterizing the unburned. Forest Ecology and Management 286:38–47. Google Scholar

  48. J. D. Kushla , and W. J. Ripple . 1998. Assessing wildfire effects with Landsat thematic mapper data. International Journal of Remote Sensing 19:2493–2507. Google Scholar

  49. A. J. Larson , and D. Churchill . 2012. Tree spatial patterns in fire-frequent forests of western North America, including mechanisms of pattern formation and implications for designing fuel reduction and restoration treatments. Forest Ecology and Management 267:74–92. Google Scholar

  50. L. B. Lentile , F. W. Smith , and W. D. Shepperd . 2005. Patch structure, fire-scar formation, and tree regeneration in a large mixed-severity fire in the South Dakota Black Hills, USA. Canadian Journal of Forest Research 35:2875–2885. Google Scholar

  51. L. B. Lentile , P. Morgan , A. T. Hudak , M. J. Bobbitt , S. A. Lewis , A. M. S. S. Smith , and P. R. Robichaud . 2007. Post-fire burn severity and vegetation response following eight large wildfires across the western United States. Fire Ecology 3:91–108. Google Scholar

  52. L. B. Lentile , Z. A. Holden , A. M. S. Smith , M. J. Falkowski , A. T. Hudak , P. Morgan , S. A. Lewis , P. E. Gessler , and N. C. Benson . 2006. Remote sensing techniques to assess active fire characteristics and post-fire effects. International Journal of Wildland Fire 15:319–345. Google Scholar

  53. J. S. Littell , and R. Gwozdz . 2011. Climatic Water Balance and Regional Fire Years in the Pacific Northwest, USA: Linking Regional Climate and Fire at Landscape Scales. In D. McKenzie , C. M. Miller and D. A. Falk (editors). The Landscape Ecology of Fire, Ecological Studies 213, Springer, Netherlands. Pp. 117–139. Google Scholar

  54. J. S. Littell , D. McKenzie , D. L. Peterson , and A. L. Westerling . 2009. Climate and wildfire area burned in western U.S. ecoprovinces, 1916–2003. Ecological Applications 19:1003–1021. Google Scholar

  55. J. A. Lutz , A. J. Larson , J. A. Freund , M. E. Swanson , and K. J. Bible . 2013. The importance of largediameter trees to forest structural heterogeneity. PloS One 8:e82784. Google Scholar

  56. J. A. Lutz , C. H. Key , C. A. Kolden , J. T. Kane , and J. W. Van Wagtendonk . 2011. Fire frequency, area burned, and severity: a quantitative approach to defining a normal fire year. Fire Ecology 7:51–65. Google Scholar

  57. J. A. Lutz , J. W. Van Wagtendonk , A. E. Thode , J. D. Miller , and J. F. Franklin . 2009. Climate, lightning ignitions, and fire severity in Yosemite National Park, California, USA. International Journal of Wildland Fire 18:765–774. Google Scholar

  58. J. A. Lutz , J. W. Van Wagtendonk , and J. F. Franklin . 2010. Climatic water deficit, tree species ranges, and climate change in Yosemite National Park. Journal of Biogeography 37:936–950. Google Scholar

  59. J. R. Marlon , P. J. Bartlein , D. G. Gavin , C. J. Long , R. S. Anderson , C. E. Briles , K. J. Brown , D. Colombaroli , D. J. Hallett , M. J. Power , E. A. Scharf , and M. K. Walsh . 2012. Long-term perspective on wildfires in the western USA. Proceedings of the National Academy of Sciences of the United States of America 109:E535–E543. Google Scholar

  60. C. I. Millar , N. L. Stephenson , and S. L. Stephens . 2007. Climate change and forests of the future: managing in the face of uncertainty. Ecological Applications 17:2145–2151. Google Scholar

  61. J. D. Miller , and A. E. Thode . 2007. Quantifying burn severity in a heterogeneous landscape with a relative version of the delta Normalized Burn Ratio (dNBR). Remote Sensing of Environment 109:66–80. Google Scholar

  62. J. D. Miller , and H. D. Safford . 2012. Trends in wildfire severity: 1984 to 2010 in the Sierra Nevada, Modoc Plateau, and southern Cascades, California, USA. Fire Ecology 8:41–57. Google Scholar

  63. J. D. Miller , B. M. Collins , J. A. Lutz , S. L. Stephens , J. W. Van Wagtendonk , and D. A. Yasuda . 2012. Differences in wildfires among ecoregions and land management agencies in the Sierra Nevada region, California, USA. Ecosphere 3:1–20. Google Scholar

  64. J. D. Miller , H. D. Safford , M. Crimmins , and A. E. Thode . 2009. Quantitative evidence for increasing forest fire severity in the Sierra Nevada and southern Cascade Mountains, California and Nevada, USA. Ecosystems 12:16–32. Google Scholar

  65. M. A. Moritz , E. Batllori , R. A. Bradstock , M. A. Gill, A. M ., J. Handmer , P. F. Hessburg , J. Leonard , S. McCaffrey , D. C. Odion , T. Schoennagel , and A. D Syphard . 2014. Learning to coexist with wildfire. Nature 515:58–66. Google Scholar

  66. P. W. Mote , A. F. Hamlet , M. P. Clark , and D. P. Lettenmaier . 2005. Declining mountain snowpack in western North America. Bulletin of the American Meteorological Society 86:39–49. Google Scholar

  67. G. Owen , J. D. McLeod , C. A. Kolden , D. B. Ferguson , T. J. Brown . 2012. Wildfire management and forecasting fire potential: the roles of climate information and social networks in the southwest United States. Weather, Climate & Society 4:90–102. Google Scholar

  68. S. A. Parks , G. K. Dillon , and C. Miller . 2014. A new Metric for quantifying burn severity: the relativized burn ratio. Remote Sensing 6:1827–1844. Google Scholar

  69. J. L. Pierce , G. A. Meyer , and A. T. Jull . 2004. Fire-induced erosion and millennial-scale climate change in northern ponderosa pine forests. Nature 432:87–90. Google Scholar

  70. S. J. Prichard , and M. C. Kennedy . 2014. Fuel treatments and landform modify landscape patterns of burn severity in an extreme fire event. Ecological Applications 24:571–590. Google Scholar

  71. S. J. Pyne 2001. The fires this time, and next. Science 294:1005–1006. Google Scholar

  72. S. J. Pyne 2010. America's Fires: A Historical Context For Policy and Practice. Forest History Society, Durham, NC. Google Scholar

  73. L. A. Ray , C. A. Kolden , F. S. Chapin III . 2012. A case for developing place-based fire management strategies from traditional ecological knowledge. Ecology & Society 17:37. Google Scholar

  74. P. R. Robichaud , J. L. Beyers , and D. G. Neary . 2000. Evaluating the effectiveness of postfire rehabilitation treatments. USDA Forest Service General Technical Report RMRS-GTR-63, Rocky Mountain Research Station, Fort Collins, CO. Google Scholar

  75. M. E. Rocca , P. M. Brown , L. H. MacDonald , and C. M. Carrico . 2014. Climate change impacts on fire regimes and key ecosystem services in Rocky Mountain forests. Forest Ecology and Management 327:290–305. Google Scholar

  76. M. G. Rollins 2009. LANDFIRE: a nationally consistent vegetation, wildland fire, and fuel assessment. International Journal of Wildland Fire 18:235–249. Google Scholar

  77. S. L. Stephens , and J. J. Moghaddas . 2005. Experimental fuel treatment impacts on forest structure, potential fire behavior, and predicted tree mortality in a California mixed conifer forest. Forest Ecology and Management 215:21–36. Google Scholar

  78. S. L. Stephens , J. K. Agee , P. Z. Fulé , M. P. North , W. H. Romme , T. W. Swetnam , and M. G. Turner . 2013. Managing forests and fire in changing climates. Science 342:41–42. Google Scholar

  79. A. B. Swengel , and S. R. Swengel . 2006. Benefit of permanent non-fire refugia for Lepidoptera conservation in fire-managed sites. Journal of Insect Conservation 11:263–279. Google Scholar

  80. T. W. Swetnam , and J. L. Betancourt . 1990. Fire-southern oscillation relations in the southwestern United States. Science 249:1017–1020. Google Scholar

  81. D. Tanré , P. Y. Deschamps , P. Duhaut , and M. Herman . 1987. Adjacency effect produced by the atmospheric scattering in thematic mapper data. Journal of Geophysical Research: Atmospheres (1984–2012) 92:12000-12006. Google Scholar

  82. A.E. Thode 2005. Quantifying the fire regime attributes of severity and spatial complexity using field and imagery data. Ph.D. Dissertation, University of California, Davis. Google Scholar

  83. A.E. Thode , J. W. Van Wagtendonk , J. D. Miller , and J. F. Quinn . 2011. Quantifying the fire regime distributions for severity in Yosemite National Park, California, USA. International Journal of Wildland Fire 20:223–239. Google Scholar

  84. M. G. Turner , W. H. Romme , R. H. Gardner , and W. W. Hargrove . 1997. Effects of fire size and pattern on early succession in Yellowstone National Park. Ecological Monographs 67:411–433. Google Scholar

  85. M. G. Turner , W. W. Hargrove , R. H. Gardner , and W. H. Romme . 1994. Effects of fire on landscape heterogeneity in Yellowstone National Park, Wyoming. Journal of Vegetation Science 5:731–742. Google Scholar

  86. C. E. Van Wagner 1987. Development and structure of the Canadian Forest Fire Weather Index System. Forest Technical Report 35, Canadian Forest Service, Ottawa. Google Scholar

  87. J. W. Van Wagtendonk , and J. A. Lutz . 2007. Fire regime attributes of wildland fires in Yosemite National Park, USA. Fire Ecology 3:34–52. Google Scholar

  88. A. L. Westerling , T. J. Brown , A. Gershunov , D. R. Cayan , and M. D. Dettinger . 2003. Climate and Wildfire in the Western United States. Bulletin of the American Meteorological Society 84:595–604. Google Scholar

  89. A. L. Westerling , H. G. Hidalgo , D. R. Cayan , and T. W. Swetnam . 2006. Warming and earlier spring increases western U.S. forest wildfire activity. Science 313:940–943. Google Scholar

  90. C. J. Willmott , C. M. Rowe , and Y. Mintz . 1985. Climatology of the terrestrial seasonal water cycle. Journal of Climatology 5:589–606. Google Scholar

  91. G. T. Zimmerman , and L. F. Neuenschwander . 1984. Livestock grazing influences on community structure, fire intensity, and fire frequency within the Douglas-fir/ninebark habitat type. Journal of Range Management 37:104–110. Google Scholar

Crystal A. Kolden, John T. Abatzoglou, James A. Lutz, C. Alina Cansler, Jonathan T. Kane, Jan W. Van Wagtendonk, and Carl H. Key "Climate Contributors to Forest Mosaics: Ecological Persistence Following Wildfire," Northwest Science 89(3), 219-238, (1 August 2015). https://doi.org/10.3955/046.089.0305
Received: 30 June 2014; Accepted: 1 May 2015; Published: 1 August 2015
JOURNAL ARTICLE
20 PAGES


SHARE
ARTICLE IMPACT
Back to Top