Over the past three decades the western U.S. observed a sharp increase in number of fires, burned area, and length of the fire season in contrast to the low fire activity during most of the middle to late 20th century (Littell et al. 2009, Jolly et al. 2015, Westerling 2016). Previous studies have shown evidence of coupling between the recent increase in fire activity and climate (Littell et al. 2009, Marlon et al. 2012, Abatzoglou and Kolden 2013, Westerling 2016); however, for the Pacific Northwest (PNW) region long-term historical records of biomass burning suggest that climate alone cannot explain the temporal variation in area burned (Heyerdahl et al. 2002, Hessl et al. 2004, Marlon et al. 2012). In the absence of strong climatic controls on fire activity, a common explanation for burned area fluctuations through time is human-induced land cover changes. Heyerdahl et al. (2002) interpreted the reduction in burned area during the early decades of the 20th century as the consequence of low, sparse fuels due to the increase in the number of domestic livestock during the European settlement. This period was followed by a strong fire suppression policy in the late 1930s that excluded fire on fire-adapted landscapes, a policy still in effect today. Thus, the recent increase in fire activity in the PNW is the likely outcome of fuel accumulation resulting from human activities, combined with the beginning of a dry and warm period (Mote 2003, Brown et al. 2004, Westerling et al. 2006, Jolly et al. 2015) that is concurrent with the onset of climate change effects (Abatzoglou et al. 2014). Despite the increase in number of fires and burned area, contemporary trends of fire severity and how they compare with historical trends remains a subject of debate (Hessburg et al. 2005, 2007, Noss et al. 2006, Miller et al. 2009, Odion et al. 2014, 2016, Sherriff et al. 2014, Baker 2015, Stevens et al. 2016). Reduction of fire severity is one of the objectives of restoration programs (U.S. Congress 2003), but restoration goals may not be accomplished if that target is estimated against a biased historical baseline. Regardless of the relative role of high-severity fire in historical fire regimes of the western U.S., fire-excluded forests in the PNW have now comparatively higher fuel loads (Hessburg et al. 2005, Thompson and Spies 2009, Merschel et al. 2014). This means that holding all other factors constant, these forests are more likely to burn with high severity than in the past, even if they only burn under severe weather conditions (Weatherspoon et al. 1992). Moreover, under most modeling scenarios of future greenhouse emissions and projected climate impacts, the likelihood of severe fire weather is expected to increase, as is burned area and fire severity (Littell et al. 2009, Spracklen et al. 2009, Halofsky et al. 2014a, Stavros et al. 2014).
The increase in burned area over the past three decades (Westerling 2016) is especially apparent on western U.S. national forests where large fire-prone landscapes are juxtaposed with expanding wildland urban interfaces (WUI), and federal land management restricts fuel management in some areas to protect biodiversity and other amenities (Williams 2013, Ager et al. 2014a, North et al. 2015). The WUI expansion and increase in highly valued resources that are susceptible to fire have led to an increase in landscape wildfire risk.
This growing threat of wildfire losses has led to a number of new U.S. federal policies and expanded investments in forest and fuel management programs on both public and private lands. However, understanding the effectiveness of these programs on mitigating large, high-severity fires is difficult because wildfire spread is a complex and nonlinear spatial process that extends beyond the spatial scale of an individual stand (Peters et al. 2004, Slocum et al. 2010). Isolated stand-scale treatments may be effective when wildfires intersect them but, given the stochastic component of wildfire (ignition and spread), a significant proportion of the landscape has to be treated to be effective in reducing fire spread and fire behavior at the landscape scale (Agee and Skinner 2005). Scaling treatments from the stand to the landscape level has been the focus of many studies addressing various methods for spatially optimizing the arrangement of fuel treatments on the landscape to minimize spread rate and fire behavior (Finney 2001, 2007, Parisien et al. 2007, Schmidt et al. 2008), maximize habitat protection (Ager et al. 2010), and analyzing trade-offs in forest restoration programs (Vogler et al. 2015). However, relatively few studies have tackled the difficult and larger problem of long-term fuel management strategies. Incorporation of time in fuel management design is critical because treatments have a lifespan, which is determined by vegetation succession. Additionally, treatments are frequently completed on the same piece of ground over a sequence of years, highlighting the importance of modeling efforts that consider both the spatial and temporal dimensions of fuel treatment effects. A number of researchers have included temporal dynamics into simulation studies of spatiotemporal predictions of fire occurrence (Collins et al. 2011, Syphard et al. 2011), forest dynamics (Strom and Fulé 2007, Gustafson et al. 2010), and carbon sequestration (Loudermilk et al. 2014). Other simulation studies tested what-if scenarios that evaluated the economic efficiency of fuel treatments (Taylor et al. 2013) and anticipated the potential impacts of climate change on fire activity and forest composition (Scheller and Mladenoff 2005, Halofsky et al. 2014a, Yospin et al. 2015). Despite these studies, no clear patterns and conclusions can be made because of the widely different models, landscape conditions, calibrations, and assumptions about forest succession dynamics and fuel treatment effects.
In this study we use a newly developed version of Envision (Bolte et al. 2007), an agent-based landscape modeling system and the widely used minimum travel time fire spread algorithm (Finney 2002) to analyze long-term fuel management dynamics on a large multiownership landscape. This study is part of the “Forests, People, Fire” project that focused on the interactions, dynamics, and adaptations in coupled social and ecological systems in central Oregon (Spies et al. 2014). Spies et al. (2017) analyze the impact of alternative forest management policies on metrics that describe wildfire, forest structure, biodiversity, and timber production. In this paper we examine how alternative fuel management scenarios affect spatial and temporal patterns of wildfire. Specifically, we examined how different intensities of fuel management on federal lands compare in terms of (1) future burned area, and (2) future probability of an ignition generating a large fire. We hypothesized that without fuel reduction and under the contemporary pattern of ignitions and fire weather, burned area and the probability of large fires are expected to increase through the 50 years of simulation as a consequence of fuel accumulation. We support this hypothesis based on the observed increase in burned area in the PNW for the past three decades, which has been attributed to both climate and land cover changes leading to landscape homogeneity and higher fuel loads that favor wildfire progression (Perry et al. 2011, Hessburg et al. 2016, Westerling 2016). On the contrary, under active management directed at reducing fuel loads, burned area (and the probability of a large fire) should remain the same or decrease through time with higher reduction for scenarios with higher treatment intensity.
The study area covers approximately 1.2 million ha of forest and range lands in the eastern Cascades of central Oregon, U.S. Vegetation types form a west to east gradient from alpine conifer forest in the higher elevations of the Cascades to moist mixed-conifer forest, dry mixed conifer forest, lodgepole pine (Pinus contorta) and ponderosa pine (Pinus ponderosa) forests, western juniper (Juniperus occidentalis), and finally arid lands in the far east end of the study area (Fig. 1A). Moist mixed-conifer forests are mostly grand fir (Abies grandis), white fir (Abies concolor), Douglas-fir (Pseudotsuga menziesii), ponderosa pine and western larch (Larix occidentalis), while dry mixed conifer includes all of the abovementioned species in addition to lodgepole pine, Engelmann spruce (Picea engelmannii), incense cedar (Calocedrus decurrens), sugar pine (Pinus lambertiana), and western white pine (Pinus monticola). Nonvegetated areas, urban centers, water bodies, and roads account for 8% of the study area.
The study area includes lands managed by the United States Forest Service (USFS, 58%, most of it corresponding to the Deschutes National Forest [DNF]), the Bureau of Land Management (BLM, 3%), the Warm Springs Tribe (21%), private forest corporations (6%), and family forest landowners (4%). State lands include the Gilchrist State Forest and account for just 2% of the landscape. Homeowners, who own tracts smaller than 2.5 ha, account for 7% of the study area (Fig. 1B).
Between 1992 and 2013, the study area had 372 ignitions per year (42% of caused by lightning, mostly during summer months) accounting for 11,423 ha burned per year (Short 2015). Recent large fires include the B&B Complex in 2003 (36,733 ha), Pole Creek in 2012 (10,844 ha), and Sunnyside Turnoff in 2013 (21,448 ha).
We used Envision (Bolte et al. 2007) to model the interactions of forest management, succession, and wildfire at the landscape scale. Envision is a spatially explicit agent-based model that simulates major biophysical and social processes that drive landscape change (Fig. A1.1). Agents in Envision are landowners who have goals for the areas they manage and act on the landscape to achieve those goals. Agents’ actions, which include forest and fuel management are constrained by forest policies and landscape condition. In this study, agents represented the major landowners in the study area: federal land managers, the Warm Springs Tribe, and private forest corporations. Envision also includes submodels that simulate autonomous landscape change processes: wildfire, forest succession, and population growth. These processes are agent-independent, but affect landscape condition and therefore the ability for agents to meet their landscape goals through management activities.
Envision operates on a yearly time step and at the spatial scale of individual decision units (IDUs) ranging between 1 and 10 ha. Each IDU was uniquely characterized in terms of ecological and social attributes that described landscape condition and quantified resource production relevant to the different agents, e.g., timber production, biodiversity, burned area. In each time step, agents make decisions, the autonomous processes operate, and the ecological characteristics of an IDU vary as a consequence of wildfire, forest management, and vegetation succession. These changes are reflected in landscape condition and feedback into the following time step. We describe specific submodels of Envision as required for our particular study below.
Vegetation succession in Envision was simulated with existing state-and-transition simulation models (STSM) developed for all major upland vegetation types in Arizona, New Mexico, Oregon, and Washington under the Integrated Landscape Assessment Project (ILAP, Halofsky et al. 2014b). ILAP models were implemented in Envision using the Vegetation Dynamics Development Tool (VDDT) and expert opinion (Hemstrom et al. 2004, 2007). STSMs are empirical simulation models whereby vegetation is classified into a structural stage defined as a standardized combination of dominant cover type, diameter class, canopy cover, and forest layering. In Envision, changes in the vegetation state of an IDU were a function of succession rules and occurred both deterministically and probabilistically. Deterministic transitions occurred after a fixed amount of time had passed without disturbance and typically reflected age-based succession in STSM. Probabilistic transitions had an associated probability of occurrence, were age-independent, and resulted from parallel vegetation growth processes such as natural regeneration and understory development. Inclusion of probabilistic transitions in modeled vegetation succession allowed the generation of a distribution of projected outcomes for the state of each IDU. Probabilistic transitions can also be used to model disturbance processes such as insects, wildfire, or forest management. However, in this work changes in vegetation states due to disturbance followed a deterministic approach, implemented through the wildfire and forest management submodels. For further details on the forest succession model see Spies et al. (2017).
We built a wildfire submodel for Envision that combines a statistical fire forecasting model with the minimum travel time algorithm (MTT, Finney 2002) to simulate fire growth and model fire behavior. The wildfire submodel is summarized here and described in detail in Preisler et al. (2004), Preisler and Ager (2013), and Ager et al. (2014b). At each annual time step, a spatiotemporal ignition prediction model sampled a given number of ignitions as a function of location, day-of-year, and daily energy release component (ERC). This model was developed based on historical (1992-2009), spatial, and temporal fire ignition patterns and associated fire sizes. Weather conditions associated with each ignition were based on ERC value of the ignition day. Wind speed was sampled from the historical distribution (1992-2009, Lava Butte Remote Automated Weather Station, RAWS) of gust speed for days with area burned larger than 500 ha for any given ERC value. Wind direction was sampled from historical gust directions from the same Lave Butte RAWS data based on day-of-year. The MTT algorithm requires a burn period for each ignition that was based on an empirical relationship between burn period and fire size developed during the calibration process that is specific for the study area.
Wildfire outputs at the end of each time step (fire season) included simulated fire perimeters and predicted flame length for IDUs affected by fire. Wildfire severity was calculated at the IDU level by comparing modeled flame length with flame length thresholds that define surface fire, mixed-severity fire, and stand-replacing fire. These thresholds were derived from simulations with the Forest Vegetation Simulator (FVS; Dixon 2002) and the Fire and Fuels Extension (FFE). Specifically, tree lists were selected to represent each of the major vegetation states in the forest succession submodel and then simulated with FVS-FFE to determine the basal area mortality associated with flame lengths in 30-cm intervals, up to 4.5 m. Low severity (surface fire) corresponded with basal area mortality less than 20%, mixed severity with 20-79% mortality, and high severity (stand-replacing fire) with 80% or greater tree mortality.
Fire effects were carried over to the next simulation time step through changes in the IDU forest and fuel structure. If any of the attributes of the IDU vegetation state, e.g., layering or canopy cover, changed as a consequence of fire, this would trigger a change in the IDU vegetation state. This was usually the case with stand-replacing fire. Fire modeling variables, canopy cover, and fuel model were updated to reflect the new, postfire vegetation state. However, a low-severity fire would reduce understory fuels without changing the IDU vegetation class. In these cases, only the IDU’s fuel model was updated to reflect changes in the understory fuel structure. The postdisturbance fuel model persisted until a new disturbance took place or through 10 years, after which fuels returned to the predisturbance baseline (Tables A2.1 and A2.2). The timing of the transition to the pre-disturbance fuel model was based on expert opinion and in some cases informed by stand-level models (Hemstrom et al. 2004).
Forest management activities were simulated in Envision through a submodel that distributed treatments across the landscape based on a set of area targets, constraints, and preferences specific to each management action and actor (Table A3.1 and Table A3.2). Area targets defined the manager’s objective in terms of maximum area treated annually and patch size. Constraints defined locations and conditions that could be selected for management and preference weights defined priorities within the selected locations. An example of a constraint was to restrict mechanical thinning to forested IDUs where diameter at breast height > 25 cm, and an example of a preference would be to prioritize treatments in the wildland-urban interface. Individual decision units (IDUs) that met the constraints were given a score based on their preference weights, then ranked and selected for treatment. The IDU with highest score was taken as the spatial kernel of a search algorithm that aggregated neighboring IDUs (that also met the constraints) and formed a treatment patch. The process of selecting kernel IDUs and expanding them into patches up to the maximum patch size was executed until the annual area target was met or the constraints inhibited further activity.
Similar to the wildfire submodel, the effects of treatment actions on an IDU’s vegetation state depended on the predisturbance state and type of management action. If the treatment induced changes in the state descriptors, then it triggered a transition into a new vegetation state. For example, thinning that reduced the number of forest layers from multi to single triggered a new, posttreatment vegetation state. Management actions that only affected surface fuels, e.g., fuels mastication, led to changes in the fuel model that lasted for five years, after which the fuel model returned to the baseline predisturbance condition (Table A2.1 and Table A2.2).
On federal lands, forest management activities were allocated to areas designated for active management. These treatable areas were identified based on forest planning documents and discussions with federal forest managers on the DNF. Treatable area corresponded to approximately 54% of federal lands and excluded high elevation forests, congressionally designated wilderness areas, biodiversity reserves established to protect the Northern Spotted Owl (Strix occidentalis caurina), and other lands classified by the USFS as unsuitable because of steep slopes (> 30%) and lack of roads (Fig. A4.1A). Forested land managed by other ownerships was considered treatable. Table 1 shows the management actions associated with each landowner. Constraints based on species and stand conditions, and preference weights used to select and rank eligible IDUs within the treatable area are shown in Table A3.1 and A3.2, respectively. Constraints and preference weights intended to simulate current management practices for each landowner and are based on the DNF land management plan (USFS 1990), information provided by federal managers and managers of other relevant ownerships in the study area.
We tested four management scenarios that modeled different area targets for fuel management on federal lands, i.e., management in tribal and corporate forests remains at current levels for all scenarios:
With the exception of the no management scenario, all active management scenarios shared the same set of constraints and preferences, differing only in terms of their annual area targets. The annual area target was allocated as follows: 50% to mechanical thinning, 30% to fuels mastication, and 20% to prescribed fire. Current management represented the status quo, corresponding to an annual target of 8500 ha, and to a treatment rate of 2.1% per year of treatable federal lands, the equivalent of 1.1% of all federal land and 0.7% of the study area. In the double and triple current management scenarios, annual area targets corresponded to 4.2% (17,000 ha) and 6.3% (25,500 ha) of treatable lands per year, respectively.
Initial vegetation states were identified from the gradient nearest neighborhood (GNN) dataset and based on year 2006 (Ohmann et al. 2011). Fuel models for the DNF were obtained from the DNF fuel model layer and in all other locations were based on fuel model mapping by LANDFIRE (Rollins 2009). The initial landscape was updated to 2012 to reflect disturbances and vegetation succession between 2006 and 2012 (Fig. A4.1B). Wildfires between 2006 and 2012 were updated using the Monitoring Trends in Burn Severity (MTBS) fire perimeter and fire severity classification tool (USFS and USGS 2009). Forest management was updated for the same period by implementing treatments on the landscape under the current management scenario (Fig. A4.1B).
We ran 15 replicates of 50 years each, based on empirical testing indicating that the percentage of burned area among landowners stabilizes around 15 replicates. Replicates represent alternative and equally probable fire scenarios that mimic historical variability in wildfire activity in the study area.
We calculated average burned area and standard errors by fire severity class on federal forested lands, i.e., excluding arid lands and juniper areas. We fitted a linear model that uses year as the explanatory variable to determine the presence of a temporal linear trend in a time series of annual area burned. To compare overall burned area within scenarios we used a pairwise version of the Kruskal-Wallis nonparametric test (alpha = 0.05). The Kruskal-Wallis statistic was used to detect differences in overall burned area between all possible combinations of the management scenarios on federal lands, SILVIS WUI, wilderness, and old-growth areas. Wilderness areas and old-growth forest patches were defined according to the DNF land and resource management plan (USFS 1990) and SILVIS WUI was delineated based on the WUI definition in Radeloff et al. (2005). The SILVIS WUI is distinct from the federal WUI, which was used as a preference to prioritize mechanical treatments on federal land. The federal WUI layer is usually defined by local federal land managers based on boundaries described in local Community Wildfire Protection Plans (CWPPs). We chose to use SILVIS WUI to assess the impact of forest management in the WUI because of its wide acceptance by researchers, thorough documentation, and consistency throughout the U.S., which enables comparison with other studies. All analyses were done using MatLab R2015a (Mathworks 2015, http://www.mathworks.com).
We used fire size associated with individual fire perimeters to calculate the conditional probability, p, of a fire of a certain size class given an ignition. Probabilities for four distinct fire size classes (> 200 ha, > 500 ha, > 1000 ha, and > 10,000 ha) were estimated using logistic regression with ERC, burn period, wind speed, and management scenario as explanatory variables. Having the explanatory variables in the model enabled the comparison among treatment scenarios when all other variables in the model were kept the same. We also calculated the change in the odds (p/[1-p]) of a fire of a given size under the current, double, and triple current management scenarios, as compared with the odds of having the same fire under the no management scenario. All analyses were performed with R statistical software (R Core Team 2014).
We reported the percentage of treatment target met in the study area in each simulation year. To quantify the temporal evolution of areas needing treatment, we calculated annual treatment availability as the area that meets the treatment allocation constraints at the end of each simulation year. Treatment availability describes the balance of disturbances and vegetation growth on the landscape. Each year, management and wildfires (depending on their location and severity) reduce the pool of area eligible for treatment, while vegetation succession adds new areas. Thus, treatment target reflects the amount of area that managers plan to treat annually, while treatment availability represents the amount of area in the landscape that could potentially be treated if resources were unlimited.
The majority (70%) of wildfire in federal forested areas was classified as high-severity (stand-replacing) for all scenarios considered (Fig. A5.1), with considerable variability among simulation replicates for all fire severity classes (Fig. A5.2). Given the predominance of high-severity fire, all further results refer to area burned by stand-replacing fire. Mean burned area in forest lands showed no significant temporal trend (upward or downward) for any management scenario (Fig. 2 and associated parameters in Table A5.3). However, all active management scenarios were associated with lower burned area than under no management as indicated by overall burned area differences over the 50-year period (Table 2). On federal lands (Fig. 3A), and compared with no management, burned area in the 50-year period was reduced by 25%, 36%, and 40% for current, double, and triple management scenarios, respectively (Table 2). However, note that differences in burned area between current management and the increased management scenarios were not statistically significant (Table A5.4). Among land designations, old-growth areas showed the highest reduction in burned area because of management: reductions of 31%, 51%, and 61% for current, double, and triple management scenarios, respectively (Fig. 3C, Table 2). This, however, was not the case for wilderness (Fig. 3B) or SILVIS WUI (Fig. 3D), where only very small differences were found in burned area between no management and any of the active management scenarios (Table 2).
Under the no management scenario and across all simulation replicates, fires larger than 200 ha accounted for 92% of overall burned area. Very large fires (> 10,000 ha) accounted for 36% of overall burned area and occurred on 5 out of 1000 ignitions over a period of 50 years (Table 3). Compared to the no management scenario, all active management scenarios were effective in reducing the odds of a fire in all size classes (Fig. 4); however, the reduction was greatest for very large fires. The odds of a very large fire were approximately 0.60 to 1 (40% reduction) under current management and 0.30 to 1 (70% reduction) under the double and triple scenarios (Fig. 4D). Differences between double and triple scenarios were only marginally significant because of the overlapping confidence intervals. Over a period of 50 years our model estimated a decrease in the probability of a very large fire, from 5 in 1000, under no management, to 3 in 1000, under current management. Under the double and triple management scenarios the probability of a very large fire was 1.5 in 1000 and there was no statistical difference detected between the odds for these two scenarios.
Under current management, the annual treatment target for thinning was met (> 90%) for the entire simulation period (Fig. 5A). Under the double and triple management scenarios the annual target was fully met every year until year 20 and 15, respectively, after which the annual target was met at 40% and 20%, respectively. Overall, total area treated with mechanical thinning during the 50-year period is similar among scenarios: 2.1x105 ha under current management vs 2.4x105 ha under the double/triple current management scenarios. Treatment availability for mechanical thinning was slightly reduced under current management and furthermore reduced under the double/triple scenarios after approximately 15/10 years, i.e., at about the same time as a sharp decline was observed in the percentage of target met (Fig. 6A).
The annual target for mastication was met throughout the entire simulation period for both current and double current management scenarios. Under the triple current management scenario, the annual target fluctuated between 100% and 50% roughly every 5 years (Fig. 5B). Treatment availability for mastication was slightly reduced under all management scenarios (Fig. 6B). Not surprisingly, in the absence of management (red line in Figs. 6A and 6B), treatment availability increased over time for both mechanical treatments.
The annual area target for prescribed fire was largely met throughout the 50 years and for all management scenarios (Fig. 5C). However, contrary to what was observed for mechanical treatments, all active management scenarios led to higher landscape availability through time (Fig. 6C, black, green, and blue lines). Only under the no management scenario (Fig. 6C, red line) did we observe significant reductions in areas eligible for prescribed fire over a 50-year period.
The simulated wildfires under our no management scenario did not lead to increasing burned area over the 50-year period, therefore contradicting our original hypothesis. However, fuel was accumulating throughout the 50 years in forested stands available for thinning and mastication, i.e., excluding those in the wilderness. In fact, at the landscape scale, there was a 10% increase of heavy forest timber fuel models (and corresponding reduction in grass-shrub fuel models), with an associated increase in fuel load per unit area. The fact that we did not observe a concurrent increase in burned area suggests that the added fuel loads from vegetation succession did not translate into faster spread rates, but consisted instead of a conversion from relatively fast spreading grass-shrub models to heavy timber-litter models with moderate spread rate (e.g., low load compact conifer litter model; Scott and Burgan 2005). This conversion was accompanied by an increase in canopy cover, thus reducing wind speed and spread rate in the understory. The proportion of other fuel models remained the same over time, suggesting that the majority of the landscape is either in a stable successional state or changes in vegetation structure leading to transitions occur in time frames longer than 50 years.
In scenarios that simulated forest and fuel management, we hypothesized that accumulation of treated area would cause a downward trend in area burned. Overall, scenarios that included some level of management had less area burned. This is not surprising because treatments are capable of hindering subsequent fire spread as long as the ignition occurs in a treated area during the time the treatment remains effective, and under relatively favorable weather conditions (Strom and Fulé 2007). We observed a higher reduction in burned area in years with more (and larger) fires throughout the simulation period. This is in agreement with Syphard et al. (2011) and Loudermilk et al. (2014) who found that fuel treatments were more effective in reducing fire severity and sequestering forest carbon under severe fire weather because of an increased number of intersections between fires and treatments.
Management also reduced the likelihood of very large fires (> 10,000 ha) and that effect increased proportionately when area treated was doubled and tripled. Yet, under the double and triple management scenarios, annual rates of mechanical treatment were only met for the first two decades of simulation. Over the entire 50-year period the actual amount of forest thinned was, remarkably similar among scenarios, as treatment allocation was limited by the availability of suitable vegetation conditions. Current management can progressively but slowly address the backlog of the very specific structural conditions targeted by thinning: multistory, closed-canopy forest stands. However, areas suitable for thinning account for a small proportion (13% at initial conditions) of the landscape, therefore the majority of the landscape remains untreated (Hampton et al. 2011, North et al. 2015). Allowing thinning in more open and/or smaller diameter stands would increase treatment availability to sustain higher targets for longer periods of time. The downside of expanding the structural restrictions on thinning is compromising the revenue that supports the cost of treatments if additional stands provide less volume per hectare treated.
Contrary to mechanical treatments, areas suitable for prescribed fire increased under all active management scenarios, and were only reduced under the no management scenario (red line, Fig. 6C). We interpret this reduction as the result of the increased density of stands (from succession) that are not suitable for prescribed fire as the first action in a restoration sequence. The opposing trends in area suitable for treatment between these two types of management action are a likely consequence of the disproportionate treatment rates between thinning and prescribed fire, roughly 4400 ha vs 1300 ha treated per year (current management), respectively. Holding other factors constant, thinning increases the amount of single-story, open-canopy stands, thereby increasing the pool of area eligible for prescribed fire. These stands are in a condition to be safely burned in the years following thinning to reduce hazardous surface fuels and that was accounted for in our model (Collins et al. 2011). We suggest that at current rates (and with current spatial allocation rules), prescribed fire is not frequent enough to take advantage of the amount of thinning implemented on the landscape. This result highlights the need for an adequate balance of treatment targets, particularly for treatments that are coordinated, e.g., when mechanical thinning is supposed to be followed by prescribed fire.
The effect of management was particularly pronounced in designated old-growth areas that burned significantly less compared to the no management scenario. Old-growth forest reserves are distributed in small, dispersed land allocations, relatively distant from areas with frequent ignitions and imbedded within large areas of managed forest, thus benefiting from protection generated from treated areas. These “off-site” benefits of fuel management have been seen with both simulations (Ager et al. 2010, Stephens et al. 2014) and actual fires (Finney et al. 2005) and complement the treatment benefits made directly on site (Reinhardt et al. 2008).
In contrast to the old growth reserves, the amount of WUI burned was similar under all scenarios. WUI occupies large patches of land, mostly concentrated along frequent-ignition areas and is only partially within treatable areas thus it experiences higher wildfire exposure. These differences highlight how the efficacy of fuel management programs is affected by landscape context, priorities for protection, and how one measures the effectiveness of management actions (Ager et al. 2013). In this study, we assessed treatment effectiveness based on changes in burned area, but fuel and forest management programs may utilize other management objectives (and associated measures for success) such as forest restoration, biodiversity protection, or economic return. However, multiple objectives and priorities can lead to conflicting treatment allocations. For example, treating to reduce burned area/severity may imply treating locations that provide no economic return, while restricting treatments to areas where management pays for itself may have only a marginal effect on area burned/severity. Moreover, at landscape scale the outcome of fuel treatments is also determined by variables that are independent of treatment objectives such as proximity to ignition sources and to managed areas. This highlights the advantage of simulation methods capable of modeling how spatial allocation rules affect treatment efficacy through time (Ager et al. 2014a).
Despite the large fires seen in our simulations and witnessed in recent decades, there is a growing wildfire deficit in the western U.S.; biomass burning is low compared with historical fire activity and out of equilibrium with current climate (Marlon et al. 2012, Parks et al. 2015). Our fire model was calibrated with recent fire history data (1992 to 2009) and incorporated spatiotemporal patterns of ignition and fire size that are the direct result of the wildfire suppression policy in effect. Well-designed fuel management programs, combined with adequate suppression can protect communities from wildfire (Collins et al. 2010, North et al. 2012). However, relying on mechanical treatments alone to address the current wildfire deficit on landscapes with expanding WUI, increasing human ignitions and under a warming future, will create a challenging management situation. Allowing significantly more prescribed fire (and wildland fire management for resource benefit, which was not modeled in this study) would contribute toward a more resilient landscape. However, air quality regulations, as well as safety and liability concerns, combined with prohibitive fuel loads after a century of fire exclusion often limit efforts to increase the amount of fire (Miller 2002).
Envision links an STSM with wildfire and forest management submodels therefore, it inherits the limitations of these models. STSMs are representations of today’s understanding of vegetation dynamics in a given ecological setting and thereby are dependent on available knowledge and data. This means that the only possible outcomes arising from an STSM are the outcomes defined by the states and transitions included in the model. To overcome such limitations users have modeled shorter temporal periods (Halofsky et al. 2014a) or relied on dynamic vegetation models to inform STSMs of likely future vegetation conditions (Kerns et al. 2011). The STSM in Envision includes dozens of vegetation states and several possible transitions between them with the total number of possible transitions in the thousands. This is a strong indication of the model’s capacity to capture different behaviors that arise from interactions of fire, succession, and management. Moreover, Halofsky et al. (2014a) found that projected changes in area of forested potential vegetation types in central Oregon under climate change scenarios did not differ much from current conditions until at least 40-60 years from present. Given the 50-year period simulated in our study, and based on Halofsky et al. (2014a) we believe that the potential vegetation changes not available in the Envision STSM do not seriously limit our results.
The MTT algorithm used in the fire submodel has been widely tested and validated in both scientific and operational contexts. However, there are known limitations to this method given the assumption of constant fuel moisture, wind speed, and wind direction for a given ignition (Finney 2002). The fire ignition model was empirically based, therefore it reflects the interaction between suppression capability, climate, and fuel conditions of the past two decades. Simulating the contemporary fire regime into the future carries the implicit assumption that this interaction will remain constant. However, under a warmer future climate, suppression effectiveness may be reduced because of severe fire weather and longer fire seasons.
Our results overestimate the proportion of high-severity fire, and we believe this is a function of the flame length thresholds used in this analysis (see Spies et al. 2017). Further work will use MTBS fire severity data to calibrate flame length thresholds, so that modeled fire severity matches historical data.
Our modeling results indicate that over the course of 50 years and compared with no management, current forest management policy on federal lands would lead to reductions in burned area up to 25%. Our results also suggest that constraints on the availability of suitable conditions limit treatment area in some federal landscapes. Doubling and tripling current levels could only be implemented for mechanical treatments over the first two decades of this simulation and led to a 40% reduction in burned area, while the likelihood of an ignition generating a fire > 10,000 ha was reduced by threefold. However, although management reduced area burned, we did not observe a decreasing trend in burned area through time. This suggests that although forest and fuel management is a viable strategy to protect highly valued resources, it may not be capable of cumulatively altering the contemporary fire regime given current landscape conditions and management constraints.
At current rates, neither wildfire nor management were frequent enough to reduce fuel on a scale and extent that persists through time. It will require significant changes in the amount and/or distribution of management and wildfire in order to have an impact on landscape-level fuels, and consequently future wildfire. Our results show that increasing the amount of forest management is effective in the near future but limited by current management guidelines on treatment allocation. Further work is needed to evaluate at what levels fire and management interact to regulate fuel availability in the long-term. The failure to reduce burned area in the WUI reinforces the need to understand how spatial context, e.g., patch size, distance to ignition sources, and distance to managed lands, affects the ability of fuel management programs to reduce fire exposure to high valued-resources. Future work should also focus on analyzing the performance of forest and fuel management programs under prospective climate change futures and explore alternative sets of management policies, ranging from the current policy to increased rates of treatment and less restrictive allocation rules.
This research was funded by the National Science Foundation, Coupled Human and Natural Systems Program (NSF Grant CHH-1013296), the USDA Forest Service, PNW Research Station. We are thankful to Stuart Brittain, Alturas Solutions, for incorporating the FlamMap DLL into the wildfire submodel into the Envision system and James Sulzman for his programming work in the forest management and vegetation submodels.
Abatzoglou, J. T., and C. A. Kolden. 2013. Relationships between climate and macroscale area burned in the western United States. International Journal of Wildland Fire 22(7):1003-1020. http://dx.doi.org/10.1071/wf13019
Abatzoglou, J. T., 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(5):2125-2142. http://dx.doi.org/10.1175/jcli-d-13-00218.1
Agee, J. K., and C. N. Skinner. 2005. Basic principles of forest fuel reduction treatments. Forest Ecology and Management 211:83-96. http://dx.doi.org/10.1016/j.foreco.2005.01.034
Ager, A. A., M. A. Day, C. W. McHugh, K. Short, J. Gilbertson-Day, M. A. Finney, and D. E. Calkin. 2014a. Wildfire exposure and fuel management on western US national forests. Journal of Environmental Management 145:54-70. http://dx.doi.org/10.1016/j.jenvman.2014.05.035
Ager, A. A., M. A. Finney, A. J. McMahan, and J. Cathcart. 2010. Measuring the effect of fuel treatments on forest carbon using landscape risk analysis. Natural Hazards and Earth System Sciences 10:2515-2526. http://dx.doi.org/10.5194/nhess-10-2515-2010
Ager, A. A., H. K. Preisler, B. Arca, D. Spano, and M. Salis. 2014b. Wildfire risk estimation in the Mediterranean area. Environmetrics 25(6):384-396. http://dx.doi.org/10.1002/env.2269
Ager, A. A., N. M. Vaillant, and A. McMahan. 2013. Restoration of fire in managed forests: a model to prioritize landscapes and analyze tradeoffs. Ecosphere 4(2):1-19. http://dx.doi.org/10.1890/ES13-00007.1
Baker, W. L. 2015. Are high-severity fires burning at much higher rates recently than historically in dry-forest landscapes of the Western USA? PLoS ONE 10(9):e0136147. http://dx.doi.org/10.1371/journal.pone.0136147
Bolte, J. P., D. W. Hulse, S. V. Gregory, and C. Smith. 2007. Modeling biocomplexity - actors, landscapes and alternative futures. Environmental Modelling & Software 22(5):570-579. http://dx.doi.org/10.1016/j.envsoft.2005.12.033
Brown, T. J., B. L. Hall, and A. L. Westerling. 2004. The impact of twenty-first century climate change on wildland fire danger in the Western United States: an application perspective. Climate Change 62:365-388. http://dx.doi.org/10.1023/B:CLIM.0000013680.07783.de
Collins, B. L., S. L. Stephens, G. Roller, and J. J. Battles. 2011. Simulating fire and forest dynamics for a landscape fuel treatment project in the Sierra Nevada. Forest Science 57(2):77-88.
Collins, B. M., S. L. Stephens, J. J. Moghaddas, and J. Battles. 2010. Challenges and approaches in planning fuel treatments across fire-excluded forested landscapes. Journal of Forestry 108:24-31.
Dixon, G. E. 2002. Essential FVS: a user’s guide to the Forest Vegetation Simulator. Internal Report, U.S. Forest Service, Forest Management Service Center, Fort Collins, Colorado, USA. [online] URL: http://www.fs.fed.us/fmsc/ftp/fvs/docs/gtr/EssentialFVS.pdf
Finney, M. A. 2001. Design of regular landscape fuel treatment patterns for modifying fire growth and behavior. Forest Science 47:219-228.
Finney, M. A. 2002. Fire growth using minimum travel time methods. Canadian Journal of Forest Research 32:1420-1424. http://dx.doi.org/10.1139/x02-068
Finney, M. A. 2007. A computational method for optimizing fuel treatment location. International Journal of Wildland Fire 16:702-711. http://dx.doi.org/10.1071/WF06063
Finney, M. A., C. W. McHugh, and I. C. Grenfell. 2005. Stand- and landscape-level effects of prescribed burning on two Arizona wildfires. Canadian Journal of Forest Research 35:1714-1722. http://dx.doi.org/10.1139/x05-090
Gustafson, E. J., A. Z. Shividenko, B. R. Sturtevant, and R. M. Scheller. 2010. Predicting global change effects on forest biomass and composition in south-central Siberia. Ecological Applications 20(3):700-715. http://dx.doi.org/10.1890/08-1693.1
Halofsky, J. E., M. K. Creutzburg, and M. A. Hemstrom. 2014b. Integrating social, economic, and ecological values across large landscapes. General Technical Report. PNW-GTR-896, U.S. Forest Service, Pacific Northwest Research Station, Portland, Oregon, USA.
Halofsky, J. S., J. E. Halofsky, T. Burcsu, and M. A. Hemstrom. 2014a. Dry forest resilience varies under simulated climate-management scenarios in a central Oregon, USA landscape. Ecological Applications 24(8):1908-1925. http://dx.doi.org/10.1890/13-1653.1
Hampton, H. M., S. E. Sesnie, J. D. Bailey, and G. B. Snider. 2011. Estimating regional wood supply based on stakeholder consensus for forest restoration in northern Arizona. Journal of Forestry 109:16-26.
Hemstrom, M., A. A. Ager, M. Vavra, B. C. Wales, and M. J. Wisdom. 2004. A state and transition approach for integrated landscape models. Pages 17-32 in J. L. Hayes, A. A. Ager, and R. J. Barbour, editors. Methods for integrating modeling of landscape change: Interior Northwest Landscape Analysis System. General Technical Report PNW-GTR-610. U.S. Forest Service, Pacific Northwest Research Station, Portland, Oregon, USA.
Hemstrom, M. A., J. Merzenich, A. Reger, and B. Wales. 2007. Integrated analysis of landscape management scenarios using state and transition models in the upper Grande Ronde River Subbasin, Oregon, USA. Landscape and Urban Planning 80(3):198-211. http://dx.doi.org/10.1016/j.landurbplan.2006.10.004
Hessburg, P. F., J. K. Agee, and J. F. Franklin. 2005. Dry forests and wildland fires of the inland Northwest USA: contrasting the landscape ecology of the pre-settlement and modern eras. Forest Ecology and Management 211(1-2):117-139. http://dx.doi.org/10.1016/j.foreco.2005.02.016
Hessburg, P. F., R. B. Salter, and K. M. James. 2007. Re-examining fire severity relations in pre-management era mixed conifer forests: inferences from landscape patterns of forest structure. Landscape Ecology 22:5-24. http://dx.doi.org/10.1007/s10980-007-9098-2
Hessburg, P. F., T. A. Spies, D. A. Perry, C. N. Skinner, A. H. Taylor, P. M. Brown, S. L. Stephens, A. J. Larson, D. J. Churchill, N. A. Povak, P. H. Singleton, B. McComb, W. J. Zielinski, B. M. Collins, R. B. Salter, J. J. Keane, J. F. Franklin, and G. Riegel. 2016. Tamm Review: management of mixed-severity fire regime forests in Oregon, Washington, and Northern California. Forest Ecology and Management 366:221-250. http://dx.doi.org/10.1016/j.foreco.2016.01.034
Hessl, A. E., D. McKenzie, and R. Schellhaas. 2004. Drought and Pacific Decadal Oscillation linked to fire occurrence in the inland Pacific Northwest. Ecological Applications 14(2):425-442. http://dx.doi.org/10.1890/03-5019
Heyerdahl, E. K., L. B. Brubaker, and J. K. Agee. 2002. Annual and decadal climate forcing of historical fire regimes in the interior Pacific Northwest, USA. Holocene 12(5):597-604. http://dx.doi.org/10.1191/0959683602hl570rp
Jolly, W. M., M. A. Cochrane, P. H. Freeborn, Z. A. Holden, T. J. Brown, G. J. Williamson, and D. M. J. S. Bowman. 2015. Climate-induced variations in global wildfire danger from 1979 to 2013. Nature Communications 6:7537. http://dx.doi.org/10.1038/ncomms8537
Kerns, B. K., M. A. Hemstrom, D. Conklin, G. I. Yospin, B. Johnson, D. Bachelet, and S. Bridgham. 2011. Approaches to incorporating climate change effects in state and transition simulation models of vegetation. Pages 161-171 in B. K. Kerns, A. J. Shlisky, and J. Colin, editors. Proceedings of the First Landscape State-and-Transition Simulation Modeling Conference. General Technical Report PNW-GTR-869. U.S. Forest Service, Pacific Northwest Research Station, Portland, Oregon, USA.
Littell, J. S., 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(4):1003-1021. http://dx.doi.org/10.1890/07-1183.1
Loudermilk, E. L., A. Stanton, R. M. Scheller, T. E. Dilts, P. J. Weisberg, C. Skinner, and J. Yang. 2014. Effectiveness of fuel treatments for mitigating wildfire risk and sequestering forest carbon: a case study in the Lake Tahoe Basin. Forest Ecology and Management 323:114-125. http://dx.doi.org/10.1016/j.foreco.2014.03.011
Marlon, J. R., 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(9):E535-E543. http://dx.doi.org/10.1073/pnas.1112839109
Merschel, A. G., T. A. Spies, and E. K. Heyerdahl. 2014. Mixed-conifer forests of central Oregon: effects of logging and fire exclusion vary with environment. Ecological Applications 24(7):1670-1688. http://dx.doi.org/10.1890/13-1585.1
Miller, C. 2002. Wildland fire use: a wilderness perspective on fuel management. Pages 379-385 in P. N. Omni and L. Joyce, editors. Fire, fuel treatments, and ecological restoration: conference proceedings. Proceedings RMRS-P-29. U.S. Forest Service, Rocky Mountain Research Station, Fort Collins, Colorado, USA.
Miller, J. D., 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(1):16-32. http://dx.doi.org/10.1007/s10021-008-9201-9
Mote, P. W. 2003. Trends in temperature and precipitation in the Pacific Northwest during the twentieth century. Northwest Science 77(4):271-282.
North, M., A. Brough, J. Long, B. Collins, P. Bowden, D. Yasuda, J. Miller, and N. Sugihara. 2015. Constraints on mechanized treatment significantly limit mechanical fuels reduction extent in the Sierra Nevada. Journal of Forestry 113(1):40-48. http://dx.doi.org/10.5849/jof.14-058
North, M., B. M. Collins, and S. Stephens. 2012. Using fire to increase the scale, benefits, and future maintenance of fuels treatments. Journal of Forestry 110(7):392-401. http://dx.doi.org/10.5849/jof.12-021
Noss, R. F., J. F. Franklin, W. L. Baker, T. Schoennagel, and P. B. Moyle. 2006. Managing fire-prone forests in the western United States. Frontiers in Ecology and the Environment 4:481-487. http://dx.doi.org/10.1890/1540-9295(2006)4[481:MFFITW]2.0.CO;2
Odion, D. C., C. T. Hanson, A. Arsenault, W. L. Baker, D. A. DellaSala, R. L. Hutto, W. Klenner, M. A. Moritz, R. L. Sherriff, T. T. Veblen, and M. A. Williams. 2014. Examining historical and current mixed-severity fire regimes in ponderosa pine and mixed-conifer forests of western North America. PLoS ONE 9(2):e87852. http://dx.doi.org/10.1371/journal.pone.0087852
Odion, D. C., C. T. Hanson, W. L. Baker, D. A. DellaSala, and M. A. Williams. 2016. Areas of agreement and disagreement regarding ponderosa pine and mixed conifer forest fire regimes: a dialogue with Stevens et al. PLoS ONE 11(5):e0154579. http://dx.doi.org/10.1371/journal.pone.0154579
Ohmann, J. L., M. J. Gregory, E. B. Henderson, and H. M. Roberts. 2011. Mapping gradients of community composition with nearest-neighbour imputation: extending plot data for landscape analysis. Journal of Vegetation Science 22(4):660-676. http://dx.doi.org/10.1111/j.1654-1103.2010.01244.x
Parisien, M.-A., D. R. Junor, and V. G. Kafka. 2007. Comparing landscape-based decision rules for placement of fuel treatments in the boreal mixedwood of western Canada. International Journal of Wildland Fire 16:664-672. http://dx.doi.org/10.1071/WF06060
Parks, S. A., C. Miller, M.-A. Parisien, L. M. Holsinger, S. Z. Dobrowski, and J. Abatzoglou. 2015. Wildland fire deficit and surplus in the western United States, 1984-2012. Ecosphere 6(12):1–13. http://dx.doi.org/10.1890/ES15-00294.1
Perry, D. A., P. F. Hessburg, C. N. Skinner, T. A. Spies, S. L. Stephen, A. H. Taylor, J. F. Franklin, B. McComb, and G. Riegel. 2011. The ecology of mixed severity fire regimes in Washington, Oregon, and northern California. Forest Ecology and Management 262:703-717. http://dx.doi.org/10.1016/j.foreco.2011.05.004
Peters, D. P. C., R. A. Pielke Sr, B. T. Bestelmeyer, C. D. Allen, S. Munson-McGee, and K. M. Havstad. 2004. Cross-scale interactions, nonlinearities, and forecasting catastrophic events. Proceeding of the National Academy of Science USA 101(42):15130-15135. http://dx.doi.org/10.1073/pnas.0403822101
Preisler, H. K. and A. A. Ager. 2013. Forest-fire models. Pages 1081-1088 in A. H. El-Shaarawi and W. Piegorsch, editors. Encyclopedia of environmetrics. John Wiley & Sons, Chichester, UK. http://dx.doi.org/10.1002/9780470057339.vaf010.pub2
Preisler, H. K., D. R. Brillinger, R. E. Burgan, and J. W. Benoit. 2004. Probability based models for estimating wildfire risk. International Journal of Wildland Fire 13:133-142. http://dx.doi.org/10.1071/WF02061
R Core Team. 2014. R: A language and environment for statistical computing. Version 3.1.1. The R Project for Statistical Computing, Vienna, Austria. [online] URL: http://www.R-project.org/
Radeloff, V. C., R. B. Hammer, S. I. Stewart, J. S. Fried, S. S. Holcomb, and J. F. McKeefry. 2005. The wildland-urban interface in the United States. Ecological Applications 15:799-805. http://dx.doi.org/10.1890/04-1413
Reinhardt, E. D., R. E. Keane, D. E. Calkin, and J. D. Cohen. 2008. Objectives and considerations for wildland fuel treatment in forested ecosystems of the interior western United States. Forest Ecology and Management 256:1997-2006. http://dx.doi.org/10.1016/j.foreco.2008.09.016
Rollins, M. G. 2009. LANDFIRE: a nationally consistent vegetation, wildland fire, and fuel assessment. International Journal of Wildland Fire 18:235-249. http://dx.doi.org/10.1071/WF08088
Scheller, R. M., and D. J. Mladenoff. 2005. A spatially interactive simulation of climate change, harvesting, wind, and tree species migration and projected changes to forest composition and biomass in northern Wisconsin, USA. Global Change Biology 11:307-321. http://dx.doi.org/10.1111/j.1365-2486.2005.00906.x
Schmidt, D. A., A. H. Taylor, and C. N. Skinner. 2008. The influence of fuels treatment and landscape arrangement on simulated fire behavior, southern Cascade Range, California. Forest Ecology and Management 255:3170-3184. http://dx.doi.org/10.1016/j.foreco.2008.01.023
Scott, J. H. and R. E. Burgan. 2005. Standard fire behavior fuel models: a comprehensive set for use with Rothermel’s surface fire spread model. General Technical Report RMRS-GTR-153, U.S. Forest Service, Rocky Mountain Research Station, Fort Collins, Colorado, USA.
Sherriff, R. L., R. V. Platt, T. T. Veblen, T. L. Schoennagel, and M. H. Gartner. 2014. Historical, observed, and modeled wildfire severity in montane forests of the Colorado Front Range. PLoS ONE 9(9):e106971. http://dx.doi.org/10.1371/journal.pone.0106971
Short, K. C. 2015. Spatial wildfire occurrence data for the United States, 1992-2013 [FPA_FOD_20150323]. Third edition. U.S. Forest Service, Rocky Mountain Research Station, Fort Collins, Colorado, USA. http://dx.doi.org/10.2737/RDS-2013-0009.3
Slocum, M. G., B. Beckage, W. J. Platt, S. L. Orzell, and W. Taylor. 2010. Effect of climate on wildfire size: a cross-scale analysis. Ecosystems 13:828-840. http://dx.doi.org/10.1007/s10021-010-9357-y
Spies, T. A., E. White, A. Ager, J. D. Kline, J. P. Bolte, E. K. Platt, K. A. Olsen, R. J. Pabst, A. M.G. Barros, J. D. Bailey, S. Charnley, J. Koch, M. M. Steen-Adams, P. H. Singleton, J. Sulzman, C. Schwartz, and B. Csuti. 2017. Using an agent-based model to examine forest management outcomes in a fire-prone landscape in Oregon, USA. Ecology and Society 22(1):25. https://doi.org/10.5751/ES-08841-220125
Spies, T. A., E. M. White, J. D. Kline, A. P. Fischer, A. Ager, J. Bailey, J. Bolte, J. Koch, E. Platt, C. S. Olsen, D. Jacobs, B. Shindler, M. M. Steen-Adams, and R. Hammer. 2014. Examining fire-prone forest landscapes as coupled human and natural systems. Ecology and Society 19(3):9. http://dx.doi.org/10.5751/ES-06584-190309
Spracklen, D. V., L. J. Mickley, J. A. Logan, R. C. Hudman, R. Yevich, M. D. Flannigan, and A. L. Westerling. 2009. Impacts of climate change from 2000 to 2050 on wildfire activity and carbonaceous aerosol concentrations in the western United States. Journal of Geophysical Research 114:1-17. http://dx.doi.org/10.1029/2008JD010966
Stavros, E. N., J. T. Abatzoglou, D. McKenzie, and N. K. Larkin. 2014. Regional projections of the likelihood of very large wildland fires under a changing climate in the contiguous Western United States. Climatic Change 126(3-4):455-468. http://dx.doi.org/10.1007/s10584-014-1229-6
Stephens, S. L., S. W. Bigelow, R. D. Burnett, B. M. Collins, C. V. Gallagher, J. Keane, D. A. Kelt, M. P. North, L. J. Roberts, P. A. Stine, and D. H. Van Vuren. 2014. California spotted owl, songbird, and small mammal responses to landscape fuel treatments. BioScience 64(10):893-906. http://dx.doi.org/10.1093/biosci/biu137
Stevens, J. T., H. D. Safford, M. P. North, J. S. Fried, A. N. Gray, P. M. Brown, C. R. Dolanc, S. Z. Dobrowski, D. A. Falk, C. A. Farris, J. F. Franklin, P. Z. Fulé, R. K. Hagmann, E. E. Knapp, J. D. Miller, D. F. Smith, T. W. Swetnam, and A. H. Taylor. 2016. Average stand age from forest inventory plots does not describe historical fire regimes in ponderosa pine and mixed-conifer forests of western North America. PLoS ONE 11(5):e0147688. 10.1371/journal.pone.0147688 http://dx.doi.org/10.1371/journal.pone.0147688
Strom, B. A. and P. Z. Fulé. 2007. Pre-wildfire fuel treatments affect long-term ponderosa pine forest dynamics. International Journal of Wildland Fire 16(1):128-138. http://dx.doi.org/10.1071/wf06051
Syphard, A. D., R. M. Scheller, B. C. Ward, W. D. Spencer, and J. R. Strittholt. 2011. Simulating landscape-scale effects of fuels treatments in the Sierra Nevada, California, USA. International Journal of Wildland Fire 20(3):364-383. http://dx.doi.org/10.1071/wf09125
Taylor, M. H., K. Rollins, M. Kobayahsi, and R. J. Tausch. 2013. The economics of fuel management: wildfire, invasive plants, and the dynamics of sagebrush rangelands in the western United States. Journal of Environmental Management 126:157-173. http://dx.doi.org/10.1016/j.jenvman.2013.03.044
Thompson, J. R., and T. A. Spies. 2009. Vegetation and weather explain variation in crown damage within a large mixed severity wildfire. Forest Ecology and Management 258:1684-1694. http://dx.doi.org/10.1016/j.foreco.2009.07.031
U.S. Congress. 2003. Healthy Forest Restoration Act of 2003. U.S. Congress, Washington, D.C., USA.
U.S. Forest Service (USFS). 1990. Deschutes National Forest: land and resource management plan. U.S. Forest Service, Pacific Northwest Region, Portland, Oregon, USA.
U.S. Forest Service and U.S. Geological Survey (USFS and USGS). 2009. Monitoring trends in burn severity (MTBS). National Geospatial Data. U.S. Forest Service, Washington, D.C., USA and U.S. Geological Survey, Reston, Virginia, USA. [online] URL: http://mtbs.gov/nationalregional/intro.html
Vogler, K. C., A. A. Ager, M. A. Day, M. Jennings, and J. D. Bailey. 2015. Prioritization of forest restoration projects: tradeoffs between wildfire protection, ecological restoration and economic objectives. Forests 6:4403-4420. http://dx.doi.org/10.3390/f6124375
Weatherspoon, C. P., S. J. Husari, and J. W. Van Wagtendonk. 1992. Fire and fuels management in relation to owl habitat in forests of the Sierra Nevada and southern California. Pages 247-260 in J. Verner, K. S. McKelvey, B. R. Noon, R. J. Gutierrez, G. I. Gould Jr., and T. W. Beck, editors. The California spotted owl: a technical assessment of its current status. General Technical. Report PSW-GTR-133. U.S. Forest Service, Pacific Southwest Research Station, Albany, California, USA.
Westerling, A. L. 2016. Increasing western US forest wildfire activity: sensitivity to changes in the timing of spring. Philosophical Transactions of the Royal Society B-Biological Sciences 371:20150178. http://dx.doi.org/10.1098/rstb.2015.0178
Westerling, A. L., H. G. Hidalgo, D. R. Cayan, and T. W. Swetnam. 2006. Warming and earlier spring increase western U.S. forest wildfire activity. Science 313:940-943. http://dx.doi.org/10.1126/science.1128834
Williams, J. 2013. Exploring the onset of high-impact mega-fires through a forest land management prism. Forest Ecology and Management 294:4-10. http://dx.doi.org/10.1016/j.foreco.2012.06.030
Yospin, G. I., S. D. Bridgham, R. P. Neilson, J. P. Bolte, D. M. Bachelet, P. J. Gould, C. A. Harrington, J. A. Kertis, C. Evers, and B. R. Johnson. 2015. A new model to simulate climate change impacts on forest succession for local land management. Ecological Applications 25(1):226-242. http://dx.doi.org/10.1890/13-0906.1