Skip to main content
NCBI home page
EPA Author Manuscripts logoLink to EPA Author Manuscripts
. Author manuscript; available in PMC: 2025 Nov 26.
Published in final edited form as: Environ Res Lett. 2024 Nov 26;19(12):124075. doi: 10.1088/1748-9326/ad8e75

Long-term patterns of post-fire harvest diverge among ownerships in the Pacific West, U.S.A

Aaron Zuspan 1,∗, Matthew J Reilly 2, E Henry Lee 3
PMCID: PMC11998924  NIHMSID: NIHMS2040047  PMID: 40241692

Abstract

Post-fire harvest (PFH) is a forest management practice designed to salvage value from burned timber, mitigate safety hazards from dead trees, reduce long-term fuels, and prepare sites for replanting. Despite public controversy and extensive ecological research, little is known about how much PFH occurs on private and public lands in the U.S. Pacific West, or how practices changed with shifting forest policy and increasing area burned over the last three decades. We mapped PFH across 2.2 M burned hectares in California, Oregon, and Washington between 1986–2017 and used time series intervention analysis to compare trends in area, rate (% of burned area harvested), and mean patch size between private (0.5 M ha) and federal (1.6 M ha) forest land and across a gradient of burn severity. Harvest rates varied by ownership (4.9% federal, 18.6% private, 8.0% overall), and practices evolved and diverged over the study period. PFH area and rate declined across all ownerships in the mid-1990s during a period of reduced fire activity. As area burned increased between the early 2000s and late 2010s, PFH area rebounded and surpassed late-1980s levels, while rates remained relatively low. On federal lands, PFH practices shifted in the early-to-mid 1990s towards lower rates (10.3%–3.8%) and smaller patches (6.0–3.3 ha), following policy changes and increased litigation. PFH rates on federal lands decreased at all levels of burn severity, with the largest decreases (6.2%–1.2%) in forests with low tree mortality (i.e. fire refugia). Conversely, private PFH rates and mean patch sizes more than doubled in forests burned at very low-to-moderate severity. Our results highlight how PFH practices have shifted with policy, socio-economic pressure, and increasing area burned over 31 years in the Pacific West. A similar area of PFH is now dispersed over larger fires, with practices diverging substantially between ownerships.

Keywords: salvage logging, Landsat, change detection, post-fire management, regional trends, time series analysis, burn severity

1. Introduction

Post-fire harvest (PFH) of trees is a common management practice globally (Leverkus et al 2021). These harvests encompass a variety of different practices with diverse objectives, including salvage logging to recoup economic losses from burned timber (Müller et al 2019), hazard tree removal and fuel treatment to mitigate safety hazards and reduce long-term reburn risks from standing dead trees (Peterson et al 2015), and site preparation for replanting (Sessions et al 2004). However, PFHs also have well-studied ecological impacts, contributing to soil compaction and runoff (Wagenbrenner et al 2015), increasing short-term reburn risk (Peterson et al 2009), and altering wildlife habitat (Reeves et al 2006, Clark et al 2013, Green et al 2022). Despite the potentially broad consequences of PFH, there is widespread uncertainty regarding where, when, and how much PFH occurs.

Annual area burned by wildfire in the western United States has approximately tripled since the mid-1980s (Higuera and Abatzoglou 2021), creating more potential for PFH as a management tool (Müller et al 2019). During this time, PFH has also become an important, and often controversial, economic and socio-ecological issue (Ryan and Hamin 2008, Lindenmayer et al 2012), serving as a lightning rod for litigation and forest policy worldwide (Müller et al 2019). With disparate management goals, shifting market pressures, and evolving public opinion and policies, the application of PFH treatments likely varies between ownerships and over time. However, there is a lack of quantitative data to compare how much burned area is harvested across ownerships, or how its use in post-fire management has changed over time with increasing fire activity and shifting forest policies and public perceptions.

Forest harvests have been mapped at broad spatial and temporal scales using remote sensing (e.g. Hansen et al 2013, Brown et al 2022, USDA Forest Service 2023). However, attempts to apply established remote sensing techniques to map harvests in burned forests have revealed unique challenges, achieving lower classification accuracy than in unburned forests (Zhao et al 2022) and struggling to distinguish harvests from direct fire effects (Schroeder et al 2012). Recent studies have made considerable progress using techniques designed for PFHs, achieving high classification accuracy (~90%) in individual fires (Green et al 2022, Li and Xu 2023), but no effort has been made to do so at broad spatial and temporal scales. Lack of a single, generalizable method for retroactively mapping PFH limits our ability to track its impact on landscape-scale patterns of subsequent burn severity, early seral habitat, and watershed health.

Given recent and projected increases in area burned and the range of potential ecological impacts, there is a need for accurate, wall-to-wall mapping and monitoring of PFHs. To that end, we developed and validated a methodology for annually mapping the presence and timing of PFHs at broad spatiotemporal scales by leveraging the Google Earth Engine cloud-computing platform (Gorelick et al 2017) to apply unsupervised change detection to Landsat composites acquired immediately after fire (pre-harvest) and in the five subsequent years, and applied it to 1007 fires that burned 2.2 million forested hectares across ownerships between 1986–2017 in California, Oregon, and Washington, USA. We used these annual PFH maps to compare long-term trends in post-fire forest harvest area, rate (percent of burned area harvested), and harvest patch size among ownerships and levels of burn severity.

2. Methods

2.1. Study region

Our study encompasses five forested ecoregions across Oregon, Washington, and California (figure 1), covering a broad range of biophysical settings and forest types from moist temperate rainforests to dry forests and woodlands, to cold subalpine forests at higher elevations. Many of these forests are dominated by long-lived conifers that constitute some of the most biodiverse and productive forests in the world and provide habitat for threatened and endangered species including the northern spotted owl (NSO; Strix occidentalis caurina) and California spotted owl (CASPO; Strix occidentalis occidentalis). The region is divided among private and public ownerships including small private landowners, industrial private timber companies, federal lands managed for multiple use by the USDA Forest Service (USFS) and Bureau of Land Management (BLM), federal lands preserved in wilderness areas and national parks, tribal lands, and public entities at the state and local levels. Harvest practices on federal lands in the region are regulated by a number of policies and guidelines, many of which were enacted throughout the 1990s (table 1). Private lands are regulated primarily by state-level policies. We excluded wilderness and national parks from the study region as they are not managed for timber production.

Figure 1.

Figure 1.

Open in a new tab

(Left) The study area in the Pacific West United States, with land ownership and study fires overlaid. Wilderness and national parks were excluded from analysis. (Right) Area burned (top two panels) and area harvested post-fire (bottom two panels) in the 1007 study fires, annually and cumulatively over the study period, by ownership. Grey bars in the top two panels show the total forested area burned in the study region, including wilderness and national parks that were excluded from analysis.

Table 1.

A partial timeline of federal policies and guidelines that impacted timber harvest practices within the study region.

Policy Year Region Summary
Dwyer court injunctions (Johnson et al 2023) 1989 and 1991 Range of the northern spotted owl in OR, WA, and northern CA Temporarily halted most federal timber sales and logging of old-growth on federal lands
Northern spotted owl listed as a threatened species 1990 Range of the northern spotted owl in OR, WA, and northern CA Mandated habitat protection under the Endangered Species Act
CASPO Interim Recommendations (Verner et al 1992) 1993 Range of the California spotted owl in Sierra Nevada and southern CA Provided recommendations for “stand retention and special treatment to maintain populations of spotted owls on public timberlands”
Northwest Forest Plan (USDA Forest Service and USDI Bureau of Land Management 1994) 1994 Range of the northern spotted owl in OR, WA, and northern CA Restricted salvage within Late Successional Reserves (LSRs) to “conservative quantities”, only in disturbances >10 acres with <40% canopy cover, where salvage would “not generally result in degeneration of currently suitable owl habitat or other late successional conditions”; commercial salvage outside of LSRs (i.e. matrix) less constrained
Eastside Screens (Johnston et al 2021) 1994 Eastern OR and WA Restricted harvest of trees >21′′ diameter; promoted leaving snags for wildlife
Salvage Rider (Dorn 1996) 1995–1996 US Temporarily exempted post-fire harvests from environmental review and court injunctions
Sierra Nevada Forest Plan Amendment 2004 Sierra Nevada Allows for ‘salvage of dead and dying trees for both economic value and fuels reduction purposes’ and provides guidelines for ‘snag and down woody material retention’
Open in a new tab

2.2. Fire selection

We identified 1170 fires that cumulatively burned 3.4 M ha of forest (Eidenshink et al 2007; available online https://mtbs.gov) between 1986–2017 across our study region. To overcome the challenge of acquiring immediate post-fire imagery without cloud, snow, and shadow contamination, we excluded late-season fires that burned beyond 1 September, leaving 1007 study fires (2.2 M ha; figure 1) that account for 66% of the total forest area burned. Within each fire, we excluded forests in wilderness and national parks, and non-forest areas based on pre-fire land cover data (USDA Forest Service 2023).

Most of the study area burned on federal lands (73%, 1.6 M ha), followed by private lands (21%, 0.5 M ha), and other ownerships (5%, 0.1 M ha; supplementary table 1). Federal lands were primarily managed by the USFS (89%, 1.4 M ha) with the remainder managed by BLM (11%, 0.2 M ha). Annual area burned in our study fires was moderate in the late 1980s, decreased and remained low through the 1990s, and increased substantially between the early 2000s and the late 2010s, consistent with patterns of overall fire activity in the region (figure 1).

2.3. Mapping PFH

We applied change detection across annual pairs of Landsat composite imagery to determine the presence and timing of PFHs for five years following each study fire (figure 2; Zuspan et al 2024; supplementary figure 1). The Landsat mission comprises a series of Earth observation satellites that have collected moderate resolution (30 m pixel size) multispectral imagery globally on a bi-weekly interval since 1984, and has been widely used in natural resource monitoring due to the quality, consistency, and length of the data record (Wulder et al 2019). Composite imagery, combining pixels from multiple different acquisitions, allows for the creation of cloud-free mosaics over large areas, and has become increasingly common with the availability of cloud-computing platforms like Google Earth Engine (Gorelick et al 2017) that can process vast quantities of data.

Figure 2.

Figure 2.

Open in a new tab

Post-fire harvest methodology applied to the 2013 North Rabbit fire in southwestern Oregon. Landsat composites from summer of each year for five years following fire were differenced to identify spectral changes (top left; increased reflectance in the red and short-wave infrared wavelengths appear bright pink or white). The magnitude and timing of the maximum change (bottom left) were used to generate maps of post-fire harvests (bottom center; color-coded by year of detected harvest, relative to the fire year) which were overlaid with ownership and burn severity maps (center). Burn severity was classified following Reilly et al (2017) using Landsat imagery from immediately post-fire and one year prior. Inset maps on the right show high-resolution NAIP aerial photography before (top) and after harvests (middle) with example classified harvest patches (bottom) on federal (A) and private lands (B).

For each of the five years following fire, we generated one cloud-free summer composite per fire and compared spectral changes in the red and short-wave infrared (SWIR) wavelengths with the following year to highlight vegetation removal and soil exposure (for details, see supplementary data S1. Methodology for Building Annual Paired Landsat Composites). We applied unsupervised classification with Otsu’s method to identify harvests based on the maximum spectral change over five years and assigned harvest years based on the timing of the largest change. Otsu’s method is an image thresholding technique that finds an optimal threshold by maximizing variance between classes, and has shown strong results in remotely sensed change detection (Melgani et al 2002). We calculated thresholds in the red and SWIR wavelengths across all fire years, and assigned harvests where the maximum change exceeded both thresholds. We chose to monitor PFHs for a five-year window as rapid decay in wood quality and timber value in the first two to three years following fire (Peterson et al 2009) makes large harvests beyond that time scale much less likely. To separate fire effects from harvests occurring immediately after fire, the composite window for the first year was shifted to coincide with the first post-fire acquisition, identified based on remotely sensed hotspots (see supplementary data S2. Methodology for Identifying the Earliest Post-fire Landsat Acquisition).

Patch sizes and areal estimates of harvested and burned areas were calculated for each fire year by ownership type (U.S. Geological Survey, Gap Analysis Project et al 2018) and burn severity class from our predicted harvest maps in Google Earth Engine. Burn severity was calculated based on changes in the normalized burn ratio between Landsat imagery acquired one-year pre-fire and immediately following fire, and was classified into four classes of basal area mortality following Reilly et al (2017): unburned/very low = <10%, low = 10%–25%, moderate = 25%–75%, and high ≤75%.

2.4. Accuracy assessment

Following PFH mapping, we randomly allocated and photo-interpreted 540 validation plots across a subset of our study fires, stratified equally between areas with and without predicted harvest. Validation fires were selected based on availability of high-resolution (0.6–1 m) aerial photography from the National Agriculture Imagery Program (NAIP), which was visually interpreted to identify effects of PFH within each 900 m2 plot, corresponding to a single 30 m Landsat pixel. Plots with any evidence of tree removal were considered harvested. During photo-interpretation, we excluded 16 plots where the presence of PFH within five years could not be confidently determined due to image quality (e.g. shadows or misregistration).

2.5. Trend analysis

We used time series intervention analysis to model trend and step changes (Box and Tiao 1975, Lee et al 2017) of annual PFH area, harvest rate, and mean harvest patch size by land ownership and burn severity. We log transformed annual area trends to linearize the relationship between area and predictor variables, and chose to evaluate all variables based on year-of-fire rather than year-of-harvest to (1) avoid introducing errors from harvest timing uncertainty and (2) allow consistent calculation of harvest rates across the study period based on a single corresponding fire year. Pulse intervention was used to account for abrupt discontinuities in the PFH area and rate time series (i.e. years with <1000 ha burned). Trend intervention was used to detect a long-term change in PFH area and rate and test for changes in trends over different time periods. Step intervention was used to detect a short- to longer-term shift in the mean level. Step and trend intervention analysis were performed using piecewise polynomial regression or grafted polynomials to partition the time series by its trend and/or step change (Fuller 1996, Lee et al 2017). The breakpoint years of the step and trend change models were determined based on optimization of statistical fit in terms of the Akaike Information Criterion (AIC) over all years. We selected the time series regression model with the lowest AIC value. For parsimony, only predictor variables that were statistically significant at the 0.10 level of significance were retained in the model. Maximum likelihood estimation of model parameter coefficients was performed using the gls function in the nlme package (Pinheiro et al 2023) in R 4.1.0 (R Core Team 2023).

3. Results

3.1. PFH mapping accuracy

PFHs observed in high-resolution imagery corresponded well with increased reflectance in Landsat composites. These changes were most noticeable when live vegetation was removed from forests that burned at lower severity but could also be seen in areas of high burn severity where harvest operations removed charcoal and dead trees, exposing bright soil underneath. Applying change thresholds across all study fires achieved high overall accuracy (90%), with the lowest performance (88%) in areas of high burn severity and highest performance (98%) in low severity plots (table 2). Errors of commission, where harvests were predicted in unharvested plots, accounted for 4% of all plots and were frequently observed in areas adjacent to harvest operations, sometimes including road construction and soil disturbance without tree removal, and in areas with coincident tree mortality not related to management, i.e. delayed mortality. Errors of omission, where harvested plots were predicted as unharvested, accounted for 6% of all plots, and were frequently observed near harvest edges or other areas where a large proportion of trees were retained. In other cases, harvests were observed with little or no corresponding spectral change, suggesting they may have occurred immediately post-fire, prior to image acquisition.

Table 2.

Classification accuracy for post-fire harvest by burn severity class, calculated from immediate post-fire Landsat imagery (pre-harvest), based on photo-interpreted plots. User’s accuracy describes the frequency of commission errors, and was calculated as the number of correctly predicted harvested plots divided by the total number of predicted harvested plots. Producer’s accuracy describes the frequency of omission errors, and was calculated as the number of correctly predicted harvested plots divided by the total number of observed harvested plots. Overall accuracy was calculated as the total number of correctly predicted plots (harvested and unharvested) divided by the total number of plots.

Burn severity User’s accuracy Producer’s accuracy Overall accuracy Plots
Unburned/very low 94% 89% 95% 114
Low 100% 95% 98% 44
Moderate 88% 87% 88% 142
High 90% 89% 88% 224
Overall 91% 89% 90% 524
Open in a new tab

3.2. Temporal trends in PFH

We mapped 178k ha of PFH over the study period (8.0% of area burned), with 80k ha on federal lands (4.9% of area burned), 89k ha on private lands (18.6% of area burned), and 9k ha distributed across other ownerships (7.6% of area burned). PFH practices varied substantially over time and by ownership. Across ownerships, the extent of mapped harvests decreased rapidly after the first year post-fire. Overall, 66% of all harvests that took place within five years of fire occurred in the first year, and 91% were completed within three years. Harvests on federal lands generally occurred later than on private lands, with 54% and 74% taking place in the first year on federal and private lands, respectively. Harvests after three years post-fire accounted for only 11% and 8% of the harvested extent on federal and private lands, respectively.

The total annual area of PFH exhibited a V-shaped trend, declining from the late 1980s until the late 1990s, then increasing through the end of the study period (table 3, figure 3). This increasing trend in harvest area from the late 1990s was reflected on both private and federal lands, with both owner groups nearing or reaching peak harvest areas at the end of the study period. Despite a larger overall area harvested, PFH rates dropped substantially in mid-late 1990s fires on federal lands (from 10.3% to 3.8%) and overall (from 13.2% to 6.5%). Private harvest rates fluctuated around 15.5% without significant trends over time. Mean harvest patch sizes showed no overall trends, but diverged sharply between owners starting with 1997 fires: federal patch sizes dropped by half from 6 ha to 3 ha, while private lands doubled from 5 ha to 10 ha over the following 20 years.

Table 3.

Times series intervention analysis results for area, rate (percent of area burned), and mean patch size of post-fire harvest by ownership across forested lands of Oregon, Washington, and California for fire years 1986–2017. All ownerships exclude wilderness and NPS. Federal lands include USFS and BLM. Temporal autocorrelation was modeled by a first-order autoregressive model (AR1). The AR1 model was retained when the model parameter ρ was significantly different from zero based on the likelihood ratio test at the 0.10 level of significance. Pulse intervention terms are 0–1 indicator variables for specific years, i.e. Pulse 1997 = 1 if Year = 1997 and =0 otherwise. Trend intervention terms are indicated by the start year of the trend, i.e. Trend 1997 = 0 if Year <1997 and =Year-1997 if Year ⩾1997. Step intervention terms are indicated by the start year of the step change, i.e. Step 1997 = 0 if Year <1997 and =1 if Year ⩾1997.

Metric Ownership Model performance Parameters Estimate Std. error p-value
All AIC = 54.2 Intercept 124.05 66.95 0.075
r2 = 0.76 Trend 1997 0.10 0.05 0.035
Pulse 1991 −1.56 0.52 0.006
Pulse 1993 −3.37 0.52 0.000
Pulse 1997 −2.01 0.54 0.001
Year −0.06 0.03 0.083
Federal AIC = 43.3 Intercept 3.46 0.15 0.000
r2 = 0.82 Step 1998 −0.66 0.23 0.009
Trend 1998 0.04 0.23 0.031
Pulse 1991 −1.56 0.44 0.002
Area (log ha) Pulse 1993 −3.46 0.44 0.000
Pulse 1995 −1.80 0.44 0.000
Pulse 1997 −2.73 0.44 0.000
Private AIC = 49.2 Intercept −81.93 33.52 0.022
r2 = 0.86 Year 0.04 0.02 0.018
Step 1997 −0.62 0.35 0.092
Pulse 1991 −2.32 0.48 0.000
Pulse 1993 −3.33 0.48 0.000
Pulse 1995 −1.53 0.49 0.005
Pulse 1997 −1.93 0.50 0.001
Pulse 2003 −3.07 0.47 0.000
All AIC = 174.1 Intercept 13.21 1.14 0.000
r2 = 0.51 Step 1996 −6.68 1.35 0.000
Pulse 1993 −13.21 3.59 0.001
Federal AIC = 153.0 Intercept 10.28 0.67 0.000
r2 = 0.81 Step 1996 −6.52 0.74 0.000
ρ = −0.59 Pulse 1991 11.03 2.32 0.000
Rate (%) Pulse 1993 −6.02 2.32 0.000
Pulse 1995 10.11 2.18 0.000
Pulse 1998 12.19 2.05 0.000
Private AIC = 217.9 Intercept 15.47 1.24 0.000
r2 = 0.32 Pulse 1993 −15.47 6.78 0.030
Pulse 1997 −12.54 6.78 0.075
Pulse 2003 −15.43 6.78 0.031
All AIC = 150.9 Intercept 6.17 0.44 0.000
r2 = 0.26 Pulse1993 −6.17 2.41 0.016
Pulse 1997 −4.80 2.41 0.056
Federal AIC = 111.2 Intercept 5.97 0.28 0.000
Mean Patch r2 = 0.83 Step 1997 −2.62 0.33 0.000
Size (ha) ρ = −0.67 Pulse 1993 −4.27 1.13 0.001
Pulse 1995 −2.76 1.13 0.020
Pulse 1998 7.78 1.04 0.000
Private AIC = 191.3 Intercept 5.14 1.11 0.000
r2 = 0.13 Trend 1997 0.25 0.12 0.044
Open in a new tab

Figure 3.

Figure 3.

Open in a new tab

Temporal trends in the log-scale area, rate (percent of area burned), and mean patch size of post-fire harvest by ownership across forested lands of Oregon, Washington, and California for fire years 1986–2017. Statistically significant trends (p < 0.05) are represented by solid lines. Dotted lines represent long term averages with no significant temporal trend. Statistically significant trend changes within each panel are noted with ∗. Statistically significant step changes are noted with ⁑. Pulse intervention terms were included in the time series intervention model to account for abrupt discontinuities in salvage area, rate, and patch size associated with low annual area burned. Refer to table 3 for details on the magnitude and timing of significant changes.

PFH rates generally increased with burn severity across ownerships, although trends again diverged over time between federal and private lands (table 4, figure 4). Federal harvest rates decreased across all burn severity classes, with lower severity areas (<25% basal area mortality) decreasing first in 1992 fires, followed by higher severity areas in 1996 fires. In contrast, harvest rates on private lands that burned at very low to moderate severity (<75% basal area mortality) showed increasing trends beginning in the early-mid 1990s, with rates doubling for forests burned at moderate severity (10.9%–24.7%) and tripling for forests burned at very low (5.7%–18.2%) and low (7.8%–21.7%) severity. Harvest rates for private forests burned at high severity remained relatively stable at 18.8%.

Table 4.

Times series intervention analysis results for post-fire harvest rate (percent of area burned) by ownership and burn severity class across forested lands of Oregon, Washington, and California for fire years 1986–2017. Federal lands include USFS and BLM and exclude wilderness. Burn severity classifications are derived from relative changes in the normalized burn ratio based on thresholds from Reilly et al (2017), using Landsat composites from immediately post-fire (pre-harvest) and one year prior.

Owner Burn severity Model performance Parameters Estimate Std. error p-value
High AIC = 189.5 Intercept 15.87 1.49 0.000
r2 = 0.69 Step 1996 −8.46 1.75 0.000
Pulse 1993 −15.87 4.48 0.002
Pulse 1995 9.87 4.48 0.036
Pulse 1998 20.92 4.32 0.000
Moderate AIC = 142.3 Intercept 8.81 0.61 0.000
r2 = 0.97 Step 1996 −6.44 0.68 0.000
ρ = −0.41 Pulse 1991 13.70 2.06 0.000
Pulse 1993 −10.26 2.06 0.000
Federal Pulse 1995 10.70 1.97 0.000
Pulse 1998 7.03 1.86 0.001
Low AIC = 133.4 Intercept 7.63 0.63 0.000
r2 = 0.96 Step 1992 −6.43 0.69 0.000
ρ = −0.27 Pulse 1991 13.36 1.84 0.000
Pulse 1998 5.36 1.71 0.004
Very Low/Unburned AIC = 140.9 Intercept 6.23 0.89 0.000
r2 = 0.69 Step 1992 −5.02 0.98 0.000
Pulse 1991 6.16 2.19 0.009
Pulse 1998 7.72 2.04 0.001
High AIC = 244.6 Intercept 18.81 1.87 0.000
r2 = 0.18 Pulse 1993 −18.81 10.41 0.081
Pulse 2003 −18.81 10.41 0.081
Moderate AIC = 223.6 Intercept 10.93 1.80 0.000
r2 = 0.41 Trend 1997 0.69 0.19 0.001
Pulse 2003 −15.06 7.40 0.051
Private Pulse 2010 −15.98 7.50 0.042
Low AIC = 224.2 Intercept 7.81 2.03 0.001
r2 = 0.35 Trend 1993 0.58 0.16 0.001
Pulse 2003 −13.64 7.57 0.082
Very Low/Unburned AIC = 216.6 Intercept 5.67 1.76 0.003
r2 = 0.29 Trend 1995 0.50 0.16 0.004
Pulse 1992 14.86 6.84 0.038
Open in a new tab

Figure 4.

Figure 4.

Open in a new tab

Temporal trends in the rate (percent of area burned) of post-fire harvest by burn severity and ownership across forested lands of Oregon, Washington, and California from 1986 to 2017. Burn severity classifications are derived from relative changes in the normalized burn ratio based on thresholds from Reilly et al (2017), using Landsat composites from immediately post-fire (pre-harvest) and one year prior. Statistically significant trends (p < 0.05) are represented by solid lines. Dotted lines represent long term averages with no significant temporal trend. Statistically significant trend changes noted with ∗. Statistically significant step changes area noted with ⁑. Refer to table 4 for details on the magnitude and timing of significant changes.

4. Discussion

We mapped 178k ha of PFH in 1007 fires that burned 2.2 million hectares of forest over 31 years across a broad, mixed-ownership region with a high degree of accuracy. Our results provide an unprecedented look at changing PFH practices between ownerships in the three-state region during a period of increasing fire activity (figure 1) and evolving federal forest policies (table 1). The area of PFH varied over time, declining from the mid-1980s and to a minimum in the late-1990s during a time of relatively low fire activity that coincided with increased public scrutiny, policy changes (Johnson et al 2023), and litigation surrounding public land management practices (Miner et al 2014). However, as area burned increased in the early 2000s, PFH area increased again. A similar area of harvests is now dispersed over larger fires and more area burned, but harvest rates and spatial patterns have diverged between ownerships.

PFH practices on federal lands shifted dramatically over the study period, decreasing in both rate and mean patch size. The earliest change we detected in federal PFH practices was a negative step change in harvest rates in forests that were either unburned or experienced very low-severity fire effects (i.e. fire refugia) in the early 1990s. These changes followed the listing of the NSO as a threatened species under the Endangered Species Act in 1990 and a 1991 court injunction that suspended harvest of late successional and old-growth trees in national forests across western Oregon, Washington, and northern California (Johnson et al 2023). Recommendations and guidelines to conserve habitat for the California spotted owl also reduced harvest in the Sierra Nevada and southern California (Verner et al 1992). While these policies were not explicitly focused on salvage logging, they effectively reduced harvest of fire refugia that maintained late successional and old-growth characteristics within burned areas.

Although the area of PFH on federal lands began increasing by the mid-to-late 1990s as fire activity increased, both rates and mean patch size of PFH rapidly declined during the same period and remained low through the end of the study period. These declines coincided with increased litigation over federal forest management practices (Miner et al 2014) and a number of policies designed to protect threatened species and old-growth forests throughout the region (table 1). The 1993 Northwest Forest Plan designated 30% of federal land within the NSO range as Late Successional Reserves where salvage was limited to ‘conservative quantities’ within larger, higher severity disturbances with ‘stand-replacing patches >10 acres’ and ‘<40% canopy cover’ retention (USDA Forest Service, USDI Bureau of Land Management 1994). Other policies and recommendations like the Eastside Screens and CASPO Interim Guidelines were enacted elsewhere in the region, limiting harvests of large trees and promoting snag retention and old-growth habitat preservation, likely leading to incidental reductions in PFHs. The only policy to lift restrictions on PFHs, the temporary 1995 Salvage Rider, was met with public outcry and protest (Goldman and Boyles 1997), and coincided with negative step changes in federal harvest area and rate, rather than increased harvest (figure 3). Despite the direct and indirect impacts of federal policy on reducing PFHs, no outright ban was ever implemented, and much of the federal land base remains open to selective PFH.

On private lands, PFH rates and mean patch sizes were consistently higher than on federal lands, reflecting fewer policy restrictions and greater pressure to recoup economic losses from burned forests (Müller et al 2019). PFHs generally occurred earlier on private lands, where state-approved emergency harvest plans facilitate recovery of merchantable timber and replanting to re-establish forest cover (e.g. Z’berg-Nejedly Forest Practice Act 1973). Harvest rates on private lands fluctuated year-to-year but rarely exceeded 25% of the total area burned or 40% of the area burned at high severity, possibly reflecting the proportion of salvageable, merchantable timber available on industrial lands managed with short rotations. Shifting harvest practices towards larger patches and higher proportions of forests burned at lower severity may reflect efforts to optimize operations and maximize returns in the face of larger fires and increasing area burned, which has more than tripled since 1997. However, other local economic factors including log value, supply, and mill capacity can also influence post-disturbance harvest practices, with complex short- and long-term implications for log markets (Prestemon and Holmes 2004).

The divergent trends in PFH practices will likely have long-lasting and potentially profound effects on landscape patterns in recently burned landscapes that will further contribute to fragmentation and structural differences at ownership boundaries (figure 2; Wimberly and Ohmann 2004, Easterday et al 2018). On privately owned lands, larger PFH patches have the potential to reduce heterogeneity in stand conditions and coarsen age structure with subsequent replanting (i.e. larger patches of even-aged forest). On federal lands where harvest rates are lower and patch sizes are smaller, early seral structural conditions with snags and dead wood are more likely to persist. Maintaining these dead components of forest structure can provide wildlife habitat (Saab et al 2007, Vogeler et al 2016) and foster the development of future structural complexity (Donato et al 2012), but coarse woody debris is a fuel concern in future reburns (Monsanto and Agee 2008, Lydersen et al 2019, Kennedy et al 2024).

The ability to accurately map PFHs at broad spatiotemporal scales opens the possibility for studying long-term ecological and economic impacts that can inform management decisions and policies in burned forests. Decades of wall-to-wall PFH data combined with existing plot networks and remotely sensed biophysical datasets can shed light on how PFH affects early seral habitat, carbon sequestration, watershed processes, and forest regeneration at unprecedented scales, allowing for more informed management decisions and mitigation strategies by both private and public landowners. Understanding when and where harvests occur can also refine burn severity mapping techniques by separating post-fire management from direct fire effects, reducing uncertainty in the extent, patterns, and drivers of both high severity fire and fire refugia. Finally, harvest maps that are both regionally comprehensive and locally explicit at the scale of individual harvests can help to disentangle the complex relationship between salvage logging and timber markets at a range of spatial scales, informing economic decisions in a future of larger fires where salvaged timber plays an increasing role in the timber supply.

While we were able to effectively map PFHs across 66% of the forested area burned in our study region, our methodology could not be applied in all cases due to limited data quality and availability. In order to detect harvests that occurred shortly after fire containment, we were forced to exclude fires where cloud- and snow-free imagery could not be acquired shortly following fire. Even so, the omission of harvests that occurred within days or weeks of fire containment is possible due to the bi-weekly acquisition schedule of Landsat imagery, and may explain why some of our misclassified plots showed little spectral change despite evidence of harvest. Adapting our methodology to include newer multispectral data sources like Sentinel-2, which began collection in 2017, could reduce this omission risk in recent fires by increasing the cadence of image acquisition. Developing an effective mapping solution utilizing synthetic aperture radar, which is largely unaffected by cloud cover, could similarly capture more harvests immediately following fire, although past attempts have shown limited success (Zhao et al 2022).

5. Conclusions

Diverging harvest practices between federal and private lands have profound implications for the current and future role of these burned landscaped, altering their utility as timber sources, wildlife habitats, and carbon sinks. On federal lands, shifts towards smaller harvest patches in forests burned at higher severity reflect changing policies and public pressure. In contrast, shifts in private harvest practices towards larger harvest patches with higher proportions of forests burned at lower severity likely represent efforts to offset increasing costs and maximize efficiency. Understanding these impacts will require accurate and scalable methods for identifying and monitoring harvests and subsequent ecosystem effects. While challenges remain in generalizing our methodology for all cases, we were able to achieve high accuracy mapping of PFHs at unprecedented spatiotemporal scales, opening the door for broader ecological analyses and shedding new light on the effects of forest policy on the complexity of post-fire landscapes.

Supplementary Material

S1
S2

Acknowledgments

Research was supported in part by an appointment to the United States Forest Service Research Participation Program administered by the Oak Ridge Institute for Science and Education (ORISE) through an interagency agreement between the US Department of Energy (DOE) and US Department of Agriculture (USDA). ORISE is managed by Oak Ridge Associated Universities (ORAU) under DOE Contract Number 20IA11261952084. Funding was provided by the USDA Forest Service Washington Office through the Right Seed, Right Place project. We are deeply grateful to Mindy Crandall, Robyn Darbyshire, Liz Dent, Jessica Halofsky, Norm Johnson, Joe Sherlock, Joe Wagenbrenner, and the anonymous reviewers for their feedback and reviews that improved the clarity of this manuscript. The views expressed in this article are those of the authors and do not necessarily reflect the views or policies of the US EPA.

Footnotes

The authors declare no competing interests.

Data availability statement

The data that support the findings of this study are openly available at the following URL/DOI: https://doi.org/10.5281/zenodo.14057799.

References

  1. Box GEP and Tiao GC 1975. Intervention analysis with applications to economic and environmental problems J. Am. Stat. Assoc. 70 70–79 [Google Scholar]
  2. Brown CF et al. 2022. Dynamic World, Near real-time global 10 m land use land cover mapping Sci. Data 9 251 [Google Scholar]
  3. Clark DA, Anthony RG and Andrews LS 2013. Relationship between wildfire, salvage logging, and occupancy of nesting territories by northern spotted owls J. Wildl. Manage. 77 672–88 [Google Scholar]
  4. Croft M, Johnson L and Prior-Magee JS (U.S. Geological Survey, Gap Analysis Project) 2018. Protected areas database of the United States (PAD-US) 2.0 ( 10.5066/P955KPLE) [DOI] [Google Scholar]
  5. Donato DC, Campbell JL, Franklin JF and Palmer M 2012. Multiple successional pathways and precocity in forest development: can some forests be born complex? J. Veg. Sci. 23 576–84 [Google Scholar]
  6. Dorn TCE 1996. Logging without laws: the 1995 salvage logging rider radically changes policy and the rule of law in the forests Tulane Environ. Law J. 9 447–82 [Google Scholar]
  7. Easterday K, McIntyre P and Kelly M 2018. Land ownership and 20th century changes to forest structure in California For. Ecol. Manage. 422 137–46 [Google Scholar]
  8. Eidenshink J, Schwind B, Brewer K, Zhu Z-L, Quayle B and Howard S 2007. A project for monitoring trends in burn severity Fire Ecol. 3 3–21 [Google Scholar]
  9. Fuller WA 1996. Introduction to Statistical Time Series 2nd edn (Wiley; ) [Google Scholar]
  10. Goldman PA and Boyles KL 1997. Forsaking the rule of law: the 1995 logging without laws rider and its legacy Environ. Law 27 1035 [Google Scholar]
  11. Gorelick N, Hancher M, Dixon M, Ilyushchenko S, Thau D and Moore R 2017. Google Earth Engine: planetary-scale geospatial analysis for everyone Remote Sens. Environ. 202 18–27 [Google Scholar]
  12. Green DS, Martin ME, Powell RA, McGregor EL, Gabriel MW, Pilgrim KL, Schwartz MK and Matthews SM 2022. Mixed-severity wildfire and salvage logging affect the populations of a forest-dependent carnivoran and a competitor Ecosphere 13 e03877 [Google Scholar]
  13. Hansen MC et al. 2013. High-resolution global maps of 21st-century forest cover change Science 342 850–3 [DOI] [PubMed] [Google Scholar]
  14. Higuera PE and Abatzoglou JT 2021. Record-setting climate enabled the extraordinary 2020 fire season in the western United States Glob. Change Biol. 27 1–2 [DOI] [PubMed] [Google Scholar]
  15. Johnson KN, Franklin JF, Reeves GH, Johnson DL and Brown SJM 2023. The Making of the Northwest Forest Plan: The Wild Science of Saving Old Growth Ecosystems (Oregon State University Press; ) [Google Scholar]
  16. Johnston JD, Greenler SM, Reilly MJ, Webb MR, Merschel AG, Johnson KN and Franklin JF 2021. Conservation of dry forest old growth in Eastern Oregon J. For. 119 647–59 [Google Scholar]
  17. Kennedy MC, Johnson MC and Harrison SC 2024. Model analysis of post-fire management and potential reburn fire behavior J. Environ. Manage. 351 119664 [DOI] [PubMed] [Google Scholar]
  18. Lee EH, Wickham C, Beedlow PA, Waschmann RS and Tingey DT 2017. A likelihood-based time series modeling approach for application in dendrochronology to examine the growth-climate relations and forest disturbance history Dendrochronologia 45 132–44 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Leverkus AB, Buma B, Wagenbrenner J, Burton PJ, Lingua E, Marzano R and Thorn S 2021. Tamm review: does salvage logging mitigate subsequent forest disturbances? For. Ecol. Manage. 481 118721 [Google Scholar]
  20. Li K and Xu E 2023. Differentiating effects of salvage logging and recovery patterns on post-fire boreal forests in Northeast China using a modified forest disturbance index GISci. Remote Sens. 60 2188674 [Google Scholar]
  21. Lindenmayer DB, Burton PJ and Franklin JF 2012. Salvage Logging and Its Ecological Consequences (Island Press; ) [Google Scholar]
  22. Lydersen JM, Collins BM, Coppoletta M, Jaffe MR, Northrop H and Stephens SL 2019. Fuel dynamics and reburn severity following high-severity fire in a Sierra Nevada, USA, mixed-conifer forest Fire Ecol. 15 43 [Google Scholar]
  23. Melgani F, Moser G and Serpico SB 2002. Unsupervised change-detection methods for remote-sensing images Opt. Eng. 41 3288 [Google Scholar]
  24. Miner AMA, Malmsheimer RW and Keele DM 2014. Twenty years of forest service land management litigation J. For. 112 32–40 [Google Scholar]
  25. Monsanto PG and Agee JK 2008. Long-term post-wildfire dynamics of coarse woody debris after salvage logging and implications for soil heating in dry forests of the eastern Cascades, Washington For. Ecol. Manage. 255 3952–61 [Google Scholar]
  26. Müller J, Noss RF, Thorn S, Bässler C, Leverkus AB and Lindenmayer D 2019. Increasing disturbance demands new policies to conserve intact forest Conserv. Lett. 12 e12449 [Google Scholar]
  27. Peterson DL, Agee JK, Aplet GH, Dykstra DP, Graham RT, Lehmkuhl JF, Pilliod DS, Potts DF, Powers RF and Stuart JD 2009. Effects of Timber Harvest following Wildfire in Western North America No. PNW-GTR-776 (U.S. Department of Agriculture, Forest Service, Pacific Northwest Research Station; ) ( 10.2737/PNW-GTR-776) [DOI] [Google Scholar]
  28. Peterson DW, Dodson EK and Harrod RJ 2015. Post-fire logging reduces surface woody fuels up to four decades following wildfire For. Ecol. Manage. 338 84–91 [Google Scholar]
  29. Pinheiro Jand Bates D (R Core Team) 2023. nlme: linear and nonlinear mixed effects models [Google Scholar]
  30. Prestemon JP and Holmes TP 2004. Market dynamics and optimal timber salvage after a natural catastrophe For. Sci. 50 495–511 [Google Scholar]
  31. R Core Team 2023. R: A Language and Environment for Statistical Computing (R Foundation for Statistical Computing; ) [Google Scholar]
  32. Reeves GH, Bisson PA, Rieman BE and Benda LE 2006. Postfire logging in riparian areas Conserv. Biol. 20 994–1004 [DOI] [PubMed] [Google Scholar]
  33. Reilly MJ, Dunn CJ, Meigs GW, Spies TA, Kennedy RE, Bailey JD and Briggs K 2017. Contemporary patterns of fire extent and severity in forests of the Pacific Northwest, USA (1985–2010) Ecosphere 8 e01695 [Google Scholar]
  34. Ryan RL and Hamin E 2008. Wildfires, communities, and agencies: stakeholders’ perceptions of postfire forest restoration and rehabilitation J. For. 106 370–9 [Google Scholar]
  35. Saab VA, Russell RE and Dudley JG 2007. Nest densities of cavity-nesting birds in relation to postfire salvage logging and time since wildfire Condor 109 97–108 [Google Scholar]
  36. Schroeder TA, Wulder MA, Healey SP and Moisen GG 2012. Detecting post-fire salvage logging from Landsat change maps and national fire survey data Remote Sens. Environ. 122 166–74 [Google Scholar]
  37. Sessions J, Bettinger P, Buckman R, Newton M and Hamann J 2004. Hastening the return of complex forests following fire: the consequences of delay J. For. 102 38–45 [Google Scholar]
  38. USDA Forest Service, USDI Bureau of Land Management 1994. Record of decision for amendments to forest service and bureau of land management planning documents within the range of the Northern Spotted Owl and standards and guidelines for management of habitat for late-successional and old-growth forest related species within the range of the Northern Spotted Owl (USDA Forest Service and USDI Bureau of Land Management; ) [Google Scholar]
  39. USDA Forest Service. USFS Landscape Change Monitoring System 2023 [Google Scholar]
  40. Verner J, McKelvey KS, Noon BR, Gutierrez RJ, Gould GIJ and Beck TW, 1992. The California spotted owl: a technical assessment of its current status Gen Tech Rep PSW-GTR-133 Albany CA Pac. Southwest Res. Stn. For. Serv. US Dep. Agric. 285 P 133 [Google Scholar]
  41. Vogeler JC, Yang Z and Cohen WB 2016. Mapping suitable Lewis’s woodpecker nesting habitat in a post-fire landscape Northwest Sci. 90 421–32 [Google Scholar]
  42. Wagenbrenner JW, MacDonald LH, Coats RN, Robichaud PR and Brown RE 2015. Effects of post-fire salvage logging and a skid trail treatment on ground cover, soils, and sediment production in the interior western United States For. Ecol. Manage. 335 176–93 [Google Scholar]
  43. Wimberly MC and Ohmann JL 2004. A multi-scale assessment of human and environmental constraints on forest land cover change on the Oregon (USA) coast range Landscape Ecol. 19 631–46 [Google Scholar]
  44. Wulder MA et al. 2019. Current status of Landsat program, science, and applications Remote Sens. Environ. 225 127–47 [Google Scholar]
  45. Z’berg-Nejedly Forest Practice Act. 1973 [Google Scholar]
  46. Zhao F, Sun R, Zhong L, Meng R, Huang C, Zeng X, Wang M, Li Y and Wang Z 2022. Monthly mapping of forest harvesting using dense time series Sentinel-1 SAR imagery and deep learning Remote Sens. Environ. 269 112822 [Google Scholar]
  47. Zuspan A, Reilly MJ and Lee EH 2024. Mapping and analyzing long-term patterns of post-fire harvest in the Pacific West, U.S.A Zenodo ( 10.5281/zenodo.14057800) [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

S1
NIHMS2040047-supplement-S1.kml (1.2MB, kml)
S2
NIHMS2040047-supplement-S2.docx (170.9KB, docx)

Data Availability Statement

The data that support the findings of this study are openly available at the following URL/DOI: https://doi.org/10.5281/zenodo.14057799.

RESOURCES