Limited tree mortality in unburned areas linked to bark beetle spillover from wildfires

Abstract

Increased fire activity in the western United States since 2000 has produced an abundance of fire-injured trees at risk to lethal attack by bark beetles. Large populations of bark beetles reproducing in fire-injured trees may disperse (or spillover) from inside the fire perimeter to adjacent, unburned forests, potentially causing extensive tree mortality. In the western United States and Canada, fire-injured Douglas-fir (DF; Pseudotsuga menziesii) are frequently colonized by Douglas-fir beetle (DFB; Dendroctonus pseudotsugae), prompting concern among land managers about elevated risk of spillover. We investigated spatiotemporal patterns of DF tree mortality from DFB in unburned areas surrounding 61 wildfires (2000-2017) with a high likelihood for spillover in the Northern Rocky Mountains, USA. We developed a multiple-scale analytical framework to examine tree mortality potentially associated with spillover following fire. Synchronous fluctuation in the amount of DF mortality within and beyond the flight distance of DFB in the region and surrounding individual fires (0-10 km) suggested that DFB activity primarily responded to a broader scale process, such as drought, rather than proximity to burned trees. Using shorter and longer range dispersal scenarios, we estimated that at <0.25 km from the fire perimeter, the dominant source of DFBs transitioned from burned to unburned sources due to the closer proximity of DFBs from unburned sources. Some fires (8%-15%; range of fires from sensitivity analysis) did exhibit evidence of DFB spillover, but spillover occurred <1 km from fires (based on our criteria) and DF tree mortality associated with spillover was 0.2%-0.3% of total DF damage area during the study period. Spillover was not associated with climate conditions that increase host tree stress, rather it was associated with greater DF mortality from DFB in the prior year in the same area (i.e., poorly linked to spillover). Site-specific monitoring of post-fire DFB populations in susceptible, unburned DF forests adjacent to fires by land managers may be necessary to determine the risk of DFB emigrating from burned areas. Our findings inform post-fire planning and the ecological implications of disturbance interactions that occurred in the early 21st century during a period of amplified wildfire and DFB activity.

Keywords: Dendroctonus pseudotsugae; Northern Rocky Mountains; climate change; disturbance interactions; fire ecology; insect susceptibility models.

FIGURE 1

Map of study area with locations of field sampling areas (5 plots per triangle), areas burned in the 61 study fires from 2000 to 2017 (orange; MTBS, 2023), Douglas‐fir (DF) mortality from Douglas‐fir beetle (DFB) from 2000 to 2021 (purple; FHP, 2023), Random Forest‐predicted DF forest susceptible to DFB (green) (see Spatial model of DF forest susceptibility to DFB in Methods ), DF forest distribution (gray; basal area >1 m² ha⁻¹; Krist et al., 2014), ecoregions (yellow lines; modified from US EPA, 2010), and the extent of the study area in the Northern Rocky Mountains (thick black line). The inset locates the study area within the western United States. The study fires burned 10,751.7 km² of susceptible DF forest (all severities).

FIGURE 2

Percent of Douglas‐fir (DF) forest susceptible to Douglas‐fir beetle (DFB) that burned from 2000 to 2017 (orange) in 61 study fires (fires after 2017 excluded from analysis). Additionally, the DF damage area (DA, in percentage) near (<1.0 km) the study fires (gray; DA < 1 km) and further (>1.0 km) from the study fires (purple; DA > 1 km) from 2002 to 2021 as a percent of susceptible DF forest. DA < 1 km is for the period 2–5 years post‐fire to correspond with the likely timing of DFB spread to unburned areas outside of the fire perimeters. One year was subtracted from the acquisition year of DFB damage area to account for the 1‐year lag between DFB attack and foliage discoloration.

FIGURE 3

(A) Average Douglas‐fir (DF) damage area (DA) from Douglas‐fir beetle (DFB) from 1 to 6 years post‐fire for 0.5 km width buffer areas from 0 to 5 km from the fire perimeter in 61 fires in the Northern Rocky Mountains, USA, illustrating the amount and variability of DF DA surrounding fire perimeters. (B) Average DF DA with distance from fire (standardized with z‐scores by fire) to compare DA among buffer areas within year (i.e., spillover analysis; see Appendix S1: Figure S4 for example of buffer areas surrounding fire). The expected pattern in DA resulting from DFB spillover from fire‐injured trees to unburned areas (i.e., DFB spillover) would be greater DA immediately surrounding the fire perimeter (e.g., 0–0.5 km or >0.5–1.0 km buffer areas) relative to further away (e.g., >1 km). (C) Average of the observed DF DA (in percentage) minus the expected DA (in percentage) from 1 to 6 years post‐fire. Observed DA was subtracted from DA in each of 15 random fire locations and then averaged by buffer and year. The vertical lines are SEs, and the dots are values for individual fires.

FIGURE 4

(A) Douglas‐fir beetle (DFB) population pressure from burned and unburned sources on areas (dots) of Douglas‐fir forest infested by DFB (i.e., aerial survey polygons) with distance from the fire for an example fire (Goat Creek), illustrating where the dominant source of DFB pressure switches from burned to unburned (~0.45 km) under a longer range DFB dispersal scenario (see Appendix S1: Figure S5 for a year‐by‐year example). Lines are generalized additive models, and gray shading is the 95% CI. (B) Percent of fires by distance from fire at which DFB pressure switched from burned to unburned sources under shorter and longer range DFB dispersal scenarios (see Methods ).

FIGURE 5

(A) Count of buffer area‐years (y‐axis) that met our criteria for Douglas‐fir beetle (DFB) spillover, sources of DFBs (burned and/or unburned; see panel B), and the count of unique fires with DFB spillover with time since fire and distance from fire (colors) based on the spatiotemporal progression of Douglas‐fir (DF) damage area from DFB. Individual fires may have multiple buffer‐years with DFB spillover. The analysis extracted DF damage area in 0.5‐km width buffer areas from 0 to 10 km from the fire perimeter by year (i.e., buffer area‐years; see Appendix S1: Figure S12 for 1.0‐km width buffer areas). (B) DFB population pressure from burned and unburned sources on individual aerial insect survey polygons for fires with evidence for DFB spillover only (see Appendix S1: Figure S11 for individual fires), and the results from pairwise tests of DFB source within buffer areas from a linear mixed model (ns, not significant; ***p < 0.001). In the boxplots, the thick horizontal line is the median, the box represents the interquartile range (IQR, 25th–75th percentiles) of the distribution, the whiskers extend no further than ±1.5 times the IQR, and the dots are outliers.

FIGURE 6

Count and percent of field plots with agents potentially confounding the attribution of Douglas‐fir (DF) mortality to spillover of Douglas‐fir beetles (DFB) from fire‐injured trees. Douglas‐fir trees affected by root disease (RD), DF trees blowdown (BD) after the fire year (2016 or 2017), and DFB infestation <5 years prior to the fire may have supported localized populations of DFB in the stand prior to or immediately after fire. Fire was included to account for undetected small fires or spot fires excluded from the fire perimeter polygon. The “none” category indicates that no potential confounding factors were identifed at the site, and DFB is assumed to have arrived from a neighboring stand, either from burned areas inside or from unburned areas outside of the fire perimeter.