Forest fires and climate-induced tree range shifts in the western US

The study addresses how wildfire influences the rate and direction of climate-driven tree range shifts in the western United States. Plant biogeography is largely determined by climate, and numerous studies document recent shifts in plant distributions, particularly upslope movements, though many species fail to keep pace with the velocity of climate change. Range edge dynamics are influenced by environmental changes, species traits, landscape characteristics, and biotic interactions, notably competition. Theory suggests that competition can constrain leading-edge expansion, yet empirical evidence at broad scales is limited. Wildfire, by reducing vegetation cover and competitor density, offers a natural experiment to test whether disturbance can facilitate climate-related range expansions. While fire can drive both successional recovery and state shifts, its overall effect on regeneration and establishment may increase the availability of niche space for colonization. The authors hypothesize that recent wildfire facilitates greater climatic displacement of seedling-only occurrences relative to tree-plus-seedling occurrences, consistent with accelerated range expansion under climate change. Using FIA plots across two EPA Level I ecoregions, they compare climatic niches of mature versus juvenile occurrences to infer potential range shifts and test whether these displacements are greater in burned versus unburned plots.

Prior work shows global and regional plant range shifts associated with warming, with average upward movements of approximately 1.5 m/year in elevation for plants. Many species have not shifted rapidly enough to track contemporary climate velocities. Biotic interactions, especially competition, are theorized to constrain range expansion at leading edges, supported by modeling but with limited broad-scale empirical confirmation; some studies suggest stronger competitive effects at trailing edges in montane systems. Disturbance regimes, particularly wildfire, are predicted to facilitate range expansion by reducing competitor abundance and opening niche space. Observational studies in western North America have documented fire-driven range contractions and reduced post-fire conifer regeneration under warmer, drier conditions, highlighting complex fire–climate–regeneration interactions. Methodological precedents include using differences between seedling and mature tree climatic niches (e.g., Zhu et al.; Dobrowski et al.) and metrics such as Schoener’s D to test niche equivalency and centroid distances in multivariate climate space to quantify displacement. Climate variables critical to plant distributions—temperatures and seasonal precipitation—have shifted over recent decades in the region, with warming generally exceeding changes in precipitation, providing a contemporary context for expected range responses.

Study area and data: The authors analyzed 74,069 USDA Forest Inventory and Analysis (FIA) plots within the Northwestern Forested Mountains and Marine West Coast Forest ecoregions of the continental U.S. (surveys from 1999–2019). FIA plots record tree species, seedlings, tree age/diameter, and recent fire disturbance (fires within 5 years prior to survey that affected ≥25% of trees over >0.4 ha). Plots are subdivided into subplots and microplots; trees with dbh >12.7 cm are tallied in subplots, seedlings in microplots. Plot grouping and species inclusion: For each species, plots were classified into four categories: (1) unburned seedling-only, (2) unburned tree-plus-seedling, (3) burned seedling-only, (4) burned tree-plus-seedling. Species with <5 plots in any category were excluded, removing 84 species. From 26 sufficiently abundant species, further vetting excluded 14 species for inconsistent displacement directions across life stages (seedlings vs saplings vs large trees) and 4 species for inconsistent directions between burned and unburned plots, leaving 8 focal species. Climate data and variables: 30-arc-second climate rasters (1981–2010 averages) from AdaptWest/ClimateNA v5.10 were extracted for plots. Starting from 9 candidate variables, collinearity was reduced (VIF < 10) to retain four: Mean Temperature of the Warmest Month (MTWM), Mean Temperature of the Coldest Month (MTCM), Mean Summer Precipitation (MSP), and Mean Winter Precipitation (MWP). SORD calculation: Seedling Only Range Displacement (SORD) was quantified as the climatic niche difference between seedling-only and tree-plus-seedling plot groups. Two metrics were used: (a) Schoener’s D to test niche equivalency via kernel-smoothed PC space of the two most informative principal components; significance was assessed via the niche equivalency test of Broennimann et al. (b) Euclidean centroid distance in 4D climate space (MTWM, MTCM, MSP, MWP) between seedling-only and tree-plus-seedling centroids, providing magnitude and direction. Directional consistency checks included: dot product (cosine) between normalized niche difference vectors of seedlings vs saplings/large trees (>0.5 threshold) and between burned and unburned SORD vectors (>0.5 threshold). Comparisons burned vs unburned: The primary comparison was the difference in centroid distances between burned and unburned samples (CD_b − CD_u). Bootstrapping with 20,000 replicates generated 90% and 95% confidence intervals. Principal component analysis was used for visualization. Component-wise analysis: For each climate variable, plots were aggregated across species that exhibited significant and directionally consistent unburned SORD for that variable. Stratified sampling equalized species contributions; 100 sampling iterations were averaged. Multiple linear regression evaluated differences in variable-specific climatic distances between burned and unburned groups. Robustness checks: Analyses were repeated with alternative climate-variable sets (VIF < 10), yielding consistent conclusions. Minimum presence thresholds of 5, 10, and 25 plots were tested; while affecting the number of species included, results remained qualitatively consistent that wildfire is associated with greater SORD for some species.

- Species set: Of 110 tree species in the study area, 26 met minimum sample sizes; after directional consistency vetting, 8 species remained for burned–unburned comparisons. - Unburned SORD existence: For all eight species, seedling-only climatic niches differed significantly from tree-plus-seedling niches in unburned plots. Schoener’s D ranged 0.64–0.85 and was significantly less than expected under niche equivalency (p < 0.05). Euclidean centroid distances exceeded 0.19 in 4D climate space and were significantly greater than 0 (Hotelling’s T^2, p < 0.05). - Burned vs unburned SORD magnitude: Centroid distance was significantly greater in burned than unburned plots for Pseudotsuga menziesii and Quercus chrysolepis (p < 0.05). For the remaining six species, differences were not significant. Averaged across all eight species, CD_b was 89% greater than CD_u. - Climatic directionality: Five species (Chrysolepis chrysophylla, Pinus ponderosa, Pseudotsuga menziesii, Quercus chrysolepis, Quercus kelloggii) had seedling-only plots with higher mean summer precipitation (MSP) and lower mean temperature of the warmest month (MTWM) than tree-plus-seedling plots; these differences were larger in burned plots. The MSP difference between seedling-only and tree-plus-seedling plots was on average 2.2× greater in burned than unburned plots (p < 0.05). Patterns for MTCM and MWP were less consistent. Subalpine species (Pinus contorta, Picea engelmannii, Pinus albicaulis) did not show the lower MTWM/higher MSP seedling trend. - Regional climate change context: Between 1961–1990 and 1981–2010, all four climate variables shifted significantly across the study area (two-sided t-tests p < 2.2e−16), with temperature increases larger in magnitude than precipitation decreases. - Approximate spatial displacement: Using a 6.5 °C/km lapse rate, non-subalpine MTWM SORD components correspond to +99 m altitudinal displacement in unburned plots and +190 m in burned plots; subalpine species correspond to −100 m and −110 m, respectively. Considering average slopes (unburned 20.2°, burned 23.6°; p < 0.05), lateral geographic displacement differences could be up to ~32% smaller than indicated by climatic niche displacement ratios. - Topography: Burned plots occurred on significantly steeper slopes (mean 23.6°) than unburned (mean 20.2°), potentially affecting climatic–spatial distance translation.

The findings indicate that wildfire can increase the magnitude of climatic displacement between seedling-only and tree-plus-seedling occurrences—interpreted as accelerated potential range shift—at least for some tree species. This supports the hypothesis that reduced competitor density following disturbance can relax biotic constraints and facilitate colonization where recent climate change has made conditions more suitable for advancing populations. Species-level responses were heterogeneous: two of eight species showed significantly greater SORD in burned plots, while others did not, reflecting differences in life-history traits, dispersal, fire adaptations, and environmental contexts. Component-wise results (higher MSP and lower MTWM in seedling-only plots for five species, amplified in burned areas) align with expectations that seedlings establish preferentially in cooler and wetter microclimates relative to existing stands under warming and drying trends, or that drought stress diminishes suitability for established dominants. Subalpine species exhibited distinct patterns, likely influenced by constrained establishment at upper-elevation margins and greater availability of suitable substrates downslope, cautioning against overinterpreting seedling–adult differences as climate-tracking in all cases. The study situates observed SORDs within documented regional climate changes and provides approximate elevational shift magnitudes consistent with other regions’ observations. Consideration of slope effects underscores complexities in mapping climatic to spatial distances but does not negate the central signal. Overall, the results add empirical support to theory that disturbance mediates climate-driven range dynamics and highlight that fire regimes and management could influence vegetation redistribution rates.

This work provides empirical evidence that recent wildfire can increase potential range shift rates of tree species responding to contemporary climate change, consistent with a mechanism of competition release following disturbance. By comparing climatic niches of seedling-only versus tree-plus-seedling occurrences across extensive FIA plots, the study shows greater seedling-only climatic displacement in burned plots for two species and an overall trend toward larger displacements. These insights underscore the role of fire and fire management in shaping the pace and pathways of vegetation redistribution under climate change. Future research should: (1) disentangle fire’s direct abiotic effects from biotic competition effects on establishment; (2) integrate finer-resolution fire severity, size, and frequency data; (3) link climatic displacement more explicitly to spatial and demographic processes; (4) examine species- and ecosystem-specific responses, especially in subalpine systems; and (5) incorporate genetic and adaptive dynamics over longer timescales.

- Inference of leading-edge from seedling-only plots: Some seedling-only occurrences may represent non-maturing populations that will not reach reproductive maturity; although species with inconsistent seedling–sapling–adult displacement directions were excluded, residual bias may remain. - Fire-related confounding: Species-specific post-fire adaptations (e.g., resprouting, serotiny) and successional dynamics could influence SORD independently of climate tracking. Species with inconsistent burned vs unburned displacement directions were excluded, but trait effects may persist. - Fire data constraints: FIA records only fires within 5 years prior to surveys and with substantial impact, omitting older/recent low-impact fires; misclassification likely reduces contrast between burned and unburned groups, biasing toward underestimation of fire effects. - Seedling/resprout identification: FIA does not distinguish seedlings from resprouts, potentially affecting composition, though resprouting alone would not typically yield seedling-only plots. - Sampling design: Seedlings are tallied in microplots only; some seedlings present in full plots may be missed, potentially reducing seedling-only counts. - Climate–space to geographic–space translation: Climatic distances are proxies for spatial displacement; topographic heterogeneity (e.g., steeper slopes in burned plots: 23.6° vs 20.2°) can alter the ratio of geographic to climatic distance, complicating direct spatial interpretations. - Incomplete fire severity/size data: Lack of granularity on fire characteristics limits mechanistic attribution and cross-fire comparisons. - Species coverage: Only eight species passed stringent vetting, limiting generality across the full tree flora of the region.