Fire suppression makes wildfires more severe and accentuates impacts of climate change and fuel accumulation

Wildfires are becoming more destructive worldwide, driven by climate change, fuel accumulation, vegetation change, and ignition patterns. Humans further shape fire regimes through initial attack and suppression decisions that act as a filter on which fires are allowed to burn. This filter removes many low-intensity events and lets the most extreme fires escape, shaping when, where, and how fires burn. Suppression also contributes to the well-known fire suppression paradox, where extinguishing fire today increases fuels and future fire intensity. The authors introduce an additional, direct consequence termed the suppression bias: suppression preferentially removes some fire types (often low- to moderate-intensity surface fires) more than others (often high-intensity crown fires), skewing the realized fire regime toward more severe burning under extreme weather. The study aims to isolate and quantify how suppression bias affects fire behavior, area burned, severity, and diversity of fire effects, and how its magnitude compares with the effects of climate change (fuel aridity) and fuel accumulation.

Prior work has documented increasing wildfire risk with climate-driven declines in fuel moisture and highlighted the wildfire paradox whereby suppression elevates future fuels and suppressibility. Studies have tangentially referenced a bias introduced by suppression but have not quantified its emergent effects due to the lack of control landscapes without suppression, limited data on initial attack outcomes, and the confounding of suppression with terrain, access, and resource availability. Empirical research on suppression effectiveness and initial attack exists, as do broader syntheses on changing fire regimes and severity trends, but isolating suppression’s direct bias has remained challenging. This study addresses that gap via simulation to disentangle suppression bias from climate change and fuel accumulation.

The authors developed a generalizable simulation framework to model individual wildfires, focusing on fundamental fire behavior and fuel moisture dynamics rather than site-specific predictions. Fires were simulated over a hypothetical 150-day season with randomly assigned ignition days. Ignitions could smolder up to three days and spread when fuel moisture fell below moisture of extinction. Daily fire growth used an elliptical growth model (Huygens’ principle), with heading spread rates informed by a functional fire spread model and converted to elliptical expansion across azimuths. Fire intensity informed flame length and was linked to the Composite Burn Index (CBI) as a proxy for ecological severity. Suppression was modeled in two stages: (1) initial attack and (2) subsequent suppression for escaped fires. The probability of initial attack escape was a function of fire intensity and size at time of engagement, based on a relationship from Hirsch et al. Regressive suppression scenarios differed by time-to-engagement (proxy for aggressiveness): Moderate = 4 h, High = 2 h, Maximum = 1 h. A Progressive suppression scenario assumed no initial attack but stronger subsequent suppression focused on higher-intensity fire. A No-suppression control was also included. For escaped fires, suppression functions related fire intensity to the proportion of fire suppressed, with Maximum suppression representing the upper bound of on-ground effectiveness and trivial impossibility above certain intensities. Daily distance burned under suppression equaled unsuppressed spread (assuming active spread for half the day) multiplied by the remaining proportion after suppression. Fireline segments were extinguished if daily spread distance was <5 m, and fires were declared out when all azimuths were extinguished or after day 150. Inputs included daily windspeed (Weibull-distributed, single direction per fire), temporally autocorrelated seasonal cycles for live and dead fuel moistures (1-, 10-, 100-h fuels; live woody and herbaceous), and a fixed terrain slope (40%) with continuous, uniform fuels (no fuel breaks). The framework varied mean seasonal fuel aridity via vapor pressure deficit (VPD) and 100-h fuel loading across ranges intended to represent approximately 240 years of modeled change under an RCP 8.5-like trajectory (e.g., increasing VPD) and historical estimates of dead wood carbon accumulation (100-h fuel load increase ≈ 0.036 Mg ha⁻¹ yr⁻¹). Simulations varied either fuel aridity (holding fuel load constant at 11.23 Mg ha⁻¹) or fuel loading (holding mean seasonal VPD constant at 1.17 kPa). Each replicate involved 1000 ignitions and multiple levels of aridity or loading, across all suppression scenarios, with 40 replicate runs. Outcomes computed for each fire included proportion burned at high severity (CBI ≥ 2.25), mean CBI severity, daily and total fire size, diversity of fire severity (mean absolute deviation of CBI), and Lorenz curves of area burned. The authors also translated differences in mean severity between suppression scenarios and a no-suppression baseline into equivalent years of change in fuel aridity or fuel loading by dividing by estimated annual change rates.

- Regressive suppression (Moderate, High, Maximum) increased wildfire ecological impacts across simulated ranges of fuel aridity and fuel loading. Under Maximum suppression, the proportion of area burned at high severity increased, and mean fire severity (CBI) increased by an average of 0.21 units relative to no suppression—equivalent to roughly 102 years of additional fuel accumulation without suppression (≈ +4.7 Mg ha⁻¹ in 100-h surface fuels). - Progressive suppression generally reduced the proportion of area burned at high severity across most fuel aridity and loading levels and lowered mean severity by ≈0.03–0.04 CBI units versus no suppression. - Area burned became more sensitive to increasing fuel aridity and fuel loading under regressive suppression. Across increasing fuel aridity, area burned under Maximum suppression increased by about 50% per year, compared to about 1.8% per year under Progressive suppression. Under increasing fuel loading, area burned increased by about 3.7% per year (Maximum) versus about 0.7% per year (No suppression). Progressive suppression had the lowest sensitivity to both drivers, with indicative doubling times of ~44 years (aridity increases) and ~133 years (fuel accumulation), slower than other strategies including no suppression. - Regressive suppression decreased the diversity of fire effects (lower variability in CBI) at all aridity and loading levels, leading to higher average severity and less heterogeneity in burn severity patterns. Maximum suppression showed minimal variability in burn severity. - Progressive suppression produced fire behavior and effects similar to unsuppressed fires burning under less-conducive conditions, effectively offsetting the impacts of climate change or fuel accumulation by approximately one to nearly two decades in the simulations. - Although regressive suppression can reduce absolute burned area and keep many fires small, it accentuates trends toward increasing severity and increases the relative rate of area burned over time under worsening climate and fuel conditions, independent of fuel accumulation and climate change effects themselves.

The results demonstrate that suppression does not merely reduce fire extent; by preferentially eliminating low- and moderate-intensity fires, regressive suppression directly biases the realized fire regime toward the most severe, high-intensity events. This suppression bias amplifies the effects of climate change and fuel accumulation, making area burned and high-severity fire more sensitive to increasing aridity and fuel loads, and reducing pyrodiversity (diversity of fire effects). Consequently, ecosystems, species, and people are disproportionately exposed to extreme fire behavior, which has ecological ramifications (e.g., selection for traits conferring resistance or resilience to high-intensity fire, reduced regeneration opportunities, and increased probability of state shifts) and social implications (public perception of wildfire shaped by extreme events, decreased support for fire use and active fire management). Conversely, progressive suppression—allowing lower-intensity fire to spread while more strongly targeting higher-intensity fire—maintains lower severity and greater diversity of fire effects and slows the rate at which climate change and fuel accumulation translate into increased area burned and severity. This approach can effectively "buy time" for ecosystems and societies to adapt. However, implementing progressive suppression faces cultural, policy, safety, and practical constraints, including narrowing operational space as conditions become more fire-conducive and increased exposure to frequent, low-level smoke. Overall, the findings indicate that integrating the suppression bias into assessments of fire regimes is essential. Management that enables fires to burn under moderate conditions, combined with intentional fire (prescribed and cultural burning), can reduce severity, increase heterogeneity, and moderate the pace of change—even as climate warms and fuels accumulate.

This study identifies and quantifies the fire suppression bias as a fundamental, direct driver of contemporary fire regimes, independent of the indirect effects of fuel accumulation. Regressive suppression shifts fire activity toward more severe, less diverse outcomes and heightens sensitivity to climate and fuel changes, while progressive suppression reduces severity, preserves diversity of fire effects, and slows increases in burned area. Managing wildfires to safely burn under low and moderate conditions, coupled with intentional fire, is a critical component of coexisting with wildfire and can be as impactful as mitigating climate change, reducing ignitions, or modifying fuels. Future research should empirically test the predicted effects of suppression bias across diverse ecosystems and social contexts and refine models to incorporate spatial heterogeneity, dynamic resource allocation, and coupled feedbacks with fuel accumulation.

The simulation is designed for generality and does not predict fire-scale behavior for specific landscapes. It omits spatial heterogeneity within fires (e.g., topography, within-fire variation in wind and fuels), does not include interannual feedbacks from prior fires (i.e., it does not model the fuel-accumulation feedback of the suppression paradox), and does not explicitly model dynamic suppression resource allocation. These choices likely make the estimates of regressive suppression impacts conservative, as real landscapes under regressive suppression would accumulate fuels faster. Practical implementation of progressive suppression is also constrained by safety, policy, public acceptance, and shrinking operational space under more fire-conducive conditions.