Bootleg Fire: evaluating the role of fuel treatments in mitigating burn severity near Sycan Marsh, Oregon

Abstract

Wildfires in western US dry forest ecosystems have increased in frequency and severity during recent decades due to a warming climate and a century of fire suppression. The 2021 Bootleg Fire in south-central Oregon provided a great opportunity to evaluate fuel treatment effectiveness near the Sycan Marsh Preserve, where pre-fire LiDAR data were also available. Within this study area, The Nature Conservancy (TNC), The Klamath Tribes (TKT), and the USDA Forest Service conducted mechanical thinning (Tx), prescribed fire (Rx), and both treatments combined (TxRx), through a collaborative partnership, over 16 years preceding the wildfire. We conducted a burn severity assessment one year after the Bootleg Fire, accounting for the local variability of top drivers, fuel treatments, and firefighting operations. We modeled the influence of burn severity drivers using Random Forests and examined their influence at a global and local scale using SHapley Additive exPlanations (SHAP) analysis.In units treated with Rx and TxRx, the percentage of area burned at low severity was 86% and 83%, respectively; for both treatments, moderate and high burn severity accounted for less than 15% and 5% of the area burned. In contrast, in units treated with Tx, the percentage of area burned at low, moderate, and high severity was 30%, 45%, and 25%, respectively. Finally, in untreated forests, the percentage of area burned at low, moderate, and high severity was 31%, 27%, and 42%, respectively. Top predictors included Rx, evaporative stress Index, 4km-window topographic position index, and canopy cover, in descending order of importance. Predictors related to firefighting operations and Tx were ranked third-to-last and last, respectively. Where fire suppression occurred, we observed a decreasing trend in predicted burn severity within units treated with Rx and more variable effects within Tx units. The local effect of each predictor varied spatially. Our results and the testimonials of local managers indicate that fuel treatments involving prescribed fire were the most effective at mitigating burn severity and facilitated firefighting operations. Top drivers that initially seemed to reflect water-stress conditions and topographic positioning turned out to be proxies for fuel structure and accumulation, further underscoring the key role of fuel characteristics in affecting burn severity. These methods allowed us to interpret relationships that would have otherwise been overlooked or potentially misunderstood in an exclusively global analysis, rather than one that considers both global and local perspectives. This study contributes to the body of knowledge by providing a reproducible framework to explain the effect of contagious processes like fire at both global and local scales. Additionally, it provides valuable lessons to guide and improve fire and fuel management practices across multi-ownership landscapes.