TY - JOUR
T1 - Drivers of fire severity shift as landscapes transition to an active fire regime, Klamath Mountains, USA
AU - Taylor, Alan H.
AU - Harris, Lucas B.
AU - Drury, Stacy A.
N1 - Funding Information:
Support for this research was provided by the USDA Forest Service Pacific Southwest Research Station (14‐JV‐11272167‐037, 17‐JV‐11272167‐048) and The Pennsylvania State University. We thank Carl N. Skinner for review of an earlier draft of this paper and Clint Isbell, Eric Knapp, and Frank Lake for discussion on recent fire patterns in the Klamath Mountains.
Publisher Copyright:
© 2021 The Authors.
PY - 2021/9
Y1 - 2021/9
N2 - Fire severity patterns are driven by interactions between fire, vegetation, and terrain, and they generate legacy effects that influence future fire severity. A century of fire exclusion and fuel buildup has eroded legacy effects, and contemporary fire severity patterns may diverge from historical patterns. In recent decades, area burned and area burned at high severity have increased and landscapes are transitioning back to an active fire regime where disturbance legacies will again play a strong role in determining fire severity. Understanding the drivers of fire severity is crucial for anticipating future fire severity patterns as active fire regimes are reestablished. We identified drivers of fire severity in the Klamath Mountains, a landscape with an active fire regime, using two machine learning statistical models: one model for non-reburns (n = 92) and one model for reburns (n = 61). Both models predicted low better than moderate or high-severity fire. Fire severity drivers contrasted sharply between non-reburns and reburns. Fire weather and fuels were dominant controls in non-reburns, while previous burn severity, fuel characteristics, and time since last fire were drivers for reburns. In reburns, areas initially burned at low (high) severity burned the same way again. This tendency was sufficiently strong that reburn fire severity could be predicted equally well with only severity of the previous fire in the model. Thus, reburn fire severity is more predictable than severity in non-reburns that are driven by the stochastic influences of fire weather. Reburn severity in aggregate was also higher than non-reburn severity suggesting a positive feedback effect that could contribute to an upward drift in fire severity as area burned increases. Terrain had low importance in both models. This indicates strong terrain controls in the past may not carry into the future. Low- and moderate-severity fire effects were prevalent in non-reburns under moderate fire weather and self-reinforcing behavior maintained these effects in reburns even under more extreme weather, particularly in reburns within 10 yr. Our findings suggest deliberate use of wildfire and prescribed fire under moderate conditions would increase fire resilience in landscapes transitioning to an active fire regime.
AB - Fire severity patterns are driven by interactions between fire, vegetation, and terrain, and they generate legacy effects that influence future fire severity. A century of fire exclusion and fuel buildup has eroded legacy effects, and contemporary fire severity patterns may diverge from historical patterns. In recent decades, area burned and area burned at high severity have increased and landscapes are transitioning back to an active fire regime where disturbance legacies will again play a strong role in determining fire severity. Understanding the drivers of fire severity is crucial for anticipating future fire severity patterns as active fire regimes are reestablished. We identified drivers of fire severity in the Klamath Mountains, a landscape with an active fire regime, using two machine learning statistical models: one model for non-reburns (n = 92) and one model for reburns (n = 61). Both models predicted low better than moderate or high-severity fire. Fire severity drivers contrasted sharply between non-reburns and reburns. Fire weather and fuels were dominant controls in non-reburns, while previous burn severity, fuel characteristics, and time since last fire were drivers for reburns. In reburns, areas initially burned at low (high) severity burned the same way again. This tendency was sufficiently strong that reburn fire severity could be predicted equally well with only severity of the previous fire in the model. Thus, reburn fire severity is more predictable than severity in non-reburns that are driven by the stochastic influences of fire weather. Reburn severity in aggregate was also higher than non-reburn severity suggesting a positive feedback effect that could contribute to an upward drift in fire severity as area burned increases. Terrain had low importance in both models. This indicates strong terrain controls in the past may not carry into the future. Low- and moderate-severity fire effects were prevalent in non-reburns under moderate fire weather and self-reinforcing behavior maintained these effects in reburns even under more extreme weather, particularly in reburns within 10 yr. Our findings suggest deliberate use of wildfire and prescribed fire under moderate conditions would increase fire resilience in landscapes transitioning to an active fire regime.
UR - http://www.scopus.com/inward/record.url?scp=85115869464&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85115869464&partnerID=8YFLogxK
U2 - 10.1002/ecs2.3734
DO - 10.1002/ecs2.3734
M3 - Article
AN - SCOPUS:85115869464
SN - 2150-8925
VL - 12
JO - Ecosphere
JF - Ecosphere
IS - 9
M1 - e03734
ER -