Best Predictors for Postfire Mortality of Ponderosa Pine Trees in the Intermountain West

• Published in: Forest science Vol. 52; no. 6; pp. 718 - 728
• Main Authors: Hull Sieg, C, McMillin, J.D, Fowler, J.F, Allen, K.K, Negron, J.F, Wadleigh, L.L, Anhold, J.A, Gibson, K.E
• Format: Journal Article
• Language: English
• Published: Bethesda Oxford University Press 01.12.2006
• Subjects: bark beetles; coniferous forests; crown scorch; fire ecology; fire severity; Fires; forest trees; Ips; logit analysis; mathematical models; Pinus ponderosa; prediction; regression analysis; sampling; statistical models; Studies; tree and stand measurements; tree crown; tree damage; tree mortality; Trees; volume; wildfires; Intermountain West region; USA, Utah, Intermountain West
• ISSN: 0015-749X; 1938-3738
• DOI: 10.1093/forestscience/52.6.718

Numerous wildfires in recent years have highlighted managers' needs for reliable tools to predict postfire mortality of ponderosa pine (Pinus ponderosa Dougl. ex Laws.) trees. General applicability of existing mortality models is uncertain, as researchers have used different sets of variables. We quantified tree attributes, crown and bole fire damage, ground fire severity, and insect presence from a total of 5,083 trees in four 2000 wildfires in four Intermountain states. Crown scorch (percentage) and consumption (percentage) volume collectively accounted for the majority of predictive capacity in all four individual models and in the pooled four-site model. The addition of tree diameter and presence of Ips beetles in the pooled model slightly improved predictive power. Four other statistically significant variables added little to the pooled model's predictive ability. The pooled model correctly classified 3-year postfire mortality of 89.9% of the trees and had a receiver operating characteristic (ROC) score of 0.96. In the external validation step, the model correctly classified 3-year postfire mortality of 96% of 1,361 trees in a 2001 wildfire. Our results and a number of previous studies suggest that a two-variable model using percentage crown scorch volume and crown consumed volume will have applicability beyond the Intermountain West.