Predicting post-fire attack of red turpentine or western pine beetle on ponderosa pine and its impact on mortality probability in Pacific Northwest forests

In ponderosa pine forests of western North America, wildfires are becoming more frequent and affecting larger areas, while prescribed fire is increasingly used to reduce fuels and mitigate potential wildfire severity. Both fire types leave trees that initially survive their burn injuries, but will eventually die. Predicting delayed tree mortality has received considerable research attention to aid in post-fire planning and management. The amount of crown scorched is recognized as the most useful variable for discriminating between trees that live or die, but models gain discrimination with additional variables such as bole scorch, bud or cambium necrosis, and post-fire bark beetle attack (Coleoptera: Curculionidae: Scolytinae). Here, logistic regression was used to determine what fire-injury variables are most associated with red turpentine beetle (RTB; Dendroctonus valens LeConte), or western pine beetle (WPB; D. brevicomis LeConte) attack within three years post-fire. This was tested on 7343 ponderosa pine representing a wide diameter range from 18 wild and prescribed fires in Oregon and Washington, and repeated on a subset of 884 large trees > 53.3 cm diameter. Bole scorch height was most associated with RTB or WPB attack on trees across all diameters, but model predictive ability was poor, whereas for pines > 53.3 cm, the models provided moderate discrimination for predicting attack by each beetle. In addition, mortality models using crown scorch proportion and bole scorch proportion were compared to models with an additional variable for RTB, or WPB attack in year one, or attack by either beetle in year one, or year three. Models including any one of these beetle variables outperformed the models using just bole scorch and crown scorch proportions. For both tree diameter groups, the models including RTB year one performed similar to, or better than models with any other beetle variable, and are preferred for predicting delayed mortality because the RTB attack functions as an additional tree injury indicator similar to cambium kill, not captured by the bole scorch proportion, or crown scorch proportion variables. Furthermore, RTB attack can be assessed within the first year post-fire, and is much faster and easier to evaluate than direct sampling of cambium necrosis.