A landscape-scale investigation into the risk of lodgepole pine mortality caused by mountain pine beetle Dendroctonus ponderosae (Coleoptera: Curculioidae: Scolytinae)
Johnson, Erik W.
MetadataShow full item record
Mountain pine beetle (MPB), Dendroctonus ponderosae Hopkins, is currently causing Pinus contorta Douglas (LP) mortality in several areas of western United States and Canada at high levels including portions of Colorado. For decades, researchers have developed models to help land managers predict when and where MPB infestation will develop based on forest structure, tree size, tree age and geographic characteristics; these models were developed at the stand-level for stand-level analysis. Land managers and planners have become increasingly interested in predicting MPB risk and susceptibility at the landscape-scale; however attempts at landscape-scale modeling have proven difficult as continuous forest mensuration datasets are often lacking. Techniques for producing low-cost, high-resolution, landscape-scale forest composition and forest structure Geographic Information System (GIS) layers were demonstrated by this study. These GIS layers were subsequently used to assess several existing MPB risk models, at the landscape-scale, and to derive a new empirical MPB model. The procedures outlined in this paper describe the generation of landscape-scale forest composition and structure GIS layers (predictive surfaces) based on recent and innovative remote sensing and spatial statistical techniques. These techniques transform a small field sample into a continuous GIS surface utilizing multiple linear regression and binary regression trees. Information derived from satellite imagery and digital elevation models are used as auxiliary variables to assist in the prediction of response variables (basal area, proportion of lodgepole pine basal area, diameter at breast height, quadratic mean diameter, percent canopy closure, and trees per acre). A carefully designed field sample, stratified by Landsat image spectral groupings, optimized sampling faculties by maximizing between-stratum variability while minimizing within-stratum variability. Forest composition (spatial distribution of tree species), basal area, proportion of lodgepole pine basal area, diameter at breast height, quadratic mean diameter, percent canopy closure, and trees per acre predictive surfaces were developed for Colorado's Fraser River Valley. These predictive surfaces were then used to assess the landscape-scale predictive capabilities of following MPB prediction models: Anhold et al., (1996), Amman et al. (1977), Shore and Safranyik (1992), and the USDA Forest Service National Insect and Disease Risk Map. Finally, a new MPB model is described based on geographic factors, the predictive surfaces, and recent occurrence of mountain pine beetle caused-tree mortality.Reich, Robin M.