Spatial and spatial-temporal analysis of mountain pine beetle infestations at a landscape scale

The impact of the current mountain pine beetle (Dendroctonus ponderosae Hopkins) epidemic in British Columbia underscores the need for scientifically informed management practices. During an epidemic it is necessary to manage large areas and an understanding of landscape scale spatial and spatial-temporal processes is required. With the recent availability of large area, multi-temporal data sets there are new opportunities for landscape scale studies of the mountain pine beetle over space and through time. In this thesis large area spatial and spatial-temporal patterns of lodgepole pine (Pinus contorta var. Latifolia) mortality are explored using point data collected through helicopter surveys. As with all large area data sets, mountain pine beetle data are prone to uncertainty. Using field measurements collected to supplement the helicopter data set, we explore the nature and amount of error in point data. Based on error estimates, a method is presented for incorporating uncertainty when visualizing data via kernel density estimation. Locations that are hot spots, or have the most intense infestations, are identified and used to explore dispersal behaviour. Comparing hot spots to various landscape characteristics allows investigation into how mountain pine beetle utilize the forest in space and through time. Locations of change are also identified and explored in terms of spatial-temporal patterns and associated landscape characteristics. The relatedness of hot spot and change locations is investigated. A randomization approach is also used to supply the spatial pattern of large area infestations by evaluating observed data relative to a null expectation conditioned on a model of forests at risk to beetle attack. Investigating the landscape characteristics associated with unexpected locations enabled exploration into the cause of differences between empirical and modelled patterns