||Context Forest insect outbreaks are influenced by ecological processes operating at multiple spatial scales, including host-insect interactions within stands
and across landscapes that are modified by regional-scale variations in climate. These drivers of outbreak dynamics are not well understood for the western
spruce budworm, a defoliator that is native to forests of western North America.
Objectives Our aim was to assess how processes across multiple spatial scales drive western spruce budworm outbreak dynamics. Our objective was to assess the relative importance and influence of a set of factors covering the stand, landscape, and regional scales for explaining spatiotemporal outbreak patterns
in British Columbia, Canada.
Methods We used generalized linear mixed effect models within a multi-model interference framework to relate annual budworm infestation mapped from Landsat time series (1996–2012) to sets of stand-, landscape-, and regional-scale factors derived from forest inventory data, GIS analyses, and climate models.
Results Outbreak patterns were explained well by our model ( R2 = 93%). The most important predictors of infestation probability were the proximity to
infestations in the previous year, landscape-scale host abundance, and dry autumn conditions. While stand characteristics were overall less important predictors,
we did find infestations were more likely amongst pure Douglas-fir stands with low site indices and high crown closure.
Conclusions Our findings add to growing empirical evidence that insect outbreak dynamics are driven by multi-scaled processes. Forest management plan-
ning to mitigate the impacts of budworm outbreaks should thus consider landscape- and regional-scale factors in addition to stand-scale factors.