Insect outbreaks are among the main disturbances shaping the structure of North American boreal forests. While the historical range and spread dynamics of such outbreaks are well documented, those dynamics are poised to change as new climatic conditions interact with the specific phenology of defoliating insects and their potential tree hosts. Precise estimates of past defoliation across different host species, forest compositions and climatic conditions are essential for the design of models that can successfully forecast the spread and impact of future outbreaks. In this study, we produce quantitative maps of defoliation intensity for the 1967-1992 spruce budworm outbreak in Quebec, Canada by combining two complementary data sources: aerial surveys, which partition the study region into discrete defoliation classes, and tree-ring data from forest inventory plots, which provide a quantitative record of cumulative defoliation via its impact on tree growth. We use a hierarchical Bayesian model that relates the underlying outbreak intensity (latent variable) to both the defoliation classes in aerial surveys (ordinal logistic regression) and the standardized tree-ring widths (autoregressive model with cumulative defoliation effects on growth).
Results/Conclusions
Consistent with previous studies, our model shows that a year of defoliation affects the growth of spruce budworm hosts for at least five successive years. Due to the limited spatial coverage of inventory plots and unmeasured factors affecting tree growth at the local scale, spatial smoothing constraints were required to stabilize defoliation estimates across the study area. This model will serve as an observation sub-model to be coupled with process sub-models as part of a larger project to forecast the spread of spruce budworm outbreaks at the landscape scale. Combining independent sources of data on outbreak occurrence will also allow us to separate the direct effects of climatic conditions on tree growth from the indirect effects through which climate modulates the severity of outbreaks.