Dynamically simulating spruce budworm in eastern Canada and its interactions with wildfire

•Eastern spruce budworm representation was integrated into a dynamic global vegetation model (LPJ-LMfire).•Cyclical temporal patterns were determined by post-outbreak regeneration of host trees.•Spatial patterns were determined by density of host trees across landscape, driven by climatic and other environmental conditions.•Outbreak enhanced fire potential through temporarily increased fuel loads but long-term effect was small due to subsequent depleted fuel load. Eastern Spruce Budworm (ESBW) is a major agent of disturbance in Eastern Canada's boreal forests. Outbreaks have historically led to widespread defoliation of its preferred host trees, fir and spruce species, leading to high rates of mortality. This in turn can result in significant economic losses and enhancement of fire potential in the region. Representation of such biotic disturbance has rarely been included in Dynamic Global Vegetation Models (DGVM), which have become essential tools in understanding and predicting forest dynamics in present and future contexts. We present novel representation of host-specific defoliation in a DGVM (LPJ-LMfire), to better represent disturbance regimes in the boreal forest of eastern Canada. Using host foliage density to trigger outbreak, we were able to calibrate and simulate general spatial patterns of defoliation relative to historical aerial sketch map data. Return intervals were thus sensitive to the growth rates of host trees. Modeled return intervals tended to be significantly longer than 30 years, the approximate observed return interval. A factorial experiment was performed on the interactions of ESBW with wildfire, which was found to be slightly enhanced in terms of burned areas after outbreaks due to increased fuel loads. Interactions between ESBW and fire were found to have higher interaction strength in the drier Western region of the boreal forest. Our study demonstrates that biotic disturbance and its interaction with wildfire can be effectively simulated in a DGVM. We show that bottom-up climatic controls are sufficient to drive simulated spatiotemporal patterns of ESBW that can be calibrated to generally match historical observations.