Epidemiological models reproduce regular insect outbreaks

Background/Question/Methods

Insect outbreaks can have a large impact on ecosystems, and understanding their underlying ecological processes is important for management. For example, the Spruce Budworm outbreaks occur roughly every 30 years and profoundly structure the dynamics of North American boreal forests. Current modeling approaches are either correlative or based on local insect population processes and so offer limited understanding of outbreaks at the landscape level, where management actions take place. Our objective is to develop a mechanistic model that reproduces insect outbreaks at relevant spatial scales. The resulting FIR (Forest - Infected - Recovering) model is inspired by epidemiology and tracks the macroscopic consequences of outbreaks. We construct two versions of the FIR model, a spatially-implicit (mean field model with global dispersal) and a spatially–explicit (cellular automata with local dispersal) version. We use the life-history of boreal forests and the spruce budworm, including the Allee effect, to parameterize our theoretical model. We then explore the model analytically and numerically to determine the conditions required for sustained and synchronous outbreaks.

Results/Conclusions

For both versions of the FIR model, the proportion of infected landscape is 10-30%, an amplitude that matches observed landscape infections of the spruce budworm. Outbreaks occur every 30-200 years depending on parameters. In the spatially-implicit version of the FIR model, density-dependent emigration (mimicking an Allee effect) is necessary to produce recurrent outbreaks. The spatially-explicit version of the FIR model with local stochastic dispersal displays regular outbreaks in time and self-organized structure in space, even without density-dependent dispersal. Our analysis reveals that spatially-explicit dispersal produces landscape outbreaks, and not the nonlinear dispersal. Outbreaks are clustered in space with a radius of up to 16 stands even though dispersal can only occur between the closest neighbors. The spatially-explicit model reproduces outbreak patterns observed at the landscape scale and is a suitable model of spruce budworm outbreaks. The FIR model produces spatio-temporally synchronous outbreaks and could provide an innovative approach to model and manage insect outbreaks. Using epidemiological models means that we can apply the experience gained from vaccines, towards managing insect outbreaks.