SPATIAL SYNCHRONY OF SPRUCE BUDWORM OUTBREAKS IN EASTERN NORTH AMERICA

Abstract

We investigated the spatial synchrony of outbreaks of the spruce budworm, Choristoneura fumiferana, over much of its outbreak range in eastern North America during the period 1945–1988. Spatial synchrony decreased with distance between local populations and approached zero near 2000 km. Investigation of the synchrony of local population time series with cluster analysis revealed a pattern of geographically distinct blocks of clusters oriented along an east–west axis. Spatial synchrony also was identified in monthly temperature and precipitation time series at 18 weather stations over the same time period and geographical range as the spruce budworm outbreaks. Cross correlations decreased linearly with distance between stations and approached zero near 3000 km and 1800 km, respectively. We developed a spatially explicit lattice model for a single species occupying multiple patches. Within patches, the model had first order logistic dynamics, and patches were linked by dispersal that depended upon their separation distances. Both local and regional stochasticity (i.e., a Moran effect) were present. The modeled lattice had the same spatial configuration as the outbreak region to facilitate investigating the relative effects of a Moran effect and dispersal on spatial synchrony. Simulations with and without a simple region-wide Moran effect and three levels of dispersal did not produce the decrease in spatial synchrony with distance observed with spruce budworm time series. However, when run at the highest dispersal rate, those simulations produced cluster maps similar to that observed for spruce budworm defoliation. Simulations with a spatially autocorrelated disturbance that had either zero or high local variability and three levels of dispersal produced decreases in spatial synchrony with distance similar to that observed in the historical data. When run at the highest dispersal rate, simulations yielded cluster maps similar to the cluster map for defoliation. We discuss the potential significance of the spatially autocorrelated disturbance factor in understanding regional insect outbreaks. We also consider the plausibility of dispersal rates used in our simulations. We suggest in conclusion that spruce budworm outbreaks were synchronized by a combination of a spatially autocorrelated Moran effect and a high dispersal rate.