COS 189-5 - Geographical variation in spatial synchrony of forest-insect outbreaks: Isolating the drivers of synchrony

Friday, August 10, 2012: 9:20 AM
E146, Oregon Convention Center
Kyle J. Haynes, Blandy Experimental Farm, University of Virginia, Boyce, VA, Ottar N. Bjornstad, Entomology, Penn State University, University Park, PA, Andrew J. Allstadt, Forest and Wildlife Ecology, University of Wisconsin, Madison, WI and Andrew M. Liebhold, Northern Research Station, USDA Forest Service, Morgantown, WV
Background/Question/Methods

Despite the pervasiveness of spatial synchrony of population fluctuations, the ability to isolate the underlying mechanisms (e.g., synchronous environmental perturbations, dispersal, or mobile predators) remains elusive.  In this study, we use a novel approach to examine the mechanisms driving spatial synchrony of population fluctuations, which is to statistically partition geographic variance in the level of synchrony of population fluctuations into potential environmental and spatial drivers.  We evaluated the extent to which the level of synchrony in gypsy moth outbreaks between paired locations was related to the level of synchrony in weather conditions, spatial proximity, and similarity of forest type.  This was accomplished by creating resemblance matrices for each variable where the entries of the matrices were the similarities (correlation or spatial proximity) between all pair-wise combinations of locations.  We then separated the roles of weather, proximity, and forest type in the synchrony of gypsy moth outbreaks using multiple regression on resemblance matrices (MRM).  Prior to conducting the MRM analysis, we reduced the dimensionality of the weather data using principle components analysis.

Results/Conclusions

Principle components analysis allowed us to reduce the dimensionality of the weather data from 36 variables to 2 principle component axes, the first primarily representing monthly average minimum and maximum temperatures and the second mainly representing monthly accumulations of precipitation.  An MRM model including only proximity as a predictor ("space model") explained 12.7% of the variation in synchrony of gypsy moth outbreaks.  A model including the resemblance matrices for synchrony in weather and similarity in forest type ("environment model") explained almost twice as much variation (25.4%) in synchrony of gypsy moth outbreaks.  Adding proximity to the "environment model" explained almost no additional variation (< 0.1%) in the synchrony of outbreaks.  The only significant predictor in the "environment model" was the level of synchrony in the scores from the second principle component axis, which largely represents precipitation.  We conclude that dispersal of gypsy moths or its natural enemies does not drive the synchrony of outbreaks based on the lack of spatial effects.  Spatial synchrony of weather, primarily precipitation, appears to be the main driver of synchrony in gypsy moth outbreaks.