Forest insects cause defoliation disturbances with complex spatial dynamics. These are difficult to measure but critical for models of disturbance risk that inform forest management. Understanding of spatial dynamics has lagged behind other disturbance processes because traditional defoliation sketch map data often suffered from inadequate precision or spatial resolution. We sought to clarify the influence of underlying habitat characteristics on outbreak patterns by combining forest plots, GIS data and defoliation intensity maps modeled from Landsat imagery. We quantified dependence of defoliation on spatial patterns of host abundance, phenology, topography, and pesticide spray for a recent gypsy moth outbreak (2000-2001), in a mixed deciduous forest in western Maryland, USA. We used semivariograms and hierarchical partitioning to quantify spatial patterns and variable importance. Habitat characteristics from plot data explained 21 % of defoliation variance in 2000 from tree density, phenological asynchrony, pesticide spray status, and landform index and 34 % of the variance in 2001 from previous-year defoliation, relative abundance of non-host species, phenological asynchrony, pesticide spray status, and relative slope position. Spatial autocorrelation in residual defoliation ranged over distances of 788 m in 2000 and 461 m in 2001, corresponding well with gypsy moth larval dispersal distances (100 m to 1 km). Un-measured processes such as predation, virus and pathogen occurrence likely contribute to unexplained variance. Because the spatial dynamics of these factors are largely unknown, our results support modeling gypsy moth defoliation as a function of dependence on significant exogenous characteristics and residual spatial pattern matching.
Bibliographical noteFunding Information:
Acknowledgments Access to the CFI plot dataset was provided by the Maryland Department of Natural Resources. Funding support was provided by NASA Carbon Cycle Science grant NNX06AD45G and a NASA earth system science graduate fellowship (NNX08AU93H). F. Zumbrun and other foresters at GRSF supported field visits and shared invaluable historical and ecological perspective on the study area. B. Thompson (Maryland Department of Agriculture) provided spray block GIS data. We thank M. Turner (Department of Zoology, UW-Madison), K. Raffa (Department of Entomology, UW-Madison), and C. Lorimer (Department of Forest and Wildlife Ecology, UW-Madison) and two anonymous reviewers for input and reviews to experimental design and this paper. N. Keuler (Department of Statistics, UW-Madison) provided early statistical guidance. Discussions with R. Scheller (Conservation Biology Institute) and B. Sturtevant (Northern Research Station, US Forest Service) were helpful for defining the scope of this study.
- Forest disturbance
- Lymantria dispar L.
- Spatial patterns