Landscape-scale spatial distribution of spruce budworm defoliation in relation to bioclimatic conditions

Two empirical statistical models were developed to describe the spatial variation in defoliation by spruce budworm (Choristoneura fumiferana Clem.), as recorded by Ontario's Forest Health Survey from 1967 to 1998. These models revealed a number of relationships between the spatial distributions of aerially detectable spruce budworm defoliation and bioclimatic conditions over the landscape. A classification tree model relates the northern and southern boundaries of defoliation to the relative abundance of different tree species that host spruce budworm. Between these boundaries, the classification tree uses the maximum winter temperature and the minimum temperature in May to describe where detectable defoliation occurred. A regression tree model uses a total of eight variables related to winter temperatures, forest composition, spring temperatures, summer temperatures, and precipitation to estimate the defoliation frequency in areas where defoliation was detected at least once from 1967 to 1998. High defoliation frequencies were associated with dry Junes (precipitation, < 86 mm) and cool springs (mean minimum temperature < -2.7 degrees C). Conversely, low frequencies were associated with cold winters (mean minimum temperature < -23.3 degrees C; mean maximum temperature > -11.0 degrees C) in the north and a low abundance of host species (percentage of the basal area occupied by balsam fir, white spruce, and black spruce, < 14.3%) in the south. Spatial autocorrelation in the bioclimatic variables had little effect on their relationships with the spatial distribution of the defoliation frequency.