Publication

A geographically variable model of hemlock woolly adelgid spread

Timothy Gregoire and 1 other contributor

On This Page

    Abstract

    The hemlock woolly adelgid (HWA) (Adelges tsugae) has spread through the range of eastern hemlock (Tsuga canadensis) leaving dead hemlock trees in its wake. We combined county and previously unavailable township level infestation records with geographic variables including plant hardiness zone, elevation, forest cover type, urbanization, precipitation, temperature, and hemlock range in a geographic information system to build models linking HWA spread to the structure and characteristics of the environment. We explored the connections between site variables and spread rate using quantile regression, Tobit regression, a parametric model for heterogeneous error variance, and classification and regression tree modeling. The results show that since 1990 HWA has spread at a rate closer to 12.5 km/yr rather than the 20-30 km/yr previously estimated, however spread rates vary significantly with environmental variables. County-based data show a faster rate of range expansion in the south, 15.6 km/yr, and very slow spread in the northern part of the current range of HWA. The rate of spread based on township records for Pennsylvania and north is 8.13 km/yr. Our study suggests that HWA spreads more slowly in colder areas, with a mean minimum temperature of -26.1 degrees C (plant hardiness zone 5B) delineating areas of slower range expansion. HWA also spreads more slowly during cold years, those with lower than average mean January temperatures. Our analysis demonstrates the value of quantile regression, Tobit regression, and classification and regression tree models for understanding the spread of invasive, exotic insects.