Cold temperature and emerald ash borer: Modelling the minimum under-bark temperature of ash trees in Canada

The potential range of forest pest species can depend on climate conditions. Predicting the effects of climate on insect species may be complicated, however, by the need to translate routinely collected meteorological data into microclimate measurements. For example, tree-boring beetles which spend the winter months in the under-bark microclimate (e.g., emerald ash borer (Agrilus planipennis Fairmaire), mountain pine beetle (Dendroctonus ponderosae Hopkins)) may experience minimum temperatures that tend to be warmer than air temperature minima. Moreover, the magnitude of the difference between under-bark and air temperature minima is variable, meaning that models of under-bark temperature cannot assume a constant level of thermal buffering.We used a Newtonian cooling model to predict the winter under-bark temperature minima of ash trees (Fraxinus spp.) in woodlots and urban centers in Canada. In order to capture a wide range of temperature extremes and patterns experienced by emerald ash borer or other deciduous-feeding tree-boring beetles, the model was parameterized and tested using hourly temperature data from two winters for 24 different ash trees in urban and woodlot sites across 6 different Ontario locations. The model was tested using a subset of the Ontario data and data from Alberta and Saskatchewan. Average model prediction errors (e.g., 1.31 °C root mean squared error for Ontario) were much smaller than errors associated with assuming a constant level of buffering of 1 °C (2.61 °C). Modifications of the model to account for elevation and solar radiation did not offer any improvement in model fit, although accounting for urban heat island effects offered some improvement.An attempt was made to model daily temperature maxima, but even when the model was modified to include solar radiation, it was not accurate at predicting daily temperature maxima, possibly because of complex shading and heating patterns that can occur during warmer parts of the day in urban and woodlot environments. We conclude that this model is suitable for modelling minimum under-bark temperature across a range of latitudes and longitudes for both deciduous and coniferous trees in urban and woodlot settings.

► We model minimum under-bark temperature using a Newtonian cooling model.
► We accurately predict minimum under-bark temperatures in similar and novel areas.
► The basic model is superior to assuming a constant level of thermal buffering.
► There is some improvement in prediction when urban heat island effects are included.
► We recommend widespread use of the model for deciduous and coniferous trees.