Exotic species risk modelling

• The problem
• Introduction to risk modelling
• Opportunities for spatially explicit risk assessments
• Canadian example applications: European larch canker
• Canadian example applications: Scleroderris
• Canadian example applications: Dung beetle
• Conclusions, acknowledgement and references

---

Canadian example applications: Scleroderris

Gremmeniella abietina (Lagerb.) (From Venier et al. 1998) Scleroderris disease, caused by the fungus Gremmeniella abietina (Lagerb.) Morelet var. abietina, has been regarded as a major pest of pine species in North America, for more than 30 years. It causes mortality to pine trees less than one metre in height. We used historical distribution data of scleroderris disease in Ontario to model its probability of occurrence as a function of climate factors. A logistic regression model of the probability of occurrence as a function of the mean temperature of the coldest quarter and the precipitation of the coldest quarter was a very good fit. We chose the two-variable model as our final model because of its relatively high concordance and its relative simplicity. According to the model, scleroderris disease is more likely to be found at lower winter temperatures and in places with more winter precipitation. The concordance (index of classification accuracy) of the model was 84%. We sub-sampled the data repeatedly, generated new parameter estimates and tested the predictions against data not included in the model. Classification accuracy was similar for each sub-sample model.  Therefore we concluded that the final model is stable. Gridded estimates of the climate variables were used to spatially extend the two-variable logistic regression model and produce a probability of occurrence map for scleroderris disease across Ontario.

Probability of occurrence, Scleroderris

The predicted map of probability of occurrence fits well with the map of the observed locations of the disease. These results lend credence to previous work that suggests that distribution of scleroderris disease is strongly influenced by climate. The classification results also suggest that this model is a useful tool for assessing the risk of scleroderris disease throughout Ontario.

---