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

---

Opportunities for spatially explicit risk assessments

One fundamental component of risk assessments of introduced species is predictions of the species possible distribution. Species distribution modeling has been a research goal of ecologists for many years. Here we present some bioclimatic analyses using the ANUCLIM modeling package and other statistically based techniques. ANUCLIM makes use of climate models, terrain models and species location information to estimate and map bioclimatic profiles and hence possible distributions. At GLFC we have created monthly mean climate surfaces for all North America using the climate interpolation techniques of Mike Hutchinson (Hutchinson 1998). Those models enable us to generate reliable estimates of many climate variables at any location. For the bioclimatic analysis as implemented here, a file is needed that lists the latitude, longitude and elevation of locations that sample the target species geographic distribution. For most of the species presented here, location data were taken from historical Canadian Forest Service insect and disease records. This location file is input to the ANUCLIM program, which then estimates a suite of climatic parameters at each species location. This provides the estimate of the climatic profile of the species. The profiles can be mapped using the BIOMAP component of the ANUCLIM program. BIOMAP compares the climatic profile of the species to a regular grid maps of the parameters. If a grid cell is within the range of the climate profile for all the selected climate variables the cell is flagged as suitable for the species.

This method has been successfully applied in the Southern Hemisphere to model range limits and to support field survey designs (e.g., Lindenmayer et al. 1996 and the references therein). Quantifying climatic domains opens the door to examining possible distribution changes in future climates. More statistical approaches like logistic regression can be used if presence/absence or abundance data are available.

---