What is it about?

Knowing how much confidence to place in the predictions of a species distribution model can be difficult, especially for invasive species that are still expanding their range and have not reached an equilibrium with the environment. The approach demonstrated in this paper uses four important techniques to address this issue: (1) k-fold cross validation of geographically independent test data relative to training data, to determine which model has the best predictive ability (2) multi-model averaging to estimate standard deviations that incorporate variance due to differences in model specification (3) using quasi-AIC to select between models because the effective degrees of freedom in a spatial model are typically much less than the number of sample units (pixels) used to construct them (4) visualising the distribution of uncertainty as a map. It is recommended that these techniques be used routinely when developing species distribution models.

Featured Image

Why is it important?

Knowing which areas of the globe are vulnerable to which invasive species can help direct biosecurity measures to where they are most needed. However the costs of a false-negative prediction are likely more costly than the costs of a false postive prediction.

Read the Original

This page is a summary of: Quantifying uncertainty in the potential distribution of an invasive species: climate and the Argentine ant, Ecology Letters, August 2006, Wiley,
DOI: 10.1111/j.1461-0248.2006.00954.x.
You can read the full text:

Read

Resources

Contributors

The following have contributed to this page