Publication at Faculty of Science |
2012
Abstract
Our study demonstrates that assembling data on both environmental characteristics and species composition of habitats is likely to be useful for predicting habitat suitability at a landscape scale. This study also demonstrates that a high AUC value is not a guarantee that a model's prediction is reliable because a cultivation experiment may provide different results.
When identifying habitats that are suitable for a species (e.g. for the purpose of a metapopulation study), the results should be subjected to a sensitivity analysis.