Proc. Natl Acad. Sci USA, doi:10.1073/pnas.0803506105 (2008)

Credit: istockphoto

A method frequently used to predict how wildlife will respond to a changing climate has been called into question. 'Climate envelope' models use current species distributions to determine how climate influences their range, and hence where they could potentially live in a warmer world. But despite their widespread application, these models have been controversial as they may overlook other potentially important drivers of distribution.

Now, Colin Beale and colleagues at the UK's Macaulay Institute of Land Use Research, Aberdeen, show that the apparent association between species distribution and climate found by these models is no better than chance association for 68 out of 100 bird species across Europe. Although the envelope models show a good fit to real distributions, Beale and colleagues found the models also fit artificial species distributions constructed with no reference to climate. In many cases models fitted the false data better than the real data.

The result raises doubts about earlier predictions made using climate envelopes, especially for species outside of extreme environments. The team urges the use of a more measured approach to assessing climate impacts on populations, accounting for factors such as species interactions, land-use change and habitat degradation.