We question Fangliang He and Stephen Hubbell's claim that species–area relationships overestimate global extinction (Nature 473, 368–371; 2011). We contend that they do not test their claims against real data on global extinction or threat. We also believe that they address only a small part of the problem.

Imagine destruction that wipes out 95% of habitat overnight — metaphorically speaking. How many species will have disappeared the following morning? He and Hubbell tell us it would be just those living only in the destroyed area, and not in the other 5%. In our view, the more important question is how many species in total, including those in the remnant habitat 'islands' (the 5%), will eventually become extinct (see M. L. Rosenzweig Species Diversity in Space and Time Cambridge Univ. Press, 1995.)

Many studies accurately verify extinction predictions based on the relationship between island area and numbers of species, which He and Hubbell dismiss. Scores of separate tests find striking agreement between the number of predicted extinctions from habitat loss and the number of consequent extinctions (or of species facing extinction). This is seen globally and within individual regions, including eastern North America, South America, Africa and southeast Asia (see, for example, S. L. Pimm and R. A. Askins Proc. Natl Acad. Sci. USA 92, 9343–9347; 1995).

Comprehensive analyses can now combine remotely sensed ecosystem changes with information on species extinction risk, distribution, habitats, threats and conservation actions from the International Union for Conservation of Nature Red List. In our opinion, it is these studies — which ask the right questions and verify the answers — that have crucial implications for the world's efforts to conserve biodiversity.