Thomas et al.1 have carried out a usefulanalysis of the extinction risk from climate warming. Their overall conclu-

sion, that a large fraction of extant speciescould be driven to extinction by expectedclimate trends over the next 50 years, iscompelling: it adds to the many other rea-sons why new energy policies are needed toreduce the pace of warming.

There is one important reason, however,why Thomas et al. could be greatly under-estimating the threat to biodiversity from climate change: genetic adaptation to cli-mate at the population level. Thomas et al.1

assume that if the future climate conditionsat any particular location within the currentrange of a species fall outside the climateenvelope determined for the entire species,then that location is no longer a suitablehabitat for the species. Implicit is theassumption that all individuals within aspecies are adapted to the same climate envelope — an envelope determined fromthe current range of the species.

But now consider a species comprisingdistinct populations that occupy a subset ofranges arrayed along a climate gradient. Ifeach population can survive only within therelatively narrow climate envelope charac-terizing its current sub-range, then everyindividual within the species could sufferstress from climate change2. If the individu-als cannot migrate, then all individualswithin the species would be at risk. By con-trast, if existing differential population-level adaptations to climate are ignored, asby Thomas et al., then individuals locatedtowards the cooler end of the current rangeare unlikely to feel the stress of warming,whether they can migrate or not.

The prevalence of such genetic adapta-tion to climate at the population level isunknown. Thomas et al. could be over-estimating extinction rates if evolutionaryadaptation to climate change occurs in thefuture, lessening the effect of warming on

species survival. But the longer generationtimes of macroscopic species will probablyprevent adaptation rates from keeping pacewith anthropogenic climate change.

Thomas et al. use a species-level approachfor estimating extinction under climatechange, in contrast to the more traditionalcommunity-level approach that uses thespecies–area relationship (SAR). Althoughtheir methods find inspiration from thepower-law form of the community-levelSAR, instead of enforcing the SAR at thecommunity level before and after climatechange, they aggregate information onspecies-level range areas into an estimate of extinction for the community. Thisapproach is based on the reasonable assump-tion that species differ in their response tothe effects of climate change. However, theirmodifications of the classical community-level approach for estimating extinction usea common species-area exponent z for allspecies, which may not be justified. Speciesthat are sparsely distributed within their climate envelope may be less likely to survivea given fractional change in range thanspecies that are found throughout their cli-mate envelope. Deriving predictions for thecommunity by applying z to all species canyield poor results3.

Species differences in the proportionalchange to their distributional area also ariseunder land-use change. If the species-levelmethods of Thomas et al. were used to esti-mate extinctions from land-use change, theywould yield different estimates from the traditionally used community-level SAR. Asnoted by Thomas et al., more work is neededto understand how extinction risk for a givenspecies depends on its reduction in range area.

Estimation methods that involve theapplication of the SAR (or SAR-inspiredspecies-level equations) to regions withholes or fragmentation typically ignore theconfounding effects of shape in the depen-dence of species richness on area4 and, for

brief communications arising

NATURE | 1 JULY 2004 | www.nature.com/nature

this reason, may be inaccurate. An alterna-tive estimation method that could be moreaccurate in some circumstances demon-strates the degree to which shape influencesthe expected number of extinctions. Thisrelies on the endemics–area relationship(EAR), which quantifies how the numberof species spatially confined to a habitatpatch depends on the area of that patch3,5,6.

Where the SAR approach asks how manyspecies are present in the remaining habitat,the EAR approach asks how many species arelost because those species were initially pre-sent only in the lost habitat. If shape did notinfluence either, these methods would yieldidentical results. However, because it does,EAR can yield significantly different answersfrom the SAR method, depending on theshape and fragmentation of the lost andremaining habitat5,6. Sorting out where theaccurate estimate falls between the EAR andSAR estimates requires as-yet unavailableknowledge of how large-scale geographicpatterns of diversity depend on shape andfragmentation.

Thomas et al. have focused attention on an important issue. The challenge is toimprove our understanding of populationgenetics and climate tolerance within species,as well as to find the most reliable way to usespatial patterns of diversity and species-levelchanges in range area for generating esti-mates of species loss under climate change.John Harte*, Annette Ostling*,Jessica L. Green†, Ann Kinzig‡*Energy and Resources Group, University ofCalifornia, Berkeley, California 94720, USA e-mail: jharte@socrates.berkeley.edu†School of Natural Sciences, University of California,Merced, PO Box 2039, California 95344, USA‡Department of Biology, Arizona State University,PO Box 871501, Tempe, Arizona 85287-1501, USAdoi:10.1038/nature027181. Thomas, C. D. et al. Nature 427, 145–148 (2003).

2. Jensen, D. Thesis, Univ. California, Berkeley (1993).

3. Green, J. L., Harte, J. & Ostling, A. Ecol. Lett. 6, 919–928 (2003).

4. Kunin, W. E. Biol. Cons. 82, 369–377 (1997).

5. Kinzig, A. & Harte, J. Ecology 81, 3305–3311 (2000).

6. Green, J. L., Harte, J. & Ostling, A. in Biotic Homogenization

(eds Lockwood, J. L. &. McKinney, M. L.) 179–200

(Kluwer Academic, New York, 2001).

Reply: Thomas et al. reply to this communication

(doi:10.1038/nature02719).

Biodiversity conservation

Climate change and extinction riskArising from: C. D. Thomas et al. Nature 427, 145–148 (2004)

© 2004 Nature Publishing Group