, 258 (2008);
et al. Robert K. Colwell,
Lowland Biotic Attrition in the Wet Tropics
Global Warming, Elevational Range Shifts, and
This copy is for your personal, non-commercial use only.
. clicking here colleagues, clients, or customers by
, you can order high-quality copies for your
If you wish to distribute this article to others
. here following the guidelines
can be obtained by
Permission to republish or repurpose articles or portions of articles
(this information is current as of February 2, 2010 ):
The following resources related to this article are available online at www.sciencemag.org
version of this article at:
including high-resolution figures, can be found in the online
Updated information and services,
can be found at:
Supporting Online Material
, 10 of which can be accessed for free:
cites 26 articles
related to this article
A list of selected additional articles on the Science Web sites
23 article(s) on the ISI Web of Science.
This article has been
5 articles hosted by HighWire Press; see:
This article has been
This article appears in the following
registered trademark of AAAS.
is aScience 2008 by the American Association for the Advancement of Science; all rights reserved. The title
Copyright American Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005.
(print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by the Science
on February 2, 2010
Global Warming, Elevational Range
Shifts, and Lowland Biotic Attrition
in the Wet Tropics
Robert K. Colwell,1* Gunnar Brehm,2† Catherine L. Cardelús,3†
Alex C. Gilman,4† John T. Longino5†
Many studies suggest that global warming is driving species ranges poleward and toward higher
elevations at temperate latitudes, but evidence for range shifts is scarce for the tropics, where the
shallow latitudinal temperature gradient makes upslope shifts more likely than poleward shifts.
Based on new data for plants and insects on an elevational transect in Costa Rica, we assess the
potential for lowland biotic attrition, range-shift gaps, and mountaintop extinctions under projected
warming. We conclude that tropical lowland biotas may face a level of net lowland biotic attrition
without parallel at higher latitudes (where range shifts may be compensated for by species from
lower latitudes) and that a high proportion of tropical species soon faces gaps between current
and projected elevational ranges.
correlated with warming climate and changes in
precipitation have been documented for a wide
spectrum of temperate and subtropical species,
and phenological changes portend poleward
ecent global climate change has already
begun to affect species’ geographic
ranges. Poleward shifts in range limits
upslope shifts in range boundaries along temper-
ate elevational gradients have also been detected
for both plants and animals (1–3).
driven, contemporary latitudinal range shifts in
the tropics. Biogeographical baseline data and
resurveys at the high resolution required to doc-
ument incipient range shifts are still scarce for
tropical regions. More fundamentally, however,
the steady latitudinal temperature gradients that
characterize temperate latitudes level off to a
broad plateau within the tropics (fig. S1). For
and limited capability for rapid range shift, the
tropics presents a daunting obstacle to poleward
range shifts (4).
In the tropics as elsewhere, however, eleva-
to 6.5°C decrease per 1000 m elevation (5)—
nearly 1000 times as much as the latitudinal rate
of decrease in temperature, per km, in the tem-
S1) and vastly greater than the modest latitudinal
temperature gradient within the tropics. Conse-
quently, for tropical species affected by warming
climates, upslope range shifts appear far more
likely than latitudinal shifts (4, 6). In fact, ele-
vational shifts provide the only published evi-
dence, to date, for contemporary range shifts
vertebrates, from Monteverde, Costa Rica (7).
To illustrate the potential for elevational range
shifts in the tropics, we analyze elevational range
data for four large survey data sets of plants and
insects [epiphytes (8), understory Rubiaceae (5),
1902 species were collected by the authors since
2001from theBarva Transect, acontinuously for-
gradient from La Selva Biological Station, near
(8) (fig.S2). Before turning to an analysisofthese
tropical species may face on such a gradient.
At temperate latitudes, resources released by
poleward or upslope range shifts may be
appropriated by species from lower latitudes or
lower elevations—species already adapted to
warmer temperatures—if suitable habitat corri-
dors and dispersal mechanisms permit (1).
Likewise, on tropical mountainsides, upslope
range shifts may be compensated by the influx
of species currently found at lower elevations or
by expansion from small nuclei left over from
previous warming episodes (4, 6). In the tropical
lowlands, however, the parallels end. No com-
is available to replace tropical lowland species
1Department of Ecology and Evolutionary Biology, Uni-
versity of Connecticut, Storrs, CT 06269, USA.2Institut für
Spezielle Zoologie und Evolutionsbiologie mit Phyletischem
Museum, Friedrich-Schiller-Universität Jena, Erbertstraße
1, 07743 Jena, Germany.3Department of Biology, Colgate
University, Hamilton, NY 13346, USA.
Ecology and Evolutionary Biology, University of California,
Los Angeles, CA 90950, USA.
Olympia, WA 98505, USA.
*To whom correspondence should be addressed. E-mail:
†These authors are listed alphabetically.
5Evergreen State College,
Fig. 1. A simple graphical model of the potential effects of climate warming
on the distribution of species ranges on a bounded elevational gradient. The
elevational range size for each species (vertical axis) is plottedas afunction of
its elevational midpoint (horizontal axis), with corresponding range limits in-
dicated by the solid horizontal lines. Because ranges cannot extend beyond
the limits of the gradient, all range-size/range-midpoint coordinate pairs lie
withinthegeometricconstraint triangle(dotted lines).(Theconstrainttriangle
is not to be confused with a mountain.) An upslope shift in isotherms with
warming climateismeasured (in meterselevation)by d,thesoleparameterof
the model. (A) Lowland biotic attrition. If lowland ranges shift d m upslope,
their new lower range limits must lie at or above d m elevation, predicting a
isotherms upslope, no projected range smaller than d m (plotted below
the dashed horizontal line) will overlap its prewarming elevational range,
posing challenges for dispersal and establishment. (C) Range contraction and
mountaintop extinction. All ranges with upper limits less than d m from the
upper domain limit are predicted to contract, and the smallest of these (those
less than d m in extent) face local extinction.
10 OCTOBER 2008VOL 322
on February 2, 2010
that shift upslope with climate change or those
that become extinct if no suitable habitat corridor
to cooler climates is accessible.
How likely is “lowland biotic attrition,” the
net loss of species richness in the tropical low-
lands from upslope range shifts and lowland
extinctions driven by climate warming? Surpris-
ingly, this question has scarcely been considered
in a broad biogeographical framework. To ex-
plore it, we must ask whether tropical lowland
species are already living near the thermal op-
which fitness would decline in the absence of
acclimation or adaptation. For plants, especially,
because temperature and precipitation interact
strongly through transpirational water loss, the
answer depends on future changes in precipita-
tion as well as in temperature. Wallace wrote that
“[i]n the equable equatorial zone there is no ...
struggle against climate. Every form of vegeta-
tion has become alike adapted to its genial heat
little even throughout geological periods” (10).
We nowknowthatlowlandtropical climates have
changed substantially and relentlessly ever since
species-rich forests resembling modern ones first
(6). Although the notion of long-term constancy
of tropical climates is nowuniversally dismissed,
on (11), underlying the apparently widespread
conviction that “[m]any tropical species may
well be able to withstand higher temperature[s]
than those in which they currently exist” (12).
Global climate has been cooling since the
Middle Miocene [14.5 million years before the
present (yr B.P.)], a trend punctuated, but not
reversed, by the repeated, dramatic temperature
fluctuations of the Quaternary glacial cycles (1.8
million yr B.P. to present). Pollen core data and
plant microfossils from tropical sites, worldwide,
Last Glacial Maximum (20,000 yr B.P.), and
back up again with the Holocene warming
(10,000 yr B.P.) (13, 14). For the Andes, similar
records span multiple Pleistocene glaciations (4).
Although the climatic changes driving them
were complex (4, 15), these range shifts attest to
the sensitivity of montane tropical species to tem-
perature regimes (indeed, it is routine to estimate
paleotemperatures from paleovegetation). Are
tropical lowland species somehow exceptional,
environments of their Tertiary ancestors? If so,
that tolerance has survived the adaptive demands
land tropics averaged some 5°C cooler than now
(a mean temperature currently characteristic of
30° N or S latitude), with much colder extreme
events in some regions (6, 13, 16). Although hab-
itat heterogeneity certainly mitigated extinction
(4), lowland species or genotypes especially well
adapted to warm climates, which “had nowhere
to go” (6) during glacial episodes, would seem the
least likely to have survived these cold periods
unchanged, because of adaptive tradeoffs (17).
The warm interglacial periods of the Pleisto-
cene, instead of Tertiary climates, might provide
more appropriate adaptive benchmarks for warm
climate tolerance. Current evidence, however,
suggests that contemporary warming, 0.25°C
decade−1since 1975 in the tropical lowlands
(15), has already driven global mean temperature
to within ~1°C of Earth’s maximum temperature
in the pastmillionyears,exceeding the Holocene
maximum (9000 to 5000 yr B.P.) of the current
interglacial (18). In short, tropical climates are
now warmer than at any time during at least the
past two million years (4).
These considerations suggest, on evolution-
ary grounds, that many lowland tropical species
elevations or to cooler, wetter microhabitats in
coming decades—and trouble may already be at
tropical ectotherms are living at temperatures al-
ready near their thermal optimum, with early
fitness declines expected under current projec-
tions for climate warming (19). The growth rate
of individual lowland tropical trees is slowing,
correlated on a year-to-year basis with increasing
mean annual temperature in Costa Rica (20) (at
La Selva) (fig. S2), and plot-level measurements
in Panamanian and Malaysian forests show
similar trends (21). In the Amazon, elevated at-
mospheric CO2itself may be interacting with
increasing temperatures to drive changes in trop-
ical forest composition (22), and distribution
models for individual species project widespread
range contractions and declines in tree viability
and species richness (23), based on current bio-
climatic envelopes. Some global climate models
suggest that Amazonian forest may be approach-
ing a “critical resiliency threshold,” beyond
the effects of warming on tropical plants (24)
Species will respond individualistically to
tropical climate change, just as they have in the
past (4, 14), and vegetation types may expand or
contract (6, 15). Nevertheless, the tropical low-
lands are likely to experience decreased species
richness, with novel plant communities (25)
composed of heat-tolerant species [including fire-
adapted, drought-tolerant species in areas subject
to decreased precipitation (14, 15, 26)]. Early
successional species, adapted to germination at
higher soil temperatures, may thrive at the
Fig. 2. The model in Fig. 1 applied to four groups of species on the Barva Transect (30 to 2900 m
elevation) in Costa Rica (fig. S2). An upslope range shift of 600 m elevation is assumed, based on 3.12°C
climate warming [the Intergovernmental Panel on Climate Change median regional rate for the next
century (32) is 3.2°C] and locally measured lapse rate (5). Each blue disc represents the midpoint-range
coordinate pair for n species (symbol size proportional to
for 555 species of epiphytes (8) sampled at six elevations. (B) Data for 82 species of understory
Rubiaceae (5) sampled at 28 elevations. (C) Data for 739 species of geometrid moths (9) sampled at six
elevations. (D) Data for 495 species of ants (5) sampled at six elevations.
) (see Fig. 1 legend for details). (A) Data
VOL 32210 OCTOBER 2008
on February 2, 2010
expense of late-successional species that require
cooler microhabitats. Anthropogenic habitat al-
teration, shifts in land use patterns, and exotics
and invasives (27, 28) can be expected to ex-
acerbate many of these effects (4, 26).
Lowland biotic attrition is not the only
biogeographic consequence of warming climate
and projected ranges (“range-shift gaps”), a con-
cern at all latitudes, are especially worrisome on
number of tropical species with narrow eleva-
tional ranges. If tropical elevational ranges are
narrower than their temperate counterparts [as
widely believed, albeit on sparse evidence (29)],
then tropical species are more likely than temper-
ate species to experience range-shift gaps for a
given upslope shift in climatic isotherms. Species
face “mountaintop extinction” unless they have
disjunct populations elsewhere on higher moun-
tains or at cooler latitudes (3, 25, 27, 30, 31).
To illustrate the potential for lowland biotic
attrition, range-shift gaps, and mountaintop extinc-
tions in the wet tropics, we analyzed data for
1902 species ofepiphytes,understory rubiaceous
plants, geometrid moths, and ants on the Barva
commonly used to project range shifts under cli-
mate change, requires data for the full geo-
graphical ranges of species and corresponding
environmental variables. Unfortunately, such
data do not exist for the vast majority of tropical
species (including those in this study), with the
principal exception of the best-known vertebrate
groups (e.g., 31). Instead, we designed a simple
graphical model (Fig. 1) that relies on temper-
ature to assess potential elevational range shifts
for transect data. In Figs. 2 and 3, we apply the
model to illustrate the potential consequences of
warming-driven, elevational range shifts for the
four groups of organisms surveyed on the Barva
Because the fourgroups of species differ sub-
stantially in statistical distribution of elevational
range extents and in patterns of range location
along the elevational gradient (Fig. 2), they pro-
ject different patterns of potential range-shift ef-
fects (Fig. 3) (5). Overall, for a 600-m upslope
shift in isotherms, driven by a 3.2°C temperature
increase over the next century (32), about half
currently reach the lowest elevations) are candi-
dates for lowland biotic attrition, and about half
(51%) may be faced with range-shift gaps. The
potential for mountaintop extinctions for these
groups on the Barva Transect is minimal for a
600-m shift in isotherms (Fig. 3A) but begins to
appear at about a 1000-m range shift (Fig. 3B).
Although mountaintop extinction has been the
focus of attention in most discussions of eleva-
tional range shifts (e.g., 2, 27, 31), in the near
term a far greater proportion of the tropical spe-
cies in this study, and elsewhere, are threatened
with lowland attrition or are challenged by early
elevational range (range-shift gaps). Many face
In many respects, the predictions illustrated in
Fig. 3 must be considered worst-case scenarios,
even if warming occurs as assumed. Estimating
likely to underestimate regional elevational range,
even accounting for local undersampling (5). Our
projections share with species distribution models
(12, 23, 30, 31) the assumptions that the funda-
mental climatic niche of each species is fully ex-
pressed by current distributions; that the effects of
climate outweigh any idiosyncratic effects of spe-
patterns, or historical contingency; that change
will be too rapid for adaptation to warmer temper-
atures at lower range limits (1, 4, 30); and that
habitats at the landscape scale are homogenous
with regard to microclimate. In fact, species that
er range limit, including lowland species, may shift
to currently cooler (and wetter) refuges at the same
elevation, in response to warming (1, 4, 14, 16).
On the other hand, our simple projections
based on temperature fail to take account of fac-
tors that may exacerbate the challenges facing
tropical species. In the lowlands, decreased pre-
cipitation and increased fire frequency may am-
plify the direct effects of increased temperature
(15, 26). Anthropogenic habitat fragmentation
and widespread interruption of elevational forest
with climate warming (4, 7, 26), cannot fail to
impede successful range shifts, particularly for
forest-dependent species with limited dispersal
potential (4, 14, 26, 28). As current local as-
semblages are shuffled by individualistic range
shifts, key species interactions may be disrupted
(1, 4, 7, 28). Narrow-ranged species faced with
range-shift gaps will have to compete success-
fully with wide-ranged species that continue to
occupy upslope portions of their current ranges
(14). Range-shift projections focus on immediate
consequences, ignoring long-term effects of de-
creased suitable habitat and smaller populations,
where land area declines with increasing eleva-
tion (27) as it does on the Barva Transect (33).
Finally, the current spectrum of climate types
may contract or novel types appear, as thermal
bands move upslope (25).
Wallace’s impression of tropical heat as
“genial” may have survived the dismissal of his
view of tropical climates as unchanging, but we
wisdom that tropical climates are biologically
benign by taking a closer look at the challenges
that climate change poses for tropical species. In
the tropics, successful latitudinal range shifts ap-
Fig. 3. (A) Proportion of species in each of the four groups of Fig. 2 subject
to decline or disappearance in the lowlands (biotic attrition), faced by gaps
between current and projected elevational range (range-shift gaps), and ex-
posed to mountaintop extinction, given a 600-m upslope shift in all ranges.
Proportions sum to greater than 1 because some species belong in two
categories. (B) Cumulative number of species facing each of these three
challenges as a function of warming-driven range shifts. The x axis repre-
sents model parameter d, measured in meters of elevation range shift, on a
continuous scale of warming-driven isotherm shifts of up to 5°C (nearly
1000 m), the upper range of projections for Central America for this century
(32). The stairstep patterns are a consequence of sampling at discrete sites on
the gradient (5).
10 OCTOBER 2008VOL 322
on February 2, 2010
pear unlikely, putting the focus on elevational Download full-text
gradients, where range-shift gaps will develop
early for the great numbers of narrow-ranged
species adapted to higher temperatures to replace
those driven upslope by warming, raising the
possibility of substantial attrition in species rich-
ness in the tropical lowlands.
References and Notes
1. C. Parmesan, Annu. Rev. Ecol. Evol. Syst. 37, 637
2. R. Wilson, D. Gutierrez, J. Gutierrez, V. Monserrat, Glob.
Change Biol. 13, 1873 (2007).
3. J. Lenoir, J. Gegout, P. Marquet, P. de Ruffray, H. Brisse,
Science 320, 1768 (2008).
4. M. B. Bush, Glob. Ecol. Biogeogr. 11, 463 (2002).
5. Materials and methods are available as supporting
material on Science Online.
6. M. B. Bush, H. Hooghiemstra, in Climate Change and
Biodiversity, T. E. Lovejoy, L. Hannah, Eds. (Yale Univ.
Press, New Haven, CT, 2005), pp. 125–137.
7. J. A. Pounds, M. P. L. Fogden, K. L. Masters, in Climate
Change and Biodiversity, T. E. Lovejoy, L. Hanna, Eds.
(Yale Univ. Press, New Haven, CT, 2005), pp. 70–74.
8. C. L. Cardelús, R. K. Colwell, J. E. Watkins, J. Ecol. 94,
9. G. Brehm, R. K. Colwell, J. Kluge, Glob. Ecol. Biogeogr.
16, 205 (2007).
10. A. R. Wallace, Tropical Nature and Other Essays
(Macmillan, London, 1878).
11. T. Dobzhansky, Am. Sci. 38, 209 (1950).
12. L. Hannah, R. A. Betts, H. H. Shugart, in Tropical
Rainforest Responses to Climatic Change, M. B. Bush,
J. R. Flenley, Eds. (Springer Praxis, 2007), pp. 351–366.
13. J. R. Flenley, Clim. Change 39, 177 (1998).
14. M. B. Bush, M. R. Silman, D. H. Urrego, Science 303, 827
15. F. E. Mayle, D. J. Beerling, W. D. Gosling, M. B. Bush,
Philos. Trans. R. Soc. London B Biol. Sci. 359, 499
16. M. Bush, M. Silman, J. Quat. Sci. 19, 677 (2004).
17. D. D. Ackerly et al., Bioscience 50, 979 (2000).
18. J. Hansen et al., Proc. Natl. Acad. Sci. U.S.A. 103, 14288
19. C. A. Deutsch et al., Proc. Natl. Acad. Sci. U.S.A. 105,
20. D. A. Clark, S. C. Piper, C. D. Keeling, D. B. Clark, Proc.
Natl. Acad. Sci. U.S.A. 100, 5852 (2003).
21. K. J. Feeley, S. Joseph Wright, M. N. Nur Supardi,
A. R. Kassim, S. J. Davies, Ecol. Lett. 10, 461 (2007).
22. O. L. Phillips, S. L. Lewis, T. R. Baker, Y. Malhi, in Tropical
Rainforest Responses to Climatic Change, M. B. Bush,
J. R. Flenley, Eds. (Springer Praxis, 2007), pp. 317–332.
23. L. Miles, A. Grainger, O. Phillips, Glob. Ecol. Biogeogr.
13, 553 (2004).
24. S. A. Cowling et al., Philos. Trans. R. Soc. London B Biol.
Sci. 359, 539 (2004).
25. J. W. Williams, S. T. Jackson, J. E. Kutzbach, Proc. Natl.
Acad. Sci. U.S.A. 104, 5738 (2007).
26. M. B. Bush, M. R. Silman, C. McMichael, S. Saatchi, Philos.
Trans. R. Soc. London B Biol. Sci. 363, 1795 (2008).
27. R. L. Peters, J. D. S. Darling, Bioscience 35, 707 (1985).
28. G.-R. Walther et al., Nature 416, 389 (2002).
29. C. Ghalambor, R. Huey, P. Martin, J. Tewksbury, G. Wang,
Integr. Comp. Biol. 46, 5 (2006).
30. R. G. Pearson, T. P. Dawson, Glob. Ecol. Biogeogr. 12,
31. S. Williams, E. E. Bolitho, S. Fox, Proc. R. Soc. London B.
Biol. Sci. 270, 1887 (2003).
32. J. H. Christensen et al., in Climate Change 2007: Fourth
Assessment Report of the Intergovernmental Panel on
Climate Change, S. Solomon et al., Eds. (Cambridge Univ.
Press, Cambridge, 2007), pp. 847–940.
33. J. Kluge, M. Kessler, R. R. Dunn, Glob. Ecol. Biogeogr.
15, 358 (2006).
34. Supported by the Organization for Tropical Studies; the
Tropical Ecology Assessment and Monitoring (TEAM) project
of Conservation International; University of Connecticut
(R.K.C.and C.L.C.);UCLA(A.C.G.); U.S. NSF (DEB-0072702:
R.K.C., G.B., and J.T.L.; DEB-0640015: J.T.L.;
DEB-0639979: R.K.C.; Research Fellowship and
Dissertation Improvement Grant: C.L.C.); Sigma Phi,
Explorer’s Club, and Steven Vavra Plant Systematics Fund
(A.C.G.); and the Deutsche Forschungsgemeinschaft
(BR 2280/1-1: G.B.). We thank M. B. Bush, R. L. Chazdon,
D. A. Clark, R. R. Dunn, K. M. Kuhn, C. Rahbek, T. F. L. V. B.
Rangel,M.R.Silman,andour peerreviewersfor comments.
Supporting Online Material
Materials and Methods
Figs. S1 and S2
30 June 2008; accepted 2 September 2008
Impact of a Century of Climate Change
on Small-Mammal Communities in
Yosemite National Park, USA
Craig Moritz,1,2* James L. Patton,1,2Chris J. Conroy,1Juan L. Parra,1,2
Gary C. White,3Steven R. Beissinger1,4
We provide a century-scale view of small-mammal responses to global warming, without
confounding effects of land-use change, by repeating Grinnell’s early–20th century survey across
a 3000-meter-elevation gradient that spans Yosemite National Park, California, USA. Using
occupancy modeling to control for variation in detectability, we show substantial (~500 meters on
average) upward changes in elevational limits for half of 28 species monitored, consistent with the
observed ~3°C increase in minimum temperatures. Formerly low-elevation species expanded their
ranges and high-elevation species contracted theirs, leading to changed community composition at
mid- and high elevations. Elevational replacement among congeners changed because species’
responses were idiosyncratic. Though some high-elevation species are threatened, protection
of elevation gradients allows other species to respond via migration.
lthough human-driven global warming
(1) has changed phenology of species
and contributed to range expansions
documented (7–10). Models of future climate-
change scenarios predict large range shifts, high
global extinction rates, and reorganized commu-
ly uncertain (13, 14). Most studies of species’
responses span only a few decades—typically
from the 1960 or 1970s, which was a relatively
cool period, to the present. Such results can be
confounded by decadal-scale climate oscillations
(15) and landscape modification (8, 16). Further-
by false absences due to limited historic sampling
between sampling periods (17, 18).
We quantified the impact of nearly a century
nity of Yosemite National Park (YNP) in Cali-
fornia, USA, by resampling a broad elevational
transect (60 to 3300 m above sea level) that
Joseph Grinnell and colleagues surveyed from
1914 to 1920 (19) (Fig. 1). Their work docu-
of the ecological niche, the importance of tem-
perature as determinant of range boundaries, and
the notion that species respond uniquely to envi-
20th century records, the “Yosemite Transect”
was densely sampled across elevations (Fig. 1)
and is amply documented by specimens (n =
species and sampling sites. From daily trapping
records, we estimated detectability of species in
historical as well as current surveys, permitting
the unbiased estimation of species’ “absences”
from elevational bands in both periods (23). The
transect spans YNP, a protected landscape since
1890, and allowed us to examine long-term re-
sponses to climate change without confounding
elevations there has been localized vegetation
records pointed to substantial increase of the av-
erage minimum monthly temperature of 3.7°C
over the past 100 years, with notable increases
from 1910 to 1945 and from 1970 to the present
(15, 22) (fig. S1).
1Museum of Vertebrate Zoology, University of California,
Berkeley, CA 94720, USA.
Biology, University of California, Berkeley, CA 94720, USA.
3Department of Fish, Wildlife, and Conservation Biology,
Colorado State University, Fort Collins, CO 80523, USA.
4Department of Environmental Science, Policy and Man-
agement, University of California, Berkeley, CA 94720,
*To whom correspondence should be addressed. E-mail:
2Department of Integrative
VOL 322 10 OCTOBER 2008
on February 2, 2010