Climate Change, Human Impacts, and the
Resilience of Coral Reefs
T. P. Hughes,
* A. H. Baird,
D. R. Bellwood,
S. R. Connolly,
J. B. C. Jackson,
J. M. Lough,
S. R. Palumbi,
J. M. Pandolﬁ,
The diversity, frequency, and scale of human impacts on coral reefs are increasing to the
extent that reefs are threatened globally. Projected increases in carbon dioxide and
temperature over the next 50 years exceed the conditions under which coral reefs have
ﬂourished over the past half-million years. However, reefs will change rather than
disappear entirely, with some species already showing far greater tolerance to climate
change and coral bleaching than others. International integration of management
strategies that support reef resilience need to be vigorously implemented, and comple-
mented by strong policy decisions to reduce the rate of global warming.
oral reefs are critically important for
the ecosystem goods and services
they provide to maritime tropical and
subtropical nations (1). Yet reefs are in
serious decline; an estimated 30% are al-
ready severely damaged, and close to 60%
may be lost by 2030 (2). There are no
pristine reefs left (3–4 ). Local successes
at protecting coral reefs over the past
30 years have failed to reverse regional-
scale declines, and global management of
reefs must undergo a radical change in
emphasis and implementation if it is to
make a real difference. Here, we review
current knowledge of the status of coral
reefs, the human threats to them now and in
the near future, and new directions for re-
search in support of management of these
vital natural resources.
Until recently, the direct and indirect
effects of overfishing and pollution from
agriculture and land development have
been the major drivers of massive and ac-
celerating decreases in abundance of coral
reef species, causing widespread changes in
reef ecosystems over the past two centuries
(3–5). With increased human populations
and improved storage and transport sys-
tems, the scale of human impacts on reefs
has grown exponentially. For example, mar-
kets for fishes and other natural resources
have become global, supplying demand for
reef resources far removed from their tropical
sources (6) (Fig. 1). On many reefs, reduced
stocks of herbivorous fishes and added nutri-
ents from land-based activities have caused
ecological shifts, from the original domi-
nance by corals to a preponderance of fleshy
seaweed (5, 7 ). Importantly, these changes to
reefs, which can often be managed success-
fully at a local scale, are compounded by the
more recent, superimposed impacts of global
The link between increased greenhouse
gases, climate change, and regional-scale
bleaching of corals, considered dubious by
many reef researchers only 10 to 20 years
ago (8), is now incontrovertible (9, 10).
Moreover, future changes in ocean chemis-
try due to higher atmospheric carbon diox-
ide may cause weakening of coral skeletons
and reduce the accretion of reefs, especially
at higher latitudes (11). The frequency and
intensity of hurricanes (tropical cyclones,
typhoons) may also increase in some re-
gions, leading to a shorter time for recovery
between recurrences (10). The most press-
ing impact of climate change, however, is
episodes of coral bleaching and disease that
have already increased greatly in frequency
and magnitude over the past 30 years
Centre for Coral Reef Biodiversity, James Cook Uni-
versity, Townsville, Qld 4811, Australia.
tal Protection Agency, Old Quarantine Station, Cape
Pallarenda, Townsville, QLD 4810, Australia.
ment of Systems Ecology, Stockholm University, SE-
106 91 Stockholm, Sweden.
Center for Population
Biology, Division of Biological Sciences, Section of
Evolution and Ecology, 1 Shields Avenue, University of
California, Davis, CA 95616, USA.
Centre for Marine
Studies, University of Queensland, St Lucia, QLD
Scripps Institution of Oceanography,
University of California San Diego, La Jolla, CA 92093,
Smithsonian Tropical Research Institute, Box
2070, Balboa, Republic of Panama.
for Atmospheric Research, Post Ofﬁce Box 3000,
Boulder, CO 80307, USA.
Australian Institute of Ma-
rine Sciences, PMB #3, Townsville, QLD 4810, Austra-
Great Barrier Reef Marine Park Authority, Post
Ofﬁce Box 1379, Townsville QLD 4810, Australia.
Department of Biological Sciences, Stanford Univer-
sity, Hopkins Marine Station, Paciﬁc Grove, CA 93950,
Department of Paleobiology, Smithsonian In-
stitution, Post Ofﬁce Box 37012, National Museum of
Natural History, Washington, DC 20013, USA.
partment of Zoology, The Natural History Museum,
Cromwell Road, London SW7 5BD, UK.
of Biological Sciences, Stanford University, Stanford,
CA 94305, USA.
*To whom correspondence should be addressed. E-
Fig. 1. Map of the Indo-
Paciﬁc Oceans showing
the scale of (A) the live-
ﬁsh trade, servicing res-
taurants in the burgeon-
ing cities of southeast
Asia, and (B) the aquari-
um tropical-ﬁsh trade
with major markets in
the United States and
www.sciencemag.org SCIENCE VOL 301 15 AUGUST 2003 929
Bleaching, Acclimation, and Adaptation
Regional-scale coral bleaching is strongly as-
sociated with elevated temperatures, particu-
larly during recurrent ENSO (El Ninˇo–
Southern Oscillation) events (8). Stressed,
overheated corals expel most of their pig-
mented microalgal endosymbionts, called
zooxanthellae, and become pale or white. If
thermal stress is severe and prolonged, most
of the corals on a reef may bleach, and many
may die. A popular model (9) shows an in-
variant bleaching “threshold” at ⬃1°C above
mean summer maximum temperatures. This
threshold will be chronically exceeded as
temperatures rise over the next 50 years, lead-
ing to predictions of massive losses of all
corals (Fig. 2A). This model is based on two
simplifying assumptions: that all corals re-
spond identically to thermal stress, and that
corals and their symbionts have inadequate
phenotypic or genetic capabilities for adapt-
ing rapidly to changes in temperature. Below,
we challenge the conventional understanding
of these key issues.
Bleaching is conspicuously patchy (15–17),
providing clear empirical evidence of the ab-
sence of a single bleaching threshold for all
locations, times, or species (contrary to the
conventional model depicted in Fig. 2A). Con-
sequently, bleached and unbleached corals are
often encountered side by side (Figs. 3A and
4B). The sources of this variation are poorly
understood and have been variously attributed
to extrinsic environmental patchiness (e.g. tem-
perature, light, turbulence), as well as intrinsic
differences (phenotypic and genetic) among
corals and their microalgal symbionts (15–19).
Whatever the mechanisms, bleaching thresh-
olds are more realistically visualized as a broad
spectrum of responses (Fig. 2B). Furthermore,
bleaching susceptibilities may also change over
time as a result of phenotypic and genetic re-
sponses (Fig. 2C). In particular, substantial geo-
graphic variation in bleaching thresholds within
coral species provides circumstantial evidence
for ongoing evolution of temperature tolerance.
Average summer water temperatures
differ enormously within the geographic
boundaries of a typical coral species’
range. Based on our current knowledge of
taxonomy, the median latitudinal extent of
coral species in the Indo-Pacific is 56° (20),
with many species’ ranges straddling the
equator and extending to or beyond the
limits of reef growth (at ⬃30°N and
⬃30°S) where water temperatures are
much cooler (Fig. 3B). Similarly, the geo-
graphic extent of 35% of coral species in
the Arabian Gulf (where the mean summer
maximum is 36°C) (21) also includes Lord
Howe Island (24°C), the southernmost cor-
al reef in the Pacific Ocean (Fig. 3C).
Importantly, corals in the Arabian Gulf do
not bleach until they experience tempera-
tures that are extreme for that location, well
over 10°C higher than summer maxima in
cooler regions elsewhere in the same spe-
cies’ ranges, providing circumstantial evi-
dence of local adaptation. Furthermore, the
lower bleaching threshold in cooler loca-
tions implies that there is strong selection
for corals and their zooxanthellae to evolve
thresholds that are near, but not too far
beyond, the expected upper temperature at
that location. This pattern points to a po-
tential trade-off between the risk of mortal-
ity from extreme temperatures versus a
high cost of thermal protective mechanisms
(e.g., antioxidant enzymes, heat shock, or
photoprotective proteins and pigments).
An emerging area of research points to the
importance of genetic variation as a determi-
nant of bleaching responses in both corals
and zooxanthellae. Corals exhibit high levels
of genetic diversity, as expected for species
with large population sizes and prodigious
sexual reproduction (22). Similarly, zooxan-
thellae (Symbiodinium spp.) cluster into a
number of groups (based on cladistic analysis
of DNA sequences), with seven clades being
recognized so far, comprising many species
(19). This recent finding raises the issue of
current and future patterns in the distribution
and relative abundance of zooxanthellae
clades. A hypothesis that bleaching is “adap-
tive,” increasing coral fitness by facilitating
expulsion of susceptible zooxanthellae spe-
cies and uptake of more resistant ones (23),
has not been supported by observations on
the fate of bleached corals. Bleaching is more
accurately described as a stress response,
which is often followed by high mortality,
reduced growth rates, and lower fecundity
(16, 24). Although adult corals may acquire a
previously undetected clade under experi-
mental conditions (25, 26), a change in the
relative proportions of zooxanthellae as a re-
sult of bleaching, like similar rearrangements
of coral assemblages (Fig, 3B), does not nec-
essarily indicate that any evolutionary re-
sponse has occurred.
A major concern is that the accelerating
rate of environmental change could exceed
the evolutionary capacity of coral and
zooxanthellae species to adapt. A common
view is that corals are too long-lived to
evolve quickly, and that geographic differ-
ences in temperature tolerances have
evolved over much longer time frames than
the decadal scale of current changes in
climate. Although some corals are indeed
very long-lived, sexual maturity is reached
within 3 to 5 years and most species at all
depths rarely live longer than 20 years (27).
Nonetheless, highly skewed fecundity dis-
tributions (where a few very large, old
individuals swamp the gene pool), strongly
overlapping generations, and high levels of
asexual reproduction are common traits
that are likely to retard rapid evolution in
many coral species. Although mortality
rates from bleaching events are often very
high, we know virtually nothing about how
much selection this exerts or the heritability
of physiological traits in corals. Further-
more, adaptive evolution could be limited if
traits under selection are negatively genet-
ically correlated (28) or if gene flow is high
enough to preclude local adaptation. On the
other hand, high gene flow or connectivity
will promote resilience and recovery from
recurrent bleaching. The available evidence
indicates that rates of gene flow in corals
vary substantially among species (22, 29),
which implies that their differential ability
to migrate in response to climate change
and to adapt will result in further changes
to community structure beyond the imme-
diate effect of selective mortality caused by
severe bleaching. In contrast, subpopula-
tions on isolated islands or archipelagoes
(e.g., Hawaii and Bermuda) may represent
genetic outposts for virtually all coral reef
species, with little input from other, distant
Fig. 2. (A) A model showing a constant coral
bleaching threshold, which is likely to be chron-
ically exceeded in the future as oceanic tem-
peratures increase (9, 14). (B) An alternative
model that incorporates differences in bleach-
ing thresholds (e.g., among species, depth, and
locations), indicated by parallel lines. (C)A
more realistic scenario where changes in
thresholds also occur over time, attributable to
acclimation and evolution.
15 AUGUST 2003 VOL 301 SCIENCE www.sciencemag.org930
localities. If isolated reefs bleach, recovery
is likely to be far slower than in more
central, interconnected populations.
Lessons from the Past: The Geological
The geological record provides the only
source of data on long-term effects of climate
change on coral reef species and assemblages
(30, 31). Many extant species of corals ex-
tend backwards in time to the Pliocene [1.8 to
5.3 million years ago (Ma)], and most scler-
actinian genera originated in the Eocene to
Miocene (55.0 to 5.3 Ma) (32). Extant species
have dominated modern reefs for the past
half-million years, providing an invaluable
baseline long before human impacts began
(3, 4 ). New assessment of past climates has
revealed unexpectedly rapid shifts over de-
cades or less, especially at high latitudes,
with ice-age transitions being linked to abrupt
changes in the North Atlantic circulation
(33). Further rapid climatic changes may
have also occurred at lower latitudes in
warmer periods since the last glacial maxi-
mum (34). Consequently, there is now some
uncertainty about the speed of expected cli-
mate change relative to the past, although we
can be certain nonetheless that the projected
increases in carbon dioxide and temperature
over the next 50 years will substantially and
very rapidly exceed the conditions under
which coral reefs have flourished over the
past half-million years (10).
During the Pleistocene and Holocene,
many extant species of tropical and subtrop-
ical organisms underwent dramatic shifts in
geographic range in response to periods of
warming and cooling (35, 36). Some species
migrated faster than others, producing rapid
shifts in species composition, especially near
faunal boundaries (35). For corals, range
boundaries of extant coral species in the
warm Late Pleistocene extended up to 500
km further south along the western Australia
coastline (to 33°S) than they do today (37).
Closer to the center of their geographic range,
however, coral diversity and species presence
or absence in eastern Papua New Guinea
changed remarkably little during nine reef-
building intervals from 125 to 30 ka (31). On
a regional scale, these same species under-
went dramatic changes in distribution and
abundance as Quaternary glacial-interglacial
cycles caused sea level to repeatedly flood
and drain from continental shelves and oce-
anic islands (38). Many marine species ex-
hibit a genetic legacy of these range shifts,
local extinctions and expansions, and the
marked population fluctuations caused by
past climatic variation (29, 39, 40). Based on
this past history, we can expect regional and
global-scale disruption to coral reefs due to
climate change to accelerate markedly in
coming decades. Already, relative abundanc-
es of corals and of other organisms are chang-
ing rapidly in response to the filtering effect
of differential mortality (from bleaching and
other, more local human impacts) and differ-
ences in rates of recovery of species from
recurrent mortality events (16, 17, 41, 42).
There are two major differences, however,
between current climate-driven changes and
the recent past. First, because the oceans
today are already at a high sea-level stand, the
projected rise [0.1 to 0.9 m in the next 100
years (10)] will be very small compared with
sea-level changes during the Pleistocene.
Second, unlike the past, the response of reef-
dwelling species to projected climatic trends
will be profoundly influenced by people. As
outlined below, human impacts and the in-
creased fragmentation of coral reef habitat
have preconditioned reefs, undermining reef
resilience and making them much more sus-
ceptible to future climate change.
Managing Coral Reef Resilience
Clearly, the capacity of coral reef ecosystems to
continue to generate the valuable goods and
services (on which human welfare depends) has
to be better understood and more actively man-
aged. Sustaining this capacity requires im-
proved protection of coral reef resilience (43).
Marine protected areas (MPAs) are currently
the best management tool for conserving coral
reefs and many other marine systems (44, 45).
MPAs range from ineffective “paper parks,” to
multiple-use areas with varying degrees of
protection, to marine reserves, or no-take
areas (NTAs). NTAs provide the most ef-
fective protection for extractive activities
such as fishing, affording a spatial refuge
for a portion of the stock from which larvae
and adults can disperse to adjoining ex-
ploited areas (44, 45).
NTAs, when properly supported and po-
liced, are effective in preserving fish stocks
because they change human behavior. They
do not, however, prevent or hold back warm
water, or stop bleaching. For example, in
1998, the biggest and most destructive
bleaching event to date killed an estimated
16% of the world’s corals, including reefs in
the western Pacific, Australia, and Indian
Ocean that are widely regarded as the best
managed and most “pristine” in the world (2).
Fig. 3. (A) Differential bleaching responses of nine species of corals in Raiatea, French Polynesia,
during May 2002. (a, Acropora anthocercis;b,A. retusa;c,Montipora tuberculosa;d,Pocillopora
verrucosa;e,M. caliculata;f,Leptastrea transversa;g,P. eydouxi;h,P. meandrina;i,L. bewickensis;
j, Porites lobata;k,L. purpurea.(B) Latitudinal extents of Indo-Paciﬁc reef corals, measured from the
northern- and southernmost point in their range. (C) Geographic range boundaries of 24 species of
Indo-Paciﬁc corals found in the Persian Gulf and at Lord Howe Island off Australia, where average
maximum summer temperatures differ by 12°C. The coloring shows temperatures in the Southern
Hemisphere summer of 1997/1978, when unprecedented mass bleaching occurred (2).
www.sciencemag.org SCIENCE VOL 301 15 AUGUST 2003 931
If NTAs do not provide a refuge from bleach-
ing, then how can they help protect coral
reefs from climate change? Overfishing, par-
ticularly of herbivorous parrotfish and sur-
geonfish, affects more than just the size of
harvestable stocks—it alters the entire dy-
namics of a reef (3–5, 46 ). Reduced her-
bivory from overfishing, increased levels of
disease, and excess nutrients can impair the
resilience of corals and prevent their recovery
following acute-disturbance events like cy-
clones or bleaching, leading instead to a
phase shift to algal-dominated reefs (Fig. 4, D
to F). Resilience is also eroded by chronic
human impacts that cause persistently elevat-
ed rates of mortality and reduced recruitment
of larvae (7, 12, 41, 43).
Although climate change is by definition a
global issue, local conservation efforts can
greatly help in maintaining and enhancing
resilience and in limiting the longer-term
damage from bleaching and related human
impacts. Managing coral reef resilience
through a network of NTAs, integrated with
management of surrounding areas, is clearly
essential to any workable solution. This re-
quires a strong focus on reducing pollution,
protecting food webs, and managing key
functional groups (such as reef constructors,
herbivores, and bioeroders) as insurance for
sustainability (7, 46).
NTAs also act to spread risk, whereby
areas that escape damage can act as sources
of larvae to aid recovery of nearby affected
areas (47 ). This highly desirable property of
NTAs raises the issue of how close they need
to be to promote connectivity—the migration
of larvae and/or adults—between them (44,
45). Critically, coral reef organisms, includ-
ing different species of corals, vary greatly in
their larval biology and potential for dispersal
(22, 29). The clear implication is that NTAs
must be substantially more numerous and
closer together than they are currently to
protect species with limited dispersal capabil-
ities. Furthermore, isolated reefs that are
largely self-seeding are unlikely to be pro-
tected by distant NTAs, and therefore will be
much less resilient to climate change.
Research and Management Challenges
Coral reefs are highly productive hotspots of
biodiversity that support social and economic
development. Their protection, therefore, is a
socioeconomic imperative, as well as an en-
vironmental one. Global warming, coupled
with preexisting human impacts, is a grave
threat that has already caused substantial
damage. However, the available evidence in-
dicates that, at a global scale, reefs will un-
dergo major changes in response to climate
change rather than disappear entirely.
There is, nonetheless, great uncertainty
whether the present economic and social ca-
pacity of coral reefs can be maintained. To
limit the damage, emerging management
strategies based on greatly expanded net-
works of NTAs, coupled with stronger pro-
tection of adjacent habitats, need to be vigor-
ously implemented. NTAs are unlikely to
prevent mortality of corals from bleaching,
but they will facilitate a partial recovery of
reefs that are reconfigured and populated by a
subset of resistant species and genotypes.
NTAs are not a panacea; their implementa-
tion needs to be complemented by heightened
protection of adjacent areas and by strong
international policy decisions to reduce the
rate of global warming.
Research in support of reef management
urgently needs to increase the scale of experi-
ments, sampling, and modeling to match the
scale of impacts and key biological processes
(e.g., dispersal, bleaching, and overfishing) and
go beyond the current emphasis on routine
monitoring and mapping. Indeed, most coral
reef research is parochial and short-term, and
provides little insight into global or longer-term
changes. For example, current knowledge of
biogeographic-scale patterns on reefs is based
on species presence or absence at local sites and
pays scant attention to temporal, regional, or
global patterns of relative abundance or func-
tional attributes of species (48) that could be
exploited for management of resilience. Simi-
larly, studies of intergenerational (genetic) re-
sponses to climate change (28) are urgently
needed for reef organisms, particularly corals
and zooxanthellae. Another crucial area for fu-
ture work is genetic dissection of population
structure and modeling of connectivity, which
could incorporate many of the unusual life-
history traits of clonal organisms, selection co-
efficients based on mortality from bleaching,
and experimental measurements of heritabili-
ties. Emerging research on marine reserves and
how they work to protect harvested stocks and
Fig. 4. (A) Aquarium ﬁsh, such as Anampses lennardi from northwest Australia, are often
endemic species and susceptible to overharvesting. (B) A bleached colony of Acropora nasuta
(bottom), and unbleached Pocillopora meandrina (top), showing contrasting responses to
thermal stress. (C) A bleached monospeciﬁc stand of the staghorn coral Acropora formosa.(D)
Parrotﬁshes, such as Scarus ferrugineus, are important herbivores. (E) Parrotﬁsh grazing-scars.
(F) Macroalgae (top) and overgrowth of corals (bottom) are promoted by overﬁshing of
herbivores and degraded water quality.
15 AUGUST 2003 VOL 301 SCIENCE www.sciencemag.org932
spread risk (44, 45) also needs to be expanded
and applied specifically to the tropics. These
approaches must be integrated with socioeco-
nomic aspects of coral reef resilience, incorpo-
rating adaptive management systems that oper-
ate locally, regionally, and globally.
International integration and scaling-up of
reef management is an urgent priority (2).
Ecological modeling studies indicate that, de-
pending on the level of exploitation outside
NTAs, at least 30% of the world’s coral reefs
should be NTAs to ensure long-term protec-
tion and maximum sustainable yield of ex-
ploited stocks (49, 50). Yet, even in affluent
countries, such as the United States and Aus-
tralia, less than 5% of reefs today are NTAs.
Wealthy countries have an obligation to take
the lead in increasing the proportion of reefs
that are NTAs, while simultaneously control-
ling greenhouse-gas emissions.
References and Notes
1. F. Moberg, C. Folke, Ecol. Econ. 29, 215 (1999).
2. C. Wilkinson, Ed., Status of Coral Reefs of the World
(Australian Institute of Marine Science, Townsville,
3. J. B. C. Jackson et al., Science 293, 629 (2001).
4. J. M. Pandolﬁ et al., Science 301, 955 (2003).
5. T. P. Hughes, Science 265, 1547 (1994).
6. Y. J. Sadovy, A. C. J. Vincent, in Coral Reef Fishes,
Dynamics and Diversity in a Complex Ecosystem,P.F.
Sale, Ed. (Academic Press, San Diego, CA, 2002), pp.
7. T. McClanahan, N. Polunin, T. Done, in Resilience and
Behavior of Large-Scale Systems, L. H. Gunderson, L.
Pritchard Jr., Eds. (Island Press, Washington, DC,
8. P. W. Glynn, Coral Reefs 12, 1 (1993).
9. O. Hoegh-Guldberg, Mar. Freshw. Res. 50, 839
10. R. T. Watson and Core Writing Team, Eds., 3rd As-
sessment Report of the Inter-Governmental Panel on
Climate Change. Climate Change 2001: Synthesis
11. J. A. Kleypas et al., Science 284, 118 (1999).
12. N. Knowlton, Proc. Natl. Acad. Sci. U.S.A. 98, 5419
13. C. D. Harvell et al., Science 296, 2158 (2002).
14. NOAA Hotspot program, www.osdpd.noaa.gov/PSB/
15. R. Berkelmans, B. L. Willis, Coral Reefs 18, 219 (1999).
16. A. H. Baird, P. A. Marshall, Mar. Ecol. Prog. Ser. 237,
17. Y. Loya et al., Ecol. Lett. 4, 122 (2001).
18. S. L. Coles, B. E. Brown, Adv. Mar. Biol., 46, 183
19. N. Knowlton, F. Rohwer, Am. Nat., in press.
20. T. P. Hughes, D. R. Bellwood, S. R. Connolly, Ecol. Lett.
5, 775 (2002).
21. S. L. Coles, Y. H. Fadlallah, Coral Reefs 9, 231 (1991).
22. D. J. Ayre, T. P. Hughes, Evolution 54, 1590 (2000).
23. R. W. Buddemeier, D. Fautin, Bioscience 43,320
24. A. E. Douglas, Mar. Pollut. Bull. 46, 385 (2003).
25. A. C. Baker, Nature 411, 765 (2001).
26. W. W. Toller, R. Rowan, N. Knowlton, Biol. Bull. 201,
27. J. H. Connell, T. P. Hughes, C. C. Wallace, Ecol.
Monogr. 67, 461 (1997).
28. J. R. Etterson, R. G. Shaw, Science 294, 151 (2001).
29. J. A. H. Benzie, Am. Zool. 39, 131 (1999)
30. J. B. C. Jackson, Am. Zool. 32, 719 (1992).
31. J. M. Pandolﬁ, Paleobiology 22, 152 (1996).
32. B. R. Rosen, Geol. J. Spec. Iss. 11, 201 (1984).
33. W. S. Broecker, Earth Sci. Rev. 51, 137 (2000).
34. J. Overpeck, R. Webb, Proc. Natl. Acad. Sci. U.S.A. 97,
35. K. Roy, D. Jablonski, J. W. Valentine, Ecol. Lett. 4, 366
36. G. Paulay, Am. Malacol. Bull. 12, 45 (1996).
37. P. E. Playford, J. R. Soc. West. Aust. 66, 10 (1983).
38. D. C. Potts, Paleobiology 10, 48 (1984)
39. M. E. Hellberg, D. P. Balch, K. Roy, Science 292, 1707
40. H. A. Lessios, B. D. Kessing, J. S. Pearse, Evolution 55,
41. T. P. Hughes, J. H. Connell, Limnol. Oceanogr. 44, 932
42. M. Nystro¨m, C. Folke, Ecosystems 4, 406 (2001).
43. M. Nystro¨m, C. Folke, F. Moberg, Trends Ecol. Evol.
15, 413 (2000).
44. J. Lubchenco, S. Palumbi, S. Gaines, S. Andelman, Ecol.
Appl. 13, S3 (2003).
45. S. R. Palumbi, Ecol. Appl. 13, S146 (2003).
46. D. R. Bellwood, A. S. Hoey, J. H. Choat, Ecol. Lett. 6,
47. J. Bascompte, H. Possingham, J. Roughgarden, Am.
Nat. 159, 128 (2002).
48. D. R. Bellwood, T. P. Hughes, Science 292, 1532
49. A. Hastings, L. W. Botsford, Ecol. Appl. 13 (suppl.),
50. J. Roughgarden, P. R. Armsworth, in Ecology: Achieve-
ment and Challenge, M. C. Press, N. J. Huntly, S. Levin,
Eds. (Blackwell Science, Oxford, 2001), pp. 337–356 .
51. We thank James Cook University and The Queens-
land Government for funding a meeting of the au-
thors, and A. Green, N. Knowlton, and B. Willis for
providing comments on a draft. This is contribution
No. 078 of the Centre for Coral Reef Biodiversity at
James Cook University.
www.sciencemag.org SCIENCE VOL 301 15 AUGUST 2003 933