REVIEW

ing a mean temperature of 25°C during the past

420,000 years (Fig. 1B). The results show a tight
cluster of points that oscillate (temperature ±3°C;
Coral Reefs Under Rapid Climate carbonate-ion concentration ±35 mmol kg−1) be-
tween warmer interglacial periods that had lower
Change and Ocean Acidification carbonate concentrations to cooler glacial pe-
riods with higher carbonate concentrations. The
overall range of values calculated for seawater
O. Hoegh-Guldberg,1* P. J. Mumby,2 A. J. Hooten,3 R. S. Steneck,4 P. Greenfield,5 E. Gomez,6 pH is ±0.1 units (10, 11). Critically, where coral
C. D. Harvell,7 P. F. Sale,8 A. J. Edwards,9 K. Caldeira,10 N. Knowlton,11 C. M. Eakin,12 reefs occur, carbonate-ion concentrations over
R. Iglesias-Prieto,13 N. Muthiga,14 R. H. Bradbury,15 A. Dubi,16 M. E. Hatziolos17 the past 420,000 years have not fallen below
240 mmol kg−1. The trends in the Vostok ice
Atmospheric carbon dioxide concentration is expected to exceed 500 parts per million and global core data have been verified by the EPICA study
temperatures to rise by at least 2°C by 2050 to 2100, values that significantly exceed those of at (6), which involves a similar range of temperatures
least the past 420,000 years during which most extant marine organisms evolved. Under conditions and [CO2]atm values and hence extends the con-
expected in the 21st century, global warming and ocean acidification will compromise carbonate clusions derived from the Vostok record to at least
accretion, with corals becoming increasingly rare on reef systems. The result will be less diverse reef 740,000 years before the present (yr B.P.). Con-

Downloaded from www.sciencemag.org on December 19, 2007

communities and carbonate reef structures that fail to be maintained. Climate change also exacerbates ditions today ([CO2]atm ~380 ppm) are significantly
local stresses from declining water quality and overexploitation of key species, driving reefs increasingly shifted to the right of the cluster points represent-
toward the tipping point for functional collapse. This review presents future scenarios for coral reefs that ing the past 420,000 years. Sea temperatures are
predict increasingly serious consequences for reef-associated fisheries, tourism, coastal protection, and warmer (+0.7°C), and pH (− 0.1 pH units) and
people. As the International Year of the Reef 2008 begins, scaled-up management intervention and carbonate-ion concentrations (~210 mmol kg−1)
decisive action on global emissions are required if the loss of coral-dominated ecosystems is to be avoided. lower than at any other time during the past
420,000 years (Fig. 1B). These conclusions match
oral reefs are among the most biologically ing global warming and ocean acidification, may recent changes reported for measurements of ocean

C diverse and economically important eco-

systems on the planet, providing ecosys-
tem services that are vital to human societies and
be the final insult to these ecosystems. Here, we
review the current understanding of how anthro-
pogenic climate change and increasing ocean acid-
temperature, pH, and carbonate concentration (8).
In addition to the absolute amount of change, the
rate at which change occurs is critical to whether
industries through fisheries, coastal protection, ity are affecting coral reefs and offer scenarios for organisms and ecosystems will be able to adapt or
building materials, new biochemical compounds, how coral reefs will change over this century. The accommodate to the new conditions (11). Notably,
and tourism (1). Yet in the decade since the in- scenarios are intended to provide a framework for rates of change in global temperature and [CO2]atm
augural International Year of the Reef in 1997 (2), proactive responses to the changes that have over the past century are 2 to 3 orders of mag-
which called the world to action, coral reefs have begun in coral reef ecosystems and to provoke nitude higher than most of the changes seen in
continued to deteriorate as a result of human in- thinking about future management and policy the past 420,000 years (Table 1). Rates of change
fluences (3, 4). Rapid increases in the atmospheric challenges for coral reef protection. under both low (B1) and high (A2) Intergovern-
carbon dioxide concentration ([CO2]atm), by driv- mental Panel on Climate Change (IPCC) emission
Warming and Acidifying Seas scenarios are even higher, as are recent measure-
1
Centre for Marine Studies, The University of Queensland, The concentration of carbon dioxide in Earth’s ments of the rate of change of [CO2]atm (9). The
St. Lucia, 4072 Queensland, Australia. 2Marine Spatial
atmosphere now exceeds 380 ppm, which is only possible exceptions are rare, short-lived
Ecology Laboratory, School of BioSciences, University of
Exeter, Prince of Wales Road, Exeter EX4 4PS, UK. 3AJH more than 80 ppm above the maximum values spikes in temperature seen during periods such
Environmental Services, 4900 Auburn Avenue, Suite 201, of the past 740,000 years (5, 6), if not 20 million as the Younger Dryas Event (12,900 to 11,500 yr
Bethesda, MD 20814, USA. 4University of Maine, School years (7). During the 20th century, increasing B.P.) (12). Given that recent and future rates of
of Marine Sciences, Darling Marine Center, Walpole, ME [CO2]atm has driven an increase in the global change dwarf even those of the ice age transitions,
04573, USA. 5The Chancellery, University of Queens-
land, St. Lucia, 4072 Queensland, Australia. 6Marine Science oceans’ average temperature by 0.74°C and sea when biology at specific locations changed dramat-
Institute, University of the Philippines, Diliman, Quezon City, level by 17 cm, and has depleted seawater car- ically, it is likely that these changes will exceed the
Philippines. 7Ecology and Evolutionary Biology, E321 Corson bonate concentrations by ~30 mmol kg−1 seawater capacity of most organisms to adapt.
Hall, Cornell University, Ithaca, NY 14853, USA. 8International and acidity by 0.1 pH unit (8). Approximately
Network on Water, Environment and Health, United Nations
University, 50 Main Street East, Hamilton, Ontario L8N 1E9, 25% (2.2 Pg C year−1) of the CO2 emitted from Ocean Acidification and Reef Accretion
Canada. 9School of Biology, Ridley Building, University of all anthropogenic sources (9.1 Pg C year−1) cur- Many experimental studies have shown that a
Newcastle, Newcastle upon Tyne, NE1 7RU, UK. 10Department of rently enters the ocean (9), where it reacts with doubling of pre-industrial [CO2]atm to 560 ppm
Global Ecology, Carnegie Institution of Washington, 260 water to produce carbonic acid. Carbonic acid decreases coral calcification and growth by up to
Panama Street, Stanford, CA 94305, USA. 11National Museum
of Natural History, Smithsonian Institution, Washington, DC
dissociates to form bicarbonate ions and protons, 40% through the inhibition of aragonite formation
20013, USA. 12National Oceanic and Atmospheric Administra- which in turn react with carbonate ions to produce (the principal crystalline form of calcium carbonate
tion, Coral Reef Watch, E/RA31, 1335 East West Highway, Silver more bicarbonate ions, reducing the availability of deposited in coral skeletons) as carbonate-ion con-
Spring, MD 20910–3226, USA. 13Unidad Académica Puerto carbonate to biological systems (Fig. 1A). De- centrations decrease (13). Field studies confirm that
Morelos, Instituto de Ciencias del Mar y Limnología, Universidad
Nacional Autónoma de México, Apdo. Postal 1152, Cancún
creasing carbonate-ion concentrations reduce the carbonate accretion on coral reefs approaches zero
77500 QR, México. 14Wildlife Conservation Society, 2300 rate of calcification of marine organisms such as or becomes negative at aragonite saturation values
Southern Boulevard, Bronx, New York, NY 10460, USA. reef-building corals, ultimately favoring erosion of 3.3 in today’s oceans (Fig. 4), which occurs
at ~200 mmol kg−1 seawater (7, 10).
15
Resource Management in Asia-Pacific Program, Australian when [CO2]atm approaches 480 ppm and carbonate-
National University, Canberra, 0200 Australia. 16Institute of We used global [CO2]atm and temperature ion concentrations drop below 200 mmol kg−1 in
Marine Sciences, University of Dar es Salaam, Tanzania. 17Envi-
ronment Department, MC5-523, The World Bank, 1818 H data from the Vostok Ice Core study (5) to ex- most of the global ocean (10, 13). These find-
Street, NW, Washington, DC 20433, USA. plore the ocean temperature and carbonate-ion ings are supported by the observation that reefs
*To whom correspondence should be addressed. E-mail: concentration (10) seen today relative to the re- with net carbonate accretion today (Fig. 4, 380 ppm)
oveh@uq.edu.au cent past for a typical low-latitude sea maintain- are restricted to waters where aragonite saturation

www.sciencemag.org SCIENCE VOL 318 14 DECEMBER 2007 1737

 REVIEW

Downloaded from www.sciencemag.org on December 19, 2007

Fig. 1. (A) Linkages between the buildup of atmospheric CO2 and the slowing (25°C), and total alkalinity (2300 mmol kg−1). Further details of these
of coral calcification due to ocean acidification. Approximately 25% of the calculations are in the SOM. Acidity of the ocean varies by ± 0.1 pH units
CO2 emitted by humans in the period 2000 to 2006 (9) was taken up by the over the past 420,000 years (individual values not shown). The thresholds for
ocean where it combined with water to produce carbonic acid, which releases a major changes to coral communities are indicated for thermal stress (+2°C) and
proton that combines with a carbonate ion. This decreases the concentration of carbonate-ion concentrations ([carbonate] = 200 mmol kg−1, approximate
carbonate, making it unavailable to marine calcifiers such as corals. (B) Tem- aragonite saturation ~Waragonite = 3.3; [CO2]atm = 480 ppm). Coral Reef
perature, [CO2]atm, and carbonate-ion concentrations reconstructed for the past Scenarios CRS-A, CRS-B, and CRS-C are indicated as A, B, and C, respectively,
420,000 years. Carbonate concentrations were calculated (54) from CO2 atm and with analogs from extant reefs depicted in Fig. 5. Red arrows pointing
temperature deviations from today’s conditions with the Vostok Ice Core data set progressively toward the right-hand top square indicate the pathway that is
(5), assuming constant salinity (34 parts per trillion), mean sea temperature being followed toward [CO2]atm of more than 500 ppm.

exceeds 3.3 (10). Geological studies report a no- have deleterious consequences for reef ecosys- density. However, erosion could be promoted
table gap in the fossil record of calcified organisms, tems. First, the most direct response is a decreased by the activities of grazing animals such as
including reef-building corals (14) and calcare- linear extension rate and skeletal density of coral parrotfish, which prefer to remove carbonates
ous algae (15), during the early Triassic when colonies. The massive coral Porites on the Great from lower-density substrates. Increasingly
[CO2]atm increased dramatically and reached levels Barrier Reef has shown reductions in linear ex- brittle coral skeletons are also at greater risk
at least five times as high as today’s (16). Phylo- tension rate of 1.02% year−1 and in skeletal den- of storm damage (21); thus, if rates of erosion
genetic studies suggest that corals as a group sity of 0.36% year−1 during the past 16 years (20). outstrip calcification, then the structural com-
survived the Permian-Triassic extinction event (14) This is equivalent to a reduction of 1.29% year−1 plexity of coral reefs will diminish, reducing
but may have done so through forms lacking cal- or a 20.6% drop in growth rate (the product of habitat quality and diversity. A loss of struc-
cified skeletons (17, 18). Although Scleractinian linear extension rate and skeletal density) over the tural complexity will also affect the ability of
(modern) corals arose in the mid-Triassic and lived 16-year period. While at present it is not possible reefs to absorb wave energy and thereby impairs
under much higher [CO2]atm, there is no evidence to confidently attribute the observed decrease in coastal protection.
that they lived in waters with low-carbonate growth and calcification to ocean acidification, it Third, corals may maintain both skeletal growth
mineral saturation. Knoll et al. succinctly state that is consistent with changes reported in oceanic pH and density under reduced carbonate saturation
“it is the rapid, unbuffered increase in [CO2]atm and carbonate-ion concentrations. by investing greater energy in calcification. A
and not its absolute values that causes impor- Second, corals may maintain their physical likely side effect of this strategy is the diversion
tant associated changes such as reduced [CO32− ], extension or growth rate by reducing skeletal of resources from other essential processes, such
pH, and carbonate as reproduction, as
saturation of sea wa- Table 1. Rates of change in atmospheric CO2 concentration ([CO2 ]atm, ppm/100 years) and global temperature seen in chronic stress
ter” (19). The rate of (°C/100 years) calculated for the past 420,000 yr B.P. using the Vostok Ice Core data (5) and compared to changes (21), which could ul-
[CO2]atm change is over the last century and those projected by IPCC for low-emission (B1) and high-emission (A2) scenarios (8). Rates timately reduce the
were calculated for each successive pair of points in the Vostok Ice Core record by dividing the difference between two
critical given that larval output from
sequential values (ppm or °C) by the time interval between them. Rates were then standardized to the change seen
modern genotypes reefs and impair the
over 100 years. Ratios of each rate relative to the mean rate seen over the past 420,000 years are also calculated.
and phenotypes of potential for recolo-
corals do not appear [CO2]atm Ratio (relative to Temperature Ratio (relative to nization following
to have the capacity Period (ppm century−1) past 420,000 years) (°C century−1) past 420,000 years) disturbances.
to adapt fast enough
to sudden environ- Past 420,000 years (99% 0.07 + 0.223 1 0.01 + 0.017 1 Resilience and
confidence interval; n = 282) Tipping Points
mental change.
Reef-building Past 136 years (1870–2006) 73.53 1050 0.7 70
Maintaining ecologi-
corals may exhibit IPCC B1 scenario: 550 ppm 170 2429 1.8 180
cal resilience is the
at 2100
several responses central plank of any
to reduced calcifi- IPCC A2 scenario: 800 ppm 420 6000 3.4 420
strategy aiming to
at 2100
cation, all of which preserve coral reef

1738 14 DECEMBER 2007 VOL 318 SCIENCE www.sciencemag.org

 REVIEW
ecosystems. Ecological resilience (4) 90
grazers like the sea urchin, Diade-
is a measure of the rate at which an ma antillarum, which essentially
ecosystem returns to a particular state disappeared from Caribbean reefs

Equilibrial coral cover after 50 years (%)

80 Coral−dominated stable equilibrium
(e.g., coral-dominated communities) in the early 1980s after a massive
after a perturbation or disturbance 70 Unstable equilibrium with 20% disease outbreak, highly produc-
(e.g., hurricane impacts). Recent reduction in coral linear extension rate tive reefs would likely require the
changes to the frequency and scale 60 Combinations of coral highest levels of parrotfish grazing
of disturbances such as mass coral cover and grazing that (i.e., ~ 40% of the reef being grazed)
bleaching, disease outbreaks, and 50 permit reef recovery for a reef to be able to recover from
destructive fishing, coupled with a between disturbance disturbance. The loss of ecological
events under reduced
decreased ability of corals to grow 40
coral growth resilience occurs because coral
and compete, are pushing reef ecosys- cover increases more slowly after
tems from coral- to algal-dominated 30
Unstable equilibrium
disturbance and competitive inter-
states (4, 22). If pushed far enough, with current coral actions with macroalgae become
the ecosystem may exceed a “tipping 20 linear extension rate more frequent and longer in dura-
point” (22) and change rapidly into tion (Fig. 3) (23) (table S1). Al-
an alternative state with its own in- 10 though the ecological model only
Algal−dominated stable equilibrium
herent resilience and stability, often represents a single Caribbean reef

Downloaded from www.sciencemag.org on December 19, 2007

making the possibility of returning 0 habitat in a very productive physical
0.1 0.2 0.3 0.4 0.5 0.6
to a coral-dominated state difficult. Grazing (proportion of reef grazed in 6 months) environment and has not incor-
To examine the ecological impli- porated several other putative con-
cations of the 20.6% reduction in Fig. 2. Reduction in the resilience of Caribbean forereefs as coral growth sequences of acidification such as
coral growth rate that Cooper et al. rate declines by 20%. Reef recovery is only feasible above or to the right of a loss of rugosity, sensitivity analy-
measured in Great Barrier Reef the unstable equilibria (open squares). The “zone of reef recovery” (pink) is ses reveal that changes to coral
Porites (20), we simulated a similar therefore more restricted under reduced coral growth rate and reefs require growth rate have a relatively large
reduction in the growth of massive higher levels of grazing to exhibit recovery trajectories. impact on model predictions (22),
brooding and spawning corals on exposed Carib- rium values of coral cover were plotted to illustrate and therefore the conclusions of a reduction in
bean forereefs specifically to investigate what hap- potential resilience (Fig. 2). The unstable equilibria resilience appear to be robust.
pens to the balance between corals and macroalgae represent thresholds, and for recovery to outweigh
in a system of high primary production (Fig. 2). mortality reefs must lie either above or to the right Thermal Stress, Synergies, and
The ecological model (22) simulated a 50-year time of the threshold. For example, if coral cover is low Ecological Feedback Loops
series for a wide range of initial coral cover and (<5%), the intensity of fish grazing on benthic algal The sensitivity of corals and their endosymbiotic
grazing rates by fish on benthic algae while hold- competitors needed to shift the reef into a state dinoflagellates (Symbiodinium spp.) to rising
ing all other factors (e.g., nutrient concentrations) where recovery is possible (i.e., to the right or ocean temperatures has been documented ex-
constant. Each time series revealed the underlying above the unstable equilibrium) moves from 30% tensively (24). Symbiodinium trap solar energy
trajectory of coral recovery, stasis, or degradation to almost a half of the reef having to be grazed. and nutrients, providing more than 95% of the
between major disturbances, and the final equilib- This implies that in the absence of invertebrate metabolic requirements of the coral host, which

Fig. 3. Ecological feedback processes on a coral reef showing pathways of first factor has a negative (decreasing) influence on the box indicated. Green
disturbance caused by climate change. Impact points associated with ocean arrows denote positive (increasing) relationships. Over time, the levels of
acidification (e.g., reduced reef rugosity, coralline algae) are indicated by the factors in hexagonal boxes will increase, whereas those in rectangular boxes
blue arrows, and impact points from global warming (e.g., bleached and will decline. Boxes with dashed lines are amenable to local management
dead corals) by the red arrows. Boxes joined by red arrows denote that the intervention.

www.sciencemag.org SCIENCE VOL 318 14 DECEMBER 2007 1739

 REVIEW
is consequently able to maintain high calcification how long we stay within each of the three sce- diversity of corals on reefs are likely to decline,
rates. When temperatures exceed summer maxima narios will depend on the CO2 emission rate, with leading to vastly reduced habitat complexity and
by 1° to 2°C for 3 to 4 weeks, this obligatory each scenario highlighting the context against which loss of biodiversity (31), including losses of coral-
endosymbiosis disintegrates with ejection of the management and policy actions must be devised. associated fish and invertebrates (32).
symbionts and coral bleaching (24). Bleaching and If conditions were stabilized at the present Coralline algae are a key settlement substrate
mortality become progressively worse as thermal [CO2]atm of 380 ppm, that is, Coral Reef Scenario for corals, but they have metabolically expensive
anomalies intensify and lengthen (24). Indeed, CRS-A (Figs. 1B and 5A), coral reefs will con- high-magnesium calcite skeletons that are very
mass coral bleaching has increased in intensity tinue to change but will remain coral dominated sensitive to pH (33). Hence, coral recruitment may
and frequency in recent decades (24–27). At the and carbonate accreting in most areas of their be compromised if coralline algal abundance de-
end of the International Year of the Reef in 1997, current distribution. Local factors—i.e., those not clines. Coral loss may also be compounded by an
mass bleaching spread from the Eastern Pacific to directly related to global climate change, such as increase in disease incidence (34). Ultimately, the
most coral reefs worldwide, accompanied by changes to water quality—affecting levels of sedi- loss of corals liberates space for the settlement of
increasing coral mortality during the following 12 ment, nutrients, toxins, and pathogens, as well as macroalgae, which in turn tends to inhibit coral
months (24). Corals may survive and recover their fishing pressure, will be important determinants of recruitment, fecundity, and growth because they
dinoflagellate symbionts after mild thermal stress, reef state and should demand priority attention in compete for space and light, and also produce anti-
but typically show reduced growth, calcification, reef-management programs. However, as we move fouling compounds that deter settlement by
and fecundity (24) and may experience greater toward higher [CO2]atm, coral-community compo- potential competitors. Together these factors allow
incidences of coral disease (28, 29). sitions will change with some areas becoming macroalgae to form stable communities that are

Downloaded from www.sciencemag.org on December 19, 2007

To illustrate the combined effects of acidifica- dominated by more thermally tolerant corals like relatively resistant to a return to coral domination
tion and bleaching on reefs, we simplified the coral the massive Porites (31) and others potentially dom- (Figs. 2 and 3) (22, 23, 35). As a result of weak-
ecosystem into the nine features required to model inated by thermally sensitive but rapidly coloniz- ening of coral growth and competitive ability, reefs
feedback mechanisms (Fig. 3). Although it is not ing genera, such as the tabulate Acropora. Under within the CRS-B scenario will be more sensitive
comprehensive, the model provides a theoretical the current rate of increase in [CO2]atm (>1 ppm to the damaging influence of other local factors,
framework indicating that acidification and bleach- year−1), carbonate-ion concentrations will drop such as declining water quality and the removal of
ing enhance all deleterious feedbacks, driving the below 200 mmol kg−1 and reef erosion will exceed key herbivore fish species.
coral ecosystems toward domination by macro- calcification at [CO2]atm = 450 to 500 ppm, i.e., Increases in [CO2]atm > 500 ppm (11) will
algae and noncoral communities (Fig. 3) (table S1). Scenario CRS-B (Figs. 1 and 5B). The density and push carbonate-ion concentrations well below

Trajectories in Response to Climate Change

Global temperatures are projected to increase rap-
idly to 1.8°C above today’s average temperature
under the low-emission B1 scenario of the IPCC,
or by 4.0°C (2.4° to 6.4°C) under the higher-
emission A1F1 scenario (Table 1) (8). Increases in
the temperature of tropical and subtropical waters
over the past 50 years (24) have already pushed
reef-building corals close to their thermal limits.
Projections for ocean acidification include reduc-
tions in oceanic pH by as much as 0.4 pH units by
the end of this century, with ocean carbonate
saturation levels potentially dropping below those
required to sustain coral reef accretion by 2050
(Fig. 4) (7, 10, 13). Changes in ocean acidity will
vary from region to region, with some regions,
such as the Great Barrier Reef and Coral Sea, and
the Caribbean Sea, attaining risky levels of arag-
onite saturation more rapidly than others (Fig. 4).
Just as carbonate coral reefs do not exist in waters
with carbonate-ion concentrations < 200 mmol kg−1
(10), they are likely to contract rapidly if future
[CO2]atm levels exceed 500 ppm. Similarly, un-
less thermal thresholds change, coral reefs will
experience an increasing frequency and severity
of mass coral bleaching, disease, and mortality
as [CO2]atm and temperatures increase (24–27).
We have projected three scenarios for coral
reefs over the coming decades and century. In
doing so, we recognize that important local threats
to coral reefs, such as deterioration of water quality Fig. 4. Changes in aragonite saturation {Waragonite = ([Ca2+].[CO32−])/Ksp aragonite)} predicted to occur as at-
arising from sediment and nutrient inputs associ- mospheric CO2 concentrations (ppm) increase (number at top left of each panel) plotted over shallow-water coral
ated with coastal development and deforestation, reef locations shown as pink dots (for details of calculations, see the SOM). Before the Industrial Revolution (280
and the overexploitation of marine fishery stocks, ppm), nearly all shallow-water coral reefs had Waragonite > 3.25 (blue regions in the figure), which is the
may produce synergies and feedbacks in concert minimum Waragonite that coral reefs are associated with today; the number of existing coral reefs with this
with climate change (30) (Fig. 3) [supporting on- minimum aragonite saturation decreases rapidly as [CO2]atm increases. Noticeably, some regions (such as the
line material (SOM)]. How quickly we arrive at or Great Barrier Reef) attain low and risky levels of Waragonite much more rapidly than others (e.g., Central Pacific).

1740 14 DECEMBER 2007 VOL 318 SCIENCE www.sciencemag.org

 REVIEW

Downloaded from www.sciencemag.org on December 19, 2007

Fig. 5. Extant examples of reefs from the Great Barrier Reef that are used the locations photographed. (A) Reef slope communities at Heron Island.
as analogs for the ecological structures we anticipate for Coral Reef (B) Mixed algal and coral communities associated with inshore reefs
Scenarios CRS-A, CRS-B, and CRS-C (see text). The [CO2]atm and tem- around St. Bees Island near Mackay. (C) Inshore reef slope around the
perature increases shown are those for the scenarios and do not refer to Low Isles near Port Douglas. [Photos by O. Hoegh-Guldberg]

200 mmol kg−1 (aragonite saturation < 3.3) and sea coral reefs as we know them today would be ex- through their impact on coastal protection, fish-
temperatures above +2°C relative to today’s val- tremely rare at higher [CO2]atm. eries, and tourism. These consequences become
ues (Scenario CRS-C, Fig. 1). These changes will We recognize that physiological acclimation or successively worse as [CO2]atm increases, and un-
reduce coral reef ecosystems to crumbling frame- evolutionary mechanisms could delay the arrival of manageable for [CO2]atm above 500 ppm.
works with few calcareous corals (Fig. 5C). The some scenarios. However, evidence that corals and Although reefs with large communities of coral
continuously changing climate, which may not their symbionts can adapt rapidly to coral bleach- reef-related organisms persist under CRS-A and
stabilize for hundreds of years, is also likely to ing is equivocal or nonexistent. Reef-building CRS-B, they become nonfunctional under CRS-C,
impede migration and successful proliferation of corals have relatively long generation times and as will the reef services that currently underpin
alleles from tolerant populations owing to con- low genetic diversity, making for slow rates of human welfare. Climate change is likely to fun-
tinuously shifting adaptive pressure. Under these adaptation. Changes in species composition are damentally alter the attractiveness of coral reefs to
conditions, reefs will become rapidly eroding also possible but will have limited impact, as even tourists (compare Fig. 5, A and C), which is an
rubble banks such as those seen in some inshore the most thermally tolerant corals will only sustain important consideration for the many low-income
regions of the Great Barrier Reef, where dense temperature increases of 2° to 3°C above their coastal countries and developing small island states
populations of corals have vanished over the long-term solar maxima for short periods (24, 31). lying within coral reef regions. Under-resourced
past 50 to 100 years. Rapid changes in sea level However, such changes come at a loss of bio- and developing countries have the lowest capacity
(+23 to 51 cm by 2100, scenario A2) (8), diversity and the removal of important redundan- to respond to climate change, but many have
coupled with slow or nonexistent reef growth, cies from these complex ecosystems. Some studies tourism as their sole income earner and thus are at
may also lead to “drowned” reefs (36) in which have shown that corals may promote one variety risk economically if their coral reefs deteriorate
corals and the reefs they build fail to keep up of dinoflagellate symbiont over another in the (40). For instance, tourism is a major foreign ex-
with rising sea levels. relatively small number of symbioses that have change earner in the Caribbean basin and in some
The types of synergistic impacts on coral and significant proportions of multiple dinoflagellate countries accounts for up to half of the gross do-
reef-dependent organisms defined for Scenario types (38). These phenotypic changes extend the mestic product (40). Coral reefs in the United
CRS-B (Fig. 5B) will be magnified substantially plasticity of a symbiosis (e.g., by 1° to 2°C) (21) States and Australia may supply smaller compo-
for CRS-C (Fig. 5C), with probably half, and pos- but are unlikely to lead to novel, long-lived as- nents of the total economy, but still generate con-
sibly more, of coral-associated fauna becoming rare sociations that would result in higher thermal siderable income (many billions of U.S. $ per year)
or extinct given their dependence on living corals tolerances (39). The potential for acclimation even from reef visitors that are increasingly responsive
and reef rugosity (37). Macroalgae may dominate to current levels of ocean acidification is also low to the quality of reefs (41).
in some areas and phytoplankton blooms may be- given that, in the many studies done to date, coral Reef rugosity is an important element for the
come more frequent in others, as water quality de- calcification has consistently been shown to de- productivity of all reef-based fisheries, whether sub-
clines owing to the collateral impact of climate crease with decreasing pH and does not recover as sistence, industrial, or to supply the aquarium trade.
change on associated coastal areas, drying catch- long as conditions of higher acidity persist (13). The density of reef fish (32) is likely to decrease as
ments and causing episodic heavy rainfall that a result of increasing postsettlement mortality aris-
transports nutrients and sediments into coastal areas. Socioeconomic Impacts of Coral Reef Decline ing from a lack of hiding places and appropriate
Whether or not one defines the transition from The scenarios presented here are likely to have se- food for newly settled juveniles (42). Regardless of
CRS-B to CRS-C and [CO2]atm of 450 to 500 ppm rious consequences for subsistence-dependent so- future climate-change influences, the total landing
as the tipping point for coral reefs, it is clear that cieties, as well as on wider regional economies of coral reef fisheries is already 64% higher than

www.sciencemag.org SCIENCE VOL 318 14 DECEMBER 2007 1741

 REVIEW
can be sustained, with an extra 156,000 km2 of reef fish, especially grazers such as parrotfish, State of Queensland Greenhouse Taskforce through the
coral reef estimated as being needed to support would be expected to result in an improved ability Department of Natural Resources and Mining, Townsville”
(2003).
anticipated population growth by 2050 (43). For of coral reefs to bounce back from disturbances (51), 27. S. D. Donner, W. J. Skirving, C. M. Little, M. Oppenheimer,
example, in Asia alone coral reefs provide about as long as other factors such as water quality are not O. Hoegh-Guldberg, Glob. Change Biol. 11, 2251 (2005).
one-quarter of the annual total fish catch and food limiting. Unfortunately, with the exception of marine 28. C. D. Harvell et al., Science 296, 2158 (2002).
to about 1 billion people (43). Climate-change im- reserves, there is negligible explicit management 29. J. F. Bruno et al., PLoS Biol. 5, e124 (2007).
30. K. Newton, I. M. Cote, G. M. Pilling, S. Jennings, N. K. Dulvy,
pacts on available habitat will only exacerbate al- of herbivores in most countries, but this could be Curr. Biol. 17, 655 (2007)
ready overstretched fisheries resources. improved by setting catch limits (52). Diversifica- 31. Y. K. Loya et al., Ecol. Lett. 4, 122 (2001).
The role of reefs in coastal protection against tion of the herbivore guild to include modest den- 32. S. K. Wilson et al., Glob. Change Biol. 12, 2220 (2006).
storms (44) has been highlighted in analyses of sities of invertebrates like sea urchins will also 33. B. Honisch, N. G. Hemming, Earth Planet. Sci. Lett. 236,
305 (2005)
exposed and reef-protected coastlines (45, 46). We enhance the resilience of coral reef ecosystems. 34. L. Mydlarz, L. Jones, C. D. Harvell, Annu. Rev. Ecol. Evol.
do not yet have estimates for how fast reef barriers Syst. 37, 251 (2006).
will disappear (47), but we can anticipate that Conclusion 35. R. S. Steneck, in Proceedings of the Colloquium on Global
decreasing rates of reef accretion, increasing rates of It is sobering to think that we have used the lower Aspects of Coral Reefs: Health, Hazards and History,
R. N. Ginsburg, Ed. (Univ. of Miami Press, FL, 1994).
bioerosion, rising sea levels, and intensifying storms range of IPCC scenarios in our analysis yet still
36. R. W. Grigg et al., Coral Reefs 21, 73 (2002).
may combine to jeopardize a wide range of coastal envisage serious if not devastating ramifications for 37. N. Knowlton, Am. Zool. 32, 674 (1992).
barriers. People, infrastructure, and lagoon and es- coral reefs. Emission pathways that include higher 38. R. Rowan, N. Knowlton, A. Baker, J. Jara, Nature 388,
tuarine ecosystems, including mangroves, seagrass [CO2]atm (600 to 1000 ppm) and global temper- 265 (1997).

Downloaded from www.sciencemag.org on December 19, 2007

meadows, and salt marshes, will become increas- atures of 3° to 6°C defy consideration as credible 39. M. Stat, D. Carter, O. Hoegh-Guldberg, Plant Ecol. Evol. Syst.
8, 23 (2006).
ingly vulnerable to growing wave and storm im- alternatives. Equally important is the fact that IPCC 40. D. L. Bryant, D. L. Burke, J. McManus, M. Spalding, Reefs
pacts. Observations of increasingly intense tropical scenarios are likely to be cautious given scientific at Risk: A Map-Based Indicator of Threats to the World’s
hurricanes and cyclones in all oceans (48) suggest reticence and the inherently conservative nature of Coral Reefs (World Resources Institute, Washington, DC,
that losses of beach sand from coastal zones are consensus seeking within the IPCC process (53). 1998).
41. H. Hoegh-Guldberg, O. Hoegh-Guldberg, Biological,
likely to increase (49). Further losses may occur Consequently, contemplating policies that result in Economic and Social Impacts of Climate Change on the
from reduced sand production, formed in many [CO2]atm above 500 ppm appears extremely risky Great Barrier Reef (World Wildlife Fund, Sydney, 2004).
cases by coral reefs, as a consequence of ocean for coral reefs and the tens of millions of people 42. M. J. Caley et al., Annu. Rev. Ecol. Syst. 27, 477 (1996).
acidification and thermal stress on calcareous algae who depend on them directly, even under the most 43. UNEP, “Marine and coastal ecosystems and human
well-being: A synthesis report based on the findings of
and other sand producers. Beaches are also under optimistic circumstances. the Millennium Ecosystem Assessment” (United Nations
threat of erosion from rising sea levels. The com- Environmental Programme, 2006).
bination of these factors will lead to less stable References and Notes 44. UNEP-WCMC, “In the front line: Shoreline protection and
beaches and impacts on other organisms, such as 1. F. Moberg, C. Folke, Ecol. Econ. 29, 215 (1999). other ecosystem services from mangroves and coral reefs
2. International Year of the Reef (IYOR), www.iyor.org/ (2008). (United Nations Environmental Programme–World
turtles and sea birds that depend on beach habitats Conservation Monitoring Centre, Cambridge, UK, 2006).
3. J. F. Bruno, S. E. Selig, PLoS ONE 2, e711 (2007).
for reproduction, as well as leading to economic 45. C. Sheppard, D. J. Dixon, M. Gourlay, A. Sheppard,
4. T. P. Hughes et al., Science 301, 929 (2003).
impacts on tourism and coastal fishing communities. 5. D. A. Petit et al., Nature 399, 429 (1999). R. Payet, Estuar. Coast. Shelf Sci. 64, 223 (2005).
6. EPICA community members, Nature 429, 623 (2004). 46. H. J. S. Fernando, J. L. McCulley, Eos 86, 301 (2005).
Opportunities for Management Intervention 7. J. Raven et al., “Acidification due to increasing carbon 47. E. H. Williams Jr., P. J. Bartels, L. Bunkley-Williams,
dioxide.” Policy Document 12/05 (The Royal Society, Glob. Change Biol. 5, 839 (1999).
The inherent inertia of the atmosphere and of our 48. P. J. Webster, G. J. Holland, J. A. Curry, H.-R. Chang,
London, 2005).
attempts to mitigate CO2 emissions suggest that 8. IPCC, Climate Change 2007: The Physical Science Basis. Science 309, 1844 (2005)
reef managers and coastal resource policies must Contribution of Working Group I to the Fourth Assessment 49. J. L. Sadd, Bull. Mar. Sci. 35, 221 (1984).
first reduce the influence of local stressors such as Report of the Intergovernmental Panel on Climate 50. P. J. Edmunds, R. C. Carpenter, Proc. Natl. Acad. Sci.
Change, S. Solomon et al., Eds. (Cambridge Univ. Press, U.S.A. 98, 5067 (2001).
declining water quality, coastal pollution, and over- 51. P. J. Mumby, Ecol. Appl. 16, 747 (2006).
Cambridge, UK, and New York, 2007).
exploitation of key functional groups such as her- 9. J. G. Canadell et al., Proc. Natl. Acad. Sci. U.S.A. 104, 52. P. J. Mumby et al., Science 311, 98 (2006).
bivores (4). These types of action are most likely to 18866 (2007). 53. J. E. Hansen, Environ. Res. Lett. 2, 024002 (2007).
assist coral reefs through the decades of stress that 10. J. A. Kleypas, J. W. McManus, L. A. B. Meñez, Am. Zool. 54. E. Lewis, D. W. R. Wallace, Program Developed for CO2
39, 146 (1999). System Calculations. ORNL/CDIAC-105 (Carbon Dioxide
inevitably face them. There may be opportunities Information Analysis Center, Oak Ridge National
11. J. H. Connell, Coral Reefs 16 (suppl.), S101 (1997).
for using coral restoration to reduce the risk that 12. IPPC, Climate Change 2001. The Scientific Basis. The Laboratory, U.S. Department of Energy, Oak Ridge, TN,
reefs will shift into a non–coral-dominated state Contribution of Working Group I to the Third Assessment 1998).
(Fig. 3); however, the efficacy of coral restoration Report of the Intergovernmental Panel on Climate 55. This project was supported by the Global Environment
Change, J. T. Houghton et al., Eds. (Cambridge Univ. Facility, the World Bank, and the University of
methods to increase rugosity and coral cover re-
Press, New York, 2001). Queensland through the Coral Reef Targeted Research
mains unclear, and further evaluation of methods is 13. J. A. Kleypas, C. Langdon, Coast. Estuar. Stud. 61, 73 (2006) Program (www.gefcoral.org). We also thank J. Dixon,
badly needed. With respect to the latter, there is a 14. G. D. Stanley Jr., Earth-Sci. Rev. 60, 195 (2003). S. Dove, and D. W. Gledhill for discussions concerning
mismatch between the feasible scale of restoration 15. J. L. Wray, Proc. N. Am. Paleontol. Conv. Pt. J, 1358 (1969) this manuscript, N. Cenacchi for assistance in rendering
(hectares) and that of the extent of degradation 16. R. A. Berner, Z. Kothavala, Am. J. Sci. 301, 182 (2001). Fig. 4, and L. Cao for helping with the analysis and
17. M. Medina, A. G. Collins, T. L. Takaoka, J. V. Kuehl, preparation of Fig. 3. The manuscript contents are solely
(many thousands of km2). Nevertheless, new tech- the opinions of the authors and do not constitute a
J. L. Boore, Proc. Natl. Acad. Sci. U.S.A. 103, 9096 (2007).
niques for the mass culture of corals from frag- 18. M. Fine, D. Tchernov, Science 315, 1811 (2007) statement of policy, decision, or position on behalf of
ments and spat may assist local restoration or the 19. A. H. Knoll, R. K. Bambach, J. L. Payne, S. Pruss, NOAA or the U.S. Government. This paper is dedicated to
culture of resistant varieties of key organisms (44). W. W. Fischer, Earth Planet. Sci. Lett. 256, 295 (2008) the memory of Kim Mitchell, who saw the value of and
20. T. F. Cooper, G. De’ath, K. E. Fabricius, J. M. Lough, worked hard for the future of the world’s natural
At the 100- to 1000-km scale of coral reefs, ecosystems.
Global Change Biol., in press.
one of the most practical interventions is to facil- 21. A. M. Szmant, N. J. Gassman, Coral Reefs 8, 217 (1990).
itate grazing by fish and invertebrate herbivores. 22. P. J. Mumby et al., Nature 450, 98 (2007).
Supporting Online Material
This is likely to play an important role in situations 23. P. J. Mumby et al., Proc. Natl. Acad. Sci. U.S.A. 104,
www.sciencemag.org/cgi/content/full/318/5857/1737/DC1
like that of the Caribbean where densities of one 8362 (2007).
24. O. Hoegh-Guldberg, Mar. Freshw. Res. 50, 839 (1999). SOM Text
important herbivore, the sea urchin Diadema 25. O. Hoegh-Guldberg, J. Geophys. Res. 110, C09S06 (2005). Table S1
antillarum, were decimated by disease in the early 26. T. J. Done et al., “Global Climate Change and Coral References
1980s (50). Clearly, the improved management of Bleaching on the Great Barrier Reef, Final report to the 10.1126/science.1152509

1742 14 DECEMBER 2007 VOL 318 SCIENCE www.sciencemag.org