Projecting Coral Reef Futures Under Global Warming and Ocean Acidification John M. Pandolfi, et al.

Science 333, 418 (2011); DOI: 10.1126/science.1204794

This copy is for your personal, non-commercial use only.

If you wish to distribute this article to others, you can order high-quality copies for your colleagues, clients, or customers by clicking here. Permission to republish or repurpose articles or portions of articles can be obtained by following the guidelines here. The following resources related to this article are available online at www.sciencemag.org (this infomation is current as of July 21, 2011 ): Updated information and services, including high-resolution figures, can be found in the online version of this article at: http://www.sciencemag.org/content/333/6041/418.full.html Supporting Online Material can be found at: http://www.sciencemag.org/content/suppl/2011/07/20/333.6041.418.DC1.html http://www.sciencemag.org/content/suppl/2011/07/21/333.6041.418.DC2.html This article cites 97 articles, 13 of which can be accessed free: http://www.sciencemag.org/content/333/6041/418.full.html#ref-list-1 This article appears in the following subject collections: Ecology http://www.sciencemag.org/cgi/collection/ecology Downloaded from www.sciencemag.org on July 21, 2011

Science (print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by the American Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005. Copyright 2011 by the American Association for the Advancement of Science; all rights reserved. The title Science is a registered trademark of AAAS.

REVIEW Projecting Coral Reef Futures Under Global Warming and Ocean Acidification
John M. Pandolfi,1,2* Sean R. Connolly,3 Dustin J. Marshall,2 Anne L. Cohen4 Many physiological responses in present-day coral reefs to climate change are interpreted as consistent with the imminent disappearance of modern reefs globally because of annual mass bleaching events, carbonate dissolution, and insufficient time for substantial evolutionary responses. Emerging evidence for variability in the coral calcification response to acidification, geographical variation in bleaching susceptibility and recovery, responses to past climate change, and potential rates of adaptation to rapid warming supports an alternative scenario in which reef degradation occurs with greater temporal and spatial heterogeneity than current projections suggest. Reducing uncertainty in projecting coral reef futures requires improved understanding of past responses to rapid climate change; physiological responses to interacting factors, such as temperature, acidification, and nutrients; and the costs and constraints imposed by acclimation and adaptation. oral reefs occupy a small part of the worlds oceans yet harbor a hugely disproportionate amount of its biodiversity. More than 450 million people from 109 countries live close to coral reefs, which provide important sources of ecosystem goods and services for these communities. But reefs have suffered degradation from human activities associated with overexploitation and pollution on centennial to millennial scales (1)degradation that has accelerated over the past ~50 years (2). Global warming and ocean acidification (OA) are compounding these threats. Indeed, past biodiversity crises in reef ecosystems do appear to coincide with episodes of rapid global warming and OA (3). Some recent projections of coral reef futures are that present day, ongoing warming and OA will cause rapid, dramatic, and global-scale losses of coral reefs (49). For example, decreasing seawater carbonate ion (CO32) concentrations because of rising atmospheric CO2 are predicted to lower rates of calcium carbonate (CaCO3) production by corals such that, within decades, rates of reef erosion will exceed rates of reef accretion across much of the tropics and subtropics [e.g., (5, 10, 11)]. Here, we summarize the most recent evidence for past, present, and predicted future responses of coral reefs to environmental change, with emphasis on rapid increases in temperature and OA and their effects on reef-building corals, which provide much of the habitat framework and implications of a deeper understanding of effects of climate change and OA alongside other anthropogenic effects, such as coastal development and overfishing, for the conservation and management of coral reefs over the next century. Coral Reef Response to Past Global Change Long-term changes in Earths atmosphere and oceans exert considerable control over biotic turnover and evolution through a number of variables, including sea surface temperature (SST), sea level, and mineral saturation state (known as W) (Fig. 1). For example, disproportionate extirpation of aragonitic or calcitic taxa from higher temperatures and lower pH during mass extinctions (12) substantially influenced calcification in the oceans as biodiversity recovered (12, 13). Long-term steady-state conditions of high atmospheric CO2 and low ocean pH are essentially decoupled from Warag (fig. S1A) because of slow negative geochemical feedbacks, which increase alkalinity, especially calcium availability (primarily caused by increased rock weathering associated with high CO2). Thus, shallow water tropical reef organisms existed throughout the past 540 million years of the Phanerozoic, through periods of temperature >7C higher than today, and under CO2 conditions >6000 parts per million (ppm) (14), more than 20 times greater than preindustrial levels (Fig. 1). In contrast to these long-term steady-state
Mya
0 100

structural complexity upon which all reef organisms depend. This new knowledge confirms that coral reefs, at least as presently structured, are indeed threatened by climate change but that current projections of global-scale collapse of reefs within the next few decades probably overestimate the rapidity and spatial homogeneity of the decline. We conclude by considering the

200

300

400

500

600 8 6 4 2 0 -2

CO2 (ppm)

7000 6000

5000 4000

1 0.8

3000 2000 1000

Total extinction rate

0.6

0.4 -4 0.2 0
CENOZOIC MESOZOIC PALEOZOIC PROTER.

1 2

Australian Research Council (ARC) Centre of Excellence for Coral Reef Studies, St. Lucia, Queensland 4072, Australia. School of Biological Sciences, University of Queensland, St. Lucia, Queensland 4072, Australia. 3ARC Centre of Excellence for Coral Reef Studies and School of Marine and Tropical Biology, James Cook University, Townsville, Queensland 4011, Australia. 4Department of Geology and Geophysics, Woods Hole Oceanographic Institution, Woods Hole, MA 02543, USA. *To whom correspondence should be addressed. E-mail: [email protected]

Fig. 1. Summary of major episodes of physical and biological change for reefs through geological time. Curves include global atmospheric parts per million by volume (ppmv) CO2 (14), global mean temperature anomaly in C (14), and total extinction rate (Paleobiology DataBase, http://paleodb.org/cgi-bin/bridge.pl). Greenhouse (green) and Icehouse (blue) periods (92) shade the geological time scale (light blue bar denotes no true polar ice caps documented). Vertical dashed gray lines indicate global mass extinction events; vertical gray bars indicate reef crises (3); and short vertical black bars indicate ocean acidification events such as ocean anoxic events (OAEs 120.5 Ma, 111 Ma, 102 Ma, and 93.5 Ma during the Cretaceous) that occurred over time scales too great to affect ocean carbonate saturation states and indeed are not associated with reef decline [after (15) and (3)]. Evidence for ocean acidification events from (3) is indicated by (derived from physical evidence of pronounced increases of pCO2), and evidence of ocean acidification events from (15) is indicated by (termed by these authors as candidate OA events). VOL 333 SCIENCE www.sciencemag.org

418

22 JULY 2011

Downloaded from www.sciencemag.org on July 21, 2011

Temperature anomaly

REVIEW
conditions, rapid increases in CO2 occur too fast potential to repeatedly warm over centennial to for such feedbacks and lead to simultaneous millennial time scales (21, 22), in one location at declines in Warag (15) (fig. S1B). Such departures rates comparable to those projected for the coming are more in line with the seemingly unprecedented century [figure 2 in (23)]. None of these post-LGM rate of CO2 rise occurring now, which would re- warming episodes appear to have interrupted reef quire a millennial-scale time lag before rock weath- growth, which was substantial throughout the tropical oceans through to the cessation of the ering restores steady-state saturation states (15). Comparably rapid declines in Warag during pre- Holocene sea level rise around 6 ka (24). Howvious geological intervals have not been documented ever, during this time, SSTs were cooler than they (16), so reef response to these conditions cannot be are today, there were no rapid changes in greenpredicted from fossil sequences. However, reefs house gases or pH comparable to current, ongoing have suffered five episodes of severe, global-scale climate change, and levels of CO2 remained below biodiversity loss or cessation of reef growth, termed 330 ppm. reef crises (3) over geologic time. The four most recent reef crises coincided with OA in conjunction Ecological Responses to Ongoing SST Increase with rapid global warming, including two of the Coral reefs are particularly sensitive to increasing Big 5 Phanerozoic mass extinctions at the end- temperatures because the major framework buildPermian [251 million years ago (Ma)] and end-Triassic (200 Ma) (3) (Fig. 1). A Acropora 0.25 An analysis of the inferred physiology palmata of victims versus that of survivors of Diploria 0.20 these biotic crises points to an imporlabyrinthiformis tant role for rapid increases in OA Stephanocoenia (12). Other episodes of increased OA intersepta 0.15 such as the Aptian Ocean Anoxic Solanastrea Event (Fig. 1) (OAE1a, about 120 Ma) bournoni 0.10 occurred on time scales over which Colpophyllia natans carbonate saturation state was well 0.05 buffered [>10,000 years (16)], so the absence of reef decline provides limited information on the consequences 0.00 of present climate change for coral 0 2 4 6 8 reefs. The most recent reef crisis, the Degree heating weeks Paleocene-Eocene Thermal Maximum Stylophora B (PETM; 55.8 Ma), was characterAcropora ized by rapid SST rise and a similar Pocillopora order of magnitude of CO2 increase as Goniastrea present, probably consisting of CO2 Leptoria Favia release in multiple pulses over 1 to 10 Montipora thousand years (fig. S1B). Although a Millepora marked faunal shift on continental Porites Ma carbonate platforms, from coral-algal Fungiidae reefs to reefs dominated by large Echinopora Platygyra benthic foraminifera, has been assoHydnophora ciated with this event (17), reef asFavites Kenya semblages in at least one oceanic Galaxea Australia setting were unaffected (18). Thus, Goniopora overall evidence from the fossil record Pavona Cyphastrea indicates that rates of change are cru0 50 100 150 200 250 cial for determining ecological outBleaching index comes, that reefs are most vulnerable when rapid increases in SSTand CO2 occur together with declines in Warag, Fig. 2. Bleaching severity differs within and between species, but and that at least some reefs have these differences vary among bleaching events. (A) Estimated probability distribution of bleaching thresholds for five species of shown greater resilience to past rapid Caribbean corals, based on logistic regression models fitted to warming and acidification than previ- bleaching data from the Florida Keys and Dry Tortugas, under ously thought. environmental conditions corresponding to the average observed Since the Last Glacial Maximum for each species (29). The width of the distribution indicates the (LGM) [~20 thousand years ago (ka)], within-population variation in the temperature at which bleaching abrupt changes in global temperatures is expected to occur. (B) Variation in bleaching intensity between have occurred repeatedly (19), espe- Kenya and Australia, during the 1998 mass bleaching event cially between 15 and 10 ka, that have [compare with figure 2 in (28)]. Bleaching intensity is a weighted affected tropical SSTs (20). Evidence average of the proportion of colonies in different bleaching catefrom high-resolution proxy records gories (28). Taxa are displayed in rank order of bleaching severity suggests that tropical SSTs had the in Kenya. Data provided by A. H. Baird and T. R. McClanahan.
Genus Bleaching threshold frequency

ers, scleractinian corals, suffer a breakdown in their symbiosis with zooxanthellae when temperatures are anomalously high. This coral bleaching (so termed because corals become white as zooxanthellae are lost), reduces the performance of the coral host, which receives most of its organic carbon from the symbiont. Mass bleaching events, when most of the coral assemblage bleaches, have become more frequent and widespread in the past few decades (25). These events are often associated with high mortality (26) and depressed colony growth and reproduction among survivors (27). Within communities, there is both taxonomic variation (28, 29) and within-species heterogeneity in bleaching susceptibility (29) (Fig. 2A). Because the extent of variability in bleaching thresholds itself varies among species, taxonomic differences in bleaching severity may vary among bleaching events (Fig. 2B). Numerous characteristics of coral hosts have the potential to confer differences in bleaching susceptibility, and these characteristics vary substantially within and among coral species (30, 31). Some coral species also harbor multiple strains of zooxanthellae, which confer differential susceptibility of their hosts to bleaching (32). As with bleaching events themselves, there is substantial variation in reef recovery in the aftermath of bleaching events (25). Where there is sufficient survival of existing colony tissue, recovery can occur within a few years (33), but in other cases no appreciable recovery of coral cover was observed even after 5 to 10 years (34, 35). Most commonly, recovery of coral cover requires about a decade (25, 36). For coral species hosting multiple symbiont strains, shifts to thermally resistant strains are sometimes observed after bleaching events (37), although reversion to domination by thermally sensitive strains may occur over several years (37), probably because of a trade-off between bleaching resistance and productivity (38). Where coral mortality is accompanied by a loss of habitat structure, the potential exists for negative effects of bleaching on the broader reef community. Mass bleaching can be followed by increases in macroalgae, when herbivores are absent or avoid macroalgal species (39), but even when macroalgae are controlled other taxa, such as encrusting sponges, can reduce the space available for coral recruitment (40). Fishes and invertebrates that consume or inhabit corals during some part of their life cycle typically decline in abundance after mass bleaching, while roving herbivores such as parrotfishes and surgeonfishes increase (25, 41). Presently, there is limited evidence of systematic changes in the abundance of mesopredators or apex predators or of declines in fisheries yields associated with bleaching, suggesting that any such effects would likely be accompanied by long time lags (34, 42). Response to Ongoing Ocean Acidification For coral reefs, the paramount concern about CO2induced OA has been its potential impact on rates of biogenic calcium carbonate production by the dominant reef calcifiers: corals and coralline algae. An earlier compilation of data from several single-

www.sciencemag.org

SCIENCE

VOL 333

22 JULY 2011

419

Downloaded from www.sciencemag.org on July 21, 2011

REVIEW
organism and mesocosm experiments and one field study focused (81,209) 160 on CaCO3 precipitation on the Great Bahama Banks (43) suggested that coral and coral reef calcification declined linearly with declining arago120 nite saturation state, reaching zero when arag = 1 (solid line in Fig. 3) (44). On the basis of these data, a 40 to 80 83% decline in reef calcification was predicted by 2065 C.E. (44). Moreover, a subsequent study based on the sensitivity of net community calcification 40 to arag on a northern Red Sea reef predicted a global shift in coral reefs (14,-59) from net accreting to net dissolving 0 20 0 20 40 60 80 100 when atmospheric CO2 doubles to 560 ppm (11). These projections fueled (arag-1) as % of (max arag-1) concern that coral reefs are under threat of imminent collapse from OA. Fig. 3. Calcification response to changes in arag observed in However, a considerably expanded experiments and in nature. Experiments using Ca to manipulate data set (4551) incorporating more arag and those reporting >80% mortality under elevated CO2 recent findings reveals that the calci- are excluded. Data are grouped as follows: (i) CO2 invasion exfication response to changing arag periments at ambient temperature [seven studies (4549, 52, 53), among individual coral species, coral red crosses], and CO2 invasion experiments at elevated temperature mesocosms, and in situ reef commu- [three studies (46, 47, 53), orange crosses]. In all CO2 invasion nities, although consistently negative, experiments, corals and coral reef organisms were exposed to is highly variable and often nonlinear treatment conditions for longer than 1 week. (ii) Non-CO2 invasion (Fig. 3). Although calcification is strong- experiments using different combinations of acid and base to manipulate arag, DIC, and pCO2 [seven studies (44, 50, 51, 54, 82, 93, 94), ly sensitive to arag in some expergray asterisks]. Periods of exposure vary from hours to years. (iii) iments, it did not change significantly Field data from Gulf of Eilat (58) (green triangles) and Great in others where corals were exposed Bahamas Bank (43) (blue diamonds). The horizontal axis is scaled to to CO2 levels between two and three ( 1) rather than because the first-order saturation-state arag arag times greater than preindustrial con- model of (44) predicts that calcification is proportional to ( 1), arag centrations (45, 46, 52) or levels that and this scaling facilitates comparing the model prediction (solid declined below the aragonite satura- line) with the empirical data (points). The range of arag conditions tion threshold (arag = 1) (45). In a for each experiment varied significantly, and, in several instances, few instances, calcification increased corals were exposed to saturation states significantly higher (and under moderately elevated partial pres- lower) than those under which they grow naturally. Here, we set sure of CO2 ( pCO2) (46, 48, 53), as has (maximum arag 1) in each experimental and field study at also been observed for some coral- 100% and the calcification response is represented as percent of line algae, crustacea, and echinoderms calcification recorded under the maximum W condition. The dash(45). Sensitivity of calcification to de- dot green line is the fitted calcification model from (58). creases in arag appears to be reduced when (i) studies are conducted over weeks or months (45, 46, 50, 53)] as opposed to cification and offset, at least partially, the impact <1 day (44, 54) or (ii) corals are reared under of elevated CO2 over a defined range of arag. In fact, the energetic cost of calcification, curnutritionally replete conditions by feeding or elevating inorganic nutrient concentrations (44, 45). rently poorly understood, may be key to predictAlthough a component of the documented ing the response of corals and other reef calcifiers variability in responses to acidification may be to future OA. Hypercalcifiers, including corals, linked to differences in experimental technique, probably utilize all forms of seawater carbon it also likely reflects real variability in the sen- by using a combination of enzymes and prositivity of coral calcification to arag and flex- ton pumps to convert CO2 (aq) and HCO3 to ibility in the ability of some corals to maintain carbonate ions for rapid calcification. However, calcification rates over a broad range of arag this process is energetically costly, and, in a conditions. There are several potential explana- high-CO2, higher dissolved inorganic carbon tions for such variable responses. For example, (DIC) ocean, well-nourished corals may be better adaptation might result in the evolution of OA- equipped to convert excess HCO3 to CO32. hardy genotypes, either highly efficient calcifiers Evidence from some experiments supports this or those that direct more of their energetic bud- (55, 44). Conversely, poorly nourished or energet to calcification. Also, given the right combina- getically depleted corals [including bleached or tion of environmental factors, such as abundance partially bleached corals (47)] will likely display of food or high levels of inorganic nutrients, some heightened sensitivity to OA. At the community corals might be able to divert excess energy to cal- level, responses of other reef organisms to OA,
% of calcification at max arag

The Role of Evolution Evolution can buffer populations from environmental change (60) or accelerate species decline. Coral reef organisms will evolve in response to the increased thermal stress and OA associated with climate change, but the impact of this evolution is unclear (61). For corals specifically, the issue is complicated by symbiosis: The interdependency of mutualists, such as the coral host and symbiont, can affect rates of evolution (60). The symbiont suffers greater mortality because of thermal stress than does the host (62), so, from an ecological perspective, the symbiont is often viewed as the weaker link [but see (63)]. From an evolutionary perspective, however, the generation times of corals are orders of magnitude greater than those of symbionts, so symbionts may show faster evolutionary responses than their hosts. Nevertheless, the generation times of many coral species do not preclude rapid genetic evolution of the host in response to climate change as well: Other organisms have undergone rapid evolution in response to anthropogenic stresses within just a few generations (60). Furthermore, the modular nature of corals means that somatic line mutations could provide additional sources of variation such that evolution can occur within a single generation (64). Phenotypic plasticity in thermal tolerances (acclimation or acclimatization) is an additional source of variation in the response of coral reef organisms to climate change (65). Because phenotypic plasticity allows organisms to cope with environmental change in the absence of genetic change, it has been argued that phenotypic plasticity will not influence reef organisms evolutionary responses to climate change (66). However, models suggest that phenotypic plasticity enhances both the persistence and the evolution of organisms facing rapid environmental change (65). Assemblages that experience greater variability in SST appear less susceptible to bleaching (67), as do assemblages that have recently

420

22 JULY 2011

VOL 333

SCIENCE

www.sciencemag.org

Downloaded from www.sciencemag.org on July 21, 2011

such as increased productivity of algae (56), could alter the intensity or consequences of corals interactions with other benthic taxa. Experimental observations appear consistent with measurements of annual calcification rates in field corals that suggest high interspecific variability in calcification rates over natural gradients in arag (57). Nevertheless, striking inconsistencies remain between experimental and field observations that must be reconciled if predictions of future OA impacts on coral reef ecosystems are to be improved. Most importantly, a field study of community calcification (58) (green symbols in Fig. 3) implies community calcification responses to changes in arag that are considerably stronger than expected from either singleorganism or mesocosm experiments. Efforts to understand this discrepancy, particularly the role of dissolution (59), are paramount to improved predictions of future OA impacts on reef ecosystems on a global scale.

REVIEW
experienced substantial bleaching (68), phenomena that may be partly due to acclimatory responses. One example is the shift to more thermally tolerant symbiont strains after bleaching events (37). This suggests that some acclimation to increasing temperatures is already occurring, and it is likely that this plasticity will also enhance adaptation to warmer conditions. However, constraints and trade-offs will undoubtedly limit the phenotypic responses of coral reef organisms. For instance, the ability of hosts to shuffle symbionts to more thermally tolerant strains may be genetically constrained (69); indeed, corals on one reef may be incapable of acquiring such strains, even when conspecifics on nearby reefs can (70). Furthermore, although phenotypic plasticity in response to climate change should facilitate adaptation, phenotypic plasticity is costly (65), with implications for physiological performance or the ability to cope with further change. Whether such costs are common on reefs remains unclear. Like acclimatory responses, evolutionary responses of reef organisms to climate change will be influenced by the nature and severity of genetic constraints. Populations can only adapt when there is sufficient genetic variation on which selection can act and the traits under selection are not correlated with other traits that affect fitness (often termed genetic trade-offs) (71). Such tradeoffs have been shown to preclude adaptation to climate change despite substantial genetic variance in the traits under direct selection (71). In coral reef organisms, there are very few estimates of standing genetic variation in traits that are likely to be under selection under climate change regimes (31, 72) and little information about genetic trade-offs associated with these traits, which have critical implications for potential adaptation to climate change (fig. S2). For example, among coral symbionts, there is a strain-level trade-off between bleaching resistance and holobiont growth rates (73), suggesting that genetic constraints may limit the scope and trajectory of evolution in response to climate change. Similarly, simultaneous adaptation to rapid warming and acidification may be constrained if these two selection pressures act on reef organisms in contrasting ways. Genetic constraints not only affect the magnitude of the response to selection via trade-offs with fitnessthey can affect the direction of the response to selection and lead to important changes in other, seemingly unaffected traits (fig. S2). Thus, as corals and reef-associated species evolve to cope with climate change, unanticipated demographic and phenotypic consequences are likely. Projecting Coral Reef Futures Because bleaching causes mortality of corals and reduces energy available for growth and reproduction among survivors, increases in its frequency and intensity can confidently be expected to reduce coral cover. The most pessimistic projection is for global-scale losses of coral reefs resulting from annual mass bleaching events (4). More recent mathematical modeling that incorporates adaptation of thermal tolerance under varying emissions scenarios suggests that a wide range of outcomes is possible, from a complete collapse of coral cover by the middle of this century to maintenance of comparable levels of cover to 2100 and beyond (74). The outcome will depend on the extent of thermal adaptation and aggressive emissions reduction: Both appear necessary to avoid extended declines in coral cover to very low levels (74). Most researchers predict coral species that are more susceptible to severe bleaching, such as the branching Acropora that form much of the habitat complexity of Indo-Pacific reefs, to be reduced in abundance relative to species that exhibit less sensitivity, such as slower-growing genera with massive or encrusting growth forms (75). Projection models that characterize interactions between idealized susceptible and resistant coral morphotypes reproduce such predicted changes (74, 76), as do observations of community change after bleaching events (77). However, because bleaching-susceptible species often have faster rates of recovery from disturbances, their relative abundances will not necessarily decline; indeed, such species could potentially increase in abundance, depending on how demographic characteristics and competitive ability are correlated with thermal tolerance (78) and on the response of other benthic taxa, such as algae. Moreover, the shorter generation times typical of moresusceptible species (79) may also confer faster rates of evolution of bleaching thresholds, which would further facilitate maintenance of or increases to the relative abundance of thermally sensitive but fasterevolving species (74). Early projections of contraction in the global distribution of reefs because of OA had their basis in part in the observation that present-day reefs are confined to regions where the neighboring open ocean has an Warag above ~3.3 (80). However, temperature and Warag are strongly correlated at global scales (80), making the attribution of geographical limits to reef growth to critical levels of Warag, rather than lower thermal limits, problematic. A more recent projection has its basis in a calcification model parameterized by field studies of calcifications response to medium-term (~2-year) fluctuations in calcification and Warag (58) (dashed green line in Fig. 3) and predicts that most of the worlds reefs will be compromised once atmospheric CO2 reaches 560 ppm (11). The model [equation 3 in (11)] assumes that calcification at the skeleton surface declines to zero as Warag approaches 1.0, as with abiogenic CaCO3 precipitation. However, corals are known to calcify at rates far exceeding the abiogenic CaCO3 precipitation rate (81); corals may still calcify when Warag < 1 (45, 82), and the calcification response underpinning these modeled projections exhibits much greater sensitivity to Warag than has been apparent in other reef localities [figure 1 in (11)] or in many laboratory experiments where much greater variability in reef and coral response to changing Warag occurs (Fig. 3). There is no unified explanation that reconciles the broad range of implications that experimental studies of individual colonies, experimental mesoSCIENCE VOL 333 cosms, and field studies have for changes in calcification rates on coral reefs over the next century. In particular, it is not clear whether the high sensitivity of calcification to Warag assumed in current projections (5) occurs because the calcification response is influenced by other physical variables that covary with arag (temperature, light, nutrients) over the temporal scale of the field studies used to calibrate the models (58) or because the whole-reef response is dominated by processes other than coral calcification (e.g., calcification by other organisms or the dissolution kinetics of the reef matrix). Evidence from other field studies does suggest geographical variation in the community calcification response [e.g., figure 1 in (11)] that may be related to these factors. An additional source of uncertainty concerns interactions between OA and other environmental variables. Any exacerbation of OA-induced reductions in calcification by high temperatures, reductions in coral cover because of mortality from bleaching, or reductions in nutrient or heterotrophic energy acquisition that reduce the energy available for calcification could accelerate declines in calcification at the whole-reef scale over the next century (47, 53). Episodic depletion of energy reserves associated with increased coral bleaching (83) and chronic reductions in ocean productivity caused by increases in stratification and reductions in mixed-layer depth in the tropics (84) are two potential drivers of such reductions. Indeed, even without coincident acidification, large-scale depletion of coral cover associated with mass bleaching has caused periods of net reef dissolution (85). Projections are useful exercises to help plan for future uncertainty in a dynamic system. There is abundant evidence indicating sensitivity of coral reefs to accelerated rates of warming and acidification conditions, including the historical record; bleaching; and the interactive effects of rising SST, OA, and reductions in oceanic productivity. This evidence has played a key role in many predictions that the disappearance of coral reefs on a global scale will be irreversibly under way within a matter of decades (46, 8, 9). However, recent evidence including studies of the late Paleocene fossil record (18) and speciesspecific responses to bleaching and OA [in experimental and field (57) settings] and recent work highlighting the role of phenotypic plasticity in evolution and the potential for rapid adaptation indicate that this view about the time scale of reef response may not adequately take account of reef organisms capacity for coping with stress and their potential for adaptation. Moreover, range expansions of corals in response to warming temperatures have been recorded (8688), but we presently lack good estimates of the potential rates of such range expansion under rapid environmental change and models that incorporate such shifts into projections. Such physiological, evolutionary, and biogeographic responses are not free from costs or constraints, but they will influence the nature of reefs responses to climate change. Thus, reef degradation resulting from climate change alone is likely

www.sciencemag.org

22 JULY 2011

421

Downloaded from www.sciencemag.org on July 21, 2011

REVIEW
to be a more spatially, temporally, and taxonomically heterogeneous process than some projections suggest (5, 8, 11) and could even be slowed where management of local pressures, such as coastal development and overfishing, provides greater opportunity for reefs to cope with increasing ocean temperature and decreasing pH. Incorporating an explicit consideration of these issues should improve our ability to project the consequences, and manage the impacts, of ongoing climate change. Management Considerations Managing for future climate change impacts on coral reefs needs integration with our understanding of other ongoing anthropogenic stresses such as overfishing and pollution (1, 75). There have been numerous calls to maximize the ecological resilience of coral reef communities to anthropogenic stress, including climate change (75). Because the rate at which the environment changes strongly affects whether and in what form coral reefs will persist (65), actions that slow the rate of climate change will diminish its impacts and maximize the potential for coral reefs to recover and even adapt. Substantial evolutionary change can occur over the decadal time scales relevant to reef managers (60), but only if there is sufficient evolutionary potential within a population. When environments change rapidly, extinction risk declines with increasing genetic variation. Consequently, actions that decrease genetic variation will erode the capacity of coral reef organisms to adapt to climate change. Larger, well-connected populations generally have greater capacities to evolve than smaller, poorly connected populations (89). Thus, human impacts, such as fishing, pollution, and habitat destruction, that fragment populations or decrease population sizes will not only have immediate ecological effects, they will also reduce the potential of coral reefs to adapt to warmer, more acidic conditions. Similarly, genetic bottlenecks, that is, populations that have reduced genetic variation because of a past period of strong selection, have reduced capacity to evolve in response to additional changes (89). The nonclimate-related threats already confronting coral reefs are likely to reduce the capacity of coral reefs to cope with climate change. For instance, bleaching susceptibility is exacerbated by coastal runoff (90). Collapses in herbivory from overfishing could also increase the risk that mass bleaching events will lead to transitions to algae-dominated reefs, but this has yet to be clearly documented (91). Thus, in addition to climate policy, local reef management strategies designed to mitigate nonclimate-related stressors, such as Marine Protected Areas, fisheries management, and marine spatial planning, may also increase the potential of coral reefs to cope with climate change. All of these actions are well aligned with current conservation management practices, and, in that regard, the best and most achievable thing we can do for coral reefs currently to deal with climate change is to seek to manage them well. However, slowing rates of climate change and reducing the strong selection imposed by anthropogenic impacts such as fishing and coastal development will remain critical to facilitating the long-term persistence of coral reef ecosystems.
References and Notes
1. J. M. Pandolfi et al., Science 301, 955 (2003). 2. C. Wilkinson, Ed., Status of Coral Reefs of the World: 2008 (Global Coral Reef Monitoring Network and Reef and Rainforest Research Centre, Townsville, Australia, 2008). 3. W. Kiessling, C. Simpson, Glob. Change Biol. 17, 56 (2011). 4. O. Hoegh-Guldberg, Mar. Freshw. Res. 50, 839 (1999). 5. O. Hoegh-Guldberg et al., Science 318, 1737 (2007). 6. S. D. Donner, PLoS ONE 4, e5712 (2009). 7. J. Kleypas, K. Yates, Oceanography 22, 108 (2009). 8. J. E. N. Veron et al., Mar. Pollut. Bull. 58, 1428 (2009). 9. C. Pelejero et al., Trends Ecol. Evol. 25, 332 (2010). 10. V. Fabry et al., ICES J. Mar. Sci. 65, 414 (2008). 11. J. Silverman et al., Geophys. Res. Lett. 36, L05606 (2009). 12. A. Knoll et al., Earth Planet. Sci. Lett. 256, 295 (2007). 13. W. Kiessling et al., Nat. Geosci. 1, 527 (2008). 14. R. Berner, Geochim. Cosmochim. Acta 70, 5653 (2006). 15. L. R. Kump et al., Oceanography 22, 94 (2009). 16. R. E. Zeebe, A. Ridgwell, in Ocean Acidification, J.-P. Gattuso, L. Hansson, Eds. (Oxford Univ. Press, Oxford, 2011). 17. C. Scheibner, R. Speijer, Earth Sci. Rev. 90, 71 (2008). 18. S. Robinson, Geology 39, 51 (2011). 19. R. B. Alley et al., Science 299, 2005 (2003). 20. C. Rhlemann et al., Nature 402, 511 (1999). 21. Y. Rosenthal, D. Oppo, B. K. Linsley, Geophys. Res. 30, 10.1029/2002GL016612 (2003). 22. M. W. Schmidt et al., Nature 428, 160 (2004). 23. D. W. Lea et al., Science 301, 1361 (2003). 24. L. Montaggioni, Earth Sci. Rev. 71, 1 (2005). 25. A. C. Baker et al., Estuar. Coast. Shelf Sci. 80, 435 (2008). 26. A. H. Baird, P. A. Marshall, Mar. Ecol. Prog. Ser. 237, 133 (2002). 27. J. Mendes, J. Woodley, Mar. Ecol. Prog. Ser. 235, 93 (2002). 28. T. R. McClanahan et al., Mar. Pollut. Bull. 48, 327 (2004). 29. S. H. Yee et al., Ecol. Modell. 218, 162 (2008). 30. A. H. Baird et al., Trends Ecol. Evol. 24, 16 (2009). 31. N. B. M. Csszr et al., PLoS ONE 5, e9751 (2010). 32. R. Rowan, Nature 430, 742 (2004). 33. G. Diaz-Pulido et al., PLoS ONE 4, e5239 (2009). 34. N. A. J. Graham et al., Conserv. Biol. 21, 1291 (2007). 35. P. Somerfield et al., Coral Reefs 27, 951 (2008). 36. C. R. C. Sheppard et al., Mar. Ecol. Prog. Ser. 362, 109 (2008). 37. D. J. Thornhill et al., Mar. Biol. 148, 711 (2006). 38. A. Jones, R. Berkelmans, PLoS ONE 5, e10437 (2010). 39. M. Ledlie et al., Coral Reefs 26, 641 (2007). 40. R. Aronson et al., Mar. Biol. 141, 435 (2002). 41. M. S. Pratchett et al., Oceanogr. Mar. Biol. 46, 251 (2008). 42. E. M. Grandcourt, H. S. J. Cesar, Fish. Res. 60, 539 (2003). 43. W. S. Broecker et al., Global Biogeochem. Cycles 15, 589 (2001). 44. C. Langdon, M. J. Atkinson, J. Geophys. Res. 110, (C9), C09S07 (2005). 45. J. B. Ries et al., Geology 37, 1131 (2009). 46. R. Rodolfo-Metalpa, S. Martin, C. Ferrier-Pags, J. Gattuso, Biogeosciences 7, 289 (2010). 47. K. R. Anthony, D. I. Kline, G. Diaz-Pulido, S. Dove, O. Hoegh-Guldberg, Proc. Natl. Acad. Sci. U.S.A. 105, 17442 (2008). 48. C. P. Jury et al., Glob. Change Biol. 16, 1632 (2010). 49. N. Leclercq et al., Limnol. Oceanogr. 47, 558 (2002). 50. F. Marubini et al., Mar. Ecol. Prog. Ser. 220, 153 (2001). 51. F. Marubini et al., Proc. R. Soc. London Ser. B Biol. Sci. 270, 179 (2003). 52. S. J. De Putron et al., Coral Reefs 30, 221 (2011). 53. S. Reynaud et al., Glob. Change Biol. 9, 1660 (2003). 54. S. Ohde, M. M. Hossain, Geochem. J. 38, 613 (2004). 55. M. Holcomb et al., J. Exp. Mar. Biol. Ecol. 386, 27 (2010). 56. I. Hendriks et al., Estuar. Coast. Shelf Sci. 86, 157 (2010). 57. D. P. Manzello, Coral Reefs 29, 749 (2010). 58. J. Silverman et al., J. Geophys. Res. 112, C05004 (2007). 59. A. Tribollet et al., Global Biogeochem. Cycles 23, GB3008 (2009). 60. C. A. Stockwell et al., Trends Ecol. Evol. 18, 94 (2003). 61. M. J. H. van Oppen, R. D. Gates, Mol. Ecol. 15, 3863 (2006). 62. K. B. Strychar, P. W. Sammarco, J. Exp. Mar. Biol. Ecol. 369, 31 (2009). 63. A. H. Baird et al., Trends Ecol. Evol. 24, 16 (2009). 64. K. Monro, A. G. B. Poore, Am. Nat. 163, 564 (2004). 65. L. M. Chevin, R. Lande, Evolution 64, 1143 (2010). 66. M. E. Visser, Proc. Biol. Sci. 275, 649 (2008). 67. R. Berkelmans, B. Willis, Coral Reefs 18, 219 (1999). 68. D. M. Thompson, R. van Woesik, Proc. Biol. Sci. 276, 2893 (2009). 69. D. J. Barshis et al., Mol. Ecol. 19, 1705 (2010). 70. R. Berkelmans, M. J. H. van Oppen, Proc. Biol. Sci. 273, 2305 (2006). 71. J. R. Etterson, R. G. Shaw, Science 294, 151 (2001). 72. E. Meyer et al., Mar. Ecol. Prog. Ser. 392, 81 (2009). 73. A. F. Little et al., Science 304, 1492 (2004). 74. M. L. Baskett et al., Ecol. Appl. 19, 3 (2009). 75. T. P. Hughes et al., Science 301, 929 (2003). 76. B. Riegl et al., Ann. N. Y. Acad. Sci. 1162, 136 (2009). 77. B. Riegl, Mar. Biol. 140, 29 (2002). 78. Consider a simple modification to the two-species Lotka-Volterra competition model to incorporate density-independent mortality from bleaching: dNi/dt = riNi(Ki Ni aijNj)/Ki BciNi, where Ni, ri, and Ki represent the abundance, intrinsic growth rate, and carrying capacity of species i (i = 1,2) and aij, the relative competitive effect of species j on species i. B represents bleaching severity, and ci is a coefficient relating bleaching severity to the species-specific mortality rate induced by bleaching. Species 2 increases in abundance as bleaching intensity increases if r2/r1 > c2K2/(c1a21K1). Species 2 can increase in abundance as bleaching intensity increases if it has a sufficiently greater ability to rapidly exploit available space (i.e., it has a sufficiently higher r), relative to its relative bleaching susceptibility (c2/c1). This occurs because increased bleaching mortality has two effects on a species: It directly reduces a species abundance, but it also increases available space for colonization because of increased mortality of the species competitor. 79. A. Baird et al., Annu. Rev. Ecol. Evol. Syst. 40, 551 (2009). 80. J. A. Kleypas et al., Am. Zool. 39, 146 (1999). 81. A. L. Cohen, T. A. McConnaughey, Rev. Mineral. Geochem. 54, 151 (2003). 82. A. L. Cohen et al., Geochem. Geophys. Geosyst. 10, Q07005 (2009). 83. A. G. Grottoli et al., Mar. Biol. 145, 621 (2004). 84. M. Steinacher et al., Biogeosciences Discuss. 6, 7933 (2009). 85. C. M. Eakin, Bull. Mar. Sci. 69, 171 (2001). 86. B. J. Greenstein, J. M. Pandolfi, Glob. Change Biol. 14, 513 (2008). 87. W. F. Precht, R. B. Aronson, Front. Ecol. Environ 2, 307 (2004). 88. H. Yamano et al., Geophys. Res. Lett. 38, L04601 (2011). 89. Y. Willi et al., Annu. Rev. Ecol. Evol. Syst. 37, 433 (2006). 90. S. A. Wooldridge, Mar. Pollut. Bull. 58, 745 (2009). 91. E. S. Darling et al., Conservation Lett. 3, 122 (2010). 92. J. Veizer, Y. Godderis, L. M. Franois, Nature 408, 698 (2000). 93. C. Langdon et al., Global Biogeochem. Cycles 14, 639 (2000). 94. K. Schneider, J. Erez, Limnol. Oceanogr. 51, 1284 (2006). Acknowledgments: Experimental data summarized in this paper can be found in the supporting online material. This review has benefited greatly from discussions with A. Baird, C. Duarte, M. Fine, B. Greenstein, C. Lovelock, P. Munday, C. Parmesan, M. Pratchett, R. Reef, and B. Sommer. Figures were drafted by M. Gomez, M. Hisano, and M. Lybolt. This work was supported by the Australian Research Council (ARC), the ARC Centre of Excellence for Coral Reef Studies, and the U.S. NSF.

Supporting Online Material

422

22 JULY 2011

VOL 333

SCIENCE

www.sciencemag.org

Downloaded from www.sciencemag.org on July 21, 2011