Forgot your password?
Document Actions
Hughes et al 07
Current Biology 17, 360–365, February 20, 2007 ª2007 Elsevier Ltd All rights reserved DOI 10.1016/j.cub.2006.12.049
Report
Phase Shifts, Herbivory,
and the Resilience of Coral Reefs
to Climate Change
Terence P. Hughes,1,* Maria J. Rodrigues,1,2 macroalgae, which suppressed the fecundity, recruit-
David R. Bellwood,1,2 Daniela Ceccarelli,1,2 ment, and survival of corals. Consequently, manage-
Ove Hoegh-Guldberg,3 Laurence McCook,1,4 ment of fish stocks is a key component in preventing
Natalie Moltschaniwskyj,5 Morgan S. Pratchett,1 phase shifts and managing reef resilience. Importantly,
Robert S. Steneck,1,6 and Bette Willis1,2 local stewardship of fishing effort is a tractable goal for
1
Australian Research Council Centre of Excellence conservation of reefs, and this local action can also
for Coral Reef Studies provide some insurance against larger-scale distur-
James Cook University bances such as mass bleaching, which are impractical
Townsville QLD 4811 to manage directly.
Australia
2
School of Marine and Tropical Biology
James Cook University Results and Discussion
Townsville QLD 4811
Australia The ecosystem goods and services provided by healthy
3
Australian Research Council Centre of Excellence coral reefs are a key component in the economic, social,
for Coral Reef Studies and Centre for Marine Science and cultural fabric of many tropical maritime countries
University of Queensland [1, 9]. Until recently, land-based pollution and overfish-
St Lucia ing were considered to be the major threats to coral
Brisbane QLD 4072 reefs. Today, reefs face additional pressure from ther-
Australia mal stress and emergent diseases that are closely linked
4
Great Barrier Reef Marine Park Authority to global warming [1–8]. In the most damaging case to
PO Box 1379 date, 16% of the world’s reefs were impacted in 1997–
Townsville QLD 4810 1998 by a regional-scale bleaching event that affected
Australia the Great Barrier Reef, vast tracts of the western Pacific,
5
School of Aquaculture the Indo-Australian Archipelago, and the Indian Ocean
University of Tasmania, Locked Bag 1-370 [1, 10–11]. Climate-change projections indicate that
Launceston, Tasmania 7250 similar events will reoccur with increased frequency in
Australia the coming decades [2, 12], highlighting the urgency of
6
School of Marine Sciences developing improved tools for managing reefs in the
University of Maine face of escalating threats [4–5, 13–15].
Darling Marine Center Here, we experimentally examine the resilience of
193 Clarks Cove Road coral-dominated assemblages on the Great Barrier
Walpole, Maine 04573 Reef and the processes underlying a phase shift to
macroalgal dominance (Figure 1A). We define resilience
as the ability of reefs to absorb recurrent disturbances
Summary (e.g., from cyclones, outbreaks of predators, or coral
bleaching events) and rebuild coral-dominated systems.
Many coral reefs worldwide have undergone phase Loss of resilience can lead to a phase or regime shift to an
shifts to alternate, degraded assemblages because of alternate assemblage that is typically characterized by
the combined effects of overfishing, declining water hyperabundances of fleshy seaweeds or other opportu-
quality, and the direct and indirect impacts of climate nistic species. The experiment was designed to simulate
change [1–9]. Here, we experimentally manipulated the depletion of large predatory and herbivorous fishes
the density of large herbivorous fishes to test their caused by chronic overfishing [16–21] and to investigate
influence on the resilience of coral assemblages in their role in the regeneration of reefs after recent mass
the aftermath of regional-scale bleaching in 1998, the bleaching and the mortality of corals (see Experimental
largest coral mortality event recorded to date. The ex- Procedures). The scale and timing of the experiment
periment was undertaken on the Great Barrier Reef, allowed us to measure the postbleaching dynamics of
within a no-fishing reserve where coral abundances a rich coral assemblage (77 species represented by
and diversity had been sharply reduced by bleaching 4569 colonies were recorded by the end of the experi-
[10]. In control areas, where fishes were abundant, ment), and its location provided us with a baseline
algal abundance remained low, whereas coral cover comparison of an unusually intact fish fauna on heavily
almost doubled (to 20%) over a 3 year period, primarily grazed reef crests within an established no-take area
because of recruitment of species that had been locally of the Great Barrier Reef Marine Park. This is the first
extirpated by bleaching. In contrast, exclusion of large replicated herbivore-exclusion experiment that explicitly
herbivorous fishes caused a dramatic explosion of examines herbivore-algae-coral interactions in the con-
text of climate change. We demonstrate that exclusion
of larger fishes profoundly erodes the resilience of coral
*Correspondence: terry.hughes@jcu.edu.au reefs and their ability to regenerate after bleaching, with
Phase Shifts and Resilience of Coral Reefs
361
Figure 1. Experimental Phase Shifts on the Great Barrier Reef
(A) Roofless cages and partial cages constructed on the seaward edge of reef crest. Each structure is 5 3 5 m in area and 4 m tall. Note the 2 m
high door in the cage in the center of the photograph, for access at low tide.
(B) Growths of Sargassum up to 3 m tall dwarf understory corals inside a fish-exclusion cage.
(C) When fishes were experimentally excluded, a foliose coralline alga, Mesophyllum purpurescens, replaced shallow-water grazer-resistant
species.
(D) Coral recruits settled on dead corals killed 5 years earlier by thermally induced bleaching in 1998. Grazing of the dead substrate by herbivores
is crucial for settlement and early survival of corals and coralline algae.
major implications for reef ecology, conservation, and censuses), ranging up to a maximum of 10% and 7%,
management. respectively (Figure 2A). In contrast, algal cover in the
cages far exceeded the two control treatments through-
Experimentally Induced Phase Shift out the experiment, reaching up to 91%, and averaging
Our experimental exclusion of fishes replicated the pau- 56% 6 21% (S.E.) after 30 months (repeated-measures
city of medium and large fishes that is characteristic of ANOVA, F = 3.82, p < 0.05; Figure 2). By the end of the
chronically overfished reefs in S.E. Asia, the Caribbean, experiment, algal biomass in the cages was 9 to 20 times
and elsewhere [16–19]. The biomass of herbivorous higher than in partial cages and open plots (1363 6 234,
fishes inside cages (Figure 1A) was reduced to levels 146 6 49, and 68 6 28 g wet weight per m2, respectively:
seven to ten times lower than in adjacent partial cages ANOVA, F = 20.8, p < 0.001). Over time, the species com-
and open plots (0.45 6 0.08 [S.E.], 4.29 6 2.81, and position of macroalgae in the cages diverged dramati-
3.12 6 1.24 kg/m2 per hr of video observation, respec- cally from the other two treatments (Figure S2). Dense
tively, F = 7.79, p < 0.001; Figure S1 in the Supplemental thickets of Sargassum, previously absent on the reef
Data available with this article online). In response to crest, grew to 3 m in height inside the cages, with maxi-
the experimental exclusion of larger herbivorous fishes, mum densities of greater than 1000 plants (holdfasts) per
benthic assemblages in the cages followed a fundamen- 25 m2 and a biomass of up to 8.55 kg wet weight per m2
tally different trajectory over time, with upright fleshy (Figure 1B and Movie S2). Cover and species composi-
macroalgae rather than corals and algal turfs becoming tion of crustose coralline algae also diverged in the three
predominant, mimicking similar responses on many experimental treatments (Figure 1C and Figure S3).
overfished and polluted reefs worldwide [4–8, 20–21].
In the aftermath of massive loss of corals on Orpheus Herbivory Boosts the Resilience of Coral
Island in 1998 [10], roving herbivorous fishes continued Assemblages to Global Warming
to suppress the biomass of macroalgae and thus facili- In tandem with the changes in macroalgae and coral-
tated the recruitment of corals (Movie S1). In the partial lines, the trajectory of coral reassembly after the 1998
cages and open plots where fish grazing was uninhib- bleaching event diverged markedly in the fish-exclusion
ited, the cover of macroalgae (primarily the calcified cages compared to the partial cages and open plots
red alga, Galaxaura subfruticulosa) averaged only (Figure 2B). Initially, the most prevalent taxa (accounting
4.1% and 1.7% during the experimental period (n = 16 for >80% of coral cover) were branching Porites
Current Biology
362
Figure 3. Demographic Responses of Corals
(A) Recruitment of corals into the three experimental treatments.
Figure 2. Contrasting Trajectories of Macroalgae and Corals after
Error bars are SE.
Exclusion of Fishes
(B) Mortality of coral colonies originally present in cages, partial
(A) Macroalgal cover. Error bars are SE. cages, and open plots. Error bars are SE.
(B) Relative coral cover over time among three experimental treat-
ments. Absolute coral cover after 130 weeks was 7.7% 6 1.0%
(S.E.), 19.2% 6 2.3%, and 20.2% 6 2.2% in the three treatments
(see text for analysis). Census dates were the same for all treatments
and are slightly staggered in the plots for clarity. Error bars are SE.
1062 new recruits from 26 coral genera were recorded
in the three treatments at the end of the experiment (Fig-
ure 1D). Overall, coral recruitment in cages was approx-
imately two-thirds lower (39 6 11 recruits per 25 m2,
cylindrica, massive Porites spp. (especially P. lobata compared to 108 6 26 for partial cages and 118 6 21
and P. rus), and massive faviids (principally heads of in the open plots; ANOVA, F = 150.9, p < 0.001; Fig-
Goniastrea, Favia, and Montastrea spp.) that had sur- ure 3A). Acropora, which was virtually eliminated in
vived the bleaching event 2 years prior to the initiation 1998 from the reef crest at Orpheus Island by bleaching
of the experiment. Alcyonacean soft corals and branch- [10], accounted for 246 of the recruits, representing 23%
ing hard corals, particularly a diverse suite of Acropora of the total. The dominant adult genus, Porites, had only
species, were virtually eliminated from shallow sites by two recruits in the cages, compared to 45 elsewhere (19
bleaching [10], and only a few small recruits were pres- in partial cages, 26 in open plots; F = 12.49, p = 0.003).
ent (<0.1% cover) when the experiment began in 2000. Similarly, Acropora recruits were three times more
In the fish-exclusion cages, total coral cover grew from abundant in partial cages and open plots (F = 7.7, p =
6.0% 6 0.8% (S.E.) to 7.7% 6 1.0% after 30 months. 0.011). In contrast, Fungia and Euphyllia were more
Coral cover increased much more quickly inside the par- abundant inside cages, where together they comprised
tial cages and open plots, reaching 19.2% 6 2.3% and 18% of the recruits compared to only 3% in each of the
20.2% 6 2.2%, respectively (RM-ANOVA, F = 3.82, p < two other treatments. A principal component analysis
0.05). In relative terms, coral cover increased by 28% in- summarizes the striking divergence in the composition
side the cages compared to 68% for partial cages and of coral recruits in cages compared to partial cages or
83% for open plots (Figure 2B). open plots (Figure 4). Recruit assemblages in the partial
The divergence in coral cover among treatments was cages and open plots were indistinguishable.
attributable to both lower recruitment and higher mortal- The suppression of coral recruitment inside cages is
ity of established corals after the experimental reduc- unlikely to have been an experimental artifact for two
tions of fish biomass (Figures 3A and 3B). A total of reasons. First, the considerable size of the cages and
Phase Shifts and Resilience of Coral Reefs
363
the cages may be due partially to reduced rates of pre-
dation or to the enhanced settlement and migration of
juveniles into the dense algal canopies that formed
after the exclusion of large roving herbivores. Ironically,
the small herbivores and detritivores that dominated the
cages may have promoted blooms of fleshy seaweeds
by removing filamentous epiphytes and sediment from
the surfaces of macroalgal that were too large or well-
defended for them to consume. These findings provide
robust experimental evidence for trophic cascades or
top-down control—changes in the structure of food-
webs and species composition (e.g., enhanced recruit-
ment of fishes and increased algal biomass) due to re-
duction in the abundance of medium and large fishes
[18, 20, 22]. After 30 months, we removed the mesh
from cages to allow entry once more to roving herbi-
vores and predators. Cover of macroalgae in the newly
Figure 4. A Principal Component Analysis Showing the Divergent
Coral Assemblages in Cages versus Other Experimental Treatments
accessible cages declined rapidly because of intense
grazing, from 53% to 13% after 12 days and to approx-
Cages are colored blue, partial cages are colored red, and open
plots are colored black. Each symbol represents one of the 4 3 25 m2 imately 0 after 30 days [23]. Juvenile fishes in the former
replicates in each experimental treatment. The first two axes cages declined much faster than the algae, by 98% after
explain 63% of the variation among the 12 experimental replicates. only 3 days, presumably because of predation. In the
Caribbean, Mumby et al. [24] tested the potential impor-
tance of marine no-take areas for safeguarding parrot-
the absence of a roof minimized caging effects (e.g., be- fish and their ability to control blooms of turf and fleshy
cause of reduced water flow or shading from the cage seaweeds. They found a greater biomass of parrotfishes
structure). Light levels in the cages supported luxuriant and less macroalgae inside a no-take reserve, consis-
algal growth, and macroalgae and juvenile fishes re- tent with the experimental results presented here (al-
cruited in great numbers into them. Second, the partial- though the abundance of adult and juvenile corals was
cage treatment did not show an intermediate reduction not reported). Our large-scale experiment provides di-
in numbers of newly recruited corals (Figure 3B). Juvenile rect evidence that overfishing of herbivores affects
Fungia and Euphyllia (and the coralline alga, Mesophyl- more than just the targeted stocks and can also influ-
lum purpurescens) are normally found in deeper water ence the resilience of coral reefs to climate change.
and on shaded vertical surfaces and are rare on shallow Process-oriented research, exemplified by the exper-
reef crests. Consequently, the divergent response by imental manipulations presented here, provides a more
juvenile corals among the experimental treatments (Fig- rigorous basis for coral-reef management than conven-
ure 4) is more likely to reflect a range of tolerances to tional approaches. In particular, the current focus on
shading by the dense stands of Sargassum than differ- descriptive mapping and monitoring of reefs needs to
ences among experimental treatments in the rate of be substantially broadened for better understanding of
delivery of larvae by currents. critical processes that underlie resilience. Our results
Mortality rates of older coral colonies, which had sur- demonstrate that loss of coral-reef resilience can be
vived bleaching and were already established when the readily quantified with several metrics (e.g., depletion
experiment began, were more than double in the cages of key functional groups of fishes, reduced rates of coral
(24.2% after 30 months compared to 9.8% for partial recruitment and population regeneration, sublethal im-
cages and 11.3% for open plots (Figure 3B; RM-ANOVA, pacts, etc.). Furthermore, our findings show that local
F = 4.29, p < 0.05). Recruitment was insufficient to management efforts in support of resilience can afford
counter these losses in the cages, where the total num- significant protection against threats that are much
ber of coral colonies decreased by an average of 72 6 32 larger in scale. Preventing coral bleaching is not a tracta-
per 25 m2 (a 26% decline). In contrast, counts of corals ble management goal at meaningful spatial or temporal
increased by 43 6 21 (16%) and 39 6 24 (14%) per scales, and a long-term solution will require global re-
25 m2 in partial cages and open plots, respectively. In ductions of greenhouse gases over decadal timeframes.
addition to changes in mortality of corals, we also re- On the other hand, supporting resilience in anticipation
corded significant differences in sublethal indices of of bleaching and other recurrent disturbances can be
coral condition attributable to indirect impacts of herbiv- achieved locally by changing destructive human activi-
orous fishes (see Supplemental Data). ties (e.g., overfishing and pollution) and thereby reduc-
ing the likelihood of undesirable phase shifts. Achieving
this outcome will require the linking of ecological resil-
Conclusions ience to social and governance structures and involve
scientists, other stakeholders, environmental man-
Implications for Coral-Reef Management agers, and policy makers [25–26]. A resilience-based
The spatial and temporal scales of our experiment approach represents a fundamental change of focus,
(300 m2, 30 months) was sufficiently large that we suc- from reactive to proactive management, aimed at sus-
cessfully generated a phase shift to macroalgal domi- taining the socioeconomic and ecological value of coral
nance. The increased numbers of small fishes inside reefs in an increasingly uncertain world.
Current Biology
364
Experimental Procedures initially in September 2000 and again in April 2003, with a grid of
100 3 0.25 m2 quadrats covering each of the 12 experimental plots.
Study Site and Experimental Treatments A comparison of the two censuses yielded data on recruitment
The fish-exclusion experiment was undertaken on the inner Great (arrival of new colonies) and coral composition.
Barrier Reef, in Pioneer Bay on the leeward coast of Orpheus Island Coral tissue thickness, an index of biomass and physiological
(18 360 S, 146 290 E), a high-island approximately 10 km offshore condition, was measured with calipers after 2 years in 64 colonies
from the Australian mainland. Like many continental reefs in Austral- of Porites cylindrica from within two cages and outside. Those col-
asia, the reef fauna is highly diverse, with a benthos dominated by onies from within cages were (1) positioned at least 10 cm away
massive and branching scleractinians and alcyonacean soft corals. from the nearest clump of macroalgae, (2) shaded or (3) partially
The water is turbid (typical horizontal visibility is 5–8 m), and the overgrown by macroalgae. Reproductive output of corals was mea-
typical tidal range is 3–3.5 m. The sheltered reef fringing the lee sured in 90 experimental fragments of Montipora digitata that were
of the island seldom experiences breaking waves except during placed 17 weeks before spawning outside and within two cages, the
rare storms and cyclones. Fishing has been banned in Pioneer Bay latter either positioned in the open or beneath clumps of macroalgae
since 1987. (principally Padina). After 14 weeks, fragments were collected and
The three experimental treatments were (1) four 5 3 5 m fully- decalcified for an estimation of egg size, number of eggs per polyp,
meshed roofless cages for excluding all large and medium fishes and number of reproductive polyps.
(Figure 1C), (2) four partially meshed cage controls that afforded ac-
cess along 50% of each perimeter to control for any effects of the Supplemental Data
caging structure, and (3) four open plots. Each of the 12 replicates Supplemental Data include additional Experimental Procedures,
was 25 m2 in area. The cage and partial cage framework consisted three figures and two movies and are available with this article online
of eight 4-m-tall vertical lengths of 50-mm-diameter tubular steel at http://www.current-biology.com/cgi/content/full/17/4/360/DC1/.
(at each corner and midway along each side), three horizontal
lengths along each side at the bottom, middle, and top, and an inter-
nal cross of tubing that connected horizontally between the four Acknowledgments
middle vertical uprights. We anchored the vertical tubes by sliding
them over a 2 m steel bar that was hammered halfway into the sub- We thank the Australian Research Council for providing financial
strate and cemented in place. Eight stays were also attached to each support, the Great Barrier Reef Marine Park Authority for granting
cage to prevent them from lifting. A door to each cage (2 3 0.8 m) a research permit, and a small army of student volunteers for help-
provided access at all tide levels. The 4 m height of the cages and ing with routine scrubbing of the experimental cages. W. Evans,
partial cages obviated the need for a roof because they extended R. Gibson, C. Hughes, J. Madin, and A. McDonald assisted with
a few decimeters above water at high tide, and the base always the construction and removal of the experiment. We are grateful
remained submerged. The plastic mesh on cages and partial cages also to M.J. Boyle, D. Gibson, and A. Hoey for technical assistance.
(1 cm2 for the bottom 2 m, and 2 cm2 for the top 3 m) was scrubbed
every 7–10 days to prevent fouling. A weighted net sealed the bot- Received: November 16, 2006
tom. After 30 months, we removed the mesh from cages and partial Revised: December 18, 2006
cages and closely followed the immediate response of fishes and Accepted: December 18, 2006
macroalgae. Diadema sea urchins are rare at this location. Three Published online: February 8, 2007
were removed from the cages (100 m2) at the start of the experiment.
References
Numbers and Sizes of Herbivorous Fishes
The abundance of herbivorous fishes in each treatment was mea- 1. Wilkinson C., ed. (2004). Status of Coral Reefs of the World
sured after 28 months with 90 hr of high-resolution remote video. (Townsville, Australia: Australian Institute of Marine Science).
In each of the 12 cages, partial cages, and open plots, five randomly 2. Hoegh-Guldberg, O. (1999). Climate change, coral bleaching
placed 1 m2 quadrats were censused for 90 min with remote video and the future of the world’s coral reefs. Mar. Freshw. Res. 50,
cameras (so that diver effects could be eliminated). Recording was 839–866.
undertaken within 90 min of high tide between 1000 and 1600, with 3. Harvell, C.D., Mitchell, C.E., Ward, J.R., Altizer, S., Dobson, A.P.,
randomly allocated times among treatments. Fish were identified Ostfeld, R.S., and Samuel, M.D. (2002). Climate warming and
to species, and their body lengths were recorded and converted disease risks for terrestrial and marine biota. Science 296,
to biomass with standard length-weight regressions. 2158–2162.
4. Hughes, T.P., Baird, A.H., Bellwood, D.R., Card, M., Connolly,
Response of Macroalgae and Coralline Algae S.R., Folke, C., Grosberg, R., Hoegh-Guldberg, O., Jackson,
Fleshy macroalgae and noncoralline crusts were identified in situ to J.B., Kleypas, J., et al. (2003). Climate change, human impacts,
genus, and their abundance in each of the 12 plots was estimated on and the resilience of coral reefs. Science 301, 929–933.
a six-point categorical scale (0 = ‘‘absent,’’ 1 = ‘‘rare,’’ 2 = ‘‘uncom- ¨
5. Bellwood, D.R., Hughes, T.P., Folke, C., and Nystrom, M. (2004).
mon,’’ 3 = ‘‘common,’’ 4 = ‘‘abundant,’’ and 5 = ‘‘dominant’’). A total Confronting the coral reef crisis. Nature 429, 827–833.
of thirteen censuses were made, and two were made after the re- 6. Pandolfi, J.M., Bradbury, R.H., Sala, E., Hughes, T.P., Bjorndal,
moval of the mesh. In addition, the percentage of macroalgal cover K.A., Cooke, R.G., McArdle, D., McClenachan, L., Newman,
was measured from 16 photographic censuses (0.25 m2 resolution) M.J., Paredes, G., et al. (2003). Global trajectories of the long-
of 2 3 2 m quadrats located centrally within each cage, partial cage, term decline of coral reef ecosystems. Science 301, 955–958.
and open plot. Abundances of crustose coralline algae were mea- 7. Knowlton, N. (2001). The future of coral reefs. Proc. Natl. Acad.
sured initially and after 26 months from photographs of 33 perma- Sci. USA 98, 5419–5425.
nently marked 10 3 10 cm quadrats (two to four quadrats distributed 8. Gardner, T.A., Cote, I., Gill, J.A., and Watkinson, A.R. (2003).
among each of the 12 replicate plots). At the final census, macroal- Long-term region-wide declines in Caribbean corals. Science
gae were first removed from the quadrats, by brushing to expose 301, 958–960.
overgrown corallines, and then the quadrats were rephotographed. 9. Moberg, F., and Folke, C. (1999). Ecological goods and services
Samples of live and dead corallines were collected for taxonomic of coral reef ecosystems. Ecol. Econ. 29, 215–233.
identification. 10. Baird, A.H., and Marshall, P.A. (2002). Mortality, growth and re-
production in scleractinian corals following bleaching on the
Responses of Corals Great Barrier Reef. Mar. Ecol. Prog. Ser. 237, 133–141.
Coral cover, survivorship of colonies, and cover of macroalgae was 11. Loya, Y., Sakai, K., Yamazato, K., Nakano, Y., Sambali, H., and
estimated from digital photographs (0.25 m2 resolution, 16 cen- van Woesik, R. (2001). Coral bleaching: The winners and losers.
suses) of 2 3 2 m areas positioned centrally within each of the 12 Ecol. Lett. 4, 122–131.
experimental areas. In addition, all corals greater than 1 cm in the ex- 12. Donner, S.D., Skirving, W.J., Little, C.M., Oppenheimer, M., and
periment were identified (to species, where possible) and mapped Hoegh-Guldberg, O. (2005). Global assessment of coral
Phase Shifts and Resilience of Coral Reefs
365
bleaching and required rates of adaptation under climate
change. Glob. Change Biol. 11, 1–15.
13. Folke, C., Carpenter, S.R., Elmqvist, T., Gunderson, L.H., Hol-
ling, C.S., Scheffer, M., and Walker, B. (2004). Phase shifts, resil-
ience and biodiversity in ecosystem management. Annu. Rev.
Ecol. Syst. 35, 1–21.
14. ¨
Nystrom, M., Folke, C., and Moberg, F. (2000). Coral reef distur-
bance and resilience in a human-dominated environment.
Trends Ecol. Evol. 15, 413–417.
15. Westmacott, S., Teleki, K., Wells, S., and West, J.M. (2000).
Management of Bleached and Severely Damaged Coral Reefs
(Gland, Switzerland: ICUN).
16. Munro J.L., ed. (1983). Caribbean Coral Reef Fisheries, 2nd
Edition, Volume 7 (Makati City, Philippines: ICLARM Studies
and Reviews), pp. 1–276.
17. Hughes, T.P. (1994). Catastrophes, phase-shifts, and large-
scale degradation of a Caribbean coral reef. Science 265,
1547–1551.
18. Jackson, J.B.C., Kirby, M.X., Berger, W.H., Bjorndal, K.A., Bots-
ford, L.W., Bourque, B.J., Bradbury, R.H., Cooke, R., Erlandson,
J., Estes, J.A., et al. (2001). Historical overfishing and the recent
collapse of coastal ecosystems. Science 293, 629–638.
19. Russ, G.R., and Alcala, A.C. (2003). Marine reserves: Rates and
patterns of recovery and decline of predatory fish, 1983–2000.
Ecol. Appl. 13, 1553–1565.
20. Steneck, R.S. (1998). Human influences on coastal ecosystems:
Does overfishing create trophic cascades? Trends Ecol. Evol.
13, 429–430.
21. Smith, S.V., Kimmerer, W.J., Laws, E.A., Brock, R.E., and Walsh,
T.W. (1981). Kaneohe Bay sewage diversion experiment: Per-
spectives on ecosystem responses to nutritional perturbation.
Pacific Science 35, 279–395.
22. Hughes, T.P., Szmant, A.M., Steneck, R.S., Carpenter, R.C., and
Miller, S. (1999). Algal blooms on coral reefs: What are the
causes? Limnol. Oceanogr. 44, 1583–1586.
23. Bellwood, D.R., Hughes, T.P., and Hoey, A.S. (2006). Sleeping
functional group drives coral-reef recovery. Curr. Biol. 16,
2434–2439.
24. Mumby, P.J., Dahlgren, C.P., Harborne, A.R., Kappel, C.V., Mi-
cheli, F., Brumbaugh, D.R., Holmes, K.E., Mendes, J.M., Broad,
K., Sanchirico, J.N., et al. (2006). Fishing, trophic cascades, and
the process of grazing on coral reefs. Science 311, 98–101.
25. Gunderson L.H. and Pritchard, L., Jr., eds. (2002). Resilience
and the Behaviour of Large-Scale Systems (Washington, D.C.:
Island Press).
26. Folke, C., Holling, C.S., and Perrings, C. (1996). Biological diver-
sity, ecosystems, and the human scale. Ecol. Appl. 4, 1018–
1024.
Report
Phase Shifts, Herbivory,
and the Resilience of Coral Reefs
to Climate Change
Terence P. Hughes,1,* Maria J. Rodrigues,1,2 macroalgae, which suppressed the fecundity, recruit-
David R. Bellwood,1,2 Daniela Ceccarelli,1,2 ment, and survival of corals. Consequently, manage-
Ove Hoegh-Guldberg,3 Laurence McCook,1,4 ment of fish stocks is a key component in preventing
Natalie Moltschaniwskyj,5 Morgan S. Pratchett,1 phase shifts and managing reef resilience. Importantly,
Robert S. Steneck,1,6 and Bette Willis1,2 local stewardship of fishing effort is a tractable goal for
1
Australian Research Council Centre of Excellence conservation of reefs, and this local action can also
for Coral Reef Studies provide some insurance against larger-scale distur-
James Cook University bances such as mass bleaching, which are impractical
Townsville QLD 4811 to manage directly.
Australia
2
School of Marine and Tropical Biology
James Cook University Results and Discussion
Townsville QLD 4811
Australia The ecosystem goods and services provided by healthy
3
Australian Research Council Centre of Excellence coral reefs are a key component in the economic, social,
for Coral Reef Studies and Centre for Marine Science and cultural fabric of many tropical maritime countries
University of Queensland [1, 9]. Until recently, land-based pollution and overfish-
St Lucia ing were considered to be the major threats to coral
Brisbane QLD 4072 reefs. Today, reefs face additional pressure from ther-
Australia mal stress and emergent diseases that are closely linked
4
Great Barrier Reef Marine Park Authority to global warming [1–8]. In the most damaging case to
PO Box 1379 date, 16% of the world’s reefs were impacted in 1997–
Townsville QLD 4810 1998 by a regional-scale bleaching event that affected
Australia the Great Barrier Reef, vast tracts of the western Pacific,
5
School of Aquaculture the Indo-Australian Archipelago, and the Indian Ocean
University of Tasmania, Locked Bag 1-370 [1, 10–11]. Climate-change projections indicate that
Launceston, Tasmania 7250 similar events will reoccur with increased frequency in
Australia the coming decades [2, 12], highlighting the urgency of
6
School of Marine Sciences developing improved tools for managing reefs in the
University of Maine face of escalating threats [4–5, 13–15].
Darling Marine Center Here, we experimentally examine the resilience of
193 Clarks Cove Road coral-dominated assemblages on the Great Barrier
Walpole, Maine 04573 Reef and the processes underlying a phase shift to
macroalgal dominance (Figure 1A). We define resilience
as the ability of reefs to absorb recurrent disturbances
Summary (e.g., from cyclones, outbreaks of predators, or coral
bleaching events) and rebuild coral-dominated systems.
Many coral reefs worldwide have undergone phase Loss of resilience can lead to a phase or regime shift to an
shifts to alternate, degraded assemblages because of alternate assemblage that is typically characterized by
the combined effects of overfishing, declining water hyperabundances of fleshy seaweeds or other opportu-
quality, and the direct and indirect impacts of climate nistic species. The experiment was designed to simulate
change [1–9]. Here, we experimentally manipulated the depletion of large predatory and herbivorous fishes
the density of large herbivorous fishes to test their caused by chronic overfishing [16–21] and to investigate
influence on the resilience of coral assemblages in their role in the regeneration of reefs after recent mass
the aftermath of regional-scale bleaching in 1998, the bleaching and the mortality of corals (see Experimental
largest coral mortality event recorded to date. The ex- Procedures). The scale and timing of the experiment
periment was undertaken on the Great Barrier Reef, allowed us to measure the postbleaching dynamics of
within a no-fishing reserve where coral abundances a rich coral assemblage (77 species represented by
and diversity had been sharply reduced by bleaching 4569 colonies were recorded by the end of the experi-
[10]. In control areas, where fishes were abundant, ment), and its location provided us with a baseline
algal abundance remained low, whereas coral cover comparison of an unusually intact fish fauna on heavily
almost doubled (to 20%) over a 3 year period, primarily grazed reef crests within an established no-take area
because of recruitment of species that had been locally of the Great Barrier Reef Marine Park. This is the first
extirpated by bleaching. In contrast, exclusion of large replicated herbivore-exclusion experiment that explicitly
herbivorous fishes caused a dramatic explosion of examines herbivore-algae-coral interactions in the con-
text of climate change. We demonstrate that exclusion
of larger fishes profoundly erodes the resilience of coral
*Correspondence: terry.hughes@jcu.edu.au reefs and their ability to regenerate after bleaching, with
Phase Shifts and Resilience of Coral Reefs
361
Figure 1. Experimental Phase Shifts on the Great Barrier Reef
(A) Roofless cages and partial cages constructed on the seaward edge of reef crest. Each structure is 5 3 5 m in area and 4 m tall. Note the 2 m
high door in the cage in the center of the photograph, for access at low tide.
(B) Growths of Sargassum up to 3 m tall dwarf understory corals inside a fish-exclusion cage.
(C) When fishes were experimentally excluded, a foliose coralline alga, Mesophyllum purpurescens, replaced shallow-water grazer-resistant
species.
(D) Coral recruits settled on dead corals killed 5 years earlier by thermally induced bleaching in 1998. Grazing of the dead substrate by herbivores
is crucial for settlement and early survival of corals and coralline algae.
major implications for reef ecology, conservation, and censuses), ranging up to a maximum of 10% and 7%,
management. respectively (Figure 2A). In contrast, algal cover in the
cages far exceeded the two control treatments through-
Experimentally Induced Phase Shift out the experiment, reaching up to 91%, and averaging
Our experimental exclusion of fishes replicated the pau- 56% 6 21% (S.E.) after 30 months (repeated-measures
city of medium and large fishes that is characteristic of ANOVA, F = 3.82, p < 0.05; Figure 2). By the end of the
chronically overfished reefs in S.E. Asia, the Caribbean, experiment, algal biomass in the cages was 9 to 20 times
and elsewhere [16–19]. The biomass of herbivorous higher than in partial cages and open plots (1363 6 234,
fishes inside cages (Figure 1A) was reduced to levels 146 6 49, and 68 6 28 g wet weight per m2, respectively:
seven to ten times lower than in adjacent partial cages ANOVA, F = 20.8, p < 0.001). Over time, the species com-
and open plots (0.45 6 0.08 [S.E.], 4.29 6 2.81, and position of macroalgae in the cages diverged dramati-
3.12 6 1.24 kg/m2 per hr of video observation, respec- cally from the other two treatments (Figure S2). Dense
tively, F = 7.79, p < 0.001; Figure S1 in the Supplemental thickets of Sargassum, previously absent on the reef
Data available with this article online). In response to crest, grew to 3 m in height inside the cages, with maxi-
the experimental exclusion of larger herbivorous fishes, mum densities of greater than 1000 plants (holdfasts) per
benthic assemblages in the cages followed a fundamen- 25 m2 and a biomass of up to 8.55 kg wet weight per m2
tally different trajectory over time, with upright fleshy (Figure 1B and Movie S2). Cover and species composi-
macroalgae rather than corals and algal turfs becoming tion of crustose coralline algae also diverged in the three
predominant, mimicking similar responses on many experimental treatments (Figure 1C and Figure S3).
overfished and polluted reefs worldwide [4–8, 20–21].
In the aftermath of massive loss of corals on Orpheus Herbivory Boosts the Resilience of Coral
Island in 1998 [10], roving herbivorous fishes continued Assemblages to Global Warming
to suppress the biomass of macroalgae and thus facili- In tandem with the changes in macroalgae and coral-
tated the recruitment of corals (Movie S1). In the partial lines, the trajectory of coral reassembly after the 1998
cages and open plots where fish grazing was uninhib- bleaching event diverged markedly in the fish-exclusion
ited, the cover of macroalgae (primarily the calcified cages compared to the partial cages and open plots
red alga, Galaxaura subfruticulosa) averaged only (Figure 2B). Initially, the most prevalent taxa (accounting
4.1% and 1.7% during the experimental period (n = 16 for >80% of coral cover) were branching Porites
Current Biology
362
Figure 3. Demographic Responses of Corals
(A) Recruitment of corals into the three experimental treatments.
Figure 2. Contrasting Trajectories of Macroalgae and Corals after
Error bars are SE.
Exclusion of Fishes
(B) Mortality of coral colonies originally present in cages, partial
(A) Macroalgal cover. Error bars are SE. cages, and open plots. Error bars are SE.
(B) Relative coral cover over time among three experimental treat-
ments. Absolute coral cover after 130 weeks was 7.7% 6 1.0%
(S.E.), 19.2% 6 2.3%, and 20.2% 6 2.2% in the three treatments
(see text for analysis). Census dates were the same for all treatments
and are slightly staggered in the plots for clarity. Error bars are SE.
1062 new recruits from 26 coral genera were recorded
in the three treatments at the end of the experiment (Fig-
ure 1D). Overall, coral recruitment in cages was approx-
imately two-thirds lower (39 6 11 recruits per 25 m2,
cylindrica, massive Porites spp. (especially P. lobata compared to 108 6 26 for partial cages and 118 6 21
and P. rus), and massive faviids (principally heads of in the open plots; ANOVA, F = 150.9, p < 0.001; Fig-
Goniastrea, Favia, and Montastrea spp.) that had sur- ure 3A). Acropora, which was virtually eliminated in
vived the bleaching event 2 years prior to the initiation 1998 from the reef crest at Orpheus Island by bleaching
of the experiment. Alcyonacean soft corals and branch- [10], accounted for 246 of the recruits, representing 23%
ing hard corals, particularly a diverse suite of Acropora of the total. The dominant adult genus, Porites, had only
species, were virtually eliminated from shallow sites by two recruits in the cages, compared to 45 elsewhere (19
bleaching [10], and only a few small recruits were pres- in partial cages, 26 in open plots; F = 12.49, p = 0.003).
ent (<0.1% cover) when the experiment began in 2000. Similarly, Acropora recruits were three times more
In the fish-exclusion cages, total coral cover grew from abundant in partial cages and open plots (F = 7.7, p =
6.0% 6 0.8% (S.E.) to 7.7% 6 1.0% after 30 months. 0.011). In contrast, Fungia and Euphyllia were more
Coral cover increased much more quickly inside the par- abundant inside cages, where together they comprised
tial cages and open plots, reaching 19.2% 6 2.3% and 18% of the recruits compared to only 3% in each of the
20.2% 6 2.2%, respectively (RM-ANOVA, F = 3.82, p < two other treatments. A principal component analysis
0.05). In relative terms, coral cover increased by 28% in- summarizes the striking divergence in the composition
side the cages compared to 68% for partial cages and of coral recruits in cages compared to partial cages or
83% for open plots (Figure 2B). open plots (Figure 4). Recruit assemblages in the partial
The divergence in coral cover among treatments was cages and open plots were indistinguishable.
attributable to both lower recruitment and higher mortal- The suppression of coral recruitment inside cages is
ity of established corals after the experimental reduc- unlikely to have been an experimental artifact for two
tions of fish biomass (Figures 3A and 3B). A total of reasons. First, the considerable size of the cages and
Phase Shifts and Resilience of Coral Reefs
363
the cages may be due partially to reduced rates of pre-
dation or to the enhanced settlement and migration of
juveniles into the dense algal canopies that formed
after the exclusion of large roving herbivores. Ironically,
the small herbivores and detritivores that dominated the
cages may have promoted blooms of fleshy seaweeds
by removing filamentous epiphytes and sediment from
the surfaces of macroalgal that were too large or well-
defended for them to consume. These findings provide
robust experimental evidence for trophic cascades or
top-down control—changes in the structure of food-
webs and species composition (e.g., enhanced recruit-
ment of fishes and increased algal biomass) due to re-
duction in the abundance of medium and large fishes
[18, 20, 22]. After 30 months, we removed the mesh
from cages to allow entry once more to roving herbi-
vores and predators. Cover of macroalgae in the newly
Figure 4. A Principal Component Analysis Showing the Divergent
Coral Assemblages in Cages versus Other Experimental Treatments
accessible cages declined rapidly because of intense
grazing, from 53% to 13% after 12 days and to approx-
Cages are colored blue, partial cages are colored red, and open
plots are colored black. Each symbol represents one of the 4 3 25 m2 imately 0 after 30 days [23]. Juvenile fishes in the former
replicates in each experimental treatment. The first two axes cages declined much faster than the algae, by 98% after
explain 63% of the variation among the 12 experimental replicates. only 3 days, presumably because of predation. In the
Caribbean, Mumby et al. [24] tested the potential impor-
tance of marine no-take areas for safeguarding parrot-
the absence of a roof minimized caging effects (e.g., be- fish and their ability to control blooms of turf and fleshy
cause of reduced water flow or shading from the cage seaweeds. They found a greater biomass of parrotfishes
structure). Light levels in the cages supported luxuriant and less macroalgae inside a no-take reserve, consis-
algal growth, and macroalgae and juvenile fishes re- tent with the experimental results presented here (al-
cruited in great numbers into them. Second, the partial- though the abundance of adult and juvenile corals was
cage treatment did not show an intermediate reduction not reported). Our large-scale experiment provides di-
in numbers of newly recruited corals (Figure 3B). Juvenile rect evidence that overfishing of herbivores affects
Fungia and Euphyllia (and the coralline alga, Mesophyl- more than just the targeted stocks and can also influ-
lum purpurescens) are normally found in deeper water ence the resilience of coral reefs to climate change.
and on shaded vertical surfaces and are rare on shallow Process-oriented research, exemplified by the exper-
reef crests. Consequently, the divergent response by imental manipulations presented here, provides a more
juvenile corals among the experimental treatments (Fig- rigorous basis for coral-reef management than conven-
ure 4) is more likely to reflect a range of tolerances to tional approaches. In particular, the current focus on
shading by the dense stands of Sargassum than differ- descriptive mapping and monitoring of reefs needs to
ences among experimental treatments in the rate of be substantially broadened for better understanding of
delivery of larvae by currents. critical processes that underlie resilience. Our results
Mortality rates of older coral colonies, which had sur- demonstrate that loss of coral-reef resilience can be
vived bleaching and were already established when the readily quantified with several metrics (e.g., depletion
experiment began, were more than double in the cages of key functional groups of fishes, reduced rates of coral
(24.2% after 30 months compared to 9.8% for partial recruitment and population regeneration, sublethal im-
cages and 11.3% for open plots (Figure 3B; RM-ANOVA, pacts, etc.). Furthermore, our findings show that local
F = 4.29, p < 0.05). Recruitment was insufficient to management efforts in support of resilience can afford
counter these losses in the cages, where the total num- significant protection against threats that are much
ber of coral colonies decreased by an average of 72 6 32 larger in scale. Preventing coral bleaching is not a tracta-
per 25 m2 (a 26% decline). In contrast, counts of corals ble management goal at meaningful spatial or temporal
increased by 43 6 21 (16%) and 39 6 24 (14%) per scales, and a long-term solution will require global re-
25 m2 in partial cages and open plots, respectively. In ductions of greenhouse gases over decadal timeframes.
addition to changes in mortality of corals, we also re- On the other hand, supporting resilience in anticipation
corded significant differences in sublethal indices of of bleaching and other recurrent disturbances can be
coral condition attributable to indirect impacts of herbiv- achieved locally by changing destructive human activi-
orous fishes (see Supplemental Data). ties (e.g., overfishing and pollution) and thereby reduc-
ing the likelihood of undesirable phase shifts. Achieving
this outcome will require the linking of ecological resil-
Conclusions ience to social and governance structures and involve
scientists, other stakeholders, environmental man-
Implications for Coral-Reef Management agers, and policy makers [25–26]. A resilience-based
The spatial and temporal scales of our experiment approach represents a fundamental change of focus,
(300 m2, 30 months) was sufficiently large that we suc- from reactive to proactive management, aimed at sus-
cessfully generated a phase shift to macroalgal domi- taining the socioeconomic and ecological value of coral
nance. The increased numbers of small fishes inside reefs in an increasingly uncertain world.
Current Biology
364
Experimental Procedures initially in September 2000 and again in April 2003, with a grid of
100 3 0.25 m2 quadrats covering each of the 12 experimental plots.
Study Site and Experimental Treatments A comparison of the two censuses yielded data on recruitment
The fish-exclusion experiment was undertaken on the inner Great (arrival of new colonies) and coral composition.
Barrier Reef, in Pioneer Bay on the leeward coast of Orpheus Island Coral tissue thickness, an index of biomass and physiological
(18 360 S, 146 290 E), a high-island approximately 10 km offshore condition, was measured with calipers after 2 years in 64 colonies
from the Australian mainland. Like many continental reefs in Austral- of Porites cylindrica from within two cages and outside. Those col-
asia, the reef fauna is highly diverse, with a benthos dominated by onies from within cages were (1) positioned at least 10 cm away
massive and branching scleractinians and alcyonacean soft corals. from the nearest clump of macroalgae, (2) shaded or (3) partially
The water is turbid (typical horizontal visibility is 5–8 m), and the overgrown by macroalgae. Reproductive output of corals was mea-
typical tidal range is 3–3.5 m. The sheltered reef fringing the lee sured in 90 experimental fragments of Montipora digitata that were
of the island seldom experiences breaking waves except during placed 17 weeks before spawning outside and within two cages, the
rare storms and cyclones. Fishing has been banned in Pioneer Bay latter either positioned in the open or beneath clumps of macroalgae
since 1987. (principally Padina). After 14 weeks, fragments were collected and
The three experimental treatments were (1) four 5 3 5 m fully- decalcified for an estimation of egg size, number of eggs per polyp,
meshed roofless cages for excluding all large and medium fishes and number of reproductive polyps.
(Figure 1C), (2) four partially meshed cage controls that afforded ac-
cess along 50% of each perimeter to control for any effects of the Supplemental Data
caging structure, and (3) four open plots. Each of the 12 replicates Supplemental Data include additional Experimental Procedures,
was 25 m2 in area. The cage and partial cage framework consisted three figures and two movies and are available with this article online
of eight 4-m-tall vertical lengths of 50-mm-diameter tubular steel at http://www.current-biology.com/cgi/content/full/17/4/360/DC1/.
(at each corner and midway along each side), three horizontal
lengths along each side at the bottom, middle, and top, and an inter-
nal cross of tubing that connected horizontally between the four Acknowledgments
middle vertical uprights. We anchored the vertical tubes by sliding
them over a 2 m steel bar that was hammered halfway into the sub- We thank the Australian Research Council for providing financial
strate and cemented in place. Eight stays were also attached to each support, the Great Barrier Reef Marine Park Authority for granting
cage to prevent them from lifting. A door to each cage (2 3 0.8 m) a research permit, and a small army of student volunteers for help-
provided access at all tide levels. The 4 m height of the cages and ing with routine scrubbing of the experimental cages. W. Evans,
partial cages obviated the need for a roof because they extended R. Gibson, C. Hughes, J. Madin, and A. McDonald assisted with
a few decimeters above water at high tide, and the base always the construction and removal of the experiment. We are grateful
remained submerged. The plastic mesh on cages and partial cages also to M.J. Boyle, D. Gibson, and A. Hoey for technical assistance.
(1 cm2 for the bottom 2 m, and 2 cm2 for the top 3 m) was scrubbed
every 7–10 days to prevent fouling. A weighted net sealed the bot- Received: November 16, 2006
tom. After 30 months, we removed the mesh from cages and partial Revised: December 18, 2006
cages and closely followed the immediate response of fishes and Accepted: December 18, 2006
macroalgae. Diadema sea urchins are rare at this location. Three Published online: February 8, 2007
were removed from the cages (100 m2) at the start of the experiment.
References
Numbers and Sizes of Herbivorous Fishes
The abundance of herbivorous fishes in each treatment was mea- 1. Wilkinson C., ed. (2004). Status of Coral Reefs of the World
sured after 28 months with 90 hr of high-resolution remote video. (Townsville, Australia: Australian Institute of Marine Science).
In each of the 12 cages, partial cages, and open plots, five randomly 2. Hoegh-Guldberg, O. (1999). Climate change, coral bleaching
placed 1 m2 quadrats were censused for 90 min with remote video and the future of the world’s coral reefs. Mar. Freshw. Res. 50,
cameras (so that diver effects could be eliminated). Recording was 839–866.
undertaken within 90 min of high tide between 1000 and 1600, with 3. Harvell, C.D., Mitchell, C.E., Ward, J.R., Altizer, S., Dobson, A.P.,
randomly allocated times among treatments. Fish were identified Ostfeld, R.S., and Samuel, M.D. (2002). Climate warming and
to species, and their body lengths were recorded and converted disease risks for terrestrial and marine biota. Science 296,
to biomass with standard length-weight regressions. 2158–2162.
4. Hughes, T.P., Baird, A.H., Bellwood, D.R., Card, M., Connolly,
Response of Macroalgae and Coralline Algae S.R., Folke, C., Grosberg, R., Hoegh-Guldberg, O., Jackson,
Fleshy macroalgae and noncoralline crusts were identified in situ to J.B., Kleypas, J., et al. (2003). Climate change, human impacts,
genus, and their abundance in each of the 12 plots was estimated on and the resilience of coral reefs. Science 301, 929–933.
a six-point categorical scale (0 = ‘‘absent,’’ 1 = ‘‘rare,’’ 2 = ‘‘uncom- ¨
5. Bellwood, D.R., Hughes, T.P., Folke, C., and Nystrom, M. (2004).
mon,’’ 3 = ‘‘common,’’ 4 = ‘‘abundant,’’ and 5 = ‘‘dominant’’). A total Confronting the coral reef crisis. Nature 429, 827–833.
of thirteen censuses were made, and two were made after the re- 6. Pandolfi, J.M., Bradbury, R.H., Sala, E., Hughes, T.P., Bjorndal,
moval of the mesh. In addition, the percentage of macroalgal cover K.A., Cooke, R.G., McArdle, D., McClenachan, L., Newman,
was measured from 16 photographic censuses (0.25 m2 resolution) M.J., Paredes, G., et al. (2003). Global trajectories of the long-
of 2 3 2 m quadrats located centrally within each cage, partial cage, term decline of coral reef ecosystems. Science 301, 955–958.
and open plot. Abundances of crustose coralline algae were mea- 7. Knowlton, N. (2001). The future of coral reefs. Proc. Natl. Acad.
sured initially and after 26 months from photographs of 33 perma- Sci. USA 98, 5419–5425.
nently marked 10 3 10 cm quadrats (two to four quadrats distributed 8. Gardner, T.A., Cote, I., Gill, J.A., and Watkinson, A.R. (2003).
among each of the 12 replicate plots). At the final census, macroal- Long-term region-wide declines in Caribbean corals. Science
gae were first removed from the quadrats, by brushing to expose 301, 958–960.
overgrown corallines, and then the quadrats were rephotographed. 9. Moberg, F., and Folke, C. (1999). Ecological goods and services
Samples of live and dead corallines were collected for taxonomic of coral reef ecosystems. Ecol. Econ. 29, 215–233.
identification. 10. Baird, A.H., and Marshall, P.A. (2002). Mortality, growth and re-
production in scleractinian corals following bleaching on the
Responses of Corals Great Barrier Reef. Mar. Ecol. Prog. Ser. 237, 133–141.
Coral cover, survivorship of colonies, and cover of macroalgae was 11. Loya, Y., Sakai, K., Yamazato, K., Nakano, Y., Sambali, H., and
estimated from digital photographs (0.25 m2 resolution, 16 cen- van Woesik, R. (2001). Coral bleaching: The winners and losers.
suses) of 2 3 2 m areas positioned centrally within each of the 12 Ecol. Lett. 4, 122–131.
experimental areas. In addition, all corals greater than 1 cm in the ex- 12. Donner, S.D., Skirving, W.J., Little, C.M., Oppenheimer, M., and
periment were identified (to species, where possible) and mapped Hoegh-Guldberg, O. (2005). Global assessment of coral
Phase Shifts and Resilience of Coral Reefs
365
bleaching and required rates of adaptation under climate
change. Glob. Change Biol. 11, 1–15.
13. Folke, C., Carpenter, S.R., Elmqvist, T., Gunderson, L.H., Hol-
ling, C.S., Scheffer, M., and Walker, B. (2004). Phase shifts, resil-
ience and biodiversity in ecosystem management. Annu. Rev.
Ecol. Syst. 35, 1–21.
14. ¨
Nystrom, M., Folke, C., and Moberg, F. (2000). Coral reef distur-
bance and resilience in a human-dominated environment.
Trends Ecol. Evol. 15, 413–417.
15. Westmacott, S., Teleki, K., Wells, S., and West, J.M. (2000).
Management of Bleached and Severely Damaged Coral Reefs
(Gland, Switzerland: ICUN).
16. Munro J.L., ed. (1983). Caribbean Coral Reef Fisheries, 2nd
Edition, Volume 7 (Makati City, Philippines: ICLARM Studies
and Reviews), pp. 1–276.
17. Hughes, T.P. (1994). Catastrophes, phase-shifts, and large-
scale degradation of a Caribbean coral reef. Science 265,
1547–1551.
18. Jackson, J.B.C., Kirby, M.X., Berger, W.H., Bjorndal, K.A., Bots-
ford, L.W., Bourque, B.J., Bradbury, R.H., Cooke, R., Erlandson,
J., Estes, J.A., et al. (2001). Historical overfishing and the recent
collapse of coastal ecosystems. Science 293, 629–638.
19. Russ, G.R., and Alcala, A.C. (2003). Marine reserves: Rates and
patterns of recovery and decline of predatory fish, 1983–2000.
Ecol. Appl. 13, 1553–1565.
20. Steneck, R.S. (1998). Human influences on coastal ecosystems:
Does overfishing create trophic cascades? Trends Ecol. Evol.
13, 429–430.
21. Smith, S.V., Kimmerer, W.J., Laws, E.A., Brock, R.E., and Walsh,
T.W. (1981). Kaneohe Bay sewage diversion experiment: Per-
spectives on ecosystem responses to nutritional perturbation.
Pacific Science 35, 279–395.
22. Hughes, T.P., Szmant, A.M., Steneck, R.S., Carpenter, R.C., and
Miller, S. (1999). Algal blooms on coral reefs: What are the
causes? Limnol. Oceanogr. 44, 1583–1586.
23. Bellwood, D.R., Hughes, T.P., and Hoey, A.S. (2006). Sleeping
functional group drives coral-reef recovery. Curr. Biol. 16,
2434–2439.
24. Mumby, P.J., Dahlgren, C.P., Harborne, A.R., Kappel, C.V., Mi-
cheli, F., Brumbaugh, D.R., Holmes, K.E., Mendes, J.M., Broad,
K., Sanchirico, J.N., et al. (2006). Fishing, trophic cascades, and
the process of grazing on coral reefs. Science 311, 98–101.
25. Gunderson L.H. and Pritchard, L., Jr., eds. (2002). Resilience
and the Behaviour of Large-Scale Systems (Washington, D.C.:
Island Press).
26. Folke, C., Holling, C.S., and Perrings, C. (1996). Biological diver-
sity, ecosystems, and the human scale. Ecol. Appl. 4, 1018–
1024.