Bellwood et al 06 Coral
Global Change Biology (2006) 12, 1587–1594, doi: 10.1111/j.1365-2486.2006.01204.x
Coral bleaching, reef fish community phase shifts
and the resilience of coral reefs
D AV I D R . B E L L W O O D *w , A N D R E W S . H O E Y *w, J O H N L . A C K E R M A N *z
and M A R T I A L D E P C Z Y N S K I *
*School of Marine Biology, James Cook University, Townsville, Qld 4811, Australia, wAustralian Research Council Centre of
Excellence for Coral Reef Studies, James Cook University, Townsville, Qld 4811, Australia, zFisheries and Marine Sciences Program,
Bureau of Rural Sciences, GPO Box 858, Canberra, ACT 2610, Australia
Abstract
The 1998 global coral bleaching event was the largest recorded historical disturbance of
coral reefs and resulted in extensive habitat loss. Annual censuses of reef fish community
structure over a 12-year period spanning the bleaching event revealed a marked phase
shift from a prebleach to postbleach assemblage. Surprisingly, we found that the
bleaching event had no detectable effect on the abundance, diversity or species richness
of a local cryptobenthic reef fish community. Furthermore, there is no evidence of
regeneration even after 5–35 generations of these short-lived species. These results have
significant implications for our understanding of the response of coral reef ecosystems to
global warming and highlight the importance of selecting appropriate criteria for
evaluating reef resilience.
Keywords: bleaching, community composition, coral reefs, fishes, habitat loss, phase shifts, resilience
Received 3 January 2006; revised version received 24 March 2006; accepted 4 April 2006
Resilience and the ability to regenerate after distur-
Introduction
bance is a central feature of coral reefs. Indeed, it is their
Coral reefs are highly dynamic systems characterized capacity to recolonize and maintain populations in the
by variable and stochastic recruitment and disturbance. face of disturbance that has underpinned their success
Our understanding of these factors has offered much to in dominating exposed and high-energy locations in
explain local variation in population numbers and the tropics (Connell et al., 1997). Recently, large-scale
community composition during periods of apparent bleaching has presented coral reefs with a new chal-
stability (Caley et al., 1996; Connell et al., 1997). Today, lenge: rapid large-scale loss of coral cover (Wilkinson,
however, coral reefs around the world are facing a 2004). This widespread disturbance has impacted nu-
scenario of steady declines in coral cover punctuated merous reefs, many of which were already weakened
by periodic large-scale perturbations. Of these, the 1998 by human activities, including systematic overfishing,
mass coral bleaching event was perhaps the most habitat destruction and pollution (Jackson et al., 2001;
noteworthy (Hoegh-Guldberg, 1999; Hughes et al., Hughes et al., 2003; Pandolfi et al., 2003). The response
2003; Wilkinson, 2004). The critical question to arise of corals to bleaching has been highly variable, with
from these changes is: to what extent can coral reefs many species showing limited evidence of short-term
recover (sensu Edwards et al., 2001) or regenerate (sensu recovery (Loya et al., 2001; Baird & Marshall, 2002;
Hughes et al., 2003) after disturbance (i.e. to what extent Hughes et al., 2003; Donner et al., 2005). Thus, from a
do they exhibit resilience; Holling, 1973; Gunderson, coral perspective the evidence, to date, suggests that
2000)? In particular, can coral reef ecosystems maintain corals show limited short-term resilience to elevated sea
the critical feedbacks, functions and processes in the surface temperatures.
face of climate change? In marked contrast, it appears that reef fishes are
relatively resilient to disturbance, with reef fish assem-
Correspondence: David Bellwood, School of Marine Biology, blages exhibiting only a limited response to the loss of
James Cook University, Townsville, Qld 4811, Australia, corals through large-scale crown of thorns starfish out-
tel. 1 61 (0)7 47814447, fax 1 61 (0)7 47251570, breaks (Williams, 1986; Hart et al., 1996; Sano, 2000) and
e-mail: David.Bellwood@jcu.edu.au coral bleaching (Kokita & Nakazono, 2001; Booth &
r 2006 The Authors
Journal compilation r 2006 Blackwell Publishing Ltd 1587
1588 D . R . B E L L W O O D et al.
Beretta, 2002). In most cases the changes are restricted 2007; Hernaman & Munday, 2005), this is a highly
to changes in a few strongly coral-associated fishes dynamic fish community in close association with the
(Williams, 1986; Kokita & Nakazono, 2001; Munday, benthos. As such, it presents a model for investigating
2004; Pratchett et al., 2004). The only studies to docu- community-level responses to disturbance. This study,
ment long-term impacts on fish communities were therefore, quantifies long-term changes in the commu-
related to a gradual decline in reef condition through nity composition of a cryptobenthic reef fish assemblage
a combination of bleaching, crown of thorns outbreaks in response to the 1998 bleaching event and evaluates
and increased terrestrial runoff (Jones et al., 2004; the resilience of this assemblage to habitat change as a
Munday, 2004). Coral reef fishes would thus appear to result of global warming.
be relatively resilient, in ecosystem terms, to short-term
perturbations. It would appear that reef fishes are able
Material and methods
to maintain ecosystem processes; the implicit assump-
tion being that no change in the community composi- Censuses have been undertaken annually from 1993 to
tion is a reasonable indication that ecosystem processes 2004 on the leeward reef slope of Orpheus Island
are intact. (18135 0 S, 146128 0 E), in the central Great Barrier Reef.
Resilience is often difficult to measure, and the extent Each year 2–4 coral bommies of approximately 2 m3
to which systems exhibit resilience is often a reflection were censused using an enclosed ichthyocide technique
of the metrics used to evaluate ecosystem ‘health’ and (Ackerman & Bellwood, 2000). The number of replicates
the status of populations (Gunderson, 2000). Central to per year was constrained by permit requirements as the
this issue is the potential for a cryptic loss of resilience censuses were in a highly protected area within a World
(i.e. changes in the ability of a system to maintain Heritage Site. Censuses were undertaken in the same
ecosystem processes which are not apparent using reef slope habitat (but not from the same coral bom-
existing monitoring metrics; Bellwood et al., 2004). Such mies) at approximately the same time of year (in late
cryptic loss of resilience may lay the foundations for March to early April; September in 1993–1995). The last
some of the ‘ecological surprises’ that beset ecosystem 10 years samples were all within the same 3-week
management and are characterized by phase shifts period during the Austral cool season, several months
or ecosystem flips (Scheffer et al., 2001; Scheffer & after the summer peak recruitment period. The 1998
Carpenter, 2003). Coral reefs are no exception. Although censuses were immediately after the December 1997–
coral reefs are highly dynamic systems with a great February 1998 bleaching event and included fishes
capacity for regeneration (Connell et al., 1997), there is living in or on bleached corals. By 1999 most bleached
increasing evidence of reefs undergoing phase shifts Acropora spp. and almost all Montipora spp. corals had
from coral-dominated to other alternate states (Nystrom died (Marshall & Baird, 2000; Baird & Marshall, 2002),
& Folke, 2001; Aronson et al., 2002; Bellwood et al., 2004; although their skeletons remained intact. In each census
McManus & Polsenberg, 2004). One critical component a small coral bommie was enclosed in a 2 mm mesh
in identifying these phase or regime shifts is the ability net of 3.5 m2 basal area before adding an ichthyocide
to separate short-term changes from long-term trends. (rotenone or high-dose clove oil). Coral bommies were
This is often difficult, as relatively few long-term eco- selected to be as similar as possible each year with
logical data sets are available for reef systems (although ‘typical’ fish assemblages and coral cover (evaluated
Connell et al., 1997, 2004; Aronson et al., 2002; Halford in Ackerman & Bellwood, 2000). Each bommie is a large
et al., 2004 provide notable exceptions; cf. Pandolfi, isolated piece of consolidated reef matrix with a num-
1999). The present study, therefore, takes advantage of ber of relatively small coral colonies growing on it; in
a 12-year census of small benthic reef fishes, using a postbleach years these corals were in place but dead.
method that accurately quantifies entire cryptobenthic The ichthyocides are nonselective and provide a rela-
reef fish assemblages, in one of the world’s best pro- tively complete census of all species within the netted
tected coral reef ecosystems. Annual censuses from area (methodological details are provided in Ackerman
1993 to 2004 broadly span the 1998 mass coral bleaching & Bellwood, 2000, 2002). Fishes in the netted area were
event. In this event approximately 75% of the corals in collected during an intensive search by 5–9 divers
the study location died, with the local extirpation of (approximately one diver-hour per m2). All specimens
some abundant habitat forming taxa (Marshall & Baird were placed in an ice-seawater slurry and transferred to
2000; D. R. Bellwood, personal observation). This fish the laboratory for identification and fixation. Larger
assemblage is particularly suited to the detection of mobile reef fishes are not sampled using this methodol-
changes in response to disturbance. With maximum ogy and are not included in the analyses. In all years,
longevities spanning from several weeks to just over the overwhelming majority of specimens were of adult
1 year (Wilson, 2004; Depczynski & Bellwood, 2005, size (cf. Ackerman & Bellwood, 2000; Ackerman et al.,
r 2006 The Authors
Journal compilation r 2006 Blackwell Publishing Ltd, Global Change Biology, 12, 1587–1594
B L E A C H I N G I M PA C T S O N C O R A L R E E F S 1589
2004). A total of 35, 3.5 m2 ichthyocide stations were (a) PC2 (14.2%)
examined, yielding a total of 3682 individuals in 145 1.5
species.
Changes to the cryptobenthic reef fish community
03
over the 12 years were evaluated using four different
metrics: species richness, species diversity (Shannon– 96 00
Weiner diversity index, H 0 ), total abundance and com- 95
munity composition. The first three are commonly 97 PC1 (38.2%)
utilized methods for evaluating changes in reef assem- −1.5 93 99 1.5
blages following disturbance (e.g. Sano, 2000; Halford 94 04
01
et al., 2004; Jones et al., 2004). Total abundance, species 98
richness and species diversity were compared among
years using three one–way ANOVA’s. Total abundance 02
was log10 transformed to meet the assumptions of
normality and homoscedasticity. Following a significant
−1.5
result, nonsignificant groupings were identified using
Tukey’s HSD tests. Long-term trends within the data (b) PC2 (14.2%)
were examined using least-squares regression of the 0.6 Neopomacentrus
untransformed annual means. All analyses were under- azysron
taken using SPSS (v. 12.0). Bathygobius
fuscus Eviota sp.
Changes in community composition were investi-
Paragobiodon spp. Eviota sp. C
gated using a principal components analysis (PCA) of Eviota sp. F
Enneapterygius spp.
the mean number of individuals per sample (i.e. per Pomacentrus Neopomacentrus bankieri
moluccensis Caesio cuning
3.5 m2); mean values are based on the 2–4 samples Gobiodon histrio Istigobius spp. PC1 (38.2%)
per year. We examined only those species with more
Eviota
than 10 individuals over the 12-year period (i.e. we −0.6 Pomacentrus 0.6
adelus queenslandica
examined the 36 most abundant species; mean Chrysiptera Trimma cf. caesiura
abundance 5 92.4 Æ 42.2 SE individuals). In three cases, rollandi Eviota pellucida
closely related species were grouped as spp. as species Cheilodipterus
quinquelineatus
separation was uncertain. The analyses were based on
the covariance matrix of log10(x 1 1) transformed data.
An ordination technique (PCA) was selected as it made − 0.6
no a priori assumptions about group membership (e.g.
pre- or post-1998). A Ward’s method cluster analysis Fig. 1 A shift in the structure of reef fish communities on the
was performed on the squared Euclidian distances of Great Barrier Reef in response to the 1998 global coral bleaching
episode. (a) A principal component analysis (PCA) reveals two
the log abundance data to provide an objective descrip-
clusters. One encompasses the prebleaching (1993–1997), bleach-
tion of yearly groupings. A multivariate analysis of
ing (1998) and immediately postbleaching (1999) years; the other
variance (MANOVA) was performed on the abundance encompasses the postbleaching years (2000–2004). (b) Species
data to examine the significance of the major groupings loadings on the above PCA showing the species that contributed
identified by the Ward’s cluster analysis. Data were to the observed shift in community structure.
log10 transformed before analyses to help fit multivari-
ate normality and homoscedasticity. To confirm statis- postbleach (1999) samples. The second cluster encom-
tical differences between the two major groups, an passes the long-term postbleach communities (2000–
analysis of similarities (ANOSIM) was undertaken using 2004). The first two axes of the PCA explained 52% of
Primer 5.22 (Clarke & Warwick, 1994), using a one-way the variation among species, a relatively high value
design with the maximum number of permutations given that 36 species were included in the analyses,
(792). with PC1 (explaining 38% of the variation) clearly
separating the two distinct groups of years. The Ward’s
cluster analysis strongly supported this division into
Results
two clusters (Fig. 2). PC1 is most clearly characterized
The community composition exhibited a major shift by a shift from coral associated species (with low
during the 12-year study period (Fig. 1). The PCA scores), such as Pomacentrus moluccensis and Gobiodon
revealed two distinct clusters, the first contains pre- histrio (which decreased in abundance by 83% and 67%,
bleached (1993–1997), bleaching (1998) and immediate respectively) to less habitat specific forms such as Eviota
r 2006 The Authors
Journal compilation r 2006 Blackwell Publishing Ltd, Global Change Biology, 12, 1587–1594
1590 D . R . B E L L W O O D et al.
1995 250 Abundance
1997 Pre-bleach
Number of individuals
200
1996
(±95%CI)
1993 150
1994
1998 Bleach/ 100
1999 immediate post-bleach
2001 50
2004 0
2000 Post-bleach
2003 35 Species richness
2002
Number of species
30
Fig. 2 A Ward’s hierarchical cluster analysis reveals two dis-
(±95%CI)
25
tinct groups in the community composition of cryptobenthic
fishes. One encompasses the 5 years prebleaching (1993–1997), 20
and the 2 years immediately postbleaching (1998–1999) years,
15
the other encompasses the 5 years postbleaching (2000–2004).
10
queenslandica and Neopomacentrus bankieri (increasing by 1.4 Diversity
240% and 138%, respectively) (cf. Allen, 1991; Depc-
1.2
zynski & Bellwood, 2005). However, a broad range of Diversity index
non-coral associated species also appear to be involved. (H′ ,±95%CI) 1.0
PC2 (explaining 14% of the variation) is largely driven
by variation in the composition of the postbleach as- 0.8
semblages, particularly in the relative abundance of 0.6
the schooling planktivore Neopomacentrus azysron. The
MANOVA revealed complete separation of the two groups 0.4
93
94
95
96
97
98
99
00
01
02
03
04
of years (F10, 1 5 802.7; P 5 0.027). The ANOSIM, likewise,
19
19
19
19
19
19
19
20
20
20
20
20
rejected the null hypothesis of no difference between Year
the two groups (prebleach and postbleach) with a
significance level of 0.3% (i.e. o0.01). The sample Fig. 3 Conventional measures of the status of coral reef fish
statistic (global R) was 0.716. assemblages. Abundance, richness and diversity are expressed
In marked contrast to the clear shift in community in terms of the mean number or diversity of fishes within
composition, analyses of the three traditional variables replicate fish censuses. The vertical shaded bar indicates the
timing and duration of the global coral bleaching event which
for monitoring reef fishes (total abundance, species rich-
killed approximately 75% of the corals in the study area. The
ness, and diversity) exhibited limited variation over the
horizontal lines indicate years that were not significantly
study period and no evidence of a change in relation to different in Tukey’s HSD tests.
the 1998 bleaching event (Fig. 3). Although all three
metrics exhibit a statistically significant difference among
years (abundance F11, 23 5 2.74, P 5 0.02; species richness
Discussion
F11, 23 5 6.36, Po0.001; diversity F11, 23 5 4.57, Po0.01),
and all contain 2–4 distinct groups based on Tukey’s There have been several reports of resilience or a
HSD tests, none of the groups lie on either side of the limited response of reef fishes to large-scale coral loss
1998 bleaching event. Indeed, all groups include at least 2 through storms (Halford et al., 2004), crown of thorns
years on either side of the 1998 event. In terms of long- starfish outbreaks (Williams, 1986; Hart et al., 1996;
term trends only abundance had a significant trend, Sano, 2001) and bleaching (Kokita & Nakazono, 2001;
marked by increasing abundance over the study period Booth & Beretta, 2002). Our results also show a limited
(r2 5 0.67, P 5 0.001). The other two parameters exhibited response to coral bleaching using traditional metrics.
no significant trend (species richness r2 5 4.9 Â 10À7, However, changes in community composition reveal an
P 5 0.998; diversity r2 5 0.33, P 5 0.052). Changes in the unexpected vulnerability of reef fish communities to
abundance of individual species after the 1998 bleaching coral bleaching. These results present a fundamentally
event reveal a marked difference between increases in different perspective, which highlights the need for
generalist omnivorous species, and the planktivore N. caution when selecting metrics to evaluate the resilience
bankieri, and decreases in coral-dependent species (Fig. 4). of coral reef ecosystems.
r 2006 The Authors
Journal compilation r 2006 Blackwell Publishing Ltd, Global Change Biology, 12, 1587–1594
B L E A C H I N G I M PA C T S O N C O R A L R E E F S 1591
Istigobius spp. N. bankieri E. queenslandica
5 120
60
4
80
3 40
Mean abundance (ind. sample −1±SE)
2
40 20
1
0 0 0
Gobiodon spp. Paragobiodon spp. P. moluccensis
16
8 3
12
6 2
8
4
1
2 4
0 0 0
94
96
98
00
02
04
94
96
98
00
02
04
94
96
98
00
02
04
19
19
19
20
20
20
19
19
19
20
20
20
19
19
19
20
20
20
Fig. 4 Changes in the abundance of six cryptobenthic fish species in relation to the 1998 coral bleaching event (marked by the vertical
grey line). The data contrast increases in three generalist species (the planktivore Neopomacentrus bankieri and the omnivorous/
detritivorous Istigobius spp. and Eviota queenslandica) and decreases in three coral-associated species (Pomacentrus moluccensis, Gobiodon
spp. and Paragobiodon spp.).
In contrast to the only other long-term studies (Jones distinct and relatively stable shift from a prebleaching
et al., 2004; Munday, 2004), we found no decrease in to a postbleaching reef fish assemblage. Our findings
diversity, richness or abundance in reef fishes over the emphasize the need for caution when evaluating the
12-year study period. Indeed, as in most previous stu- resilience of reef ecosystems, and highlight the potential
dies, the three main metrics, species richness, diversity insensitivity of common metrics used for measuring
(H 0 ) and total abundance showed no response to a major responses of fishes to bleaching.
disturbance (in this case the 1998 coral bleaching event). By looking at small species with high turnover rates,
At face value this could be taken as strong evidence for and using annual censuses over 12 years, we are able to
resilience – the reef fish fauna remained intact in the face distinguish responses to bleaching from a background
of what was arguably the greatest impact on global coral pattern of interannual variation. Furthermore, our ob-
reefs in living memory. This single event resulted in servations avoid the storage effects of long-lived species
extensive bleaching, with approximately 40% of the where recruitment failure and declines may be masked
world’s reefs exhibiting serious damage. This damage by the presence of long-lived individuals. Many of the
contributed significantly to the widespread decline in conspicuous reef fishes that are included in fish cen-
coral cover on reefs in all tropical oceans (Wilkinson, suses live for one to two decades (Choat & Axe, 1996;
2004; Agardy et al., 2005). Our study site suffered an Choat et al., 1996). The resilience reported in previous
estimated 75% mortality in Acropora species and almost studies (e.g. Williams, 1986; Hart et al., 1996; Sano, 2000;
total loss of the dominant structure providing species Kokita & Nakazono, 2001; Booth & Beretta, 2002;
Montipora (Marshall & Baird, 2000), yet, the reef fishes Halford et al., 2004) may, at least in part, reflect the
appeared to fare well. Indeed, in terms of overall abun- ability of adults of long-lived species to withstand
dance, fish numbers in the study area have almost disturbance (by movement among habitats, using
doubled since the 1998 bleaching event. In terms of reef stored reserves or prey switching, Pratchett et al.,
processes, one may thus assume that little has changed 2004). In contrast, most of the species in the community
and that the critical controlling functions and processes that we examined had maximum longevities of less
provided by fishes remain intact. than 1 year (Depczynski & Bellwood, 2005, 2007; Herna-
This apparent stasis or resilience is misleading: There man & Munday, 2005). This means that these species
were marked changes in community structure. The reveal the effects of bleaching on the entire life history
traditional metrics did not detect a major, yet cryptic, including settlement, recruitment, juvenile and adult
change in community composition, which marked a survival. Previously, our ability to detect changes in reef
r 2006 The Authors
Journal compilation r 2006 Blackwell Publishing Ltd, Global Change Biology, 12, 1587–1594
1592 D . R . B E L L W O O D et al.
systems may have been compromised by both limita- The proximate basis for the changes is uncertain
tions in the methods and the speed with which changes although loss of habitat, if not structure, certainly
are likely to be manifest. If species with a maximum played a role (soft coral colonies were rapidly lost
longevity of a few months take 3 years to reorganize leaving only solid rounded bases, many hard coral
into a relatively stable postbleach community, compar- skeletons are still intact). Live coral (hard and soft)
able reorganization in larger taxa with longevities of 5– was lost with the inevitable loss of habitat for coral
20 years may take several decades. Indeed, if coral dwelling or feeding specialists (cf. Munday, 2004).
bleaching is manifested in coral reef fishes through a Changes in the abundance of coral-associated fishes
disruption of recruitment patterns then the full impact of with decreasing coral cover has been recorded in a
the 1998 bleaching event may have not even begun to number of studies (e.g. Booth & Beretta, 2002; Jones
appear in censuses of long-lived species. et al., 2004). Furthermore, the presence of living coral
This raises the question of whether the observed shift in tissue may be a critical factor in shaping patterns of
cryptobenthic community structure represents a stable recruitment or early postsettlement survivorship
state or merely an alternate or even transitional state. (Beukers & Jones, 1997; Ohman et al., 1998; Holbrook
The postbleaching community is quite distinct and after et al., 2000; McCormick & Hoey, 2006). The stable or
5 years and up to 35 generations later the assemblage gradually increasing fish abundance combined with a
has shown little signs of returning to anything resembling shift in community composition as documented the
the prebleached condition. If the frequency and intensity current study contrasts markedly with the only other
of thermal anomalies and bleaching continues the pre- long-term study in Papua New Guinea (PNG) (Jones
bleach community may never return (cf. Donner et al., et al., 2004) which reported a gradual decline in fish
2005). However, the apparent stability of the postbleach abundance over an 8-year period (changes in commu-
assemblage does not necessarily indicate a stable state. It nity composition in PNG are likely but were not ana-
is certainly plausible that with a return of significant coral lyzed). This disparity between the two studies appears
cover the fish community will change again; the current to mark a fundamental difference in the disturbance
community may thus represent a transitional state. Never- regimes. Our observations mark a distinct shift, primarily
theless, the nature of the shift in community composition in species composition, in response to a single massive
suggests that the time frame for regeneration is likely to bleaching event. Subsequent bleaching events had negli-
be decades rather than years and that there remains a gible impact, as susceptible corals were absent; there were
strong possibility that the path to regeneration will exhibit no noticeable changes in other parameters over the 12-
significant hysteresis (cf. Hughes et al., 2005). year period. In contrast, the reefs examined in PNG were
The timing and nature of the observed changes in fish marked by ongoing bleaching events, crown-of-thorns
community structure strongly suggest that it was a outbreaks and concurrent increases in terrestrial runoff.
direct result of the 1998 bleaching event, even though The latter factor may be particularly important, as it
the shift was marked by changes to a wide range of fish could account for the decline in noncoral species in
taxa. Coral associated species were still living in PNG, reflecting the impacts of terrestrial runoff on
bleached corals in 1998 and in small remnants in 1999, benthic dynamics including algal and detrital resources.
but they largely disappeared thereafter. The loss of coral Finally, one must ask the question of what these
remnants was probably exacerbated by changes in changes mean for ecosystem function, and the resilience
density-dependent coral mortality rates as a result of of reefs to further change. Although our knowledge
excessive predation by the remaining fish coralivores of the ecology of the component taxa is limited, the shift
(up to eight Labrichthys unilineatus were observed feed- in the fish community structure appears to reflect a
ing on a single isolated Acropora colony). What was distinct move from habitat and/or diet specialists to
most striking, apart from a decrease in species with habitat or dietary generalists (cf. Depczynski &
no known association with corals, (e.g. Eviota sp. F Bellwood, 2003, 2004). Prebleaching communities were
(from 1.3 Æ 0.1 to 0.4 Æ 0.1 fish per sample Æ SE) and characterized by the presence of several coral dwelling
Pomacentrus adelus (1.6 Æ 0.5 to 0.6 Æ 0.2 fish per or associated species (P. moluccensis, Gobiodon spp.,
sample Æ SE)), was the marked increase in abundance Paragobiodon spp.; cf. Munday et al., 1997) and included
in number of common, habitat and trophic generalist Amblygobius rainfordi, the only herbivore in this crypto-
species (e.g. Istigobius spp. and E. queenslandica (Fig. 4); benthic community. The density of detritivorous blen-
Depczynski & Bellwood, 2003, 2004). This is particularly nies (Wilson et al., 2003) changed little over the sample
striking as coral represents only a small, but apparently period. The loss of several coral dwelling specialists
important, component of the benthic fauna (throughout reduces the potential for positive synergism between
the period mean live coral cover on GBR reefs was in reef corals and commensal fishes (Pratchett, 2001).
the region of 20–30%; Bellwood et al., 2004). Postbleaching communities were dominated by the
r 2006 The Authors
Journal compilation r 2006 Blackwell Publishing Ltd, Global Change Biology, 12, 1587–1594
B L E A C H I N G I M PA C T S O N C O R A L R E E F S 1593
highly abundant habitat generalist and planktivore identifications; J. Pandolfi, R. Winterbottom and colleagues in the
N. bankieri and the benthic omnivores/detritivores Isti- ARC Centre of Excellence for Coral Reef Studies for helpful
gobius spp. and E. queenslandica. The latter is marked by comments or discussions. Supported by the Australian Research
Council (DRB).
having one of the broadest dietary ranges in the assem-
blage (Depczynski & Bellwood, 2003). The change in
community structure does indeed appear to represent a References
phase shift in that fish-mediated ecosystem processes Ackerman JL, Bellwood DR (2000) Reef fish assemblages: a
have probably been modified. This is not simply a case re-evaluation using enclosed rotenone stations. Marine Ecology
of functional redundancy, responding to change by Progress Series, 206, 227–237.
replacing one species with a functional equivalent; there Ackerman JL, Bellwood DR (2002) Comparative efficiency of
is clear evidence of trophic simplification. clove oil and rotenone for sampling tropical reef fish assem-
This phase shift reveals the limited resilience of the blages. Journal of Fish Biology, 60, 893–901.
prebleach community to coral bleaching. Resilience Ackerman JL, Bellwood DR, Brown JH (2004) The contribution of
small individuals to density–body size relationships: examination
would have been characterized by either no change in
of energetic equivalence in reef fishes. Oecologia, 139, 568–571.
the community composition (resistance) or regeneration
Agardy T, Alder J, Dayton P et al. (2005) Coastal Systems. In:
and the return to a community with similar ecological
Ecosystems and Human Well-Being: Vol. 1 Current State and
characteristics (in terms of ecosystem processes) to Trends (eds Hassan R, Scholes R, Ash N), pp. 513–549. Island
those exhibited by the prebleach community. However, Press, Washington.
the community composition and ecosystem processes Allen GR (1991) Damselfishes of the world. Mergus, Hans A.
both appeared to change as a result of the 1998 coral- Baensch, Melle.
bleaching event. This shift also marked the beginning of Aronson RB, Macintyre IG, Precht WF et al. (2002) The expanding
a new and potentially more resilient postbleach assem- scale of species turnover events on coral reefs in Belize.
blage. The postbleach assemblage is characterized by Ecological Monographs, 72, 233–249.
many generalist species (cf. Bellwood et al., 2005), and Baird AH, Marshall PA (2002) Mortality, growth and reproduc-
as such, may be capable of withstanding future bleach- tion in scleractinian corals following bleaching on the Great
Barrier Reef. Marine Ecology Progress Series, 237, 133–141.
ing events with little further change. We may, thus, be
Bellwood DR, Hughes TP, Folke C et al. (2004) Confronting the
faced with an undesirable yet resilient state: a perni-
coral reef crisis. Nature, 429, 827–833.
cious postbleach fish assemblage. Bellwood DR, Wainwright PC, Fulton CJ et al. (2005) Functional
It is sobering to note that this is the first record of a versatility supports coral reef biodiversity. Proceedings of the
clear shift in a GBR reef fish community in response Royal Society B, 273, 101–107.
to the 1998 bleaching event and that has only been Beukers JS, Jones GP (1997) Habitat complexity modifies the
revealed by examination of exceptionally short-lived impact of piscivores on a coral reef fish population. Oecologia,
species. Furthermore, the study location lies within 114, 50–59.
one of the best protected areas of the Great Barrier Reef Booth DJ, Beretta GA (2002) Changes in a fish assemblage after a
Marine Park, with a relatively intact large fish fauna coral bleaching event. Marine Ecology Progress Series, 245,
(Graham et al., 2003), yet, this phase shift has persisted 205–212.
Caley MJ, Carr MH, Hixon MA et al. (1996) Recruitment and the
for over 5 years, representing 5–35 generations of these
local dynamics of open marine populations. Annual Review of
cryptobenthic fishes. The implications of global warm-
Ecology and Systematics, 27, 477–500.
ing for coral reefs are poorly understood. However, Choat JH, Axe LM (1996) Growth and longevity in acanthurid
our observations provide a clear example of the ex- fishes. An analysis of otolith increments. Marine Ecology Pro-
ceptional sensitivity of coral reefs to environmental gress Series, 134, 15–26.
change and highlight the difficulty of measuring the Choat JH, Axe LM, Lou DC (1996) Growth and longevity in fishes
resilience of coral reef ecosystems. The need for sensi- of the family Scaridae. Marine Ecology Progress Series, 145, 33–41.
tive metrics for evaluating resilience is of paramount Clarke KR, Warwick RM (1994) Changes in Marine Communities:
importance and presents an ongoing challenge in coral An Approach to Statistical Analysis and Interpretation. Natural
reef management. Environmental Research Council, UK.
Connell JH, Hughes TP, Wallace CC (1997) A 30-year study of
coral abundance, recruitment, and disturbance at several
scales in space and time. Ecological Monographs, 67, 461–488.
Acknowledgements Connell JH, Hughes TP, Wallace CC et al. (2004) A long-
term study of competition and diversity of corals. Ecological
We thank the 500 1 JCU MB3160 goby-pickers for their enthu-
siastic assistance; the Great Barrier Reef Marine Park Authority, Monographs, 74, 179–210.
National Parks, Environmental Protection Agency and Depart- Depczynski M, Bellwood DR (2003) The role of cryptobenthic
ment of Primary Industries for permission to collect fishes; reef fishes in coral reef trophodynamics. Marine Ecology
H. Larson, P. Munday and R. Winterbottom for assistance with Progress Series, 256, 183–191.
r 2006 The Authors
Journal compilation r 2006 Blackwell Publishing Ltd, Global Change Biology, 12, 1587–1594
1594 D . R . B E L L W O O D et al.
Depczynski M, Bellwood DR (2004) Microhabitat utilization Marshall PA, Baird AH (2000) Bleaching of corals on the Great
patterns in cryptobenthic reef fish communities. Marine Barrier Reef: differential susceptibilities among taxa. Coral
Biology, 145, 455–463. Reefs, 19, 155–163.
Depczynski M, Bellwood DR (2005) Shortest recorded vertebrate McCormick MI, Hoey AS (2006) Biological and physical corre-
lifespan found in a coral reef fish. Current Biology, 15, 288–289. lates of settlement and survival for a coral reef fish, Pomacen-
Depczynski M, Bellwood DR (2007) Extremes, plasticity and trus amboinensis (Pomacentridae). Proceedings of the 10th
invariance in vertebrate life history traits: insights from coral International Coral Reef Symposium, Okinawa, 425–430.
reef fishes. Ecology (in press). McManus JW, Polsenberg JF (2004) Coral-algal phase shifts on
Donner SD, Skirving WJ, Little CM et al. (2005) Global assess- coral reefs: ecological and environmental aspects. Progress in
ment of coral bleaching requires rates of adaptation under Oceanography, 60, 263–279.
climate change. Global Change Biology, 11, 2251–2265. Munday PL (2004) Habitat loss, resource specialization, and
Edwards AJ, Clark S, Zahir H et al. (2001) Coral bleaching and extinction on coral reefs. Global Change Biology, 10, 1642–1647.
mortality on artificial and natural reefs in Maldives in 1998, Munday PL, Jones GP, Caley MJ (1997) Habitat specialisation
sea surface temperature anomalies and initial recovery. Marine and the distribution and abundance of coral-dwelling gobies.
Pollution Bulletin, 42, 7–15. Marine Ecology Progress Series, 152, 227–239.
Graham NAJ, Evans RD, Russ GR (2003) The effects of marine Nystrom M, Folke C (2001) Spatial resilence of coral reefs.
reserve protection on the trophic relationships of reef fishes on Ecosystems, 4, 406–417.
the Great Barrier Reef. Environmental Conservation, 30, 200–208. Ohman MC, Munday PL, Jones GP et al. (1998) Settlement
Gunderson LH (2000) Ecological resilience – In theory and appli- strategies and distribution patterns of coral-reef fishes. Journal
cation. Annual Review of Ecology and Systematics, 31, 425–439. of Experimental Marine Biology and Ecology, 225, 219–238.
Halford A, Cheal AJ, Ryan D et al. (2004) Resilience to large-scale Pandolfi JM (1999) Response of Pleistocene coral reefs to envir-
disturbance in coral and fish assemblages on the Great Barrier onmental change over long temporal scales. American Zoolo-
Reef. Ecology, 85, 1892–1905. gist, 39, 113–130.
Hart AM, Klumpp DW, Russ GR (1996) Response of herbivorous Pandolfi JM, Bradbury RH, Sala E et al. (2003) Global trajectories
fishes to crown-of-thorns starfish Acanthaster planci outbreaks. of the long-term decline of coral reef ecosystems. Science, 301,
2. Density and biomass of selected species of herbivorous fish 955–958.
and fish-habitat correlations. Marine Ecology Progress Series, Pratchett MS (2001) Influence of coral symbionts on feeding
132, 21–30. preferences of crown-of-thorns starfish Acanthaster planci in
Hernaman V, Munday PL (2005) Life history characteristics of the west Pacific. Marine Ecology Progress Series, 214, 111–119.
coral reef gobies: I. Growth and life-span. Marine Ecology Pratchett MS, Wilson SK, Berumen ML et al. (2004) Sublethal
Progress Series, 290, 207–211. effects of coral bleaching on an obligate coral feeding butter-
Hoegh-Guldberg O (1999) Climate change, coral bleaching and flyfish. Coral Reefs, 23, 352–356.
the future of the world’s coral reefs. Marine and Freshwater Sano M (2000) Stability of reef fish assemblages: responses to
Research, 50, 839–866. coral recovery after catastrophic predation by Acanthaster
Holbrook SJ, Forrester GE, Schmitt RJ (2000) Spatial patterns in planci. Marine Ecology Progress Series, 198, 121–130.
abundance of a damselfish reflect availability of suitable Sano M (2001) Short-term responses of fishes to macroalgal
habitat. Oecologia, 122, 109–120. overgrowth on coral rubble on a degraded reef at Iriomote
Holling CS (1973) Resilience and stability of ecological systems. Island, Japan. Bulletin of Marine Science, 68, 543–556.
Annual Review of Ecology and Systematics, 4, 1–23. Scheffer M, Carpenter SR (2003) Catastrophic regime shifts in
Hughes TP, Baird AH, Bellwood DR et al. (2003) Climate change, ecosystems: linking theory to observation. Trends in Ecology and
human impacts, and the resilience of coral reefs. Science, 301, Evolution, 18, 648–656.
929–933. Scheffer M, Carpenter S, Foley JA et al. (2001) Catastrophic shifts
Hughes TP, Bellwood DR, Folke C et al. (2005) New paradigms in ecosystems. Nature, 413, 591–596.
for supporting resilience of marine ecosystems. Trends in Wilkinson C (2004) Status of Coral Reefs of the World: 2004. Global
Ecology and Evolution, 20, 380–386. Coral Reef Monitoring Network and Australian Institute of
Jackson JBC, Kirby MX, Berger WH et al. (2001) Historical over- Marine Science, Townsville.
fishing and the recent collapse of coastal ecosystems. Science, Williams DMcB (1986) Temporal variation in the structure of reef
293, 629–638. slope fish communities (central Great Barrier Reef): short-term
Jones GP, McCormick MI, Srinivasan M et al. (2004) Coral decline effects of Acanthaster planci infestation. Marine Ecology Progress
threatens fish biodiversity in marine reserves. Proceedings of the Series, 28, 157–164.
National Academy of Sciences of the USA, 101, 8251–8253. Wilson SK (2004) Growth, mortality and turnover rates of a small
Kokita T, Nakazono A (2001) Rapid response of an obligately detritivorous fish. Marine Ecology Progress Series, 284, 253–259.
corallivorous filefish Oxymonacanthus longirostris (Monacanthi- Wilson SK, Bellwood DR, Choat JH et al. (2003) Detritus in
dae). Coral Reefs, 20, 155–158. the epilithic algal matrix and its use by coral reef fishes.
Loya Y, Sakai K, Yamazato K et al. (2001) Coral bleaching: the Oceanography and Marine Biology, An Annual Review, 41,
winners and the losers. Ecology Letters, 4, 122–131. 279–309.
r 2006 The Authors
Journal compilation r 2006 Blackwell Publishing Ltd, Global Change Biology, 12, 1587–1594
Coral bleaching, reef fish community phase shifts
and the resilience of coral reefs
D AV I D R . B E L L W O O D *w , A N D R E W S . H O E Y *w, J O H N L . A C K E R M A N *z
and M A R T I A L D E P C Z Y N S K I *
*School of Marine Biology, James Cook University, Townsville, Qld 4811, Australia, wAustralian Research Council Centre of
Excellence for Coral Reef Studies, James Cook University, Townsville, Qld 4811, Australia, zFisheries and Marine Sciences Program,
Bureau of Rural Sciences, GPO Box 858, Canberra, ACT 2610, Australia
Abstract
The 1998 global coral bleaching event was the largest recorded historical disturbance of
coral reefs and resulted in extensive habitat loss. Annual censuses of reef fish community
structure over a 12-year period spanning the bleaching event revealed a marked phase
shift from a prebleach to postbleach assemblage. Surprisingly, we found that the
bleaching event had no detectable effect on the abundance, diversity or species richness
of a local cryptobenthic reef fish community. Furthermore, there is no evidence of
regeneration even after 5–35 generations of these short-lived species. These results have
significant implications for our understanding of the response of coral reef ecosystems to
global warming and highlight the importance of selecting appropriate criteria for
evaluating reef resilience.
Keywords: bleaching, community composition, coral reefs, fishes, habitat loss, phase shifts, resilience
Received 3 January 2006; revised version received 24 March 2006; accepted 4 April 2006
Resilience and the ability to regenerate after distur-
Introduction
bance is a central feature of coral reefs. Indeed, it is their
Coral reefs are highly dynamic systems characterized capacity to recolonize and maintain populations in the
by variable and stochastic recruitment and disturbance. face of disturbance that has underpinned their success
Our understanding of these factors has offered much to in dominating exposed and high-energy locations in
explain local variation in population numbers and the tropics (Connell et al., 1997). Recently, large-scale
community composition during periods of apparent bleaching has presented coral reefs with a new chal-
stability (Caley et al., 1996; Connell et al., 1997). Today, lenge: rapid large-scale loss of coral cover (Wilkinson,
however, coral reefs around the world are facing a 2004). This widespread disturbance has impacted nu-
scenario of steady declines in coral cover punctuated merous reefs, many of which were already weakened
by periodic large-scale perturbations. Of these, the 1998 by human activities, including systematic overfishing,
mass coral bleaching event was perhaps the most habitat destruction and pollution (Jackson et al., 2001;
noteworthy (Hoegh-Guldberg, 1999; Hughes et al., Hughes et al., 2003; Pandolfi et al., 2003). The response
2003; Wilkinson, 2004). The critical question to arise of corals to bleaching has been highly variable, with
from these changes is: to what extent can coral reefs many species showing limited evidence of short-term
recover (sensu Edwards et al., 2001) or regenerate (sensu recovery (Loya et al., 2001; Baird & Marshall, 2002;
Hughes et al., 2003) after disturbance (i.e. to what extent Hughes et al., 2003; Donner et al., 2005). Thus, from a
do they exhibit resilience; Holling, 1973; Gunderson, coral perspective the evidence, to date, suggests that
2000)? In particular, can coral reef ecosystems maintain corals show limited short-term resilience to elevated sea
the critical feedbacks, functions and processes in the surface temperatures.
face of climate change? In marked contrast, it appears that reef fishes are
relatively resilient to disturbance, with reef fish assem-
Correspondence: David Bellwood, School of Marine Biology, blages exhibiting only a limited response to the loss of
James Cook University, Townsville, Qld 4811, Australia, corals through large-scale crown of thorns starfish out-
tel. 1 61 (0)7 47814447, fax 1 61 (0)7 47251570, breaks (Williams, 1986; Hart et al., 1996; Sano, 2000) and
e-mail: David.Bellwood@jcu.edu.au coral bleaching (Kokita & Nakazono, 2001; Booth &
r 2006 The Authors
Journal compilation r 2006 Blackwell Publishing Ltd 1587
1588 D . R . B E L L W O O D et al.
Beretta, 2002). In most cases the changes are restricted 2007; Hernaman & Munday, 2005), this is a highly
to changes in a few strongly coral-associated fishes dynamic fish community in close association with the
(Williams, 1986; Kokita & Nakazono, 2001; Munday, benthos. As such, it presents a model for investigating
2004; Pratchett et al., 2004). The only studies to docu- community-level responses to disturbance. This study,
ment long-term impacts on fish communities were therefore, quantifies long-term changes in the commu-
related to a gradual decline in reef condition through nity composition of a cryptobenthic reef fish assemblage
a combination of bleaching, crown of thorns outbreaks in response to the 1998 bleaching event and evaluates
and increased terrestrial runoff (Jones et al., 2004; the resilience of this assemblage to habitat change as a
Munday, 2004). Coral reef fishes would thus appear to result of global warming.
be relatively resilient, in ecosystem terms, to short-term
perturbations. It would appear that reef fishes are able
Material and methods
to maintain ecosystem processes; the implicit assump-
tion being that no change in the community composi- Censuses have been undertaken annually from 1993 to
tion is a reasonable indication that ecosystem processes 2004 on the leeward reef slope of Orpheus Island
are intact. (18135 0 S, 146128 0 E), in the central Great Barrier Reef.
Resilience is often difficult to measure, and the extent Each year 2–4 coral bommies of approximately 2 m3
to which systems exhibit resilience is often a reflection were censused using an enclosed ichthyocide technique
of the metrics used to evaluate ecosystem ‘health’ and (Ackerman & Bellwood, 2000). The number of replicates
the status of populations (Gunderson, 2000). Central to per year was constrained by permit requirements as the
this issue is the potential for a cryptic loss of resilience censuses were in a highly protected area within a World
(i.e. changes in the ability of a system to maintain Heritage Site. Censuses were undertaken in the same
ecosystem processes which are not apparent using reef slope habitat (but not from the same coral bom-
existing monitoring metrics; Bellwood et al., 2004). Such mies) at approximately the same time of year (in late
cryptic loss of resilience may lay the foundations for March to early April; September in 1993–1995). The last
some of the ‘ecological surprises’ that beset ecosystem 10 years samples were all within the same 3-week
management and are characterized by phase shifts period during the Austral cool season, several months
or ecosystem flips (Scheffer et al., 2001; Scheffer & after the summer peak recruitment period. The 1998
Carpenter, 2003). Coral reefs are no exception. Although censuses were immediately after the December 1997–
coral reefs are highly dynamic systems with a great February 1998 bleaching event and included fishes
capacity for regeneration (Connell et al., 1997), there is living in or on bleached corals. By 1999 most bleached
increasing evidence of reefs undergoing phase shifts Acropora spp. and almost all Montipora spp. corals had
from coral-dominated to other alternate states (Nystrom died (Marshall & Baird, 2000; Baird & Marshall, 2002),
& Folke, 2001; Aronson et al., 2002; Bellwood et al., 2004; although their skeletons remained intact. In each census
McManus & Polsenberg, 2004). One critical component a small coral bommie was enclosed in a 2 mm mesh
in identifying these phase or regime shifts is the ability net of 3.5 m2 basal area before adding an ichthyocide
to separate short-term changes from long-term trends. (rotenone or high-dose clove oil). Coral bommies were
This is often difficult, as relatively few long-term eco- selected to be as similar as possible each year with
logical data sets are available for reef systems (although ‘typical’ fish assemblages and coral cover (evaluated
Connell et al., 1997, 2004; Aronson et al., 2002; Halford in Ackerman & Bellwood, 2000). Each bommie is a large
et al., 2004 provide notable exceptions; cf. Pandolfi, isolated piece of consolidated reef matrix with a num-
1999). The present study, therefore, takes advantage of ber of relatively small coral colonies growing on it; in
a 12-year census of small benthic reef fishes, using a postbleach years these corals were in place but dead.
method that accurately quantifies entire cryptobenthic The ichthyocides are nonselective and provide a rela-
reef fish assemblages, in one of the world’s best pro- tively complete census of all species within the netted
tected coral reef ecosystems. Annual censuses from area (methodological details are provided in Ackerman
1993 to 2004 broadly span the 1998 mass coral bleaching & Bellwood, 2000, 2002). Fishes in the netted area were
event. In this event approximately 75% of the corals in collected during an intensive search by 5–9 divers
the study location died, with the local extirpation of (approximately one diver-hour per m2). All specimens
some abundant habitat forming taxa (Marshall & Baird were placed in an ice-seawater slurry and transferred to
2000; D. R. Bellwood, personal observation). This fish the laboratory for identification and fixation. Larger
assemblage is particularly suited to the detection of mobile reef fishes are not sampled using this methodol-
changes in response to disturbance. With maximum ogy and are not included in the analyses. In all years,
longevities spanning from several weeks to just over the overwhelming majority of specimens were of adult
1 year (Wilson, 2004; Depczynski & Bellwood, 2005, size (cf. Ackerman & Bellwood, 2000; Ackerman et al.,
r 2006 The Authors
Journal compilation r 2006 Blackwell Publishing Ltd, Global Change Biology, 12, 1587–1594
B L E A C H I N G I M PA C T S O N C O R A L R E E F S 1589
2004). A total of 35, 3.5 m2 ichthyocide stations were (a) PC2 (14.2%)
examined, yielding a total of 3682 individuals in 145 1.5
species.
Changes to the cryptobenthic reef fish community
03
over the 12 years were evaluated using four different
metrics: species richness, species diversity (Shannon– 96 00
Weiner diversity index, H 0 ), total abundance and com- 95
munity composition. The first three are commonly 97 PC1 (38.2%)
utilized methods for evaluating changes in reef assem- −1.5 93 99 1.5
blages following disturbance (e.g. Sano, 2000; Halford 94 04
01
et al., 2004; Jones et al., 2004). Total abundance, species 98
richness and species diversity were compared among
years using three one–way ANOVA’s. Total abundance 02
was log10 transformed to meet the assumptions of
normality and homoscedasticity. Following a significant
−1.5
result, nonsignificant groupings were identified using
Tukey’s HSD tests. Long-term trends within the data (b) PC2 (14.2%)
were examined using least-squares regression of the 0.6 Neopomacentrus
untransformed annual means. All analyses were under- azysron
taken using SPSS (v. 12.0). Bathygobius
fuscus Eviota sp.
Changes in community composition were investi-
Paragobiodon spp. Eviota sp. C
gated using a principal components analysis (PCA) of Eviota sp. F
Enneapterygius spp.
the mean number of individuals per sample (i.e. per Pomacentrus Neopomacentrus bankieri
moluccensis Caesio cuning
3.5 m2); mean values are based on the 2–4 samples Gobiodon histrio Istigobius spp. PC1 (38.2%)
per year. We examined only those species with more
Eviota
than 10 individuals over the 12-year period (i.e. we −0.6 Pomacentrus 0.6
adelus queenslandica
examined the 36 most abundant species; mean Chrysiptera Trimma cf. caesiura
abundance 5 92.4 Æ 42.2 SE individuals). In three cases, rollandi Eviota pellucida
closely related species were grouped as spp. as species Cheilodipterus
quinquelineatus
separation was uncertain. The analyses were based on
the covariance matrix of log10(x 1 1) transformed data.
An ordination technique (PCA) was selected as it made − 0.6
no a priori assumptions about group membership (e.g.
pre- or post-1998). A Ward’s method cluster analysis Fig. 1 A shift in the structure of reef fish communities on the
was performed on the squared Euclidian distances of Great Barrier Reef in response to the 1998 global coral bleaching
episode. (a) A principal component analysis (PCA) reveals two
the log abundance data to provide an objective descrip-
clusters. One encompasses the prebleaching (1993–1997), bleach-
tion of yearly groupings. A multivariate analysis of
ing (1998) and immediately postbleaching (1999) years; the other
variance (MANOVA) was performed on the abundance encompasses the postbleaching years (2000–2004). (b) Species
data to examine the significance of the major groupings loadings on the above PCA showing the species that contributed
identified by the Ward’s cluster analysis. Data were to the observed shift in community structure.
log10 transformed before analyses to help fit multivari-
ate normality and homoscedasticity. To confirm statis- postbleach (1999) samples. The second cluster encom-
tical differences between the two major groups, an passes the long-term postbleach communities (2000–
analysis of similarities (ANOSIM) was undertaken using 2004). The first two axes of the PCA explained 52% of
Primer 5.22 (Clarke & Warwick, 1994), using a one-way the variation among species, a relatively high value
design with the maximum number of permutations given that 36 species were included in the analyses,
(792). with PC1 (explaining 38% of the variation) clearly
separating the two distinct groups of years. The Ward’s
cluster analysis strongly supported this division into
Results
two clusters (Fig. 2). PC1 is most clearly characterized
The community composition exhibited a major shift by a shift from coral associated species (with low
during the 12-year study period (Fig. 1). The PCA scores), such as Pomacentrus moluccensis and Gobiodon
revealed two distinct clusters, the first contains pre- histrio (which decreased in abundance by 83% and 67%,
bleached (1993–1997), bleaching (1998) and immediate respectively) to less habitat specific forms such as Eviota
r 2006 The Authors
Journal compilation r 2006 Blackwell Publishing Ltd, Global Change Biology, 12, 1587–1594
1590 D . R . B E L L W O O D et al.
1995 250 Abundance
1997 Pre-bleach
Number of individuals
200
1996
(±95%CI)
1993 150
1994
1998 Bleach/ 100
1999 immediate post-bleach
2001 50
2004 0
2000 Post-bleach
2003 35 Species richness
2002
Number of species
30
Fig. 2 A Ward’s hierarchical cluster analysis reveals two dis-
(±95%CI)
25
tinct groups in the community composition of cryptobenthic
fishes. One encompasses the 5 years prebleaching (1993–1997), 20
and the 2 years immediately postbleaching (1998–1999) years,
15
the other encompasses the 5 years postbleaching (2000–2004).
10
queenslandica and Neopomacentrus bankieri (increasing by 1.4 Diversity
240% and 138%, respectively) (cf. Allen, 1991; Depc-
1.2
zynski & Bellwood, 2005). However, a broad range of Diversity index
non-coral associated species also appear to be involved. (H′ ,±95%CI) 1.0
PC2 (explaining 14% of the variation) is largely driven
by variation in the composition of the postbleach as- 0.8
semblages, particularly in the relative abundance of 0.6
the schooling planktivore Neopomacentrus azysron. The
MANOVA revealed complete separation of the two groups 0.4
93
94
95
96
97
98
99
00
01
02
03
04
of years (F10, 1 5 802.7; P 5 0.027). The ANOSIM, likewise,
19
19
19
19
19
19
19
20
20
20
20
20
rejected the null hypothesis of no difference between Year
the two groups (prebleach and postbleach) with a
significance level of 0.3% (i.e. o0.01). The sample Fig. 3 Conventional measures of the status of coral reef fish
statistic (global R) was 0.716. assemblages. Abundance, richness and diversity are expressed
In marked contrast to the clear shift in community in terms of the mean number or diversity of fishes within
composition, analyses of the three traditional variables replicate fish censuses. The vertical shaded bar indicates the
timing and duration of the global coral bleaching event which
for monitoring reef fishes (total abundance, species rich-
killed approximately 75% of the corals in the study area. The
ness, and diversity) exhibited limited variation over the
horizontal lines indicate years that were not significantly
study period and no evidence of a change in relation to different in Tukey’s HSD tests.
the 1998 bleaching event (Fig. 3). Although all three
metrics exhibit a statistically significant difference among
years (abundance F11, 23 5 2.74, P 5 0.02; species richness
Discussion
F11, 23 5 6.36, Po0.001; diversity F11, 23 5 4.57, Po0.01),
and all contain 2–4 distinct groups based on Tukey’s There have been several reports of resilience or a
HSD tests, none of the groups lie on either side of the limited response of reef fishes to large-scale coral loss
1998 bleaching event. Indeed, all groups include at least 2 through storms (Halford et al., 2004), crown of thorns
years on either side of the 1998 event. In terms of long- starfish outbreaks (Williams, 1986; Hart et al., 1996;
term trends only abundance had a significant trend, Sano, 2001) and bleaching (Kokita & Nakazono, 2001;
marked by increasing abundance over the study period Booth & Beretta, 2002). Our results also show a limited
(r2 5 0.67, P 5 0.001). The other two parameters exhibited response to coral bleaching using traditional metrics.
no significant trend (species richness r2 5 4.9 Â 10À7, However, changes in community composition reveal an
P 5 0.998; diversity r2 5 0.33, P 5 0.052). Changes in the unexpected vulnerability of reef fish communities to
abundance of individual species after the 1998 bleaching coral bleaching. These results present a fundamentally
event reveal a marked difference between increases in different perspective, which highlights the need for
generalist omnivorous species, and the planktivore N. caution when selecting metrics to evaluate the resilience
bankieri, and decreases in coral-dependent species (Fig. 4). of coral reef ecosystems.
r 2006 The Authors
Journal compilation r 2006 Blackwell Publishing Ltd, Global Change Biology, 12, 1587–1594
B L E A C H I N G I M PA C T S O N C O R A L R E E F S 1591
Istigobius spp. N. bankieri E. queenslandica
5 120
60
4
80
3 40
Mean abundance (ind. sample −1±SE)
2
40 20
1
0 0 0
Gobiodon spp. Paragobiodon spp. P. moluccensis
16
8 3
12
6 2
8
4
1
2 4
0 0 0
94
96
98
00
02
04
94
96
98
00
02
04
94
96
98
00
02
04
19
19
19
20
20
20
19
19
19
20
20
20
19
19
19
20
20
20
Fig. 4 Changes in the abundance of six cryptobenthic fish species in relation to the 1998 coral bleaching event (marked by the vertical
grey line). The data contrast increases in three generalist species (the planktivore Neopomacentrus bankieri and the omnivorous/
detritivorous Istigobius spp. and Eviota queenslandica) and decreases in three coral-associated species (Pomacentrus moluccensis, Gobiodon
spp. and Paragobiodon spp.).
In contrast to the only other long-term studies (Jones distinct and relatively stable shift from a prebleaching
et al., 2004; Munday, 2004), we found no decrease in to a postbleaching reef fish assemblage. Our findings
diversity, richness or abundance in reef fishes over the emphasize the need for caution when evaluating the
12-year study period. Indeed, as in most previous stu- resilience of reef ecosystems, and highlight the potential
dies, the three main metrics, species richness, diversity insensitivity of common metrics used for measuring
(H 0 ) and total abundance showed no response to a major responses of fishes to bleaching.
disturbance (in this case the 1998 coral bleaching event). By looking at small species with high turnover rates,
At face value this could be taken as strong evidence for and using annual censuses over 12 years, we are able to
resilience – the reef fish fauna remained intact in the face distinguish responses to bleaching from a background
of what was arguably the greatest impact on global coral pattern of interannual variation. Furthermore, our ob-
reefs in living memory. This single event resulted in servations avoid the storage effects of long-lived species
extensive bleaching, with approximately 40% of the where recruitment failure and declines may be masked
world’s reefs exhibiting serious damage. This damage by the presence of long-lived individuals. Many of the
contributed significantly to the widespread decline in conspicuous reef fishes that are included in fish cen-
coral cover on reefs in all tropical oceans (Wilkinson, suses live for one to two decades (Choat & Axe, 1996;
2004; Agardy et al., 2005). Our study site suffered an Choat et al., 1996). The resilience reported in previous
estimated 75% mortality in Acropora species and almost studies (e.g. Williams, 1986; Hart et al., 1996; Sano, 2000;
total loss of the dominant structure providing species Kokita & Nakazono, 2001; Booth & Beretta, 2002;
Montipora (Marshall & Baird, 2000), yet, the reef fishes Halford et al., 2004) may, at least in part, reflect the
appeared to fare well. Indeed, in terms of overall abun- ability of adults of long-lived species to withstand
dance, fish numbers in the study area have almost disturbance (by movement among habitats, using
doubled since the 1998 bleaching event. In terms of reef stored reserves or prey switching, Pratchett et al.,
processes, one may thus assume that little has changed 2004). In contrast, most of the species in the community
and that the critical controlling functions and processes that we examined had maximum longevities of less
provided by fishes remain intact. than 1 year (Depczynski & Bellwood, 2005, 2007; Herna-
This apparent stasis or resilience is misleading: There man & Munday, 2005). This means that these species
were marked changes in community structure. The reveal the effects of bleaching on the entire life history
traditional metrics did not detect a major, yet cryptic, including settlement, recruitment, juvenile and adult
change in community composition, which marked a survival. Previously, our ability to detect changes in reef
r 2006 The Authors
Journal compilation r 2006 Blackwell Publishing Ltd, Global Change Biology, 12, 1587–1594
1592 D . R . B E L L W O O D et al.
systems may have been compromised by both limita- The proximate basis for the changes is uncertain
tions in the methods and the speed with which changes although loss of habitat, if not structure, certainly
are likely to be manifest. If species with a maximum played a role (soft coral colonies were rapidly lost
longevity of a few months take 3 years to reorganize leaving only solid rounded bases, many hard coral
into a relatively stable postbleach community, compar- skeletons are still intact). Live coral (hard and soft)
able reorganization in larger taxa with longevities of 5– was lost with the inevitable loss of habitat for coral
20 years may take several decades. Indeed, if coral dwelling or feeding specialists (cf. Munday, 2004).
bleaching is manifested in coral reef fishes through a Changes in the abundance of coral-associated fishes
disruption of recruitment patterns then the full impact of with decreasing coral cover has been recorded in a
the 1998 bleaching event may have not even begun to number of studies (e.g. Booth & Beretta, 2002; Jones
appear in censuses of long-lived species. et al., 2004). Furthermore, the presence of living coral
This raises the question of whether the observed shift in tissue may be a critical factor in shaping patterns of
cryptobenthic community structure represents a stable recruitment or early postsettlement survivorship
state or merely an alternate or even transitional state. (Beukers & Jones, 1997; Ohman et al., 1998; Holbrook
The postbleaching community is quite distinct and after et al., 2000; McCormick & Hoey, 2006). The stable or
5 years and up to 35 generations later the assemblage gradually increasing fish abundance combined with a
has shown little signs of returning to anything resembling shift in community composition as documented the
the prebleached condition. If the frequency and intensity current study contrasts markedly with the only other
of thermal anomalies and bleaching continues the pre- long-term study in Papua New Guinea (PNG) (Jones
bleach community may never return (cf. Donner et al., et al., 2004) which reported a gradual decline in fish
2005). However, the apparent stability of the postbleach abundance over an 8-year period (changes in commu-
assemblage does not necessarily indicate a stable state. It nity composition in PNG are likely but were not ana-
is certainly plausible that with a return of significant coral lyzed). This disparity between the two studies appears
cover the fish community will change again; the current to mark a fundamental difference in the disturbance
community may thus represent a transitional state. Never- regimes. Our observations mark a distinct shift, primarily
theless, the nature of the shift in community composition in species composition, in response to a single massive
suggests that the time frame for regeneration is likely to bleaching event. Subsequent bleaching events had negli-
be decades rather than years and that there remains a gible impact, as susceptible corals were absent; there were
strong possibility that the path to regeneration will exhibit no noticeable changes in other parameters over the 12-
significant hysteresis (cf. Hughes et al., 2005). year period. In contrast, the reefs examined in PNG were
The timing and nature of the observed changes in fish marked by ongoing bleaching events, crown-of-thorns
community structure strongly suggest that it was a outbreaks and concurrent increases in terrestrial runoff.
direct result of the 1998 bleaching event, even though The latter factor may be particularly important, as it
the shift was marked by changes to a wide range of fish could account for the decline in noncoral species in
taxa. Coral associated species were still living in PNG, reflecting the impacts of terrestrial runoff on
bleached corals in 1998 and in small remnants in 1999, benthic dynamics including algal and detrital resources.
but they largely disappeared thereafter. The loss of coral Finally, one must ask the question of what these
remnants was probably exacerbated by changes in changes mean for ecosystem function, and the resilience
density-dependent coral mortality rates as a result of of reefs to further change. Although our knowledge
excessive predation by the remaining fish coralivores of the ecology of the component taxa is limited, the shift
(up to eight Labrichthys unilineatus were observed feed- in the fish community structure appears to reflect a
ing on a single isolated Acropora colony). What was distinct move from habitat and/or diet specialists to
most striking, apart from a decrease in species with habitat or dietary generalists (cf. Depczynski &
no known association with corals, (e.g. Eviota sp. F Bellwood, 2003, 2004). Prebleaching communities were
(from 1.3 Æ 0.1 to 0.4 Æ 0.1 fish per sample Æ SE) and characterized by the presence of several coral dwelling
Pomacentrus adelus (1.6 Æ 0.5 to 0.6 Æ 0.2 fish per or associated species (P. moluccensis, Gobiodon spp.,
sample Æ SE)), was the marked increase in abundance Paragobiodon spp.; cf. Munday et al., 1997) and included
in number of common, habitat and trophic generalist Amblygobius rainfordi, the only herbivore in this crypto-
species (e.g. Istigobius spp. and E. queenslandica (Fig. 4); benthic community. The density of detritivorous blen-
Depczynski & Bellwood, 2003, 2004). This is particularly nies (Wilson et al., 2003) changed little over the sample
striking as coral represents only a small, but apparently period. The loss of several coral dwelling specialists
important, component of the benthic fauna (throughout reduces the potential for positive synergism between
the period mean live coral cover on GBR reefs was in reef corals and commensal fishes (Pratchett, 2001).
the region of 20–30%; Bellwood et al., 2004). Postbleaching communities were dominated by the
r 2006 The Authors
Journal compilation r 2006 Blackwell Publishing Ltd, Global Change Biology, 12, 1587–1594
B L E A C H I N G I M PA C T S O N C O R A L R E E F S 1593
highly abundant habitat generalist and planktivore identifications; J. Pandolfi, R. Winterbottom and colleagues in the
N. bankieri and the benthic omnivores/detritivores Isti- ARC Centre of Excellence for Coral Reef Studies for helpful
gobius spp. and E. queenslandica. The latter is marked by comments or discussions. Supported by the Australian Research
Council (DRB).
having one of the broadest dietary ranges in the assem-
blage (Depczynski & Bellwood, 2003). The change in
community structure does indeed appear to represent a References
phase shift in that fish-mediated ecosystem processes Ackerman JL, Bellwood DR (2000) Reef fish assemblages: a
have probably been modified. This is not simply a case re-evaluation using enclosed rotenone stations. Marine Ecology
of functional redundancy, responding to change by Progress Series, 206, 227–237.
replacing one species with a functional equivalent; there Ackerman JL, Bellwood DR (2002) Comparative efficiency of
is clear evidence of trophic simplification. clove oil and rotenone for sampling tropical reef fish assem-
This phase shift reveals the limited resilience of the blages. Journal of Fish Biology, 60, 893–901.
prebleach community to coral bleaching. Resilience Ackerman JL, Bellwood DR, Brown JH (2004) The contribution of
small individuals to density–body size relationships: examination
would have been characterized by either no change in
of energetic equivalence in reef fishes. Oecologia, 139, 568–571.
the community composition (resistance) or regeneration
Agardy T, Alder J, Dayton P et al. (2005) Coastal Systems. In:
and the return to a community with similar ecological
Ecosystems and Human Well-Being: Vol. 1 Current State and
characteristics (in terms of ecosystem processes) to Trends (eds Hassan R, Scholes R, Ash N), pp. 513–549. Island
those exhibited by the prebleach community. However, Press, Washington.
the community composition and ecosystem processes Allen GR (1991) Damselfishes of the world. Mergus, Hans A.
both appeared to change as a result of the 1998 coral- Baensch, Melle.
bleaching event. This shift also marked the beginning of Aronson RB, Macintyre IG, Precht WF et al. (2002) The expanding
a new and potentially more resilient postbleach assem- scale of species turnover events on coral reefs in Belize.
blage. The postbleach assemblage is characterized by Ecological Monographs, 72, 233–249.
many generalist species (cf. Bellwood et al., 2005), and Baird AH, Marshall PA (2002) Mortality, growth and reproduc-
as such, may be capable of withstanding future bleach- tion in scleractinian corals following bleaching on the Great
Barrier Reef. Marine Ecology Progress Series, 237, 133–141.
ing events with little further change. We may, thus, be
Bellwood DR, Hughes TP, Folke C et al. (2004) Confronting the
faced with an undesirable yet resilient state: a perni-
coral reef crisis. Nature, 429, 827–833.
cious postbleach fish assemblage. Bellwood DR, Wainwright PC, Fulton CJ et al. (2005) Functional
It is sobering to note that this is the first record of a versatility supports coral reef biodiversity. Proceedings of the
clear shift in a GBR reef fish community in response Royal Society B, 273, 101–107.
to the 1998 bleaching event and that has only been Beukers JS, Jones GP (1997) Habitat complexity modifies the
revealed by examination of exceptionally short-lived impact of piscivores on a coral reef fish population. Oecologia,
species. Furthermore, the study location lies within 114, 50–59.
one of the best protected areas of the Great Barrier Reef Booth DJ, Beretta GA (2002) Changes in a fish assemblage after a
Marine Park, with a relatively intact large fish fauna coral bleaching event. Marine Ecology Progress Series, 245,
(Graham et al., 2003), yet, this phase shift has persisted 205–212.
Caley MJ, Carr MH, Hixon MA et al. (1996) Recruitment and the
for over 5 years, representing 5–35 generations of these
local dynamics of open marine populations. Annual Review of
cryptobenthic fishes. The implications of global warm-
Ecology and Systematics, 27, 477–500.
ing for coral reefs are poorly understood. However, Choat JH, Axe LM (1996) Growth and longevity in acanthurid
our observations provide a clear example of the ex- fishes. An analysis of otolith increments. Marine Ecology Pro-
ceptional sensitivity of coral reefs to environmental gress Series, 134, 15–26.
change and highlight the difficulty of measuring the Choat JH, Axe LM, Lou DC (1996) Growth and longevity in fishes
resilience of coral reef ecosystems. The need for sensi- of the family Scaridae. Marine Ecology Progress Series, 145, 33–41.
tive metrics for evaluating resilience is of paramount Clarke KR, Warwick RM (1994) Changes in Marine Communities:
importance and presents an ongoing challenge in coral An Approach to Statistical Analysis and Interpretation. Natural
reef management. Environmental Research Council, UK.
Connell JH, Hughes TP, Wallace CC (1997) A 30-year study of
coral abundance, recruitment, and disturbance at several
scales in space and time. Ecological Monographs, 67, 461–488.
Acknowledgements Connell JH, Hughes TP, Wallace CC et al. (2004) A long-
term study of competition and diversity of corals. Ecological
We thank the 500 1 JCU MB3160 goby-pickers for their enthu-
siastic assistance; the Great Barrier Reef Marine Park Authority, Monographs, 74, 179–210.
National Parks, Environmental Protection Agency and Depart- Depczynski M, Bellwood DR (2003) The role of cryptobenthic
ment of Primary Industries for permission to collect fishes; reef fishes in coral reef trophodynamics. Marine Ecology
H. Larson, P. Munday and R. Winterbottom for assistance with Progress Series, 256, 183–191.
r 2006 The Authors
Journal compilation r 2006 Blackwell Publishing Ltd, Global Change Biology, 12, 1587–1594
1594 D . R . B E L L W O O D et al.
Depczynski M, Bellwood DR (2004) Microhabitat utilization Marshall PA, Baird AH (2000) Bleaching of corals on the Great
patterns in cryptobenthic reef fish communities. Marine Barrier Reef: differential susceptibilities among taxa. Coral
Biology, 145, 455–463. Reefs, 19, 155–163.
Depczynski M, Bellwood DR (2005) Shortest recorded vertebrate McCormick MI, Hoey AS (2006) Biological and physical corre-
lifespan found in a coral reef fish. Current Biology, 15, 288–289. lates of settlement and survival for a coral reef fish, Pomacen-
Depczynski M, Bellwood DR (2007) Extremes, plasticity and trus amboinensis (Pomacentridae). Proceedings of the 10th
invariance in vertebrate life history traits: insights from coral International Coral Reef Symposium, Okinawa, 425–430.
reef fishes. Ecology (in press). McManus JW, Polsenberg JF (2004) Coral-algal phase shifts on
Donner SD, Skirving WJ, Little CM et al. (2005) Global assess- coral reefs: ecological and environmental aspects. Progress in
ment of coral bleaching requires rates of adaptation under Oceanography, 60, 263–279.
climate change. Global Change Biology, 11, 2251–2265. Munday PL (2004) Habitat loss, resource specialization, and
Edwards AJ, Clark S, Zahir H et al. (2001) Coral bleaching and extinction on coral reefs. Global Change Biology, 10, 1642–1647.
mortality on artificial and natural reefs in Maldives in 1998, Munday PL, Jones GP, Caley MJ (1997) Habitat specialisation
sea surface temperature anomalies and initial recovery. Marine and the distribution and abundance of coral-dwelling gobies.
Pollution Bulletin, 42, 7–15. Marine Ecology Progress Series, 152, 227–239.
Graham NAJ, Evans RD, Russ GR (2003) The effects of marine Nystrom M, Folke C (2001) Spatial resilence of coral reefs.
reserve protection on the trophic relationships of reef fishes on Ecosystems, 4, 406–417.
the Great Barrier Reef. Environmental Conservation, 30, 200–208. Ohman MC, Munday PL, Jones GP et al. (1998) Settlement
Gunderson LH (2000) Ecological resilience – In theory and appli- strategies and distribution patterns of coral-reef fishes. Journal
cation. Annual Review of Ecology and Systematics, 31, 425–439. of Experimental Marine Biology and Ecology, 225, 219–238.
Halford A, Cheal AJ, Ryan D et al. (2004) Resilience to large-scale Pandolfi JM (1999) Response of Pleistocene coral reefs to envir-
disturbance in coral and fish assemblages on the Great Barrier onmental change over long temporal scales. American Zoolo-
Reef. Ecology, 85, 1892–1905. gist, 39, 113–130.
Hart AM, Klumpp DW, Russ GR (1996) Response of herbivorous Pandolfi JM, Bradbury RH, Sala E et al. (2003) Global trajectories
fishes to crown-of-thorns starfish Acanthaster planci outbreaks. of the long-term decline of coral reef ecosystems. Science, 301,
2. Density and biomass of selected species of herbivorous fish 955–958.
and fish-habitat correlations. Marine Ecology Progress Series, Pratchett MS (2001) Influence of coral symbionts on feeding
132, 21–30. preferences of crown-of-thorns starfish Acanthaster planci in
Hernaman V, Munday PL (2005) Life history characteristics of the west Pacific. Marine Ecology Progress Series, 214, 111–119.
coral reef gobies: I. Growth and life-span. Marine Ecology Pratchett MS, Wilson SK, Berumen ML et al. (2004) Sublethal
Progress Series, 290, 207–211. effects of coral bleaching on an obligate coral feeding butter-
Hoegh-Guldberg O (1999) Climate change, coral bleaching and flyfish. Coral Reefs, 23, 352–356.
the future of the world’s coral reefs. Marine and Freshwater Sano M (2000) Stability of reef fish assemblages: responses to
Research, 50, 839–866. coral recovery after catastrophic predation by Acanthaster
Holbrook SJ, Forrester GE, Schmitt RJ (2000) Spatial patterns in planci. Marine Ecology Progress Series, 198, 121–130.
abundance of a damselfish reflect availability of suitable Sano M (2001) Short-term responses of fishes to macroalgal
habitat. Oecologia, 122, 109–120. overgrowth on coral rubble on a degraded reef at Iriomote
Holling CS (1973) Resilience and stability of ecological systems. Island, Japan. Bulletin of Marine Science, 68, 543–556.
Annual Review of Ecology and Systematics, 4, 1–23. Scheffer M, Carpenter SR (2003) Catastrophic regime shifts in
Hughes TP, Baird AH, Bellwood DR et al. (2003) Climate change, ecosystems: linking theory to observation. Trends in Ecology and
human impacts, and the resilience of coral reefs. Science, 301, Evolution, 18, 648–656.
929–933. Scheffer M, Carpenter S, Foley JA et al. (2001) Catastrophic shifts
Hughes TP, Bellwood DR, Folke C et al. (2005) New paradigms in ecosystems. Nature, 413, 591–596.
for supporting resilience of marine ecosystems. Trends in Wilkinson C (2004) Status of Coral Reefs of the World: 2004. Global
Ecology and Evolution, 20, 380–386. Coral Reef Monitoring Network and Australian Institute of
Jackson JBC, Kirby MX, Berger WH et al. (2001) Historical over- Marine Science, Townsville.
fishing and the recent collapse of coastal ecosystems. Science, Williams DMcB (1986) Temporal variation in the structure of reef
293, 629–638. slope fish communities (central Great Barrier Reef): short-term
Jones GP, McCormick MI, Srinivasan M et al. (2004) Coral decline effects of Acanthaster planci infestation. Marine Ecology Progress
threatens fish biodiversity in marine reserves. Proceedings of the Series, 28, 157–164.
National Academy of Sciences of the USA, 101, 8251–8253. Wilson SK (2004) Growth, mortality and turnover rates of a small
Kokita T, Nakazono A (2001) Rapid response of an obligately detritivorous fish. Marine Ecology Progress Series, 284, 253–259.
corallivorous filefish Oxymonacanthus longirostris (Monacanthi- Wilson SK, Bellwood DR, Choat JH et al. (2003) Detritus in
dae). Coral Reefs, 20, 155–158. the epilithic algal matrix and its use by coral reef fishes.
Loya Y, Sakai K, Yamazato K et al. (2001) Coral bleaching: the Oceanography and Marine Biology, An Annual Review, 41,
winners and the losers. Ecology Letters, 4, 122–131. 279–309.
r 2006 The Authors
Journal compilation r 2006 Blackwell Publishing Ltd, Global Change Biology, 12, 1587–1594