Content uploaded by Nicholas K Dulvy
Author content
All content in this area was uploaded by Nicholas K Dulvy
Content may be subject to copyright.
Flattening of Caribbean coral reefs:
region-wide declines in architectural
complexity
Lorenzo Alvarez-Filip1,*, Nicholas K. Dulvy3, Jennifer A. Gill1,4,
Isabelle M. Co
ˆte
´3and Andrew R. Watkinson2
1
Centre for Ecology, Evolution and Conservation, School of Biological Sciences, and
2
School of Environmental Sciences, University of East Anglia, Norwich NR4 7TJ, UK
3
Department of Biological Sciences, Simon Fraser University, Burnaby, British Columbia V5A 1S6, Canada
4
Tyndall Centre for Climate Change Research, Norwich NR4 7TJ, UK
Coral reefs are rich in biodiversity, in large part because their highly complex architecture provides shelter and
resources for a wide range of organisms. Recent rapid declines in hard coral cover have occurred across the
Caribbean region, but the concomitant consequences for reef architecturehave notbeen quantified on a large
scale to date. We provide, to our knowledge, the first region-wide analysis of changes in reef architectural
complexity, using nearly 500 surveys across 200 reefs, between 1969 and 2008. The architectural complexity
of Caribbean reefs has declined nonlinearly with the near disappearance of the most complex reefs over the
last 40 years. The flattening of Caribbean reefs was apparent by the early 1980s, followed by a period of stasis
between 1985 and 1998 and then a resumption of the decline in complexity to the present. Rates of loss are
similar on shallow (,6 m), mid-water (6–20 m) and deep (.20 m) reefs and are consistent across all five
subregions. The temporal pattern of declining architecture coincides with key events in recent Caribbean
ecological history: the loss of structurally complex Acropora corals, the mass mortality of the grazing
urchin Diadema antillarum and the 1998 El Nino Southern Oscillation-induced worldwide coral bleaching
event. The consistently low estimates of current architectural complexity suggest regional-scale degradation
and homogenization of reef structure. The widespread loss of architectural complexity is likely to have serious
consequences for reef biodiversity, ecosystem functioning and associated environmental services.
Keywords: climate change; ecosystem degradation; ecosystem services; foundation species;
habitat complexity; vulnerability
1. INTRODUCTION
The physical structure of a habitat profoundly influences
its associated biodiversity and ecosystem functioning
(MacArthur & MacArthur 1961), with more complex
habitats facilitating species coexistence through niche par-
titioning and the provision of refuges from predators and
environmental stressors (Bruno & Bertness 2001;Willis
et al. 2005). In tropical shallow waters, the calcium
carbonate skeletons of stony corals contribute to reef
frameworks that sustain the most diverse ecosystem in
our seas (Spalding et al. 2001). However, coral reefs
have been heavily impacted worldwide by a combination
of local and global stressors, including overfishing,
climate change-induced coral bleaching, eutrophication
and disease (Hughes et al. 2003). The marked declines
in live hard coral cover documented over recent decades
throughout the Caribbean and the Indo-Pacific regions
(Gardner et al. 2003;Bruno & Selig 2007) exceed those
reported for many other foundation species in terrestrial
or marine ecosystems (Balmford et al. 2003). However,
in contrast to other ecosystems where degradation usually
indicates reductions in habitat area (e.g. deforestation),
decreases in live coral cover on coral reefs do not immedi-
ately result in loss of available habitat because the reef
framework can persist long after the death of corals.
In the Caribbean, declines in live coral cover began in
the late 1970s, when substantial loss of the major reef-
forming corals Acropora palmata and Acropora cervicor nis
occurred as a result of white-band disease (Aronson &
Precht 2001). Coral mortality, in combination with
the mass mortality of the black sea urchin (Diadema
antillarum), which was a major remover of algae, and
the long-term depletion of herbivorous fishes through
overfishing, facilitated phase shifts to macro-algal dominance
on many reefs (Carpenter 1988;Precht & Aronson 2006).
In the Caribbean and elsewhere, reef-building corals now
face new threats from climate change, particularly in the
form of thermally induced coral bleaching and mortality,
which are becoming increasingly frequent and extensive
as thermal anomalies intensify and lengthen (Hughes
et al. 2003;McWilliams et al. 2005).
A potential consequence of the widespread reduction
in Caribbean coral cover is a reversal of the historic net
accretion of calcium carbonate, resulting in a decrease
in calcification and erosion of the reef framework. At
local scales, hard coral mortality is associated with the
loss of architectural complexity and ‘reef flattening’ after
direct impacts such as hurricanes through the breakage
of coral skeletons (e.g. Rogers et al. 1982). Reefs may
*Author for correspondence (l.alvarez@uea.ac.uk).
Electronic supplementary material is available at http://dx.doi.org/10.
1098/rspb.2009.0339 or via http://rspb.royalsocietypublishing.org.
Proc. R. Soc. B (2009) 276, 3019–3025
doi:10.1098/rspb.2009.0339
Published online 10 June 2009
Received 26 February 2009
Accepted 13 May 2009 3019 This journal is q2009 The Royal Society
on 14 July 2009rspb.royalsocietypublishing.orgDownloaded from
also erode gradually owing to the natural activity of host
organisms, such as herbivorous fishes and sea urchins,
and by physical abrasion or geochemical shifts. However,
widespread mortality of hard corals, for example, after
severe bleaching events, moves the balance towards net
reef erosion (Sheppard et al. 2002). These impacts
could be exacerbated in the future by ocean acidification,
which is expected to enhance calcium carbonate dissol-
ution with negative consequences, initially for coral
growth and eventually for the entire reef framework
(Hoegh-Guldberg et al. 2007).
The ecological and socio-economic consequences of
declining architectural complexity are likely to be sub-
stantial (Pratchett et al. 2008). For many reef organisms,
risk of predation is influenced by access to refuges, and
the densities of herbivores and grazing rates typically
increase with architectural complexity (Beukers & Jones
1997;McClanahan 1999;Almany 2004;Lee 2006).
Consequently, the species richness, abundance and
biomass of coral reef fishes and invertebrates are all
influenced by architectural complexity (e.g. Gratwicke &
Speight 2005;Idjadi & Edmunds 2006;Wilson et al.
2007). The loss of architectural complexity may therefore
drive declines in diversity, particularly of habitat special-
ists, and compromise fisheries productivity through
elevated post-settlement mortality (Beukers & Jones
1997;Graham et al. 2007). Reef architectural complexity
also plays a key role in providing important environmental
services to humans, including enhancing coastal protec-
tion through the dissipation of wave energy transmitted
over reefs (Lugo-Fernandez et al. 1998).
While recent regional-scale analyses have revealed
declines in hard coral cover (Gardner et al.2003;Bruno &
Selig 2007), the consequences for reef habitat complexity
on a similar large scale have not been quantified. The
capacity of reefs to continue to perform key functions of
refuge provision and coastal protection will depend on
whether reef architecture persists for a substantial period
of time following the loss of live coral. Here we collate
published and unpublished estimates of reef complexity
spanning four decades from reefs across the Caribbean,
a region with clear evidence of recent declines in coral
cover. We explore the rate and timing of changes in reef
architecture in relation to region-wide events such as the
demise of Acropora corals and grazing urchins. As the
drivers of reef degradation are apparent throughout the
Caribbean, we also examine whether the patterns are
consistent throughout the entire region.
2. MATERIAL AND METHODS
(a)Estimating architectural complexity
Habitat complexity on coral reefs has been measured using a
variety of methods that differ in the attributes measured, the
scale of measurement and the degree of subjectivity (with
attendant variation in inter-observer comparability). The
rugosity index is by far the most widely used method for
measuring reef architectural complexity (see electronic sup-
plementary material for further details) and is generally
highly correlated with other methods (Wilson et al. 2007).
Studies reporting the rugosity index were therefore chosen
to quantify spatial and temporal variation in the architectural
complexity of reefs across the Caribbean.
The rugosity index is expressed as the ratio between the
total length of a chain and the length of the same chain
when moulded to a reef surface. A perfectly flat surface
would have a rugosity index of 1, with larger numbers indicat-
ing a greater degree of architectural complexity (figure 1). The
index tends towards infinity with increasing architectural com-
plexity; however, rugosity estimates greater than 3 are very rare.
(b)Data search
A database of quantitative surveys that measured reef rugos-
ity within the wider Caribbean was compiled. We searched
online ISI Web of Science, Google Scholar and other relevant
(a)
(b)
(c)
Figure 1. Examples of three different values of rugosity index
of architectural complexity on Caribbean reefs. The values of
theindexare(a)1.2,(b)1.5and(c) 2.5. Source for photos:
L. Alvarez-Filip, M. Uyarra and W. Henry Marine Photobank.
3020 L. Alvarez-Filip et al. The flattening of Caribbean coral reefs
Proc. R. Soc. B (2009)
on 14 July 2009rspb.royalsocietypublishing.orgDownloaded from
databases (e.g. Reefbase) for peer-reviewed and grey litera-
ture using several search terms (see electronic supplementary
material for examples). We also searched for papers that used
the rugosity index in all issues of the journals Coral Reefs,
Bulletin of Marine Science,Atoll Research Bulletin,Caribbean
Journal of Science and in all Proceedings of the International
Coral Reef Symposium. Additionally, we directly contacted
coral reef scientists, site managers and those responsible for
reef monitoring programmes throughout the Caribbean,
asking for any available data pertaining to their study sites.
A total of 464 records from 200 reefs surveyed between
1969 and 2008 across the Caribbean were obtained
(figure 2a,b). The database includes reefs that were surveyed
only once (n¼214) and reefs where repeated measures of
rugosity were collected over more than 1 year (n¼250).
Both datasets provide highly consistent results (table S2,
electronic supplementary material). We therefore present
findings only from the whole dataset, because they offer a
wider spatial and temporal representation.
(c)Analyses
To assess the temporal pattern of change in region-wide
architectural complexity, we calculated annual estimates of
rugosity averaged across all available sites for each year
from 1969 to 2008. We fitted a range of linear and nonlinear
models to represent increasing degrees of complexity in the
rate of change in rugosity over time and used the
small-sample adjusted Akaike information criterion (AIC
c
)
to evaluate the models (Burnham & Anderson 2002).
Linear models were fitted using both simple regressions
and robust regression, to reduce the influence of outliers.
We contrasted these linear models, which represent a hypoth-
esis of constant change in rugosity over the whole time
period, against segmented models that assumed piecewise
linear relationships (i.e. two or more straight lines connected
by breakpoints) and a general additive model (GAM) of an
unspecified nonlinear (spline) function, which assumed that
the rate of change in rugosity varied over time (Venables &
Ripley 2002;Muggeo 2003). In addition, because the
number of sites contributing to each annual rugosity estimate
varied, with more sites available towards the end of the time
period, we ran all models with annual estimates unweighted
and weighted by sample size. Weighted models consistently
provided a significantly better fit (lower AIC and higher var-
iance explained) than unweighted models. All analyses were
implemented in R (R 2008).
We used randomization techniques to evaluate whether
the pattern and rate of change were sensitive to the inclusion
of any particular site or year. For the best-supported model
identified in the AIC
c
analysis, we tested whether the rate
of decline in rugosity was biased by the inclusion of any
particular year, using a jackknife method to calculate the
distribution of annual decline rates while sequentially
removing each individual year. To evaluate any potential
site selection bias, we used a bootstrap method to compare
the annual decline rate with the range of possible decline
rates for 10 000 random combinations of year, rugosity and
weighting.
To explore whether the trends of changing architectural
complexity varied with depth and within the region, we
aggregated the data by decade to maximize the signal relative
to interannual variation while retaining sufficient power. To
evaluate the change in rugosity at different depths, we
divided the data into three zones: (i) ,6 m, which represents
the optimal range of A. palmata (and therefore Acropora
reefs); (ii) 6 – 20 m, to include the range of other reef-
building scleractinian corals, including A. cervicornis; and
(iii) .20 m, to reflect sites where hard corals are present
but do not necessarily form complex three-dimensional
structures. We also aggregated the data within five subregions
to explore spatial variation in changes in rugosity within the
Caribbean region (figure 2a).
A key question is whether the regional change in reef
structure has produced more structurally homogeneous habi-
tats throughout the Caribbean. We classified each reef into
one of five rugosity index categories (1.0–1.49, 1.5–1.99,
2.0–2.49, 2.5– 2.99 and greater than 3.0) to explore the
change in the region-wide representation of complex (rugos-
ity greater than 2.0) and flatter (rugosity less than or equal to
1.5) reefs for the four different decades.
3. RESULTS
There has been a marked decline in the architectural
complexity of Caribbean reefs over the past four decades
(figure 3). The best-supported model of change in rugos-
ity over time was a weighted segmented model (table 1),
0
15
30
45
(a)
(b)
90°0'0'' W
0 500 1000
km
N
10°0'0'' N 20°0'0'' N 30°0'0'' N
80°0'0'' W 70°0'0'' W 60°0'0'' W
1965 1980 1995 2010
number of studies
Figure 2. (a) Regional distribution of locations from which
rugosity values were obtained. Grey circles, Central America;
white circles, South America; black circles, lesser Antilles;
circles with vertical lines, greater Antilles; circles with
crosses, southwest North Atlantic. (b) Number of studies
from which rugosity data were collated per year, from 1969
to 2008.
The flattening of Caribbean coral reefs L. Alvarez-Filip et al. 3021
Proc. R. Soc. B (2009)
on 14 July 2009rspb.royalsocietypublishing.orgDownloaded from
which suggests that the decline in rugosity has three
distinct phases of change (figure 3). Architectural
complexity declined steeply early in the time series
(1969–1985), from reefs with indices of approximately
2.5 to much flatter reefs with indices of approximately 1.5.
This period of decline ended in 1985 (+2.4 years s.e.),
and architectural complexity throughout the region then
remained relatively stable until the late 1990s. However,
since 1998 (+2.8 years s.e.), the declining trend has
resumed, with rugosity indices after the mid-2000s reach-
ing the lowest levels recorded in the time series (approx.
1.2; see example in figure 1). The pattern of change is
robust to the inclusion or exclusion of individual years
(jackknife) and individual sites (bootstrap) (figure S1,
electronic supplementary material).
The decline in architectural complexity is widespread.
The temporal pattern of change was consistent across all
three depth intervals (figure 4a) and across the three sub-
regions for which the available data spanned the whole
time period, and the two regions with patchier data
(Central America and southwest North Atlantic; figure 4b).
Caribbean reefs are becoming both flatter and
more structurally homogeneous across the region. The
proportion of complex reefs (rugosity greater than 2) has
declined from approximately 45 per cent of sites to
approximately 2 per cent in the past four decades (figure 5).
4. DISCUSSION
The architectural complexity of coral reefs has declined
drastically over the last 40 years throughout the
Caribbean. Structurally complex reefs with a rugosity of
greater than 2 have been virtually lost from the entire
region. Today, the flattest reefs (rugosity less than 1.5)
comprise approximately 75 per cent of the total compared
with approximately 20 per cent in the 1970s, with most of
the increase in the proportion of flattest reefs occurring in
the 2000s. The high proportion of complex reefs in the
1960s and 1970s is unlikely to result from researchers
tending to visit just the most pristine reefs at this time,
because less architecturally-complex categories were also
well represented during this period. The loss of architec-
tural complexity is nonlinear and has occurred over three
distinct phases that coincide closely with large-scale
events that have affected Caribbean reef ecosystems.
The rate of decline was steepest prior to 1985. The
sample sizes are small and variance high during the
1960s and 1970s, hence it is unclear whether the decline
began prior to the early 1980s, when widespread loss of
acroporid corals began (Precht & Aronson 2006). After
this period, average architectural complexity changed
little until the late 1990s, when a new episode of decline
began. The pattern of decline is consistent across depth
zones and subregions. The widespread occurrence of
flatter reefs could have serious implications for reef-
associated biodiversity and reef-based environmental
services.
The nonlinearity in the loss of architectural complexity
suggests that different drivers operating at different times
have influenced components of the reef community. Dis-
turbances on reefs range in scale and intensity, from local
tropical storms that can break and displace coral skel-
etons, to widespread events such as climate-induced
bleaching and diseases that kill coral tissue without
immediately compromising the reef structure (Pratchett
et al. 2008). In the late 1970s, one key event is likely to
have had a major role in the early, steep decline in
Caribbean reef architecture. White-band disease killed
approximately 90 per cent of the shallow-water, structu-
rally dominant acroporid corals, exposing their fragile
branching skeletons to erosion and hurricanes that prob-
ably led to their collapse in subsequent years (Aronson &
Precht 2001). However, declines also occurred at depths
greater than those at which acroporids were dominant,
suggesting that the systematic loss of Caribbean reef
corals was more widespread than previously thought
during the 1970s and early 1980s.
After 1985, the main driver(s) of declining architec-
tural complexity appear to cease; by this time, acroporids
had disappeared almost entirely from the Caribbean, and
the sea urchin D. antillarum had experienced a region-
wide disease-induced mass mortality in 1983 – 1984
(Carpenter 1988). The loss of this important source of
bioerosion may have slowed the decline following the
first phase of reef flattening. This intermediate stable
period of architectural complexity in the region persisted
in the face of several disturbance events, including the
first large-scale bleaching events and several major
hurricanes (Gardner et al. 2005;McWilliams et al. 2005).
Around 1998, Caribbean reefs were tipped into a new
phase of structural decline, following the most
intense and widespread coral bleaching event to date
(McWilliams et al. 2005). The coral mortality and
reductions in growth rates that typically follow such
bleaching events are likely to have precipitated the
resumption of loss of architectural complexity. The low
levels of coral cover, and presumably reef accretion, at
this time (Gardner et al. 2003) may also have increased
rates of erosion of underlying geological structures that
were no longer shielded by actively growing hard corals.
Since 1998, further mass bleaching events have occurred
1
2
3
4
1965 1980 1995 2010
rugosity
Figure 3. Changes in reef rugosity on reefs across the
Caribbean from 1969 to 2008. Black line represents
the best fitting model—a segmented regression weighted by
the number of sites contributing to each annual rugosity
estimate (mean +95% confidence intervals). Black dots at
the top of the figure indicate the significant breakpoint in
1985 and 1998 (+1 s.e.) for the segmented regression.
Model slopes: 1969–1984, 20.054; 1985–1997, 0.008;
1998–2008, 20.038.
3022 L. Alvarez-Filip et al. The flattening of Caribbean coral reefs
Proc. R. Soc. B (2009)
on 14 July 2009rspb.royalsocietypublishing.orgDownloaded from
regularly (McWilliams et al. 2005), probably contributing
to the continued decline in reef complexity.
All of the major events that are likely to have impacted
reef complexity have occurred against a backdrop of
changes not only in coral abundance but also in commu-
nity composition. Following the disappearance of
acroporids, massive species with slower growth rates,
such as Montastrea spp., remained as the primary reef
framework builders, and weedy corals, such as Porites
spp. and Agaricia spp., that form rapidly growing, small
colonies that are short-lived and quickly replaced, started
to increase in abundance (Green et al. 2008). The shift
from major reef-building species to weedy species that
contribute less to maintenance of the reef framework,
combined with increases in macro-algae (Co
ˆte
´et al.
2006) that compete for space with coral recruits
(Mumby et al. 2007), probably reduced the rates of
coral accretion on Caribbean reefs.
The loss of reef architecture is likely to have profound
ecological, social and economic impacts. A growing body
of evidence indicates severe repercussions for biodiversity
of the loss of architectural complexity. On Indo-Pacific
reefs, major changes in fish community composition
have resulted from the long-term loss of structure
following coral bleaching events (Pratchett et al.2008and
references therein). The effects of bleaching are first manifest
in obligate coral-dwelling species, followed by impacts on
other small-bodied fishes (both small adults and juveniles
of larger species) when the physical matrix of the reef col-
lapses (Pratchett et al.2008). In the Caribbean, the greatest
impacts on biodiversity are expected to occur only with the
breakdown of the reef matrix because no fish species feed
1
2
3
4
1970s 1980s 1990s 2000s
rugosity
1.0
1.5
2.0
2.5
(a)
(b)
rugosity
Figure 4. Change in Caribbean reef rugosity in four different
decades (a) at three depth intervals (filled circle, 0–6 m;
open diamond, 6 –20 m; filled triangle, 20 m) and (b)in
five subregions (filled square, southwest North Atlantic;
grey circle, greater Antilles; open diamond, lesser Antilles;
filled triangle, South America; open circle, Central America)
(mean index value +95% confidence intervals).
0
0.2
0.4
0.6
0.8
1.0
1970s 1980s 1990s 2000s
proportion of reefs
Figure 5. Proportion of reefs in five rugosity index categories
across the Caribbean between 1969 and 2008. Number of
studies for each decade: 1970s: n¼32; 1980s: n¼52;
1990s: n¼136 and 2000s: n¼167. Black, .3; dark grey,
2.5– 3; mid grey, 2– 2.5; pale grey, 1.5 –2; white, 1– 1.5.
Table 1. Model structure and the temporal pattern of change in Caribbean architectural complexity. Summary of AIC
c
analysis of linear and nonlinear models of change in yearly mean rugosity (derived from all 464 estimates), ordered by
decreasing weight. (Models in which annual rugosity estimates have been weighted by sample size are indicated (wt). df,
degrees of freedom of the model (for GAM, we use the estimated degrees of freedom). AIC
c
is the Akaike information
criterion corrected for small sample size; Dis the difference in AIC
c
between a given model and the best-supported model
(indicated in bold); and Wis the Akaike weight, which represents the probability that a given model is the best of those
models considered. The asterisks indicate the average slope of the different model segments.)
model R
2
slope df AIC
c
DAIC
c
AIC
c
W
segmented model (wt) 0.64 20.028*26 225.8 0 0.8695
linear model (wt) 0.53 20.019 30 217.1 8.7 0.0112
robust linear model (wt) — 20.018 30 216.9 8.8 0.0107
segmented model 0.65 20.038*26 22.9 22.9 0.0000
generalized additive model (wt) 0.99 20.033*3.6 0.1 25.8 0.0000
linear model 0.49 20.026 30 9.4 35.2 0.0000
robust linear model — 20.021 30 11.2 37 0.0000
generalized additive model 0.59 20.044*3.3 22.8 48.6 0.0000
The flattening of Caribbean coral reefs L. Alvarez-Filip et al. 3023
Proc. R. Soc. B (2009)
on 14 July 2009rspb.royalsocietypublishing.orgDownloaded from
exclusively on live coral, although many reef-associated
species depend highly on rugose substrata to feed, recruit
and hide (Gratwicke & Speight 2005). In this context, declin-
ing reef complexity may explain the onset of a decline in
Caribbean reef fishes that has occurred since approximately
1996 (Paddack et al.2009). Given that the loss of reef archi-
tecture began much earlier, our analysis supports the notion
of a degradation debt for Caribbean reef fishes. Reduced
recruitment resulting from a lack of settlement sites and
refuges for species with commercial importance, such as lob-
sters and large fishes (Graham et al.2007;Wynne & Co
ˆte
´
2007), may compromise the long-term sustainability of fish-
eries and fishing communities. Collapsing reef structures
may also lead to the loss of important environmental services
such as coastal protection. Simulation models predict that a
reduction in reef surface roughness of approximately 50 per
cent could produce a doubling of the wave energy reaching
the shores behind those reefs (Sheppard et al.2005). The vul-
nerability of coastal human communities in the Caribbean to
projected increases in the intensity of Atlantic Ocean hurri-
canes and sea levels (Hopkinson et al.2008) will therefore
probably be compounded by the reduced wave dissipation
function of architecturally simpler reefs.
Reversing declines in reef architecture will be a major
challenge for scientists and policy-makers concerned
with maintaining reef ecosystems and the security and
wellbeing of Caribbean coastal communities. Although
recent evidence suggests increases in coral cover on
some Caribbean reefs (e.g. Cho & Woodley 2000;Idjadi
et al. 2006), the effect of coral recovery on architectural
complexity is unknown. If weedy corals dominate this
recovery in the long term, future reef complexity is unli-
kely to mirror any improvement in coral condition. To
regain the levels of architectural complexity that were
prevalent prior to 1980, the recovery of large branching
corals (i.e. Acropora spp.) and the maintenance of healthy
populations of massive robust species (e.g. Montastrea
spp.) are essential within the region. Not meeting these
challenges will most probably result in a continued flat-
tening of reefs throughout the region and seriously
compromised biodiversity and environmental services.
We are grateful to Peter Edmunds, Michelle Paddack, Philip
Molloy, Renata Goodridge and Hazel Oxenford
(CARICOMP Barbados), Francisco Geraldes (Centro de
Investigaciones de Biologı
´a Marina de la Universidad
Auto
´noma de Santo Domingo and CARICOMP) and
Simon Pittman and the NOAA Biogeography Branch for
contributing unpublished data. L.A.-F. was supported by a
scholarship from the CONACYT (171864) Mexico. I.M.C.
and N.K.D. are supported by Discovery grants from the
Natural Sciences and Engineering Research Council of
Canada.
REFERENCES
Almany, G. R. 2004 Differential effects of habitat complex-
ity, predators and competitors on abundance of juvenile
and adult coral reef fishes. Oecologia 141, 105–113.
(doi:10.1007/s00442-004-1617-0)
Aronson, R. B. & Precht, W. F. 2001 White-band disease and
the changing face of Caribbean coral reefs. Hydrobiologia
460, 25–38. (doi:10.1023/A:1013103928980)
Balmford, A., Green, R. E. & Jenkins, M. 2003 Measuring
the changing state of nature. Trends Ecol. Evol. 18,
326–330. (doi:10.1016/S0169-5347(03)00067-3)
Beukers, J. S. & Jones, G. P. 1997 Habitat complexity mod-
ifies the impact of piscivores on a coral reef fish
population. Oecologia 114, 50–59. (doi:10.1007/
s004420050419)
Bruno, J. F. & Bertness, M. D. 2001 Habitat modification
and facilitation in benthic marine communities. In
Marine community ecology (eds M. D. Bertness, S. D.
Gaines & M. E. Hay), pp. 201– 218. Sunderland, MA:
Sinauer.
Bruno, J. F. & Selig, E. Z. 2007 Regional decline of coral
cover in the Indo-Pacific: timing, extent, and subregional
comparisons. PLoS ONE 2, e711. (doi:10.1371/journal.
pone.0000711)
Burnham, K. P. & Anderson, D. R. 2002 Model selection and
multimodel inference: a practical information–theoretic
approach. New York, NY: Springer-Verlag.
Carpenter, R. C. 1988 Mass mortality of a Caribbean sea
urchin: immediate effects on community metabolism and
other herbivores. Proc. Natl Acad. Sci. USA 85, 511–514.
Cho, L. L. & Woodley, J. D. 2000 Recovery of reefs at Dis-
covery Bay, Jamaica and the role of Diadema antillarum.
In 9th Int. Coral Reef Symp. vol. 1 (eds M. K. Moosa,
S. Soemodihardjo, A. Soegiarto, K. Romimohtarto,
A. Nontji, Soekarno & Suharsono), pp. 331 – 338. Bali,
Indonesia: Ministry of Environment, Indonesian Institute
of Sciences and International Society for Reef Studies.
Co
ˆte
´, I. M., Gardner, T. A., Gill, J. A., Hutchinso, D. J. &
Watkinson, A. R. 2006 New approaches to estimating
recent ecological changes on coral reefs. In Coral reef con-
servation (eds I. M. Co
ˆte
´& D. J. Reynolds), pp. 293– 313.
Cambridge, UK: Cambridge University Press.
Gardner, T. A., Co
ˆte
´, I. M., Gill, J. A., Grant, A. &
Watkinson, A. R. 2003 Long-term region-wide declines
in Caribbean corals. Science 301, 958 –960. (doi:10.
1126/science.1086050)
Gardner, T. A., Co
ˆte
´, I. M., Gill, J. A., Grant, A. &
Watkinson, A. R. 2005 Hurricanes and Caribbean coral
reefs: impacts, recovery patterns, and role in long-term
decline. Ecology 85, 174–184. (doi:10.1890/04-0141)
Graham, N. A. J., Wilson, S. K., Jennings, S., Polunin,
N. V. C., Robinson, J., Bijoux, J. P. & Daw, T. M. 2007
Lag effects in the impacts of mass coral bleaching on
coral reef fish, fisheries, and ecosystems. Conserv. Biol. 21,
1291– 1300. (doi:10.1111/j.1523-1739.200700754.x)
Gratwicke, B. & Speight, M. R. 2005 Effects of habitat com-
plexity on Caribbean marine fish assemblages. Mar. Ecol.
Prog. Ser. 292, 301–310. (doi:10.3354/meps292301)
Green, D. H., Edmunds, P. J. & Carpenter, R. C. 2008
Increasing relative abundance of Porites astreoides on
Caribbean reefs mediated by an overall decline in coral
cover. Mar. Ecol. Prog. Ser. 359, 1 – 10. (doi:10.3354/
meps07454)
Hoegh-Guldberg, O. et al. 2007 Coral reefs under rapid
climate change and ocean acidification. Science 318,
1737–1742. (doi:10.1126/science.1152509)
Hopkinson, C. S., Lugo, A. E., Alber, M., Covich, A. P. &
Van Bloem, S. J. 2008 Forecasting effects of sea-level
rise and windstorms on coastal and inland ecosystems.
Front. Ecol. Environ. 6, 255–263. (doi:10.1890/070153)
Hughes, T. P. et al. 2003 Climate change, human impacts,
and the resilience of coral reefs. Science 301, 929 –933.
(doi:10.1126/science.1085046)
Idjadi, J. A. & Edmunds, P. J. 2006 Scleractinian corals as
facilitators for other invertebrates on a Caribbean reef.
Mar. Ecol. Prog. Ser. 319, 117– 127. (doi:10.3354/
meps319117)
Idjadi, J. A., Lee, S. C., Bruno, J. F., Precht, W. F.,
Allen-Requa, L. & Edmunds, P. J. 2006 Rapid phase-
shift reversal on a Jamaican coral reef. Coral Reefs 25,
209–211. (doi:10.1007/s00338-006-0088-7)
3024 L. Alvarez-Filip et al. The flattening of Caribbean coral reefs
Proc. R. Soc. B (2009)
on 14 July 2009rspb.royalsocietypublishing.orgDownloaded from
Lee, S. C. 2006 Habitat complexity and consumer-mediated
positive feedbacks on a Caribbean coral reef. Oikos
112, 442–447. (doi:10.1111/j.0030-1299.2006.14247.x)
Lugo-Fernandez, A., Roberts, H. H. & Suhayda, J. N. 1998
Wave transformations across a Caribbean fringing-barrier
coral reef. Cont. Shelf Res. 18, 1099–1124. (doi:10.1016/
S0278-4343(97)00020-4)
MacArthur, R. H. & MacArthur, J. W. 1961 On bird species
diversity. Ecology 42, 594–598.
McClanahan, T. R. 1999 Predation and the control of the sea
urchin Echinometra viridis and fleshy algae in the patch
reefs of Glovers Reef, Belize. Ecosystems 2, 511– 523.
(doi:10.1007/s100219900099)
McWilliams, J. P., Co
ˆte
´, I. M., Gill, J. A., Sutherland, W. J. &
Watkinson, A. R. 2005 Accelerating impacts of tempera-
ture-induced coral bleaching in the Caribbean. Ecology
86, 2055–2060. (doi:10.1890/04-1657)
Muggeo, V. M. R. 2003 Estimating regression models with
unknown break-points. Stat. Med. 22, 3055– 3071.
(doi:10.1002/sim.1545)
Mumby, P. J. et al. 2007 Trophic cascade facilitates coral
recruitment in a marine reserve. Proc. Natl Acad. Sci.
104, 8362–8367. (doi:10.1073/pnas.0702602104)
Paddack, M. J. et al. 2009 Recent region-wide declines in
Caribbean reef fish abundance. Curr. Biol. 19, 590–595.
(doi:10.1016/j.cub.2009.02.041)
Pratchett, M. S., Munday, P. L., Wilson, S. K., Graham,
N. A. J., Cinner, J. E., Bellwood, D. R., Jones, G. P.,
Polunin, N. & McClanahan, T. 2008 Effects of
climate-induced coral bleaching on coral-reef-fishes:
ecological and economic consequences. Oceanogr. Mar.
Biol. Annu. Rev. 46, 251– 296.
Precht, W. F. & Aronson, R. B. 2006 Death and resurrection
of Caribbean coral reefs: a paleoecological perspective.
In Coral reef conservation (eds I. M. Co
ˆte
´& D. J. Reynolds),
pp. 40–77. Cambridge, UK: Cambridge University Press.
R. 2008 R: a language and environment for statistical
computing. Vienna, Austria: R Foundation for Statistical
Computing.
Rogers, C. S., Suchanek, T. H. & Pecora, F. A. 1982 Effects
of hurricanes David and Frederic (1979) on shallow Acro-
pora palmata reef communities: St-Croix, United States
Virgin Islands. Bull. Mar. Sci. 32, 532– 548.
Sheppard, C. R. C., Spalding, M., Bradshaw, C. &
Wilson, S. 2002 Erosion vs. recovery of coral reefs
after 1998 El nino: Chagos reefs, Indian Ocean. Ambio
31,40–48.(doi:10.1579/0044-7447-31.1.40)
Sheppard, C., Dixon, D. J., Gourlay, M., Sheppard, A. &
Payet, R. 2005 Coral mortality increases wave energy
reaching shores protected by reef flats: examples from
the Seychelles. Estuar. Coast. Shelf Sci. 64, 223–234.
(doi:10.1016/j.ecss.2005.02.016)
Spalding, M., Ravilious, C. & Green, E. P. 2001 World atlas of
coral reefs. Berkeley, CA: University of California Press.
Venables, W. N. & Ripley, B. D. 2002 Modern and applied
statistics. New York, NY: Springer-Verlag.
Willis, S., Winemiller, K. & Lopez-Fernandez, H. 2005
Habitat structural complexity and morphological diversity
of fish assemblages in a Neotropical floodplain river. Oeco-
logia 142, 284–295. (doi:10.1007/s00442-004-1723-z)
Wilson, S. K., Graham, N. A. J. & Polunin, N. V. C. 2007
Appraisal of visual assessments of habitat complexity
and benthic composition on coral reefs. Mar. Biol. 151,
1069–1076. (doi:10.1007/s00227-006-0538-3)
Wynne, S. P. & Co
ˆte
´, I. M. 2007 Effects of habitat quality
and fishing on Caribbean spotted spiny lobster popu-
lations. J. Appl. Ecol. 44, 488– 494. (doi:10.1111/
j.1365-2664.2007.01312.x)
The flattening of Caribbean coral reefs L. Alvarez-Filip et al. 3025
Proc. R. Soc. B (2009)
on 14 July 2009rspb.royalsocietypublishing.orgDownloaded from
