Content uploaded by Nicholas K Dulvy
Author content
All content in this area was uploaded by Nicholas K Dulvy on Jan 19, 2018
Content may be subject to copyright.
Region-wide temporal and spatial variation in Caribbean
reef architecture: is coral cover the whole story?
LORENZO ALVAREZ-FILIP
*
w, ISABELLE M. CO
ˆTE
´w,JENNIFERA.GILL
*
z,
ANDREW R. WATKINSON§ and NICHOLAS K. DULVYw
*
Centre for Ecology, Evolution, and Conservation, School of Biological Sciences, University of East Anglia, Norwich NR4 7TJ, UK,
wEarth to Ocean Research Group, Department of Biological Sciences, Simon Fraser University, Burnaby, BC, Canada V5A 1S6,
zTyndall Centre for Climate Change Research, Norwich NR4 7TJ, UK, §School of Environmental Sciences, University of East
Anglia, Norwich NR4 7TJ, UK
Abstract
The architectural complexity of coral reefs is largely generated by reef-building corals, yet the effects of current
regional-scale declines in coral cover on reef complexity are poorly understood. In particular, both the extent to which
declines in coral cover lead to declines in complexity and the length of time it takes for reefs to collapse following coral
mortality are unknown. Here we assess the extent of temporal and spatial covariation between coral cover and reef
architectural complexity using a Caribbean-wide dataset of temporally replicated estimates spanning four decades.
Both coral cover and architectural complexity have declined rapidly over time, with little evidence of a time-lag.
However, annual rates of change in coral cover and complexity do not covary, and levels of complexity vary greatly
among reefs with similar coral cover. These findings suggest that the stressors influencing Caribbean reefs are
sufficiently severe and widespread to produce similar regional-scale declines in coral cover and reef complexity, even
though reef architectural complexity is not a direct function of coral cover at local scales. Given that architectural
complexity is not a simple function of coral cover, it is important that conservation monitoring and restoration give
due consideration to both architecture and coral cover. This will help ensure that the ecosystem services supported by
architectural complexity, such as nutrient recycling, dissipation of wave energy, fish production and diversity, are
maintained and enhanced.
Keywords: climate change, ecosystem services, foundation species, habitat loss, reef degradation
Received 1 October 2010; revised version received 2 December 2010 and accepted 4 December 2010
Introduction
In some ecosystems, complex structural or functional
attributes can be provided by a single taxon. For exam-
ple, particular tree species within forests or kelp in
temperate seas can provide shelter and living space
for a wide variety of other species (Jones et al., 1997). In
addition, foundation species can underpin fundamental
ecosystem processes such as productivity and nutrient
cycling (Bruno & Bertness, 2001; Ellison et al., 2005). In
tropical shallow waters, hard corals increase the archi-
tectural heterogeneity of the seascape considerably,
providing suitable habitat and microclimatic conditions
for a myriad of species and contributing substantially to
ecosystem dynamics (Hatcher, 1997). Loss of hard corals
on reefs is therefore likely to have severe repercussions
for associated biodiversity, ecosystem structure, func-
tion and stability.
Hard corals are increasingly threatened worldwide
by direct and indirect effects of human activities (Pan-
dolfi et al., 2003; Carpenter et al., 2008; Mora, 2008). As
result, live coral cover has decreased rapidly on tropical
reefs throughout the world (Gardner et al., 2003; Bruno
& Selig, 2007). However, many of the services that
coral reefs provide to humans and other species are
mediated not by the cover of live hard corals but by
the three-dimensional architectural complexity of the
underlying reef structure. For instance, reef complexity
is strongly related to the availability of shelter
and refugia, and consequently to fish and invertebrate
richness, abundance and biomass (e.g. Idjadi &
Edmunds, 2006; Wilson et al., 2007). Reef architecture
also plays a key role in providing additional important
environmental services to humans such as coastal pro-
tection. Wave energy transmitted over reefs is signifi-
cantly dissipated by the friction exerted by bottom
roughness (Lugo-Fernandez et al., 1998; Sheppard
et al., 2005).
Correspondence: Lorenzo Alvarez-Filip, Centre for Ecology,
Evolution, and Conservation, School of Biological Sciences,
University of East Anglia, Norwich NR4 7TJ, UK, tel. 144 1603
591346, fax 144 1603 593901, e-mail: lalvarez@sfu.ca
Global Change Biology (2011) 17, 2470–2477, doi: 10.1111/j.1365-2486.2010.02385.x
2470 r2011 Blackwell Publishing Ltd
Reductions in the architectural complexity of Carib-
bean reefs have also occurred in recent decades (Alvarez-
Filip et al., 2009), but the extent to which the declines in
coral cover and reef complexity are directly linked will
depend on the nature of the any relationship between
coral cover and complexity, and extent of any time lag
between the coral mortality and the subsequent erosion
of coral skeletons. At large scales, direct relationships
between changing coral cover and reef architecture
have been suggested based on the aftermath of wide-
spread coral mortality following mass bleaching events
on some Indo-Pacific reefs (Wilson et al., 2006; Pratchett
et al., 2008). ‘Before-and-after’ comparisons of reefs
either side of the 1998 ENSO bleaching event show that
declines in architectural complexity appeared to lag
behind bleaching-induced coral mortality by more than
5 years (Graham et al., 2007, 2008). In contrast, there is
little evidence in the Caribbean of a region-wide lag in
loss of reef complexity following declines in coral cover,
but turning points in the regional trajectory of declining
architectural complexity coincide closely with the loss
of structurally complex Acropora corals in the late 1980s
and with bleaching-induced coral mortality in 1998
(Aronson & Precht, 2006; Alvarez-Filip et al., 2009).
Consequently, fundamental questions regarding the
exact nature of the relationship between coral cover
and reef architecture, including the generality of a
5-year lagged response, remain unanswered.
Here we use a Caribbean-wide dataset of temporally
replicated coral cover and reef architecture estimates
that spans four decades to explore the regional covar-
iance in coral cover and reef architectural complexity.
First, we test whether the change in architectural com-
plexity lags behind the change in live coral cover.
Second, we then explore the relationships between the
annual rates of change in coral cover and architectural
complexity. Third, we examine the variation among
individual sites in live coral cover and reef architecture
across the region.
Material and methods
Data collation
We collated all available site-specific data on the percentage
cover of live hard coral and associated architectural complex-
ity for reefs within the wider Caribbean Basin. We focused on
studies that used the rugosity index to describe reef architec-
ture, as this is the most commonly used method for measuring
reef complexity in the region (Alvarez-Filip et al., 2009). The
rugosity index of reef complexity is a descriptor of small-scale
reef relief and thus the patterns reported here relate primarily
to fine-scale architectural complexity. However, the positive
correlations between fine-scale reef rugosity and reefscape-
scale visual and remote sensing estimates of complexity re-
ported in some studies (Kuffner et al., 2007; Wilson et al., 2007;
Alvarez-Filip, 2010) suggest that larger-scale patterns of com-
plexity may, to some degree, follow the patterns depicted at
finer scales. Reef rugosity is less frequently measured than
coral cover, thus we first searched for studies reporting rug-
osity of specific sites, and then from this dataset we selected all
studies that also reported information on coral cover.
The database was compiled by searching online ISI Web of
Science, Google Scholar and other relevant databases (e.g.
Reefbase) for peer-reviewed and grey literature. We searched
for pertinent papers in all issues of the journals Coral Reefs,
Bulletin of Marine Sciences,Atoll Research Bulletin,Caribbean
Journal of Science and in all Proceedings of the International Coral
Reef Symposium. Additionally, we directly contacted scientists
and site managers asking for any available data pertaining to
their study sites. The search resulted in a total of 81 studies
that includes 312 records from 139 reefs surveyed between
1977 and 2008 across the Caribbean. From this larger database,
we identified 24 studies with repeated measures (i.e. data
collected over more than 1 year; Fig. S1). This subset included
96 records reporting information for 37 reef sites between 1978
and 2004 (Table S1), and ranging in duration from 2 to 12 years
(mean 55.01 3.41 SD years).
Time-lags in the loss of reef architecture
To test for a delayed response in architectural complexity to
changing coral cover, we first used all available studies (re-
peated and unrepeated) to calculate regional annual averaged
estimates of live coral cover and reef rugosity. We then
calculated the coefficient of correlation between annual aver-
age coral cover and architectural complexity for lags of up to
15 years. This technique provides a broad overview of the
temporal correlations between coral cover and architectural
complexity. However, each iteration includes data from differ-
ent sites and thus spatial variation in either coral cover or
architectural complexity could reduce the strength of the
correlation and the power to detect lags.
To address this potential problem we used two methods to
test for lagged effects using only repeated-measures studies, in
which both coral cover and architectural complexity measures
were available for more than 1 year. First, the coefficients of
correlation between coral cover and architectural complexity
were calculated for lags of up to 4 years for each separate time
series (only for those eight time series with at least 4 years of
overlapping data) and the overall mean correlation coefficient
for each lag was calculated (Fig. S2). Second, we calculated
lagged-correlations between the matrices of coral cover and
architectural complexity of all sites with coral cover and
architectural complexity estimates for each specific lag
(equivalent to a ‘lagged’ Mantel test). This method allowed
us to explore lags of up to 6 years across at least 12 sites. Our
finding was robust to the choice of method (see Figs 1d and
S2), so we present only the findings from the analysis of all 37
time series, as they provide much greater spatial and temporal
representation.
CORAL COVER DOES NOT PREDICT REEF COMPLEXITY 2471
r2011 Blackwell Publishing Ltd, Global Change Biology,17, 2470–2477
In all analyses, the largest significant coefficient of correla-
tion is considered as the best estimate of the number of years
between the death of coral and a decline in architectural
complexity. False discovery rates were used to correct for
multiple tests (Benjamini & Hochberg, 1995).
Annual rates of change in coral cover and rugosity
To investigate the nature of the relationship between coral
cover and reef architecture, we examined how annual region-
wide changes in coral cover were related to the corresponding
1.0
1.5
2.0
2.5
3.0
3.5
4.0
01020304050
Mean rugosity
Mean coral cover
1.0
1.5
2.0
2.5
3.0
3.5
4.0
0 102030405060
Site rugosity
Site coral cover
(e) (f)
Repeated and unrepeated measures Time series (repeated measures)
1
2
3
4
0
15
30
45
60
1977 1982 1987 1992 1997 2002
Reef rugosity
Coral cover
Coral
Rugosity
1.0
1.5
2.0
2.5
3.0
3.5
0
15
30
45
60
1977 1987 1997 2007
Mean rugosity (SE)
Mean coral cover
Coral
Rugosity
(a) (b)
1
–0.2
0.0
0.2
0.4
0.6
0.8
0 23456789101112131415
Pearson’s correlation
Lagged years
0
25
50
75
100
125
150
–0.2
–0.1
0.0
0.1
0.2
0.3
0123456
Number of records
Pearson’s correlation
Lagged years
(c) (d)
Fig. 1 Temporal relationships between coral cover and reef rugosity for Caribbean reef sites. Left panels are constructed with regional
annual averaged (SE) estimates of live coral cover and reef rugosity, showing (a) changes in mean coral cover and reef rugosity across
the Caribbean from 1977 to 2008; (c) changes in Pearson correlation coefficients for lagged relationships between regional average live
coral cover and reef rugosity for the same time period; and (e) the strongest correlation for lagged iterations (i.e. no-lag). Right panels are
constructed with temporally replicated series, and show (b) live coral cover and reef rugosity for each of the 37 time series, (d) changes in
Pearson’s correlation coefficients for lagged relationships between live coral cover and reef rugosity on the same sites, in which each time
series was lagged by one additional year in each iteration; and (f) the strongest correlation for those lagged iterations (i.e. no-lag). (b, e)
Significant correlations, corrected for false discovery rates, are indicated with filled circles. The grey triangles in (b) indicate the number
of records used in each iteration. (c, f) The decade that each estimate represents is indicated (black: 1970s, dark grey: 1980s, light grey:
1990s, white: 2000s).
2472 L. ALVAREZ-FILIP et al.
r2011 Blackwell Publishing Ltd, Global Change Biology,17, 2470–2477
annual change in architectural complexity between 1974 and
2004. Similar rates of change (and minimal lag) would indicate
a strong direct relationship between coral cover and architec-
tural complexity, whereas different rates of change would
indicate that the relationship between the two is less direct
and or potentially governed by different causal mechanisms.
We used a weighted meta-analytic approach to estimate an-
nual rates of change in live coral cover and reef architecture in
the temporally replicated studies (Rosenberg et al., 2000). The
standardized effect size was the annual rate of change in coral
cover and reef rugosity for each study, calculated as
annual rate of change ¼ðlog End log StartÞ=d;
where End and Start represent the final and initial coral cover
or reef rugosity of the time series, respectively, and dis the
number of years elapsed between the two estimates. Calculat-
ing annual rates of change between the initial and final
estimates for time series that both increase and decrease over
time (see Fig. 1b) may not fully capture the overall change. We
therefore performed a second analysis to calculate the annual
rate of change from the average of the differences between
each pair of years (note that some time series are not contin-
uous) to produce site-specific estimates of change for coral
cover and reef rugosity. Results from both analyses were very
similar (see Figs 2 and S3), thus we only present findings from
the first analysis, as this methodology has been previously
used in studies of ecological change on coral reefs (Co
ˆte
´et al.,
2005; Paddack et al., 2009) and its properties as an effect size
have been thoroughly investigated (Co
ˆte
´et al., 2005).
We weighted effect sizes using the natural logarithm of the
total transect length surveyed (see also Mosqueira et al., 2000;
Co
ˆte
´et al., 2001; Molloy et al., 2008). Statistically significant
effect sizes were identified by 95% bias-corrected bootstrapped
confidence intervals (CI; generated from 9999 iterations) that
did not encompass zero. The Q
M
statistic was used to test for
differences in rates of change in live coral cover and architec-
tural complexity. A significant Q
M
implies that there are
differences in mean effect sizes among groups, but a nonsigni-
ficant Q
M
does not preclude individual groups showing sig-
nificant effect sizes (i.e. individual CIs that do not overlap
zero). The meta-analyses were conducted in METAWIN Version
2.0 (Rosenberg et al., 2000). Annual rates of change and
confidence intervals are presented back-transformed to per-
centages to facilitate their interpretation.
Spatial relationships between coral cover and rugosity
Finally, we explored the variation in live coral cover and reef
architecture among sites throughout the region using (i) all
available data (i.e. repeated and unrepeated studies) and (ii) a
smaller dataset of unrepeated measures (which avoids includ-
ing more than one estimate per site). As both datasets provide
very similar results (see Figs 3 and S4), we only present
findings for all the available data because of the wider spatial
and temporal representation. Preliminary analysis demon-
strated that the variation in reef rugosity was unequal along
the gradient of coral cover, so we used quantile regression to
describe the relationship between coral cover and architectural
complexity (Koenker & Bassett, 1978). Quantile regression
differs from ordinary least squares regression in that it mini-
mizes the sum of absolute values of residual errors around a
specified quantile of the dependent variable, rather than just
changes in the mean (Cade & Noon, 2003). Exploring the full
range of quantile responses provides a more complete view of
the relationship between variables than those captured by
individual (median) quantile regression functions (Knight &
Ackerly, 2002), hence we estimated the complete series of
quantile regression functions from the 1st to the 99th quantile
for the regional relationship between coral cover and reef
architecture. Analyses were carried out in Rand using the
QUANTREG package (R, 2009).
Results
What is the time-lag between coral cover loss and reef
architecture decline?
The annual mean estimates of coral cover and reef
rugosity from sites throughout the Caribbean and the
trajectories of individual time series both indicate de-
clines in coral cover and architectural complexity over
the last 30 years (Fig. 1a and b). However, there was
little evidence for a time-lag of more than 2 years
between the onset of a change in coral cover and a
subsequent change in architectural complexity, either
for sites throughout the region or for individual time
series, (Figs 1c and d and S2). In all analyses, the
strongest correlation between architectural complexity
and coral cover was found when the data were un-
lagged (Figs 1e and f and S2). After correction for
multiple tests, only the relationships with no lag and
the 2-year lag of all sites were statistically significant
(a50.05, Tables S2 and S3). However, two relatively
–60
–40
–20
0
20
–40 –20 0 20
Rugosity change per year (%)
Coral cover chan
g
e
p
er
y
ear (%)
Fig. 2 The absence of a relationship between annual rates of
change in live coral cover and architectural complexity on 37 reef
sites across the Caribbean region (R
2
50.07, P50.12).
CORAL COVER DOES NOT PREDICT REEF COMPLEXITY 2473
r2011 Blackwell Publishing Ltd, Global Change Biology,17, 2470–2477
high correlation values also occur in the regional ana-
lysis at 9 and 15 years (Fig. 1c), both of which are
significant or marginally significant before the correc-
tion factors are applied (Table S2).
Are the rates of change in coral cover and reef architecture
similar?
There is no consistent relationship between the annual
rates of change in coral cover and annual changes in
reef rugosity across the 37 sites with repeated measures
(Fig. 2). The overall meta-analysis showed that both live
coral cover and reef architecture have declined signifi-
cantly (i.e. the confidence intervals do not encompass
zero) but at different overall rates (Q
M
53.68,
P50.054). The annual rate of change in coral cover
has been 8.6% (95% CI 511.9% to 5.2%) whereas
the annual change in rugosity has been 4.0% (95%
CI 57.8% to 1.3%).
Is reef architecture a function of coral cover?
Across 31 years and 139 reef sites throughout the
Caribbean, architectural complexity varies greatly
among sites with similar levels of coral cover (Fig. 3a).
For example, the rugosity indices of reefs with 10%
coral cover varies from 1.05 (i.e. relatively flat) to 2.0
(i.e. moderately complex reefs) whereas, at 40% coral
cover, rugosity ranges more widely from 1.05 to 3.5 (i.e.
highly complex reefs; Fig. 3a). Quantile regression ana-
lyses indicated consistently positive relationships be-
tween coral cover and reef architecture, with steeper
relationships for higher quantiles of architectural com-
plexity (Fig. 3b). In the lower quantiles, rugosity is low
across a wide range of coral cover estimates, indicating
that reefs with relatively high coral cover may still be
quite flat (Fig. 3b). By contrast, the steepest coral cover–
architectural complexity relationships are associated
with high rugosity even at the lowest levels of coral
cover (Fig. 3b and c), with slopes of the highest quan-
tiles lying far above the mean and 90% CI of the overall
relationship (Fig. 3a and b). Thus, some reefs with lower
coral cover may still have some level of architectural
complexity.
Discussion
The region-wide decline in coral cover represents both
an absolute loss and a reduction in the quality of reef
habitat, but the implications of coral loss for reef archi-
tecture will depend on whether coral skeletons erode
rapidly following coral mortality. Our findings indicate
that, across the Caribbean, reductions in coral cover
have been rapidly followed by the loss of architectural
0.0 0.2 0.4 0.6 0.8 1.0
2.0
Intercept
1.8
1.6
1.4
1.2
1.0
0.0 0.2 0.4 0.6 0.8 1.0
0.04
Slope
0.03
0.02
0.01
0.00
Quantile
4.0
Rugosity
3.0
2.0
1.0
0.0 20 40 60
Coral cover
(a)
(b)
(c)
Fig. 3 (a) Relationship between coral cover and reef rugosity on
139 reef sites (323 surveys from 1977 to 2008) throughout the
Caribbean. The decade in which each study was conducted is
indicated (circles 51970s, triangles 51980s, squares 51990s,
diamonds 52000s). Nine estimated quantile regression lines
(0.01, 0.05, 0.1, 0.25, 0.5, 0.75, 0.9, 0.95, 0.99 quantile) are super-
imposed on the scatterplot; the median (0.5 quantile) is indicated
with a black dashed line and the others are indicated with grey
dotted lines. The least square estimate of the mean function is
indicated by the black solid line (R
2
50.11, Slope 50.009,
Po0.001). The (b) slopes and (c) intercepts of the quantile
regressions are shown from the 0.01 quantile to the 0.99 quantile,
with 90% confidence bands (grey shading), and the mean (solid
line) 90% confidence intervals (dashed lines) from the ordin-
ary least squares regression.
2474 L. ALVAREZ-FILIP et al.
r2011 Blackwell Publishing Ltd, Global Change Biology,17, 2470–2477
complexity, despite variation among sites in the rela-
tionship between coral cover and architectural com-
plexity and in their rates of change. This suggests that
the scale, frequency and magnitude of stressors on
Caribbean reefs have been sufficiently severe to cause
declines in architectural complexity and live coral cover
that are apparent at regional scales, despite small-scale
variation in reef community composition and structure.
The trajectories of coral cover and architectural com-
plexity are likely to be influenced by the scale, fre-
quency and intensity of disturbances to reefs, and the
degree to which drivers of change in coral cover and
complexity co-occur in space and time. Events such as
coral disease and bleaching can produce widespread
coral mortality without immediately modifying the reef
framework (beyond halting carbonate accretion; Glynn,
1997; Aronson & Precht, 2001; Sheppard et al., 2002). By
contrast, hurricanes and persistent direct human im-
pacts can both kill coral and degrade underlying reef
structures (Hughes, 1994; Hughes & Connell, 1999;
Gardner et al., 2005). In the Caribbean, drivers of both
coral mortality and erosion have operated as virtually
chronic pressures throughout the entire region in recent
decades (Pandolfi et al., 2003; Gardner et al., 2005;
Aronson & Precht, 2006; Mora, 2008). The similar timing
and scale of the region-wide declines in coral cover and
complexity are thus likely to be a consequence of these
chronic pressures.
The absence of time-lags reported here may also be
influenced by declines in coral cover and/or reef com-
plexity that began some time before the onset of the
time series. Declines in coral cover before the first year
of our time series could influence subsequent declines
in reef complexity during the first years of the time
series, in which case longer periods of time may be
needed to detect evidence of lags in reef erosion follow-
ing coral mortality. The suggestion of high region-wide
correlations between coral cover and architectural com-
plexity after 9–15 years (Fig. 1c), suggests that there may
be some evidence for lags in the erosion of architectural
complexity on time-scales that may be expected for
some biological disturbances, such as bleaching or
diseases (Pratchett et al., 2008). As these time series
accumulate, so too will the power to evaluate the
significance of these longer lags.
The rate of decline in Caribbean reef complexity over
the last three decades may be influenced by the relative
abundance of different coral morpho-functional types,
which can vary in both structure and relative suscept-
ibility to erosion following mortality. In the Caribbean,
the largest changes in coral cover occurred as a result of
the disease-induced die-off of acroporids in the late
1970s and early 1980s (Aronson & Precht, 2006; Schutte
et al., 2010). Before this the erect branching structures of
Acropora corals contributed disproportionately to reef
complexity. Although the robust skeletons of A. palmata
may have persisted longer than the fragile framework
of A. cervicornis in some locations, the regional trends of
declining coral cover and architectural complexity sug-
gest that most dead Acropora were relatively rapidly
broken down and eroded following mortality (Aronson
& Precht, 2006; Alvarez-Filip et al., 2009). In our study,
similarly rapid annual rates of change of both coral
cover and architectural complexity were apparent dur-
ing the Acropora die-off period (1978–1985; coral cover
523.19, bias-corrected 95% CI 52.73 to 1.49;
rugosity 527.03, bias-corrected 95% CI 514.64 to
9.57), suggesting a rapid response of architectural
complexity to coral cover loss during this period. How-
ever, only three studies are available for these years,
and therefore the temporal pattern of decline in reef
complexity reported here refers primarily to the years
since the demise of Acropora, in which Caribbean reefs
have been dominated by a combination of massive and
weedy corals.
The weak relationship between rates of change in
coral cover and architectural complexity (Fig. 2) sug-
gests that additional factors have influenced the re-
sponse of reef complexity to the loss of coral cover.
Caribbean coral communities have changed continu-
ously since the mass mortality of acroporids. Important
reef-building corals such as Montastrea have been de-
clining throughout the region and stress-resistant coral
species that contribute relatively little to the reef frame-
work and architectural complexity have increasingly
dominated Caribbean reefs (e.g. Agaricia and Porites;
Hughes, 1994; Edmunds & Carpenter, 2001; Aronson
et al., 2002; Green et al., 2008). Thus changes in coral
composition leading to ‘flatter’ reef communities, to-
gether with possible changes in carbonate budgets as a
consequence of higher amounts of bare substrata (Ea-
kin, 1996; Glynn, 1997), can occur in the absence of
declines in coral cover. Previous studies do indeed
suggest that rates of loss of Caribbean reef architecture
have remained high in recent years (Alvarez-Filip et al.,
2009), while coral loss has almost ceased (Schutte et al.,
2010).
For any given level of coral cover, reef rugosity can
vary markedly (Fig. 3). Although variation in other reef
organisms, such as sponges and soft corals, may con-
tribute to local reef rugosity (e.g. Diaz & Rutzler, 2001;
Halford et al., 2004), it is likely that much of the varia-
tion in architectural complexity reflects different habitat
types and variation in coral species assemblages (Chit-
taro, 2004; Alvarez-Filip, 2010). Coral identity may
therefore be an important mediator of reef complexity
and, consequently, the impact of coral loss on reef
architecture will differ among sites, with sites domi-
CORAL COVER DOES NOT PREDICT REEF COMPLEXITY 2475
r2011 Blackwell Publishing Ltd, Global Change Biology,17, 2470–2477
nated by architecturally complex reef-building corals
bearing the greatest losses in rugosity following coral
mortality. Unfortunately, coral species composition and
reef type were seldom reported in the studies included
here, hence we could not explore their influence on
patterns of change in architectural complexity.
The loss of architectural complexity following de-
clines in coral cover in the Caribbean differs from the
pattern reported in the Indo-Pacific region, where a
lagged response in the aftermath of widespread coral
mortality following mass bleaching events was appar-
ent (Wilson et al., 2006; Graham et al., 2007, 2008;
Pratchett et al., 2008). However, these studies encom-
passed different temporal scales; our Caribbean ana-
lyses explore year-by-year changes throughout a
multidecadal period of continual coral and reef archi-
tecture loss, whereas the Indo-Pacific studies spanned
either side of a discrete catastrophic coral mortality
event. In addition, there are important historical and
ecological differences between these two regions that
are likely to influence these processes, with Caribbean
reefs typically having fewer coral species, less ecologi-
cal redundancy and frequent hurricane impacts (Bell-
wood et al., 2004; Briggs, 2005). To determine whether
our findings can be broadly generalized would require
similar longitudinal and spatial studies of Indo-Pacific
reefs.
Architectural complexity is clearly not a simple
function of coral cover. Therefore, to restore the eco-
system services that Caribbean reefs provide to other
species, including humans, these two critical reef attri-
butes may need to be considered separately and at
different spatial scales. Much of coral reef conservation
at present focuses on ecological management and
control of the cover of coral and algae (Gardner et al.,
2003; Co
ˆte
´et al., 2006; Bruno & Selig, 2007; Mumby
et al., 2007; Bruno et al., 2009). However, restoring coral
cover on reefs may not necessarily provide the archi-
tectural complexity that underpins important coral
reef ecosystem services relating to nutrient recycling,
dissipation of wave energy and fish production
(Szmant, 1997; Lugo-Fernandez et al., 1998; Sheppard
et al., 2005). Understanding the range of the biotic and
abiotic drivers of architectural complexity may there-
fore provide an effective means of targeting reef man-
agement in relation to the services provided by
different reef structures.
Acknowledgements
We are grateful to Peter Edmunds, Renata Goodridge and Hazel
Oxenford (CARICOMP Barbados) and Francisco Geraldes (Cen-
tro de Investigaciones de Biologı
´a Marina de la Universidad
Auto
´noma de Santo Domingo and CARICOMP) for contributing
unpublished data. The manuscript was further improved by the
comments of five anonymous reviewers. This research was
founded by the Mexican scholarships from the CONACYT
(171864) and S. E. P. to L. A-F. I. M. C. and N. K. D. are supported
by Natural Science and Engineering Research Council of Canada
Discovery grants.
References
Alvarez-Filip L (2010) Habitat complexity in coral reefs: patterns of degradation and
consequences for biodiversity. PhD thesis, University of East Anglia, Norwich,
124 pp.
Alvarez-Filip L, Dulvy NK, Gill JA, Co
ˆte
´IM, Watkinson AR (2009) Flattening of
Caribbean coral reefs: region-wide declines in architectural complexity. Proceedings
of the Royal Society B: Biological Sciences,276, 3019–3025.
Aronson RB, Macintyre IG, Precht WF, Murdoch TJT, Wapnick CM (2002) The
expanding scale of species turnover events on coral reefs in Belize. Ecological
Monographs,72, 233–249.
Aronson RB, Precht WF (2001) White-band disease and the changing face of Caribbean
coral reefs. Hydrobiologia,460, 25–38.
Aronson RB, Precht WF (2006) Conservation, precaution, and Caribbean reefs. Coral
Reefs,25, 441–450.
Bellwood DR, Hughes TP, Folke C, Nystrom M (2004) Confronting the coral reef crisis.
Nature,429, 827–833.
Benjamini Y, Hochberg Y (1995) Controlling the false discovery rate – a practical and
powerful approach to multiple testing. Journal of the Royal Statistical Society Series B
– Methodological,57, 289–300.
Briggs JC (2005) Coral reefs: conserving the evolutionary sources. Biological Conserva-
tion,126, 297–305.
Bruno JF, Bertness MD (2001) Habitat modification and facilitation in benthic marine
communities. In Marine Community Ecology (eds Bertness MD, Gaines SD, Hay ME),
pp. 201–218. Sinauer, Sunderland, MA.
Bruno JF, Selig EZ (2007) Regional decline of coral cover in the Indo-Pacific: timing,
extent, and subregional comparisons. PLoS One,2, e711.
Bruno JF, Sweatman H, Precht WF, Selig ER, Schutte VGW (2009) Assessing evidence
of phase shifts from coral to macroalgal dominance on coral reefs. Ecology,90,
1478–1484.
Cade BS, Noon BR (2003) A gentle introduction to quantile regression for ecologists.
Frontiers in Ecology and the Environment,1, 412–420.
Carpenter KE, Abrar M, Aeby G et al. (2008) One-third of Reef-building Corals face
elevated extinction risk from climate change and local impacts. Science,321,
560–563.
Chittaro PM (2004) Fish–habitat associations across multiple spatial scales. Coral Reefs,
23, 235–244.
Co
ˆte
´IM, Gardner TA, Gill JA, Hutchinso DJ, Watkinson AR (2006) New approaches
to estimating recent ecological changes on coral reefs. In Coral Reef Conservation
(eds Co
ˆte
´IM, Reynolds DJ), pp. 293–313. Cambridge University Press, Cambridge,
UK.
Co
ˆte
´IM, Gill JA, Gardner TA, WatkinsonAR (2005) Measuring coral reef decline meta-
analyses. Philosophical Transactions of the Royal Society,360, 385–395.
Co
ˆte
´IM, Mosqueira I, Reynolds JD (2001) Effects of marine reserve characteristics
on the protection of fish populations: a meta-analysis. Journal of Fish Biology,59,
178–189.
Diaz MC, Rutzler K (2001) Sponge: an essential component of Caribbean coral reefs.
Bulletin of Marine Science,69, 535–546.
Eakin CM (1996) Where have all the carbonates gone? A model comparison of calcium
carbonate budgets before and after the 1982–1983 El Nino at Uva Island in the
eastern Pacific. Coral Reefs,15, 109–119.
Edmunds PJ, Carpenter RC (2001) Recovery of Diadema antillarum reduces macroalgal
cover and increases abundance of juvenile corals on a Caribbean reef. Proceedings of
the National Academy of Sciences,98, 5067–5071.
Ellison AM, Bank MS, Clinton BD et al. (2005) Loss of foundation species: conse-
quences for the structure and dynamics of forested ecosystems. Frontiers in Ecology
and the Environment,3, 479–486.
Gardner TA, Co
ˆte
´IM, Gill JA, Grant A, Watkinson AR (2003) Long-term region-wide
declines in Caribbean corals. Science,301, 958–960.
Gardner TA, Co
ˆte
´IM, Gill JA, Grant A, Watkinson AR (2005) Hurricanes and
Caribbean coral reefs: impacts, recovery patterns, and role in long-term decline.
Ecology,85, 174–184.
2476 L. ALVAREZ-FILIP et al.
r2011 Blackwell Publishing Ltd, Global Change Biology,17, 2470–2477
Glynn PW (1997) Bioerosion and coral reef growth: a dynamic balance. In Life and
Death of Coral Reefs (ed Birkeland CE), pp. 68–95. Chapman & Hall, New York.
Graham NAJ, Mcclanahan TR, Macneil MA et al. (2008) Climate warming, marine
protected areas and the ocean-scale integrity of coral reef ecosystems. PLoS One,3,
e3039.
Graham NAJ, Wilson SK, Jennings S, Polunin NVC, Robinson J, Bijoux JP, Daw TM
(2007) Lag effects in the impacts of mass coral bleaching on coral reef fish, fisheries,
and ecosystems. Conservation Biology,21, 1291–1300.
Green DH, Edmunds PJ, Carpenter RC (2008) Increasing relative abundance of Porites
astreoides on Caribbean reefs mediated by an overall decline in coral cover. Marine
Ecology Progress Series,359, 1–10.
Halford A, Cheal AJ, Ryan D, Williams DM (2004) Resilience to large-scale disturbance
in coral and fish assemblages on the Great Barrier Reef. Ecology,85, 1892–1905.
Hatcher BG (1997) Coral reef ecosystems: how much greater is the whole than the sum
of the parts? Coral Reefs,16, S77–S91.
Hughes TP (1994) Catastrophes, phase shifts, and large-scale degradation of a
Caribbean coral reef. Science,256, 1547–1551.
Hughes TP, Connell JH (1999) Multiple stressors on coral reefs: a long-term perspec-
tive. Limnology and Oceanography,44, 932–940.
Idjadi JA, Edmunds PJ (2006) Scleractinian corals as facilitators for other invertebrates
on a Caribbean reef. Marine Ecology Progress Series,319, 117–127.
Jones CG, Lawton JH, Shachak M (1997) Positive and negative effects of organisms as
physical ecosystem engineers. Ecology,78, 1946–1957.
Knight CA, Ackerly DD (2002) Variation in nuclear DNA content across environmental
gradients: a quantile regression analysis. Ecology Letters,5, 66–76.
Koenker R, Bassett G (1978) Regression quantiles. Econometrica,46, 33–50.
Kuffner IB, Brock JC, Grober-Dunsmore R, Bonito VE, Hickey TD, Wright CW (2007)
Relationships between reef fish communities and remotely sensed rugosity mea-
surements in Biscayne National Park, Florida, USA. Environmental Biology of Fishes,
78, 71–82.
Lugo-Fernandez A, Roberts HH, Suhayda JN (1998) Wave transformations across a
Caribbean fringing-barrier Coral Reef. Continental Shelf Research,18, 1099–1124.
Molloy PP, Reynolds JD, Gage MJG, Mosqueirac L, Co
ˆte
´IM (2008) Links between sex
change and fish densities in marine protected areas. Biological Conservation,141,
187–197.
Mora C (2008) A clear human footprint in the coral reefs of the Caribbean. Proceedings
of the Royal Society B: Biological Sciences,275, 767–773.
Mosqueira I, Co
ˆte
´IM, Jennings S, Reynolds JD (2000) Conservation benefits of marine
reserves for fish populations. Animal Conservation,3, 321–332.
Mumby PJ, Hastings A, Edwards HJ (2007) Thresholds and the resilience of Caribbean
coral reefs. Nature,450, 98–101.
Paddack MJ, Reynolds JD, Aguilar C et al. (2009) Recent region-wide declines in
Caribbean Reef fish abundance. Current Biology,19, 590–595.
Pandolfi JM, Bradbury RH, Sala E et al. (2003) Global trajectories of the long-term
decline of Coral Reef ecosystems. Science,301, 955–958.
Pratchett MS, Munday PL, Wilson SK et al. (2008) Effects of climate-induced coral
bleaching on coral-reef-fishes ecological and economic consequences. Oceanography
and Marine Biology: An Annual Review,46, 251–296.
R (2009) R: A Language and Environment for Statistical Computing. R Foundation for
Statistical Computing, Vienna, Austria.
Rosenberg MS, Adams DC, Gurevitch J (2000) MetaWin: Statistical Software for Meta-
Analysis, Version 2.0. Sinauer, Sunderland, MA.
Schutte VGW, Selig ER, Bruno JF (2010) Regional spatio-temporal trends in Caribbean
coral reef benthic communities. Marine Ecology Progress Series,402, 115–122.
Sheppard C, Dixon DJ, Gourlay M, Sheppard A, Payet R (2005) Coral mortality
increases wave energy reaching shores protected by reef flats: examples from the
Seychelles. Estuarine Coastal and Shelf Science,64, 223–234.
Sheppard CRC, 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.
Szmant AM (1997) Nutrient effects on coral reefs: a hypothesis on the importance of
topographic and trophic complexity to reef nutrient dynamics. Proceedings of the 8th
International Coral Reef Symposium,2, 1527–1532.
Wilson SK, Graham NAJ, Polunin NVC (2007) Appraisal of visual assessments of habitat
complexity and benthic composition on coral reefs. Marine Biology,151, 1069–1076.
Wilson SK, Graham NAJ, Pratchett MS, Jones GP, Polunin NVC (2006) Multiple
disturbances and the global degradation of coral reefs: are reef fishes at risk or
resilient? Global Change Biology,12, 2220–2234.
Supporting Information
Additional Supporting Information may be found in the
online version of this article:
Figure S1. Regional distribution of reef sites with both live
coral cover and rugosity data replicated over more than one
year.
Figure S2. Pearson correlation coefficients for lagged re-
lationships between live coral cover and reef rugosity on
eight reef sites (grey circles). Black triangles represent the
average across all time series.
Figure S3. Relationship between the averages ( SE) of the
annual rates of change in live coral cover and architectural
complexity on 37 reef sites across the Caribbean region
(R
2
50.09, P50.08). At each site, rates of change were cal-
culated between each pair of years and then averaged to
produce a site estimate of change for coral cover and reef
rugusity.
Figure S4. (a) Relationship between coral cover and reef
rugosity on 140 unreplicated reef surveys through the Car-
ibbean from 1977 to 2008. The decade in which each study
was conducted is indicated (circles 51970s, triang-
les 51980s, squares 51990s, diamonds 52000s). Nine esti-
mated quantile regression lines (0.01, 0.05, 0.1, 0.25, 0.5, 0.75,
0.9, 0.95, 0.99 quantile) are superimposed on the scatterplot;
the median (0.5 quantile) is indicated with a black dashed
line and the others are indicated with grey dotted lines. The
least square estimate of the mean function is indicated by the
black solid line (R
2
50.19, Slope 50.013, Po0.001). The (b)
slopes and (c) intercepts of the quantile regressions are
shown from the 0.01 quantile to the 0.99 quantile, with90%
confidence bands (grey shading), and the mean (solid
line) 90% confidence intervals (dashed lines) from the or-
dinary least squares regression.
Table S1. Supplementary site information of the time series
reporting coral cover and reef rugosity for Caribbean reefs.
Table S2. Unadjusted and corrected (using false discovery
rates) Pvalues of the correlations between the regional
average live coral cover and reef rugosity from 1978 to 2008,
with time-lags ranging from 0 to 15 years.
Table S3. Unadjusted and corrected (using false discovery
rates) Pvalues of regional correlation between site-lagged
live coral cover and reef rugosity, withtime-lags ranging from
0 to 6 years.
Please note: Wiley-Blackwell are not responsible for the
content or functionality of any supporting materials supplied
by the authors. Any queries (other than missing material)
should be directed to the corresponding author for the
article.
CORAL COVER DOES NOT PREDICT REEF COMPLEXITY 2477
r2011 Blackwell Publishing Ltd, Global Change Biology,17, 2470–2477
