Maliao et al 08
Mar Biol (2008) 154:841–853
DOI 10.1007/s00227-008-0977-0
ORIGINAL PAPER
Phase-shift in coral reef communities in the Florida Keys National
Marine Sanctuary (FKNMS), USA
Ronald J. Maliao · Ralph G. Turingan · Junda Lin
Received: 17 September 2007 / Accepted: 7 April 2008 / Published online: 18 April 2008
© Springer-Verlag 2008
Abstract Characterizing the Florida Keys National structuring of coral reef communities at a spatial scale
Marine Sanctuary (FKNMS), USA, has gained much atten- larger than the individual reef. Furthermore, it is conceiv-
tion over the past several decades because of apparent able that these predictor variables exerted inXuence for a
changes in the benthic community structure over space and long time rather than being a recent event. Results also
time representative of patterns occurring in the Caribbean revealed a pattern showing reduction in hard coral cover
region. We used a 5-year dataset (1996–2000) of macroal- and species richness, and subsequent proliferation of mac-
gal and sponge cover and water quality measurements as roalgae and sponges during the study period. Our analyses
predictor variables of hard coral community structure in the of the Florida Keys present a pattern that is consistent with
FKNMS. The 16 water quality variables were summarized the characteristics of a reef that has undergone a “phase-
into 4 groups by principal component analysis (PCA). Hier- shift,” a phenomenon that is widely reported in the Carib-
archical agglomerative cluster analysis of the mean and bean region.
standard deviation (SD) of the principal component scores
of water quality variables separated the reef sites into two
main groups (and Wve sub-groups), referred to as reefs of Introduction
similar inXuence (RSI). The main groups corresponded
with their geographical locations within the Florida Keys: The community structure of coral reefs in the Caribbean,
the reefs in the Upper and Middle Keys being homoge- including the Florida Keys, have remained stable for thou-
neous and collectively, having lower water quality scores sands of years before the region suVered a catastrophic
relative to reefs in the Lower Keys. Canonical correspon- coral mass mortality in the 1980s (Hughes 1994; Aronson
dence analysis (CCA) between hard coral cover and key et al. 1998). Since the 1980s die oV, coral reef deterioration
predictor variables (i.e., water quality, macroalgal cover has been characterized by a reduction in coral cover by as
and sponge cover) also separated the reefs in the Lower much as 40%, a shift in the composition of surviving spe-
Keys from reefs in the Upper–Middle Keys, consistent with cies (Hughes 1994; Greenstein and PandolW 1997; Gardner
results of the cluster analysis, which categorized reefs et al. 2003), and a phase-shift from coral-dominated to
based on RSI. These results suggest that the prevailing macroalgal-dominated communities (Dustan 1977; Dustan
gradient of predictor variables may have inXuenced the and Halas 1987; Porter and Meier 1992; Hughes 1994;
McClanahan and Muthiga 1998; Porter et al. 2002; Gardner
et al. 2003, 2005).
Communicated by R. Cattaneo-Vietti.
The magnitude, scale and cause of the deterioration of
Caribbean coral reefs remain controversial (Hughes 1994;
R. J. Maliao (&) · R. G. Turingan · J. Lin Ginsburg 1994; Murdoch and Aronson 1999; Bellwood
Department of Biological Sciences, et al. 2004). Hypothesized mechanisms working at small
Florida Institute of Technology,
150 West University Blvd.,
spatio-temporal scales (meters to kilometers and days to
Melbourne, FL 32901-6975, USA years) include hurricanes (Gardner et al. 2003), point-
e-mail: rmaliao@Wt.edu source nutrient loading (Ginsburg and Shinn 1994; Leichter
123
842 Mar Biol (2008) 154:841–853
et al. 2003), impacts of upstream land use patterns 40 cm above the reef at a constant swim speed of about
(Bellwood et al. 2004), and macroalgal overgrowth (Hughes 4 m/min yielding approximately 9,000 video frames per
1994). Other hypotheses suggest the importance of meso- transect. Image analysis used a custom software application
to large-scale processes (tens to thousands of kilometers PointCount for coral reefs, developed speciWcally for the
and decades to millennia), such as sea urchin mass mortal- CREMP (Wheaton et al. 2001). Percent cover of hard cor-
ity events (Carpenter 1990), larval transport (Galindo et al. als, macroalgae (Xeshy and Wlamentous and non-coralline)
2006), herbivore reduction due to overWshing (Jackson and sponges was calculated from the images. Water quality
et al. 2001; PandolW et al. 2003), disease outbreaks (Aron- monitoring under WQMP was undertaken quarterly from
son and Precht 2001; Porter et al. 2001) and climate change 154 stations. Analyses only included those WQPP stations
(Walther et al. 2002; Hughes et al. 2003). It is unclear that overlapped with reef sites monitored by CREMP. Six-
whether the current state of the Caribbean coral reefs repre- teen water quality measurements were included in our anal-
sents a permanent phase-shift or that the community yses: temperature (°C), salinity (PSU), dissolved oxygen
structure would rebound (Petraitis and Dudgeon 2004). (DO, mg l¡1), and light attenuation (Kd, m¡1) were col-
Long-term monitoring of coral reefs in the Florida Keys lected in the Weld; organic carbon (TOC), total nitrogen
(1981–1991) and US Virgin Islands (1989–2003) have (TN), total phosphorus (TP), silicate (Si (OH)4), nitrite
indicated no evidence of recovery in terms of coral cover or (NO2¡), dissolved nitrate (NO3¡), ammonium (NH4+),
reduction of algal abundance over a decadal time scale soluble reactive phosphate (SRP), chlorophyll a (chl a),
(Porter and Meier 1992; Rogers and Miller 2006). dissolved inorganic nitrogen (DIN), total unWltered
The main goal of this study is to Wnd evidence of a possi- concentrations of organic nitrogen (TON) and turbidity
ble phase-shift in coral reef communities in the Florida (NTU) were determined in the laboratory from the water
Keys National Marine Sanctuary (FKNMS), USA. SpeciW- collected in situ (reported in g l¡1) (see Keller and Dona-
cally, this study is designed to determine the correspon- hue 2006 for the complete sampling protocol).
dence between hard coral cover, macroalgal and sponge
cover and water quality variables using the 1996–2000 data Data analyses
of the FKNMS Water Quality Protection Program (WQPP).
Data analyses were conducted at the level of reef site using
the 1996¡2000 dataset. Average values of percent cover of
Materials and methods benthic biota data (hard corals, macroalgae and sponges)
and water quality measurements were used for those reef
Study site and data collection sites with stations at varying depths. In the multivariate
analyses, benthic biota data and water quality measure-
The reef tract of the Florida Keys, which represents the ments were arcsine square-root and square-root trans-
third largest barrier reef in the world and the only living formed, respectively, and then data matrices were screened
coral reef in North America (Lapointe and Matzie 1996), for outliers. Outliers are extreme data values relative to
was designated as a marine sanctuary in 1990, and became others in a sample and have unduly large eVects on the
known as the FKNMS (Keller and Donahue 2006; Fig. 1). resultant probabilities (McCune and Grace 2002; Quinn and
The reef tract of the FKNMS, referred to here as the Florida Keough 2006). For example, outliers may distort estimates
Keys, is partitioned into three regions based on geographic and P values and inXate sums-of-squares, all of which
and environmental criteria: Upper Keys, Middle Keys, and would result in faulty conclusions. In our analyses, we deW-
Lower Keys (Ginsburg and Shinn 1994). The WQPP was ned outliers as those reefs, coral species or water quality
implemented in 1996 to monitor both benthic biota (under variables with SD (standard deviation) >2 (Quinn and
the Coral Reef Evaluation and Monitoring Project or Keough 2006) and/or those with sample points <3 (i.e., rare
CREMP) and water quality (under the Water Quality Moni- samples) (McCune and Grace 2002). Outliers were
toring Project or WQMP) in the Florida Keys until 2002 excluded from all data analyses.
(Keller and Donahue 2006). The CREMP includes 40 reef Principal component analysis (PCA) was used to extract
sites; benthic biota in two to four permanent stations the underlying patterns of the water quality variables using
(22 £ 2 m transects) were monitored quarterly on each reef a correlation matrix based on the average values of the 5-
site. Field monitoring consists of station species inventories year data. Water quality variables with component loading
(SSI) and video transects conducted in four permanent sta- ¸ §0.5 were retained in the description of each principal
tions (22 £ 2 m transect) in each reef site. SSI consists of component. The mean and SD of component scores for
timed (15 min) counts of stony coral species (Milleporina each reef site were then used as input variables in a hierar-
and Scleractinia) present in each station to provide data on chical agglomerative cluster analysis by unweighed
hard coral species richness. Video recordings were taken pair-group method (UPGMA) linkage rule using Euclidean
123
Mar Biol (2008) 154:841–853 843
Fig. 1 Map of the Florida Keys showing only the 20 reef sites included in the analyses. Reef abbreviations (enclosed): Western Head (Western),
West Washer Women (West)
distance (ED). Groups of reefs produced by the cluster CCA, the set of water quality variables identiWed by the
analysis were linked objectively using the similarity proWle BIOENV routine, and macroalgal and sponge cover were
(SIMPROF) routine (Clarke and Gorley 2006). Reef clus- used as the predictors of hard coral cover. In addition, the
ters were further merged at 0.5 ED to facilitate interpreta- variation in hard coral cover and predictor variables over
tion and were referred to as reefs of similar inXuence (RSI). the 5-year period was examined using a Kruskal–Wallis
The outcome of this analysis reveals the variation in water test. The variation in hard coral, macroalgal and sponge
quality across reefs in the Florida Keys. cover, as well as indices of Shannon diversity and species
Canonical correspondence analysis (CCA) (ter Braak richness of hard corals between 1996 and 2000 were further
1986), the most widely used direct gradient ordination examined using Mann–Whitney U test.
method (Palmer 1993; ter Braak and Verdonschot 1995; The robustness of the result of CCA was determined by
McCune 1997; GraVelman 2001; McCune and Grace Monte–Carlo permutation test (ter Braak 1986). Ordination
2002), was employed to determine the multivariate correla- scores in the CCA biplots were optimized by coral species
tion between hard coral community structure and predictor using the linear combination (LC scores) of predictor vari-
variables. The minimal set of water quality variables ables. LC scores were used because they best represent the
entered in the CCA was identiWed by the BIOENV routine relative contribution of each predictor variable to the varia-
(Clarke and Ainsworth 1993; Clarke and Gorley 2006). The tion in hard coral cover (McCune and Grace 2002) and
water quality variables were screened for multicollinearity have been shown to perform well even in skewed distribu-
prior to running the BIOENV routine and only a representa- tions (Palmer 1993). In the ordination biplot, predictor
tive of highly correlated water quality variables (i.e., those variables are represented as arrows and the length of the
with Spearman’s = ¸ §0.8) were used. Pruning of pre- arrow indicates the relative importance of each predictor
dictor variables to a minimal set that accounts for the larg- variable (ter Braak 1986). The angle between arrows
est variation in the hard coral community structure was indicates the degree of correlation between predictor vari-
necessary because the result of CCA becomes less robust as ables; the location of the reef or species relative to the
the number of predictor variables increases (McCune and arrow indicates site characteristics or species preferences
Grace 2002) and to minimize the problem associated with (Palmer 1993). Since ordination scores were calculated
multicollinearity (ter Braak and Looman 1994). In our using LC scores, the relative contribution of the predictor
123
844 Mar Biol (2008) 154:841–853
variables in each axis was determined from the intraset cor- ond main group (Fig. 2a). The physical property component
relation output (McCune and Grace 2002). A predictor var- (PC4: salinity, temperature and DO) and the algal bloom
iable with correlation coeYcient of ¸ §0.4 is considered component (PC2: chl a, TP, and turbidity) appeared to be
ecologically relevant (Rakocinski et al. 1996). The main the main components driving Florida Keys-wide variation
advantage of CCA is that it allowed simultaneous explora- in water quality, with higher values in the Lower Keys and
tion of relationships among diVerent reef sites or coral lower values in the Middle and Upper Keys (Fig. 2b).
species with multiple predictor variables. Results of running the BIOENV routine revealed that
Cluster analysis and SIMPROF and BIOENV routines TN, chl a and salinity are the principal set of water quality
were conducted using PRIMER version 6.1.9 (Plymouth variables signiWcantly explaining the variation in hard coral
Routines Multivariate Ecological Research, PRIMER-E community structure ( w = 0.393, P < 0.05, global BEST
Ltd, Plymouth, UK; Clarke and Gorley 2006) while the permutation test). TN Xuctuated signiWcantly across the
CCA was conducted using PC-Ord version 5 (MjM Soft- 5-year monitoring period (Kruskal–Wallis test, P < 0.01,
ware, Oregon, USA; McCune and MeVord1999; McCune Fig. 3a); chl a and salinity remained stable (Kruskal–Wallis
and Grace 2002). PCA and univariate tests were conducted test, P > 0.05, Fig. 3b, c).
using SPSS 14 (SPSS Inc., Chicago, USA).
Reef community structure
Results Macroalgal cover signiWcantly increased from 1996 to 2000
(Kruskal–Wallis test, P < 0.001, Fig. 3d), with an almost
Reefs of similar inXuence triple increase from the 1996 (5.7%) to 1998 (16.5%) level.
The macroalgal cover in 1996 had signiWcantly increased
The 16 water quality variables were reduced into four prin- by 68% in 2000 (9.6%) (Mann–Whitney U test, P < 0.05).
cipal components (PC), explaining 82.3% of the total varia- Although sponge cover had varied signiWcantly across
tion in the data set (Table 1). PC 1 to 4 are referred to as years (Kruskal–Wallis test, P < 0.01, Fig. 3e), the diVer-
organic component, algal bloom component, inorganic ence between 1996 and 2000 was not signiWcant (Mann–
nitrogen component and physical properties component, Whitney U test, P > 0.05).
respectively. Cluster analysis of the mean and SD of the Forty-two coral species were identiWed in the Florida Keys.
component scores separated the reefs into two main groups In terms of percent cover, the Montastraea annularis complex
and Wve subgroups; the Upper and Middle Keys reefs form dominated the hard coral species in the Florida Keys (Fig. 4a).
the Wrst main group and the Lower Keys reefs form the sec- Collectively, hard coral cover showed a decreasing trend
Table 1 Result of the PCA with
Water quality parameters Component
Varimax rotation for 16 water
quality variables PC1 PC2 PC3 PC4
Total organic nitrogen (TON) 0.865 0.373 0.163 ¡0.010
Total nitrogen (TN) 0.860 0.380 0.194 ¡0.017
Silicate SI (OH)4 0.819 0.335 0.322 0.141
Light attenuation (Kd) 0.781 ¡0.249 0.044 0.053
Total organic carbon (TOC) 0.752 0.482 0.326 ¡0.149
Nitrite (NO2¡) 0.572 0.017 0.501 ¡0.402
Chlorophyll a (Chl a) 0.051 0.843 0.050 0.439
Total phosphorus (TP) 0.153 0.834 ¡0.045 0.257
Turbidity 0.361 0.725 0.260 ¡0.082
Nitrate (NO3¡) 0.232 0.250 0.887 0.070
Soluble reactive phosphorus (SRP) 0.038 ¡0.176 0.847 0.113
Dissolved inorganic nitrogen (DIN) 0.384 0.276 0.829 ¡0.176
Water quality variables with Temperature 0.208 0.267 0.099 0.873
component loadings ¸ §0.5 Dissolved oxygen (DO) 0.322 ¡0.151 0.047 ¡0.816
(bolded text) were retained to
describe each component (PCA Salinity 0.477 0.258 0.147 0.760
PC1 = organic; PC2 = algal Ammonium (NH4+) 0.320 0.236 0.278 ¡0.475
bloom; PC3 = inorganic Percent variance explained 28.118 18.791 18.141 17.211
nitrogen; PC4 = physical (Total = 82.261)
properties)
123
Mar Biol (2008) 154:841–853 845
Fig. 2 a Dendogram of reef sites based on the cluster analysis of mean Wve subgroups (1–5); b mean of four PCA component scores of water
and SD of PCA component scores of water quality variables. Objective quality variables across RSI; error bars indicate §standard error of
grouping of clusters (dashed line) was based on SIMPROF routine. mean. RSI numbers refer to reef clusters in a
Clusters were further merged at 0.5 ED resemblance, forming a total of
1.4 4.5
a Total nitrogen b Chl a
1.2 4.0
3.5
1.0
3.0
0.8 2.5
mg l-1
mg l-1
0.6 2.0
n=102 140 141 143 143 n=102 140 141 143 143
1.5
0.4
1.0
0.2
0.5
0.0 0.0
0.16 0.14 0.13 0.13 0.11
0.315 0.260 0.289 0.312 0.278
39 80 e 35
c n=102 140 141 143 143 d n=136 in each year Macroalgae n=136 in each year Sponge
38 70 30
Salinity
37 60
25
36 50
20
cent cover
35 40
PSU
15
34 30
10
Per
33 20
32 10 5
36.3 36.3 35.6 36.2 36.3
31 0 0
5.7 4.9 16.5 6.1 9.6 2.2 1.9 1.3 1.4 2.2
30
1996 1997 1998 1999 2000 1996 1997 1998 1999 2000 1996 1997 1998 1999 2000
Year
Fig. 3 Box-plots of predictor variables entered in CCA : a TN, b chl percentiles. Points outside the whiskers are the outliers. Numbers
a, c salinity, d macroalgae, and e sponge. The center line of the box is below the box are the median values. These values were calculated
the median, the bottom and top of the box are the 25th and 75th percen- from the 20 reefs included in the analyses
tiles and the whiskers below and above of the box are the 10th and 90th
across the 5-year monitoring period, with the lowest level Correspondence between predictor variables and hard coral
occurring in 1999 (Kruskal–Wallis test, P < 0.01; Fig. 4b). community structure
Hard coral percent cover had signiWcantly decreased by
almost 50%, from 8.1% in 1996 to 4.6% in 2000 (Mann– Twenty out of the 25 aggregated reef sites and 35 out of the
Whitney U test, P < 0.05). DiVerences in species richness and 42 coral species were included in the CCA after data
Shannon diversity index over the two periods were not signiW- screening. The CCA revealed two gradients in the coral
cant (Mann–Whitney U test, P > 0.05, Fig. 5a, b). community structure, with a total variance (inertia) of
123
846 Mar Biol (2008) 154:841–853
Fig. 4 a Percent cover of a 50 b
Montastraea annularis complex Total hard coral
Montastraea annularis
complex, the most dominant 40 n=136 in each year n=136 in each year
hard coral species in the Florida
Percent cover
Keys, and b total hard coral 30
percent cover. These values
were calculated from the 20 20
reefs included in the analyses
10
0
0.77 % 8.10 % 8.20 % 5.70 % 4.42 % 4.60 %
1.30 % 1.51 % 1.11 % 0.97 %
1996 1997 1998 1999 2000 1996 1997 1998 1999 2000
Year
Fig. 5 a Shannon’s diversity a 2.5
index and b species richness
of hard corals across reefs
2.0
Shannon diversity index
between 1996 and 2000
1.5
1.0
1996
0.5 2000
0.0
30
b
25
Species richness
20
15
10
5
0
Western
Jaap
Looe
Smith
Tennessee
Conch
West
Sambo
Sand
Rock
Molasses
Carysfort
Admiral
Porter
Dustan
Grecian
Alligator
Long
Turtle
Sombrero
Lower Keys Middle Keys Upper Keys
Reef
76.52% (Table 2). The Monte–Carlo permutation test The strength and direction of the relationships between
revealed that the CCA was robust (P < 0.01), indicating a the predictor variables and reef sites and coral species are
strong correspondence between hard coral community illustrated in the CCA biplots (Fig. 6a, b). Macroalgae had
structure and predictor variables. Correlation between hard a negative relationship with chl a (T = ¡0.451), as reXected
coral species and the two canonical axes was statistically in their opposing position in the canonical space. Chl a and
signiWcant (Monte–Carlo permutation test, P < 0.01, sponge cover run parallel with salinity because of the posi-
Table 2). These canonical axes explained 32% of the varia- tive relationships between the two former predictor vari-
tion in coral community structure. Macroalgal and sponge ables and salinity (chl a–salinity, T = 0.64; sponge–salinity,
cover accounted for high negative contribution to this vari- T = 0.44). Reefs in the Upper and Middle Keys aggregated
ation. in the upper left corner of the canonical space and toward
123
Mar Biol (2008) 154:841–853 847
Table 2 Results of CCA of 35
Axes p (Monte–Carlo
coral species in 20 reef sites in
permutation test)
relation to the Wve predictor 1 2
variables
a. Summary statistics of ordination axes
Eigenvalues (total inertia = 0.765) 0.151 0.094 0.01
Species-predictor correlation (T) 0.611 0.684 0.01
Cumulative % variance explained 19.7 32.0
b. Axis intraset correlation coeYcient of predictor variables (coeYcient ¸ §0.4 are considered ecologically
relevant and are in bolded text)
Macroalgae ¡0.851 ¡0.082
Salinity 0.778 ¡0.056
Chlorophyll a 0.713 0.376
Sponge 0.577 ¡0.647
Total Nitrogen 0.390 0.310
Fig. 6 Biplot of a 20 reef sites and b 35 coral species in relation to the (mea), Millepora alcicornis (mia), Millepora complanata (mil), Mon-
Wve predictor variables entered in the CCA. Species abbreviations (en- tastraea annularis complex (moa), Montastraea cavernosa (moc),
closed): Acropora cervicornis (arc), Acropora palmata (acr), Agaricia Mussa angulosa (mus), Mycetophyllia aliciae (myc), Mycetophyllia
agaricites complex (agg), Agaricia fragilis (agf), Agaricia lamarcki danaana (myd), Mycetophyllia feroz (myf), Mycetophyllia lamarcki-
(agl), Colpophyllia natans (col), Dendrogyra cylindrus (den), Dicho- ana (myl), Oculina diVusa (ocu), Porites astreoides (pos), Porites
coenia stokesi (dis), Diploria clivosa (dic), Diploria labyrinthiformis porites (por), Scolymia cubensis (sco), Siderastrea radians (sic),
(dil), Diploria strigosa (dis), Eusmilia fastigiata (eus), Favia fragum Siderastrea siderea (sid), Solenastrea bournoni (sol) and Stephano-
(fav), Leptoseris cucullata (lep), Madracis decactis (mad), Madracis coenia michelinii (ste)
mirabilis (mam), Manicina areolata (man), Meandrina meandrites
the macroalgal space; in contrast, reefs in the Lower Keys Lower Keys). This pattern suggests that the inXuence of
were spread out (Fig. 6a). Most of the coral species are water quality in aVecting the structure of the benthic biota
aggregated toward the sponge space; in contrast, none in the Florida Keys occurs in two levels. On the one hand,
aggregated near the chl a space (Fig. 6b). there is similarity in water quality within each locale (e.g.,
within the Lower Keys) and on the other, there is a diVer-
ence in water quality between the two regions (e.g.,
Discussion between the Lower Keys and the Upper–Middle Keys).
Previous studies oVer some bases for the intra and inter-
The separation of reefs into two main RSI based on water regional variations in water quality in the Florida Keys. The
quality measurements corresponded with their geological lower physical component scores (i.e., temperature) in the
locations in the Florida Keys (Upper-Middle Keys and Upper–Middle Keys reefs maybe a reXection of the cold
123
848 Mar Biol (2008) 154:841–853
freshwater Xowing from Florida Bay through the island Third, diVerential recruitment of corals between regions
passages in the Middle Keys (Smith and Lee 2003; Keller could help explain the prevailing regional variation in coral
and Donahue 2006) and to some extent from Biscayne Bay community structure in the Florida Keys (Hughes and Tan-
(Jones and Boyer 2002). Water quality measurements in ner 2000). It is well known that the recruitment of corals is
reefs within the Upper–Middle Keys were similar, perhaps aVected by several factors, including larval supply, substra-
because of the mixing of water masses in these areas tum availability, disturbance and regional oceanography
through wind-forced currents and gravity-driven transport (Chiappone and Sullivan 1996). Recently, Moulding (2005)
produced by cross-Key sea level diVerences (Smith and Lee demonstrated that the density and diversity of recruits var-
2003). The strength of these water exchanges exhibit sea- ies between regions, being signiWcantly lower in the Upper
sonal variations in response to prevailing winds and other Keys than in the Lower Keys. Finally, it is possible that the
oceanographic factors, for example, the Loop current (Kel- Upper Keys is exposed to severe environmental stresses
ler and Donahue 2006). The higher salinity and algal bloom and selective pressures due to the fact that this region
component scores (i.e., for chl a, TP and turbidity) in the is closer to the northern limit of the Florida reef tract
Lower Keys relative to the rest of the FKNMS could be (Moulding 2005).
attributed to the onshore movement of high-saline and Coral reef degradation in the Florida Keys was charac-
nutrient-rich deep water into this region (Leichter et al. terized by marked reduction in coral cover, and subsequent
2003) brought about by cyclonic gyres spun oV of the Flor- proliferation of macroalgae from 1996 to 2000. In the Flor-
ida Current (Szmant and Forrester 1996; Jones and Boyer ida Keys, the concurrent dominance of macroalgae relative
2002). Higher nutrient concentration in the Lower Keys to hard corals over time were documented previously by
could also have an anthropogenic origin, considering that several authors, including Dustan (1977), Dustan and Halas
the density of human population in the Lower Keys is (1987), Porter and Meier (1992), Murdoch and Aronson
higher than the rest of the Florida Keys (Ward-Paige et al. (1999), and Porter et al. (2002) (Table 3). Such dramatic
2005). Alternatively, the lower nutrient concentration in the changes in coral reef community structure are widely
Upper–Middle Keys relative to the Lower Keys could be reported in the Caribbean region (Done 1992; Hughes
attributed to the much higher likelihood of mixing between 1994; McClanahan and Muthiga 1998; Gardner et al. 2003,
waters in these regions and the relatively clean Atlantic 2005). This phenomenon is often considered an indicator of
Ocean waters (Boyer and Jones 2002). a coral reef community that is undergoing a phase-shift
The separation of reefs in the Lower Keys from the (McCook 1999; McManus and Polsenberg 2004; Rogers
Upper–Middle Keys in CCA corroborates the clustering of and Miller 2006). In the Florida Keys, the phase-shift from
reefs based on RSI, further illustrating the strong inXuence coral-dominated to macroalgal-dominated reef communi-
of the predictor variables (macroalgae, sponge, chl a, TN ties may be attributed to the diminished resilience of corals
and salinity) in the structuring of reef communities at a against perturbations (Jackson 2001; Aronson and Precht
spatial scale larger than the individual reef. This pattern is 2006), hypothesized to be caused by top-down mechanisms
consistent with the Wndings of Ogden et al. (1994) and such as herbivore reduction (Hughes 1994; Jackson et al.
Murdoch and Aronson (1999) who attributed the inter-reef 2001) and bottom-up processes such as nutrient enrichment
variations in coral community structure in the Florida Keys (Lapointe 1999; Leichter et al. 2003) (Fig. 7 for a diagram-
to the region-wide diVerences in environmental characteris- matic presentation of how we perceive these interacting
tics. The inXuence of environmental gradients in the struc- factors that result in a phase-shift).
turing of reef communities in the Florida Keys may be More recent evidence elucidates the potential mecha-
explained by several possible mechanisms. First, anteced- nisms underlying macroalgal-coral interactions that may
ent topography determine the suitability of present sub- lead to a localized phase-shift in the Florida Keys. Miller
strate for coral settlement and development. For example, and Hay (1998) demonstrated that Dictyota spp. and Hali-
Ginsburg and Shinn (1994) reported that the predominance meda opuntia, the two most abundant macroalgal species in
of mobile sand substrates in the Middle Keys and seaward the Florida Keys, inhibited the growth of Porites porites.
of Biscayne Bay may render these areas unsuitable for coral Jompa and McCook (2003a, b) have also shown that some
recruitment, thereby causing reefs in these areas to become algae (Anotrichium tenue, Corallophila huysmansii) can
impoverished. Second, Florida Bay waters that pass directly cause coral tissue death. Furthermore, KuVner et al.
through the Middle Keys may potentially inhibit coral (2006) have provided empirical evidence that macroalgae
growth, survival and recruitment into this region and adja- (e.g., Dictyota) can directly inhibit the settlement of coral
cent reefs by introducing pulses of waters with extremely recruits (e.g., Porites astreoides). The decline in coral
variable temperature and salinity as well as with high nutri- diversity and cover has also been associated with an
ent and sediment loads (Ginsburg and Shinn 1994; Chiap- increase in coral disease prevalence (Porter et al. 2001).
pone and Sullivan 1994; Szmant and Forrester 1996). Increased physical contact between corals and macroalgae
123
Table 3 Studies in the Caribbean region (Western Atlantic, including Florida and the Bahamas) documenting changes in benthic community structure in the region
Sites Period Mean % cover Proposed direct and indirect agents of change Source
Coral Macroalgae Sponge
Initial After Initial After Initial After
Caribbean region 1970 s–2000 s 50 10 Increasing Increasing White band disease; hurricanes; Gardner et al. (2003) and
indirectly attributed to increased López-Victoria
Mar Biol (2008) 154:841–853
abundance of macroalgae following and Zea (2005)
mass mortality of the herbivorous
D. antillarum(DA) in 1983
Florida Keys 1996–2000 8.1 4.6 5.7 9.6 (16.5 in 1998) 2.2 2.2 Disease; bleaching; anomalous extreme This study; also see
salinity and temperature events; eutrophication Porter et al. (2002)
and sedimentation; low herbivory due to
overWshing and mass mortality of DA;
high cover of macroalgae and probably sponges
Florida Keys (Carysfort Reef) 1975–1982 Boat grounding and anchor damage; hurricanes; Dustan and Halas (1987)
Mean percent cover (all species) 31.6 32.2 sedimentation; coral diseases; eutrophication
Acropora palmata 63 33
(mean colony size in cm)
Agaricia agaricites cm 4 6
(mean colony size in cm)
Florida Keys 1984–1991 28.4 6.6 Disease; bleaching Porter and Meier (1992)
(Carysfort and Looe Reefs) and Connell (1997)
Gulf of Mexico 1980–1990 55.2 8.7 Disease; bleaching Connell (1997)
(Flower Gardens Reef)
St. John, US Virgin Islands 1988–2003 43.4 12.3 Reported macroalgae competing Hurricane; high sea surface temperature (SST); Edmunds and Elahi (2007)
(Yawzi Point) with corals for space coral disease; macroalgae
St John, US Virgin Islands 1999–2000 (coral) Before- after 18 14 7.3 33.5 Hurricane; diseases; low herbivory due Rogers and Miller (2006)
Hurricane Hugo to overWshing and DA mass mortality;
in 1989 (macroalgae) failure of settlement
Panama (Punta de San Blas) 1983–1990 43–45 12–26 <2 10–32 Bleaching; DA mass mortality; sedimentation Shulman and
and eutrophication Robertson (1996)
Jamaica 1977–1980 vs. 1990–1993 52 3 4 92 OverWshing of herbvivores; hurricane; Hughes (1994)
disease; DA mass mortality
Jamaica (Dairy Bull) 1995–2004 <1 11 90 45–6 Long-lived colonies of Montastraea annularis Idjadi et al. (2006)
provided structural refugia
Belize (Glovers Reef lagoon) 1970–71 vs. 1996–1997 80 20 20 80 Diseases; low herbivory due to overWshing McClanahan and
and DA mass mortality Muthiga (1998)
Belize (Channel Cay) 1997–2001 »43 »0 <10 (no change) »15 43 Bleaching due to high SST Aronson et al. (2000, 2002)
(mainly Agaricia tenuifolia)
Costa Rica (Cahuita) 1980–1993 40.4 29.2 Natural disturbances; human impacts Connell (1997)
A majority of these studies indicate a decrease in coral cover (or a change in species composition), followed by subsequent increase in macroalgal and sponge cover over time
This list in not meant to be exhaustive. For worldwide summary of the impacts of diVerent array of disturbances, see Connell (1997) and Goreau et al. (2000)
123
849
850 Mar Biol (2008) 154:841–853
Fig. 7 Diagram showing an array of factors that potentially contrib- human disturbances. Regional-global factors include f increase sever-
uted to the phase-shift in Florida Keys coral reefs from coral-domi- ity and incidence of marine diseases, which is partly attributed to the
nated to algal/sponge-dominated community. Only dominant mass mortality of g Diadema antillarum in 1983–1984 and regional-
pathways are presented in the diagram. Solid and dashed lines indicate wide coral bleaching events; h global climate change, which has been
negative and positive interactions, respectively, and the arrowhead of associated with increased severity and frequency of anomalous high
the line indicates the trajectory of the interaction. Inside the dashed cir- SST causing regional-wide bleaching events; and i and historical and
cle are the dominant benthic organisms (corals, macroalgae and spong- natural disturbances such as hurricanes. Symbols courtesy of the Inte-
es). Inside the dashed square are array of local factors that potentially gration and Application Network (ian.umces.edu/symbols/), Univer-
contributed to the phase-shift in the Florida Keys coral reefs: a Wshing sity of Maryland Center for Environmental Science, USA. For a
indirectly impact corals through anchor damage and extraction of b comprehensive review of phase-shift in coral reefs, see Done (1992);
herbivores and c spongivores Wshes; d extreme seasonal changes of McCook (1999); McManus et al. (2000); McManus and Polsenberg
temperature and salinity; and e eutrophication, sedimentation and other (2004); and Aronson et al. (2006)
due to macroalgal bloom has been hypothesized to provoke in sponge cover from 1996 to 2000; this agrees with earlier
coral diseases (Green and Bruckner 2000). This observation reports of Aronson et al. (2002) and López-Victoria and
was empirically supported by the Wndings of Nugues et al. Zea (2005) in the Caribbean region. Cliona delitrix,
(2004) that the increasing physical contact between Hali- C. lampa, and C. caribboea, known to be aggressive
meda opuntia triggered a virulent disease known as white bioeroders, were among the dominant species of sponges in
plague type II in Montastraea faveolata. The combination the Florida Keys (Keller and Donahue 2006). Ward-Paige
of these mechanisms may have contributed to the negative et al. (2005) reported that the Lower Keys had the highest
relationship between coral and macroalgal cover in the sponge abundance and size, and they attributed this to
Florida Keys over the 5-year study period, as demonstrated sewage contamination based on higher 15 N levels in the
by the CCA. sponge tissues. Reduction in abundance of spongivorous
Sponge cover was also negatively correlated with hard Wshes as postulated by Hill (1998) could also play a role in
coral cover in the Florida Keys. There was a subtle increase the increase of sponge cover in the Florida Keys. Spongivory
123
Mar Biol (2008) 154:841–853 851
was previously demonstrated by Pawlik (1998) to be an factors (e.g., source of propagules), frequency of occurrence
important agent in controlling the population of at least of physical disturbance, and historical processes may inXu-
some sponges. Sponge proliferation could aVect corals ence the community structure of corals in the Florida Keys.
through several mechanisms: allelopathy (Engel and Future investigations that focus on mechanistic processes
Pawlik 2005; Pawlik et al. 2007), physical smothering and need to address whether these factors act alone or interac-
cellular digestion (Hill 1998) as well as direct space tively, either through mitigative or exacerbative pathways.
competition (Lopez-Victoria et al. 2006). These mecha-
nisms would lead to increased bioerosion, resulting in a net Acknowledgments R. J. Maliao is supported by the Fulbright-
Philippines Agriculture Scholarship Program. We thank the FKNMS
loss of carbonate, thus compromising the integrity of the
management through its coordinator, Fred McManus, for allowing us
coral skeletal structure. However, the positive and negative to use their monitoring database. Edward Webb, Kathe Jensen, J. R.
relationships between hard coral and sponge in the Wrst and Kerfoot, Matt Wittenrich, Zan Didoha, Vutheary Hean, Justin Anto,
second CCA axis, respectively, indicate that sponge-coral and Bernice Polohan provided valuable comments on an earlier version
of this manuscript. We particularly appreciate the advice of Bruce
relationships are species-speciWc. This pattern is consistent McCune on the use of CCA. The valuable comments and recommen-
with the Wndings of Aerts (1998) who reported that the out- dations of three anonymous reviewers greatly improved this manu-
come of coral-sponge interactions depends on the species script.
of both corals and sponges and the frequency of previous
encounters between coral and aggressive sponge species.
The positive relationships of both sponge and chl a with References
salinity is indicative of their wide tolerance to salinity Xuc-
tuations. Moreover, the negative relationship between mac- Aerts LAM (1998) Sponge/coral interactions in Caribbean reefs: anal-
roalgae and chl a is probably caused by chl a shading in the ysis of overgrowth patterns in relation to species identity and
water column, thereby reducing macroalgal productivity. cover. Mar Ecol Prog Ser 175:241–249
Aronson RB, Precht WF (2001) White-band disease and the changing
The chl a concentration in the FKNMS was around face of Caribbean coral reefs. Hydrobiologia 460(1–3):25–38
0.4 mg¡1, a value that can be considered eutrophic in reef Aronson RB, Precht WF (2006) Conservation, precaution, and Carib-
communities (Lapointe et al. 2004). In the biplot, there bean reefs. Coral Reefs 25:441–450
were no coral species that aggregated in the chl a space, Aronson RB, Precht WF, Macintrye IG (1998) Extrinsic control of spe-
cies replacement on a Holocene reef in Belize: the role of coral
consistent with the Wndings of van Woesik et al. (1999) and disease. Coral Reefs 17(3):223–230
Tomascik and Sander (1985), who demonstrated the nega- Aronson RB, Precht WF, Macintyre IG, Murdoch TJT (2000) Coral
tive correlation between chl a concentration and coral bleach-out in Belize. Nature 405:36
cover, growth and diversity. However, it is diYcult to sepa- Aronson RB, Precht WF, Toscano MA, Koltes KH (2002) The 1998
bleaching event and its aftermath on a coral reef in Belize. Mar
rate the eVects of chl a on the structuring of coral communi- Biol 141(3):435–447
ties because it is collinear with a variety of other water Bellwood DR, Hughes TP, Folke C, Nystrom M (2004) Confronting
quality variables. the coral reef crisis. Nature 429:827–833
In summary, results of our analyses revealed a strong Boyer JN, Jones RD (2002) A view from the bridge: external and inter-
nal forces aVecting the ambient water quality of the Florida Keys
correspondence between key predictor variables (e.g., mac- National Marine Sanctuary (FKNMS). In: Porter JW, Porter KG
roalgae, sponge and chl a) and characteristics of the hard (eds) The Everglades, Florida Bay, and coral reefs of the Florida
coral community structure in the Florida Keys. Water qual- Keys. CRC Press, Florida, pp 609–628
ity variables divide the reef tract of the Florida Keys into Carpenter RC (1990) Mass mortality of Diadema antillarum I. Long-
term eVects on sea urchin population-dynamics and coral reef
two major groups: the Upper–Middle Keys reefs, character- algal communities. Mar Biol 104(1):67–77
ized by lower values of water quality variables and the Chiappone M, Sullivan KM (1994) Patterns of coral abundance deWn-
Lower Keys, characterized by higher values of water qual- ing nearshore hardbottom communities of the Florida Keys. Fla
ity variables. The negative relationship between hard coral Sci 57:108–125
Chiappone M, Sullivan KM (1996) Distribution, abundance and spe-
cover and macroalgal/sponge cover suggests a phase-shift cies composition of juvenile Scleractinian corals in the Florida
in the coral reef structure in the Florida Keys from coral- reef tract. Bull Mar Sci 58(2):555–569
dominated to macroalgae and sponge-dominated communi- Clarke KR, Ainsworth M (1993) A method of linking multivariate
ties. However, the correlative nature of our analyses community structure to environmental variables. Mar Ecol Prog
Ser 92:205–219
precludes the identiWcation of precise causal mechanisms Clarke KR, Gorley RN (2006) PRIMER v6: User manual/tutorial.
that underlie the structuring of the hard coral community in PRIMER-E Ltd, Plymouth
the Florida Keys. It is also probable that the concurrent Connell JH (1997) Disturbance and recovery of coral assemblages.
increase of macroalgal and sponge cover in the Florida Coral Reefs 16:101–113
Done TJ (1992) Phase shifts in coral reef communities and their
Keys reXects an alternative state as a result, rather than the ecological signiWcance. Hydrobiologia 247:121–132
cause of coral demise. We recognize that other factors such Dustan P (1977) Vitality of reef coral populations oV Key Largo,
as resource availability, herbivory intensity, supply-side Florida: recruitment and mortality. Environ Geol 2:51–58
123
852 Mar Biol (2008) 154:841–853
Dustan P, Halas JC (1987) Changes in the reef-coral community of Keller BD, Donahue S (eds) (2006) 2002–03 sanctuary science report:
Carysfort Reef, Key Largo, Florida: 1974–1982. Coral Reefs an ecosystem report card after Wve years of marine zoning. US
6:91–106 Department of Commerce, National Oceanic and Atmospheric
Edmunds PJ, Elahi R (2007) The demographics of a 15-year decline in Administration, National Ocean Service, OYce of National
cover of the Caribbean reef coral Montastraea annularis. Ecol Marine Sanctuaries, Florida Keys National Marine Sanctuary,
Monogr 77:3–18 Marathon
Engel S, Pawlik JR (2005) Interactions among Florida sponges. I. Reef KuVner IB, Walters LJ, Becerro MA, Paul VJ, Ritson-Williams R,
habitats. Mar Ecol Prog Ser 303:133–144 Beach KS (2006) Inhibition of coral recruitment by macroalgae
Galindo HM, Olson DB, Palumbi SR (2006) Seascape genetics: a cou- and cyanobacteria. Mar Ecol Prog Ser 323:107–117
pled oceanographic-genetic model predicts population structure Lapointe BE (1999) Simultaneous top-down and bottom-up forces
of Caribbean corals. Curr Biol 16(16):1622–1626 control macroalgal blooms on coral reefs. Limnol Oceanogr
Gardner TA, Gill JA, Grant A, Watkinson AR (2003) Long-term 44:1586–1592
region-wide declines in Caribbean corals. Science 301:958–960 Lapointe BE, Matzie WR (1996) EVects of stormwater nutrient dis-
Gardner TA, Cote IM, Gill IA, Grant A, Watkinson AR (2005) Hurri- charges on eutrophication processes in nearshore waters of the
canes and Caribbean coral reefs: impacts, recovery, patterns, and Florida Keys. Estuaries 19:422–435
role in long-term decline. Ecology 86(1):174–184 Lapointe BE, Barile PJ, Matzie WR (2004) Anthropogenic nutrient
Ginsburg RN (ed) (1994) Proceedings of the colloquium on global as- enrichment of seagrass and coral reef communities in the Lower
pects of coral reefs: health, hazards and history. Rosenstiel School Florida Keys: discrimination of local versus regional nitrogen
of Marine and Atmospheric Science, University of Miami, Miami sources. J Exp Mar Biol Ecol 308:23–58
Ginsburg, RN, Shinn EA (1994) Preferential distribution of reefs in the Leichter JJ, Stewart HL, Miller SL (2003) Episodic nutrient transport
Florida reef tract: the past is the key to the present. In: Ginsburg to Florida coral reefs. Limnol Oceanogr 48:1394–1407
RN (ed) Proceedings of the colloquium on global aspects of coral López-Victoria M, Zea S (2005) Current trends of space monopoliza-
reefs: health, hazards and history. Rosenstiel School of Marine tion by encrusting excavating sponges on Colombian coral reefs.
and Atmospheric Science, University of Miami, pp 21–26 Mar Ecol 26:33–41
GraVelman J (2001) Quality statistics in canonical correspondence Lopez-Victoria M, Zea S, Weil E (2006) Competition for space
analysis. Environmetrics 12:485–497 between encrusting excavating Caribbean sponges and other coral
Green EP, Bruckner AW (2000) The signiWcance of coral disease epizo- reef organisms. Mar Ecol Prog Ser 312:113–121
otiology for coral reef conservation. Biol Conserv 96(3):347–361 McClanahan TR, Muthiga NA (1998) An ecological shift in a remote
Greenstein BJ, PandolW JM (1997) Preservation of community struc- coral atoll of Belize over 25 years. Environ Conserv 25(2):122–130
ture in modern reef coral life and death assemblages of the Florida McCook LJ (1999) Macroalgae, nutrients and phase shifts on coral
Keys: implications for the quaternary fossil record of coral reefs. reefs: scientiWc issues and management consequences for the
Bull Mar Sci 61:431–452 Great Barrier Reef. Coral Reefs 18:357–367
Hill MS (1998) Spongivory on Caribbean reefs releases corals from McCune B (1997) InXuence of noisy environmental data on canonical
competition with sponges. Oecologia 117:143–150 correspondence analysis. Ecology 78(8):2617–2623
Hughes TP (1994) Catastrophes, phase shifts, and large-scale degrada- McCune B, MeVord MJ (1999) PC-ORD ver. 5. Multivariate analysis
tion of a Caribbean coral reef. Science 265(5178):1547–1551 of ecological data. MjM Software, Gleneden Beach, Oregon
Hughes TP, Tanner JE (2000) Recruitment failure, life histories, and McCune B, Grace JB (2002) Analysis of ecological communities.
long-term decline of Caribbean corals. Ecology 81:2250–2263 MJM Software Design, Gleneden Beach, Oregon
Hughes TP, Baird AH, Bellwood DR, Card M, Connolly SR, Folke C, McManus JW, Polsenberg JF (2004) Coral-algal phase shifts on coral
Grosberg R, Hoegh-Guldberg O, Jackson JBC, Kleypas J, Lough reefs: ecological and environmental aspects. Prog Oceanogr
JM, Marshall P, Nystroem M, Palumbi SR, PandolW JM, Rosen B, 60:263–279
Roughgarden J (2003) Climate change, human impacts, and the McManus JW, Menez LAB, Kesner-Reyes KN, Vergara SG, Ablan
resilience of coral reefs. Science 301(5635):929–933 MC (2000) Coral reef Wshing and coral-algal phase shifts: impli-
Idjadi JA, Lee SC, Bruno JF, Precht WF, Allen-Requa L, Edmunds PJ cations for global reef status. ICES J Mar Sci 57:572
(2006) Rapid phase-shift reversal on a Jamaican coral reef. Coral Miller MW, Hay ME (1998) EVects of Wsh predation and seaweed
reefs 25:209–211 competition on the survival and growth of corals. Oecologia
Jackson JBC (2001) What was natural in the coastal oceans? Proc Natl 113:231–238
Acad Sci USA 98:5411–5418 Moulding AL (2005) Coral recruitment patterns in the Florida Keys.
Jackson JBC, Kirby MX, Wolfgang HB, Bjorndal KA, Botsford LW, Rev Biol Trop 25(Suppl 1):75–82
Bourque BJ, Bradbury RH, Cooke R, Erlandson J, Estes JA, Murdoch TJT, Aronson RB (1999) Scale-dependent spatial variability
Hughes TP, Kidwell S, Lange LB, Lenizan HS, PandolW JM, Pett- of coral assemblages along the Florida Reef Tract. Coral Reefs
erson CH, Steneck RS, Tegner MJ, Warner RR (2001) Historical 18:341–351
overWshing and the recent collapse of coastal ecosystems. Science Nugues MM, Smith GW, Hooidonk RJ, Seabra MI, Bak RPM (2004)
293:629–638 Algal contact as a trigger for coral disease. Ecol Lett 7:919–923
Jompa J, McCook LJ (2003a) Coral-algal competition: macroalgae Ogden JC, Porter JW, Smith NP, Szmant AM, Jaap WC, Forcucci D
with diVerent properties have diVerent eVects on corals. Mar Ecol (1994) A long-term interdisciplinary study of the Florida Keys
Prog Ser 258:87–95 seascape. Bull Mar Sci 54(3):1059–1071
Jompa J, McCook LJ (2003b) Contrasting eVects of turf algae on cor- Palmer MW (1993) Putting things in even better order: the advantages
als: massive Porites spp. are unaVected by mixed-species turfs, of canonical correspondence analysis. Ecology 74(8):2215–2230
but killed by the red alga Anotrichium tenue. Mar Ecol Prog Ser PandolW JM, Bradbury RH, Sala E, Hughes TP, Bjorndal KA, Cooke
258:79–86 RG, McArdle D, McClenachan L, Newman MJH, Paredes G,
Jones RD, Boyer JN (2002) FY2002 annual report of the water quality Warner RR, Jackson JBC (2003) Global trajectories of the long-
monitoring project for the water quality protection program in the term decline of coral reef ecosystems. Science 301(5635):955–
Florida Keys National Marine Sanctuary. Technical Report # 958
T192 of the Southeast Environmental Research Center, Florida Pawlik JR (1998) Coral reef sponges: do predatory Wshes aVect their
International University, Miami distribution? Limnol Oceanogr 43:1396–1399
123
Mar Biol (2008) 154:841–853 853
Pawlik JR, Steindler L, Henkel TP, Beer S, Ilan M (2007) Chemical Szmant AM, Forrester A (1996) Water column and sediment nitrogen
warfare on coral reefs: sponge metabolites diVerentially aVect and phosphorus distribution patterns in the Florida Keys, USA.
coral symbiosis in situ. Limnol Oceanogr 52(2):907–911 Coral Reefs 15:21–41
Petraitis PS, Dudgeon SR (2004) Detection of alternative stable states ter Braak CJF (1986) Canonical correspondence analysis: a new eigen-
in marine communities. J Exp Mar Biol Ecol 300:343–371 vector technique for multivariate direct gradient analysis. Ecol-
Porter JW, Meier OW (1992) QuantiWcation of loss and change of Flo- ogy 67:1167–1179
ridian reef coral populations. Am Zool 32:625–640 ter Braak CJF, Looman CWN (1994) Biplots in reduced-rank regres-
Porter JW, Dustan P, Jaap WC, Patterson KL, Kosmynin V, Meier sion. Biometrical J 36(8):983–1003
OW, Patterson ME, Parsons M (2001) Patterns of spread of coral ter Braak CFJ, Verdonschot PFM (1995) Canonical correspondence
disease in the Florida Keys. Hydrobiologia 460:1–24 analysis and related multivariate methods in aquatic ecology.
Porter JW, Kosmynin V, Patterson KL, Porter KG, Jaap WC, Wheaton Aquat Sci 57(3):255–289
JL, Hackett K, Lybolt M, Tsokos CP, Yanev G, Marcinek DM, Tomascik T, Sander F (1985) EVects of eutrophication on reef building
Dotten J, Eaken D, Patterson M, Meier OW, Brill M, Dustan P corals. I. EVects of eutrophication on the reef building coral Mon-
(2002) Detection of coral reef change by the Florida Keys coral tastraea annularis. Mar Biol 87:143–155
reef monitoring project. In: Porter JW, Porter KG (eds) The Ever- van Woesik R, Tomascik T, Blake S (1999) Coral assemblages and
glades, Florida Bay, and coral reefs of the Florida Keys: an eco- physico-chemical characteristics of the Whitsunday Islands: evi-
system sourcebook. CRC, Boca Raton, pp 749–769 dence of recent community changes. Mar Freshw Res 50:427–440
Quinn GP, Keough MJ (2006) Experimental design and data analysis Walther GR, Post E, Convey P, Menzel A, Parmesan C, Beebee TJC,
for biologists. Cambridge University Press, Cambridge Fromentin JM, Hoegh-Guldberg O, Bairlein F (2002) Ecological
Rakocinski CF, Lyczkowski-Shultz J, Richardson SL (1996) Ichtyoplank- responses to recent climate change. Nature 416:389–395
ton assemblage structure in Mississippi Sound as revealed by canon- Ward-Paige CA, Risk MJ, Sherwood OA, Jaap WC (2005) Clionid
ical correspondence analysis. Estuar Coast Shelf Sci 43:237–257 sponge surveys on the Florida reef tract suggest land-based nutri-
Rogers CS, Miller J (2006) Permanent ‘phase shifts’ or reversible ent inputs. Mar Pollut Bull 51:570–579
declines in coral cover? Lack of recovery of two coral reefs in Wheaton J, Jaap WC, Porter JW, Kosminyn V, Hackett K, Lybolt M,
St John, US Virgin Islands. Mar Ecol Prog Ser 306:103–114 Callahan MK, Kidney J, Kupfner S, Tsokos C, Yanev G (2001)
Shulman MJ, Robertson DR (1996) Changes in the coral reefs of San EPA/FKNMS Coral reef monitoring project: executive summary.
Blas, Caribbean Panama: 1983 to 1990. Coral Reefs 15:231–236 FKNMS symposium: an ecosystem report card, Washington DC
Smith NP, Lee TN (2003) Volume transports through tidal channels in
the middle Florida Keys. J Coast Res 19:254–260
123
DOI 10.1007/s00227-008-0977-0
ORIGINAL PAPER
Phase-shift in coral reef communities in the Florida Keys National
Marine Sanctuary (FKNMS), USA
Ronald J. Maliao · Ralph G. Turingan · Junda Lin
Received: 17 September 2007 / Accepted: 7 April 2008 / Published online: 18 April 2008
© Springer-Verlag 2008
Abstract Characterizing the Florida Keys National structuring of coral reef communities at a spatial scale
Marine Sanctuary (FKNMS), USA, has gained much atten- larger than the individual reef. Furthermore, it is conceiv-
tion over the past several decades because of apparent able that these predictor variables exerted inXuence for a
changes in the benthic community structure over space and long time rather than being a recent event. Results also
time representative of patterns occurring in the Caribbean revealed a pattern showing reduction in hard coral cover
region. We used a 5-year dataset (1996–2000) of macroal- and species richness, and subsequent proliferation of mac-
gal and sponge cover and water quality measurements as roalgae and sponges during the study period. Our analyses
predictor variables of hard coral community structure in the of the Florida Keys present a pattern that is consistent with
FKNMS. The 16 water quality variables were summarized the characteristics of a reef that has undergone a “phase-
into 4 groups by principal component analysis (PCA). Hier- shift,” a phenomenon that is widely reported in the Carib-
archical agglomerative cluster analysis of the mean and bean region.
standard deviation (SD) of the principal component scores
of water quality variables separated the reef sites into two
main groups (and Wve sub-groups), referred to as reefs of Introduction
similar inXuence (RSI). The main groups corresponded
with their geographical locations within the Florida Keys: The community structure of coral reefs in the Caribbean,
the reefs in the Upper and Middle Keys being homoge- including the Florida Keys, have remained stable for thou-
neous and collectively, having lower water quality scores sands of years before the region suVered a catastrophic
relative to reefs in the Lower Keys. Canonical correspon- coral mass mortality in the 1980s (Hughes 1994; Aronson
dence analysis (CCA) between hard coral cover and key et al. 1998). Since the 1980s die oV, coral reef deterioration
predictor variables (i.e., water quality, macroalgal cover has been characterized by a reduction in coral cover by as
and sponge cover) also separated the reefs in the Lower much as 40%, a shift in the composition of surviving spe-
Keys from reefs in the Upper–Middle Keys, consistent with cies (Hughes 1994; Greenstein and PandolW 1997; Gardner
results of the cluster analysis, which categorized reefs et al. 2003), and a phase-shift from coral-dominated to
based on RSI. These results suggest that the prevailing macroalgal-dominated communities (Dustan 1977; Dustan
gradient of predictor variables may have inXuenced the and Halas 1987; Porter and Meier 1992; Hughes 1994;
McClanahan and Muthiga 1998; Porter et al. 2002; Gardner
et al. 2003, 2005).
Communicated by R. Cattaneo-Vietti.
The magnitude, scale and cause of the deterioration of
Caribbean coral reefs remain controversial (Hughes 1994;
R. J. Maliao (&) · R. G. Turingan · J. Lin Ginsburg 1994; Murdoch and Aronson 1999; Bellwood
Department of Biological Sciences, et al. 2004). Hypothesized mechanisms working at small
Florida Institute of Technology,
150 West University Blvd.,
spatio-temporal scales (meters to kilometers and days to
Melbourne, FL 32901-6975, USA years) include hurricanes (Gardner et al. 2003), point-
e-mail: rmaliao@Wt.edu source nutrient loading (Ginsburg and Shinn 1994; Leichter
123
842 Mar Biol (2008) 154:841–853
et al. 2003), impacts of upstream land use patterns 40 cm above the reef at a constant swim speed of about
(Bellwood et al. 2004), and macroalgal overgrowth (Hughes 4 m/min yielding approximately 9,000 video frames per
1994). Other hypotheses suggest the importance of meso- transect. Image analysis used a custom software application
to large-scale processes (tens to thousands of kilometers PointCount for coral reefs, developed speciWcally for the
and decades to millennia), such as sea urchin mass mortal- CREMP (Wheaton et al. 2001). Percent cover of hard cor-
ity events (Carpenter 1990), larval transport (Galindo et al. als, macroalgae (Xeshy and Wlamentous and non-coralline)
2006), herbivore reduction due to overWshing (Jackson and sponges was calculated from the images. Water quality
et al. 2001; PandolW et al. 2003), disease outbreaks (Aron- monitoring under WQMP was undertaken quarterly from
son and Precht 2001; Porter et al. 2001) and climate change 154 stations. Analyses only included those WQPP stations
(Walther et al. 2002; Hughes et al. 2003). It is unclear that overlapped with reef sites monitored by CREMP. Six-
whether the current state of the Caribbean coral reefs repre- teen water quality measurements were included in our anal-
sents a permanent phase-shift or that the community yses: temperature (°C), salinity (PSU), dissolved oxygen
structure would rebound (Petraitis and Dudgeon 2004). (DO, mg l¡1), and light attenuation (Kd, m¡1) were col-
Long-term monitoring of coral reefs in the Florida Keys lected in the Weld; organic carbon (TOC), total nitrogen
(1981–1991) and US Virgin Islands (1989–2003) have (TN), total phosphorus (TP), silicate (Si (OH)4), nitrite
indicated no evidence of recovery in terms of coral cover or (NO2¡), dissolved nitrate (NO3¡), ammonium (NH4+),
reduction of algal abundance over a decadal time scale soluble reactive phosphate (SRP), chlorophyll a (chl a),
(Porter and Meier 1992; Rogers and Miller 2006). dissolved inorganic nitrogen (DIN), total unWltered
The main goal of this study is to Wnd evidence of a possi- concentrations of organic nitrogen (TON) and turbidity
ble phase-shift in coral reef communities in the Florida (NTU) were determined in the laboratory from the water
Keys National Marine Sanctuary (FKNMS), USA. SpeciW- collected in situ (reported in g l¡1) (see Keller and Dona-
cally, this study is designed to determine the correspon- hue 2006 for the complete sampling protocol).
dence between hard coral cover, macroalgal and sponge
cover and water quality variables using the 1996–2000 data Data analyses
of the FKNMS Water Quality Protection Program (WQPP).
Data analyses were conducted at the level of reef site using
the 1996¡2000 dataset. Average values of percent cover of
Materials and methods benthic biota data (hard corals, macroalgae and sponges)
and water quality measurements were used for those reef
Study site and data collection sites with stations at varying depths. In the multivariate
analyses, benthic biota data and water quality measure-
The reef tract of the Florida Keys, which represents the ments were arcsine square-root and square-root trans-
third largest barrier reef in the world and the only living formed, respectively, and then data matrices were screened
coral reef in North America (Lapointe and Matzie 1996), for outliers. Outliers are extreme data values relative to
was designated as a marine sanctuary in 1990, and became others in a sample and have unduly large eVects on the
known as the FKNMS (Keller and Donahue 2006; Fig. 1). resultant probabilities (McCune and Grace 2002; Quinn and
The reef tract of the FKNMS, referred to here as the Florida Keough 2006). For example, outliers may distort estimates
Keys, is partitioned into three regions based on geographic and P values and inXate sums-of-squares, all of which
and environmental criteria: Upper Keys, Middle Keys, and would result in faulty conclusions. In our analyses, we deW-
Lower Keys (Ginsburg and Shinn 1994). The WQPP was ned outliers as those reefs, coral species or water quality
implemented in 1996 to monitor both benthic biota (under variables with SD (standard deviation) >2 (Quinn and
the Coral Reef Evaluation and Monitoring Project or Keough 2006) and/or those with sample points <3 (i.e., rare
CREMP) and water quality (under the Water Quality Moni- samples) (McCune and Grace 2002). Outliers were
toring Project or WQMP) in the Florida Keys until 2002 excluded from all data analyses.
(Keller and Donahue 2006). The CREMP includes 40 reef Principal component analysis (PCA) was used to extract
sites; benthic biota in two to four permanent stations the underlying patterns of the water quality variables using
(22 £ 2 m transects) were monitored quarterly on each reef a correlation matrix based on the average values of the 5-
site. Field monitoring consists of station species inventories year data. Water quality variables with component loading
(SSI) and video transects conducted in four permanent sta- ¸ §0.5 were retained in the description of each principal
tions (22 £ 2 m transect) in each reef site. SSI consists of component. The mean and SD of component scores for
timed (15 min) counts of stony coral species (Milleporina each reef site were then used as input variables in a hierar-
and Scleractinia) present in each station to provide data on chical agglomerative cluster analysis by unweighed
hard coral species richness. Video recordings were taken pair-group method (UPGMA) linkage rule using Euclidean
123
Mar Biol (2008) 154:841–853 843
Fig. 1 Map of the Florida Keys showing only the 20 reef sites included in the analyses. Reef abbreviations (enclosed): Western Head (Western),
West Washer Women (West)
distance (ED). Groups of reefs produced by the cluster CCA, the set of water quality variables identiWed by the
analysis were linked objectively using the similarity proWle BIOENV routine, and macroalgal and sponge cover were
(SIMPROF) routine (Clarke and Gorley 2006). Reef clus- used as the predictors of hard coral cover. In addition, the
ters were further merged at 0.5 ED to facilitate interpreta- variation in hard coral cover and predictor variables over
tion and were referred to as reefs of similar inXuence (RSI). the 5-year period was examined using a Kruskal–Wallis
The outcome of this analysis reveals the variation in water test. The variation in hard coral, macroalgal and sponge
quality across reefs in the Florida Keys. cover, as well as indices of Shannon diversity and species
Canonical correspondence analysis (CCA) (ter Braak richness of hard corals between 1996 and 2000 were further
1986), the most widely used direct gradient ordination examined using Mann–Whitney U test.
method (Palmer 1993; ter Braak and Verdonschot 1995; The robustness of the result of CCA was determined by
McCune 1997; GraVelman 2001; McCune and Grace Monte–Carlo permutation test (ter Braak 1986). Ordination
2002), was employed to determine the multivariate correla- scores in the CCA biplots were optimized by coral species
tion between hard coral community structure and predictor using the linear combination (LC scores) of predictor vari-
variables. The minimal set of water quality variables ables. LC scores were used because they best represent the
entered in the CCA was identiWed by the BIOENV routine relative contribution of each predictor variable to the varia-
(Clarke and Ainsworth 1993; Clarke and Gorley 2006). The tion in hard coral cover (McCune and Grace 2002) and
water quality variables were screened for multicollinearity have been shown to perform well even in skewed distribu-
prior to running the BIOENV routine and only a representa- tions (Palmer 1993). In the ordination biplot, predictor
tive of highly correlated water quality variables (i.e., those variables are represented as arrows and the length of the
with Spearman’s = ¸ §0.8) were used. Pruning of pre- arrow indicates the relative importance of each predictor
dictor variables to a minimal set that accounts for the larg- variable (ter Braak 1986). The angle between arrows
est variation in the hard coral community structure was indicates the degree of correlation between predictor vari-
necessary because the result of CCA becomes less robust as ables; the location of the reef or species relative to the
the number of predictor variables increases (McCune and arrow indicates site characteristics or species preferences
Grace 2002) and to minimize the problem associated with (Palmer 1993). Since ordination scores were calculated
multicollinearity (ter Braak and Looman 1994). In our using LC scores, the relative contribution of the predictor
123
844 Mar Biol (2008) 154:841–853
variables in each axis was determined from the intraset cor- ond main group (Fig. 2a). The physical property component
relation output (McCune and Grace 2002). A predictor var- (PC4: salinity, temperature and DO) and the algal bloom
iable with correlation coeYcient of ¸ §0.4 is considered component (PC2: chl a, TP, and turbidity) appeared to be
ecologically relevant (Rakocinski et al. 1996). The main the main components driving Florida Keys-wide variation
advantage of CCA is that it allowed simultaneous explora- in water quality, with higher values in the Lower Keys and
tion of relationships among diVerent reef sites or coral lower values in the Middle and Upper Keys (Fig. 2b).
species with multiple predictor variables. Results of running the BIOENV routine revealed that
Cluster analysis and SIMPROF and BIOENV routines TN, chl a and salinity are the principal set of water quality
were conducted using PRIMER version 6.1.9 (Plymouth variables signiWcantly explaining the variation in hard coral
Routines Multivariate Ecological Research, PRIMER-E community structure ( w = 0.393, P < 0.05, global BEST
Ltd, Plymouth, UK; Clarke and Gorley 2006) while the permutation test). TN Xuctuated signiWcantly across the
CCA was conducted using PC-Ord version 5 (MjM Soft- 5-year monitoring period (Kruskal–Wallis test, P < 0.01,
ware, Oregon, USA; McCune and MeVord1999; McCune Fig. 3a); chl a and salinity remained stable (Kruskal–Wallis
and Grace 2002). PCA and univariate tests were conducted test, P > 0.05, Fig. 3b, c).
using SPSS 14 (SPSS Inc., Chicago, USA).
Reef community structure
Results Macroalgal cover signiWcantly increased from 1996 to 2000
(Kruskal–Wallis test, P < 0.001, Fig. 3d), with an almost
Reefs of similar inXuence triple increase from the 1996 (5.7%) to 1998 (16.5%) level.
The macroalgal cover in 1996 had signiWcantly increased
The 16 water quality variables were reduced into four prin- by 68% in 2000 (9.6%) (Mann–Whitney U test, P < 0.05).
cipal components (PC), explaining 82.3% of the total varia- Although sponge cover had varied signiWcantly across
tion in the data set (Table 1). PC 1 to 4 are referred to as years (Kruskal–Wallis test, P < 0.01, Fig. 3e), the diVer-
organic component, algal bloom component, inorganic ence between 1996 and 2000 was not signiWcant (Mann–
nitrogen component and physical properties component, Whitney U test, P > 0.05).
respectively. Cluster analysis of the mean and SD of the Forty-two coral species were identiWed in the Florida Keys.
component scores separated the reefs into two main groups In terms of percent cover, the Montastraea annularis complex
and Wve subgroups; the Upper and Middle Keys reefs form dominated the hard coral species in the Florida Keys (Fig. 4a).
the Wrst main group and the Lower Keys reefs form the sec- Collectively, hard coral cover showed a decreasing trend
Table 1 Result of the PCA with
Water quality parameters Component
Varimax rotation for 16 water
quality variables PC1 PC2 PC3 PC4
Total organic nitrogen (TON) 0.865 0.373 0.163 ¡0.010
Total nitrogen (TN) 0.860 0.380 0.194 ¡0.017
Silicate SI (OH)4 0.819 0.335 0.322 0.141
Light attenuation (Kd) 0.781 ¡0.249 0.044 0.053
Total organic carbon (TOC) 0.752 0.482 0.326 ¡0.149
Nitrite (NO2¡) 0.572 0.017 0.501 ¡0.402
Chlorophyll a (Chl a) 0.051 0.843 0.050 0.439
Total phosphorus (TP) 0.153 0.834 ¡0.045 0.257
Turbidity 0.361 0.725 0.260 ¡0.082
Nitrate (NO3¡) 0.232 0.250 0.887 0.070
Soluble reactive phosphorus (SRP) 0.038 ¡0.176 0.847 0.113
Dissolved inorganic nitrogen (DIN) 0.384 0.276 0.829 ¡0.176
Water quality variables with Temperature 0.208 0.267 0.099 0.873
component loadings ¸ §0.5 Dissolved oxygen (DO) 0.322 ¡0.151 0.047 ¡0.816
(bolded text) were retained to
describe each component (PCA Salinity 0.477 0.258 0.147 0.760
PC1 = organic; PC2 = algal Ammonium (NH4+) 0.320 0.236 0.278 ¡0.475
bloom; PC3 = inorganic Percent variance explained 28.118 18.791 18.141 17.211
nitrogen; PC4 = physical (Total = 82.261)
properties)
123
Mar Biol (2008) 154:841–853 845
Fig. 2 a Dendogram of reef sites based on the cluster analysis of mean Wve subgroups (1–5); b mean of four PCA component scores of water
and SD of PCA component scores of water quality variables. Objective quality variables across RSI; error bars indicate §standard error of
grouping of clusters (dashed line) was based on SIMPROF routine. mean. RSI numbers refer to reef clusters in a
Clusters were further merged at 0.5 ED resemblance, forming a total of
1.4 4.5
a Total nitrogen b Chl a
1.2 4.0
3.5
1.0
3.0
0.8 2.5
mg l-1
mg l-1
0.6 2.0
n=102 140 141 143 143 n=102 140 141 143 143
1.5
0.4
1.0
0.2
0.5
0.0 0.0
0.16 0.14 0.13 0.13 0.11
0.315 0.260 0.289 0.312 0.278
39 80 e 35
c n=102 140 141 143 143 d n=136 in each year Macroalgae n=136 in each year Sponge
38 70 30
Salinity
37 60
25
36 50
20
cent cover
35 40
PSU
15
34 30
10
Per
33 20
32 10 5
36.3 36.3 35.6 36.2 36.3
31 0 0
5.7 4.9 16.5 6.1 9.6 2.2 1.9 1.3 1.4 2.2
30
1996 1997 1998 1999 2000 1996 1997 1998 1999 2000 1996 1997 1998 1999 2000
Year
Fig. 3 Box-plots of predictor variables entered in CCA : a TN, b chl percentiles. Points outside the whiskers are the outliers. Numbers
a, c salinity, d macroalgae, and e sponge. The center line of the box is below the box are the median values. These values were calculated
the median, the bottom and top of the box are the 25th and 75th percen- from the 20 reefs included in the analyses
tiles and the whiskers below and above of the box are the 10th and 90th
across the 5-year monitoring period, with the lowest level Correspondence between predictor variables and hard coral
occurring in 1999 (Kruskal–Wallis test, P < 0.01; Fig. 4b). community structure
Hard coral percent cover had signiWcantly decreased by
almost 50%, from 8.1% in 1996 to 4.6% in 2000 (Mann– Twenty out of the 25 aggregated reef sites and 35 out of the
Whitney U test, P < 0.05). DiVerences in species richness and 42 coral species were included in the CCA after data
Shannon diversity index over the two periods were not signiW- screening. The CCA revealed two gradients in the coral
cant (Mann–Whitney U test, P > 0.05, Fig. 5a, b). community structure, with a total variance (inertia) of
123
846 Mar Biol (2008) 154:841–853
Fig. 4 a Percent cover of a 50 b
Montastraea annularis complex Total hard coral
Montastraea annularis
complex, the most dominant 40 n=136 in each year n=136 in each year
hard coral species in the Florida
Percent cover
Keys, and b total hard coral 30
percent cover. These values
were calculated from the 20 20
reefs included in the analyses
10
0
0.77 % 8.10 % 8.20 % 5.70 % 4.42 % 4.60 %
1.30 % 1.51 % 1.11 % 0.97 %
1996 1997 1998 1999 2000 1996 1997 1998 1999 2000
Year
Fig. 5 a Shannon’s diversity a 2.5
index and b species richness
of hard corals across reefs
2.0
Shannon diversity index
between 1996 and 2000
1.5
1.0
1996
0.5 2000
0.0
30
b
25
Species richness
20
15
10
5
0
Western
Jaap
Looe
Smith
Tennessee
Conch
West
Sambo
Sand
Rock
Molasses
Carysfort
Admiral
Porter
Dustan
Grecian
Alligator
Long
Turtle
Sombrero
Lower Keys Middle Keys Upper Keys
Reef
76.52% (Table 2). The Monte–Carlo permutation test The strength and direction of the relationships between
revealed that the CCA was robust (P < 0.01), indicating a the predictor variables and reef sites and coral species are
strong correspondence between hard coral community illustrated in the CCA biplots (Fig. 6a, b). Macroalgae had
structure and predictor variables. Correlation between hard a negative relationship with chl a (T = ¡0.451), as reXected
coral species and the two canonical axes was statistically in their opposing position in the canonical space. Chl a and
signiWcant (Monte–Carlo permutation test, P < 0.01, sponge cover run parallel with salinity because of the posi-
Table 2). These canonical axes explained 32% of the varia- tive relationships between the two former predictor vari-
tion in coral community structure. Macroalgal and sponge ables and salinity (chl a–salinity, T = 0.64; sponge–salinity,
cover accounted for high negative contribution to this vari- T = 0.44). Reefs in the Upper and Middle Keys aggregated
ation. in the upper left corner of the canonical space and toward
123
Mar Biol (2008) 154:841–853 847
Table 2 Results of CCA of 35
Axes p (Monte–Carlo
coral species in 20 reef sites in
permutation test)
relation to the Wve predictor 1 2
variables
a. Summary statistics of ordination axes
Eigenvalues (total inertia = 0.765) 0.151 0.094 0.01
Species-predictor correlation (T) 0.611 0.684 0.01
Cumulative % variance explained 19.7 32.0
b. Axis intraset correlation coeYcient of predictor variables (coeYcient ¸ §0.4 are considered ecologically
relevant and are in bolded text)
Macroalgae ¡0.851 ¡0.082
Salinity 0.778 ¡0.056
Chlorophyll a 0.713 0.376
Sponge 0.577 ¡0.647
Total Nitrogen 0.390 0.310
Fig. 6 Biplot of a 20 reef sites and b 35 coral species in relation to the (mea), Millepora alcicornis (mia), Millepora complanata (mil), Mon-
Wve predictor variables entered in the CCA. Species abbreviations (en- tastraea annularis complex (moa), Montastraea cavernosa (moc),
closed): Acropora cervicornis (arc), Acropora palmata (acr), Agaricia Mussa angulosa (mus), Mycetophyllia aliciae (myc), Mycetophyllia
agaricites complex (agg), Agaricia fragilis (agf), Agaricia lamarcki danaana (myd), Mycetophyllia feroz (myf), Mycetophyllia lamarcki-
(agl), Colpophyllia natans (col), Dendrogyra cylindrus (den), Dicho- ana (myl), Oculina diVusa (ocu), Porites astreoides (pos), Porites
coenia stokesi (dis), Diploria clivosa (dic), Diploria labyrinthiformis porites (por), Scolymia cubensis (sco), Siderastrea radians (sic),
(dil), Diploria strigosa (dis), Eusmilia fastigiata (eus), Favia fragum Siderastrea siderea (sid), Solenastrea bournoni (sol) and Stephano-
(fav), Leptoseris cucullata (lep), Madracis decactis (mad), Madracis coenia michelinii (ste)
mirabilis (mam), Manicina areolata (man), Meandrina meandrites
the macroalgal space; in contrast, reefs in the Lower Keys Lower Keys). This pattern suggests that the inXuence of
were spread out (Fig. 6a). Most of the coral species are water quality in aVecting the structure of the benthic biota
aggregated toward the sponge space; in contrast, none in the Florida Keys occurs in two levels. On the one hand,
aggregated near the chl a space (Fig. 6b). there is similarity in water quality within each locale (e.g.,
within the Lower Keys) and on the other, there is a diVer-
ence in water quality between the two regions (e.g.,
Discussion between the Lower Keys and the Upper–Middle Keys).
Previous studies oVer some bases for the intra and inter-
The separation of reefs into two main RSI based on water regional variations in water quality in the Florida Keys. The
quality measurements corresponded with their geological lower physical component scores (i.e., temperature) in the
locations in the Florida Keys (Upper-Middle Keys and Upper–Middle Keys reefs maybe a reXection of the cold
123
848 Mar Biol (2008) 154:841–853
freshwater Xowing from Florida Bay through the island Third, diVerential recruitment of corals between regions
passages in the Middle Keys (Smith and Lee 2003; Keller could help explain the prevailing regional variation in coral
and Donahue 2006) and to some extent from Biscayne Bay community structure in the Florida Keys (Hughes and Tan-
(Jones and Boyer 2002). Water quality measurements in ner 2000). It is well known that the recruitment of corals is
reefs within the Upper–Middle Keys were similar, perhaps aVected by several factors, including larval supply, substra-
because of the mixing of water masses in these areas tum availability, disturbance and regional oceanography
through wind-forced currents and gravity-driven transport (Chiappone and Sullivan 1996). Recently, Moulding (2005)
produced by cross-Key sea level diVerences (Smith and Lee demonstrated that the density and diversity of recruits var-
2003). The strength of these water exchanges exhibit sea- ies between regions, being signiWcantly lower in the Upper
sonal variations in response to prevailing winds and other Keys than in the Lower Keys. Finally, it is possible that the
oceanographic factors, for example, the Loop current (Kel- Upper Keys is exposed to severe environmental stresses
ler and Donahue 2006). The higher salinity and algal bloom and selective pressures due to the fact that this region
component scores (i.e., for chl a, TP and turbidity) in the is closer to the northern limit of the Florida reef tract
Lower Keys relative to the rest of the FKNMS could be (Moulding 2005).
attributed to the onshore movement of high-saline and Coral reef degradation in the Florida Keys was charac-
nutrient-rich deep water into this region (Leichter et al. terized by marked reduction in coral cover, and subsequent
2003) brought about by cyclonic gyres spun oV of the Flor- proliferation of macroalgae from 1996 to 2000. In the Flor-
ida Current (Szmant and Forrester 1996; Jones and Boyer ida Keys, the concurrent dominance of macroalgae relative
2002). Higher nutrient concentration in the Lower Keys to hard corals over time were documented previously by
could also have an anthropogenic origin, considering that several authors, including Dustan (1977), Dustan and Halas
the density of human population in the Lower Keys is (1987), Porter and Meier (1992), Murdoch and Aronson
higher than the rest of the Florida Keys (Ward-Paige et al. (1999), and Porter et al. (2002) (Table 3). Such dramatic
2005). Alternatively, the lower nutrient concentration in the changes in coral reef community structure are widely
Upper–Middle Keys relative to the Lower Keys could be reported in the Caribbean region (Done 1992; Hughes
attributed to the much higher likelihood of mixing between 1994; McClanahan and Muthiga 1998; Gardner et al. 2003,
waters in these regions and the relatively clean Atlantic 2005). This phenomenon is often considered an indicator of
Ocean waters (Boyer and Jones 2002). a coral reef community that is undergoing a phase-shift
The separation of reefs in the Lower Keys from the (McCook 1999; McManus and Polsenberg 2004; Rogers
Upper–Middle Keys in CCA corroborates the clustering of and Miller 2006). In the Florida Keys, the phase-shift from
reefs based on RSI, further illustrating the strong inXuence coral-dominated to macroalgal-dominated reef communi-
of the predictor variables (macroalgae, sponge, chl a, TN ties may be attributed to the diminished resilience of corals
and salinity) in the structuring of reef communities at a against perturbations (Jackson 2001; Aronson and Precht
spatial scale larger than the individual reef. This pattern is 2006), hypothesized to be caused by top-down mechanisms
consistent with the Wndings of Ogden et al. (1994) and such as herbivore reduction (Hughes 1994; Jackson et al.
Murdoch and Aronson (1999) who attributed the inter-reef 2001) and bottom-up processes such as nutrient enrichment
variations in coral community structure in the Florida Keys (Lapointe 1999; Leichter et al. 2003) (Fig. 7 for a diagram-
to the region-wide diVerences in environmental characteris- matic presentation of how we perceive these interacting
tics. The inXuence of environmental gradients in the struc- factors that result in a phase-shift).
turing of reef communities in the Florida Keys may be More recent evidence elucidates the potential mecha-
explained by several possible mechanisms. First, anteced- nisms underlying macroalgal-coral interactions that may
ent topography determine the suitability of present sub- lead to a localized phase-shift in the Florida Keys. Miller
strate for coral settlement and development. For example, and Hay (1998) demonstrated that Dictyota spp. and Hali-
Ginsburg and Shinn (1994) reported that the predominance meda opuntia, the two most abundant macroalgal species in
of mobile sand substrates in the Middle Keys and seaward the Florida Keys, inhibited the growth of Porites porites.
of Biscayne Bay may render these areas unsuitable for coral Jompa and McCook (2003a, b) have also shown that some
recruitment, thereby causing reefs in these areas to become algae (Anotrichium tenue, Corallophila huysmansii) can
impoverished. Second, Florida Bay waters that pass directly cause coral tissue death. Furthermore, KuVner et al.
through the Middle Keys may potentially inhibit coral (2006) have provided empirical evidence that macroalgae
growth, survival and recruitment into this region and adja- (e.g., Dictyota) can directly inhibit the settlement of coral
cent reefs by introducing pulses of waters with extremely recruits (e.g., Porites astreoides). The decline in coral
variable temperature and salinity as well as with high nutri- diversity and cover has also been associated with an
ent and sediment loads (Ginsburg and Shinn 1994; Chiap- increase in coral disease prevalence (Porter et al. 2001).
pone and Sullivan 1994; Szmant and Forrester 1996). Increased physical contact between corals and macroalgae
123
Table 3 Studies in the Caribbean region (Western Atlantic, including Florida and the Bahamas) documenting changes in benthic community structure in the region
Sites Period Mean % cover Proposed direct and indirect agents of change Source
Coral Macroalgae Sponge
Initial After Initial After Initial After
Caribbean region 1970 s–2000 s 50 10 Increasing Increasing White band disease; hurricanes; Gardner et al. (2003) and
indirectly attributed to increased López-Victoria
Mar Biol (2008) 154:841–853
abundance of macroalgae following and Zea (2005)
mass mortality of the herbivorous
D. antillarum(DA) in 1983
Florida Keys 1996–2000 8.1 4.6 5.7 9.6 (16.5 in 1998) 2.2 2.2 Disease; bleaching; anomalous extreme This study; also see
salinity and temperature events; eutrophication Porter et al. (2002)
and sedimentation; low herbivory due to
overWshing and mass mortality of DA;
high cover of macroalgae and probably sponges
Florida Keys (Carysfort Reef) 1975–1982 Boat grounding and anchor damage; hurricanes; Dustan and Halas (1987)
Mean percent cover (all species) 31.6 32.2 sedimentation; coral diseases; eutrophication
Acropora palmata 63 33
(mean colony size in cm)
Agaricia agaricites cm 4 6
(mean colony size in cm)
Florida Keys 1984–1991 28.4 6.6 Disease; bleaching Porter and Meier (1992)
(Carysfort and Looe Reefs) and Connell (1997)
Gulf of Mexico 1980–1990 55.2 8.7 Disease; bleaching Connell (1997)
(Flower Gardens Reef)
St. John, US Virgin Islands 1988–2003 43.4 12.3 Reported macroalgae competing Hurricane; high sea surface temperature (SST); Edmunds and Elahi (2007)
(Yawzi Point) with corals for space coral disease; macroalgae
St John, US Virgin Islands 1999–2000 (coral) Before- after 18 14 7.3 33.5 Hurricane; diseases; low herbivory due Rogers and Miller (2006)
Hurricane Hugo to overWshing and DA mass mortality;
in 1989 (macroalgae) failure of settlement
Panama (Punta de San Blas) 1983–1990 43–45 12–26 <2 10–32 Bleaching; DA mass mortality; sedimentation Shulman and
and eutrophication Robertson (1996)
Jamaica 1977–1980 vs. 1990–1993 52 3 4 92 OverWshing of herbvivores; hurricane; Hughes (1994)
disease; DA mass mortality
Jamaica (Dairy Bull) 1995–2004 <1 11 90 45–6 Long-lived colonies of Montastraea annularis Idjadi et al. (2006)
provided structural refugia
Belize (Glovers Reef lagoon) 1970–71 vs. 1996–1997 80 20 20 80 Diseases; low herbivory due to overWshing McClanahan and
and DA mass mortality Muthiga (1998)
Belize (Channel Cay) 1997–2001 »43 »0 <10 (no change) »15 43 Bleaching due to high SST Aronson et al. (2000, 2002)
(mainly Agaricia tenuifolia)
Costa Rica (Cahuita) 1980–1993 40.4 29.2 Natural disturbances; human impacts Connell (1997)
A majority of these studies indicate a decrease in coral cover (or a change in species composition), followed by subsequent increase in macroalgal and sponge cover over time
This list in not meant to be exhaustive. For worldwide summary of the impacts of diVerent array of disturbances, see Connell (1997) and Goreau et al. (2000)
123
849
850 Mar Biol (2008) 154:841–853
Fig. 7 Diagram showing an array of factors that potentially contrib- human disturbances. Regional-global factors include f increase sever-
uted to the phase-shift in Florida Keys coral reefs from coral-domi- ity and incidence of marine diseases, which is partly attributed to the
nated to algal/sponge-dominated community. Only dominant mass mortality of g Diadema antillarum in 1983–1984 and regional-
pathways are presented in the diagram. Solid and dashed lines indicate wide coral bleaching events; h global climate change, which has been
negative and positive interactions, respectively, and the arrowhead of associated with increased severity and frequency of anomalous high
the line indicates the trajectory of the interaction. Inside the dashed cir- SST causing regional-wide bleaching events; and i and historical and
cle are the dominant benthic organisms (corals, macroalgae and spong- natural disturbances such as hurricanes. Symbols courtesy of the Inte-
es). Inside the dashed square are array of local factors that potentially gration and Application Network (ian.umces.edu/symbols/), Univer-
contributed to the phase-shift in the Florida Keys coral reefs: a Wshing sity of Maryland Center for Environmental Science, USA. For a
indirectly impact corals through anchor damage and extraction of b comprehensive review of phase-shift in coral reefs, see Done (1992);
herbivores and c spongivores Wshes; d extreme seasonal changes of McCook (1999); McManus et al. (2000); McManus and Polsenberg
temperature and salinity; and e eutrophication, sedimentation and other (2004); and Aronson et al. (2006)
due to macroalgal bloom has been hypothesized to provoke in sponge cover from 1996 to 2000; this agrees with earlier
coral diseases (Green and Bruckner 2000). This observation reports of Aronson et al. (2002) and López-Victoria and
was empirically supported by the Wndings of Nugues et al. Zea (2005) in the Caribbean region. Cliona delitrix,
(2004) that the increasing physical contact between Hali- C. lampa, and C. caribboea, known to be aggressive
meda opuntia triggered a virulent disease known as white bioeroders, were among the dominant species of sponges in
plague type II in Montastraea faveolata. The combination the Florida Keys (Keller and Donahue 2006). Ward-Paige
of these mechanisms may have contributed to the negative et al. (2005) reported that the Lower Keys had the highest
relationship between coral and macroalgal cover in the sponge abundance and size, and they attributed this to
Florida Keys over the 5-year study period, as demonstrated sewage contamination based on higher 15 N levels in the
by the CCA. sponge tissues. Reduction in abundance of spongivorous
Sponge cover was also negatively correlated with hard Wshes as postulated by Hill (1998) could also play a role in
coral cover in the Florida Keys. There was a subtle increase the increase of sponge cover in the Florida Keys. Spongivory
123
Mar Biol (2008) 154:841–853 851
was previously demonstrated by Pawlik (1998) to be an factors (e.g., source of propagules), frequency of occurrence
important agent in controlling the population of at least of physical disturbance, and historical processes may inXu-
some sponges. Sponge proliferation could aVect corals ence the community structure of corals in the Florida Keys.
through several mechanisms: allelopathy (Engel and Future investigations that focus on mechanistic processes
Pawlik 2005; Pawlik et al. 2007), physical smothering and need to address whether these factors act alone or interac-
cellular digestion (Hill 1998) as well as direct space tively, either through mitigative or exacerbative pathways.
competition (Lopez-Victoria et al. 2006). These mecha-
nisms would lead to increased bioerosion, resulting in a net Acknowledgments R. J. Maliao is supported by the Fulbright-
Philippines Agriculture Scholarship Program. We thank the FKNMS
loss of carbonate, thus compromising the integrity of the
management through its coordinator, Fred McManus, for allowing us
coral skeletal structure. However, the positive and negative to use their monitoring database. Edward Webb, Kathe Jensen, J. R.
relationships between hard coral and sponge in the Wrst and Kerfoot, Matt Wittenrich, Zan Didoha, Vutheary Hean, Justin Anto,
second CCA axis, respectively, indicate that sponge-coral and Bernice Polohan provided valuable comments on an earlier version
of this manuscript. We particularly appreciate the advice of Bruce
relationships are species-speciWc. This pattern is consistent McCune on the use of CCA. The valuable comments and recommen-
with the Wndings of Aerts (1998) who reported that the out- dations of three anonymous reviewers greatly improved this manu-
come of coral-sponge interactions depends on the species script.
of both corals and sponges and the frequency of previous
encounters between coral and aggressive sponge species.
The positive relationships of both sponge and chl a with References
salinity is indicative of their wide tolerance to salinity Xuc-
tuations. Moreover, the negative relationship between mac- Aerts LAM (1998) Sponge/coral interactions in Caribbean reefs: anal-
roalgae and chl a is probably caused by chl a shading in the ysis of overgrowth patterns in relation to species identity and
water column, thereby reducing macroalgal productivity. cover. Mar Ecol Prog Ser 175:241–249
Aronson RB, Precht WF (2001) White-band disease and the changing
The chl a concentration in the FKNMS was around face of Caribbean coral reefs. Hydrobiologia 460(1–3):25–38
0.4 mg¡1, a value that can be considered eutrophic in reef Aronson RB, Precht WF (2006) Conservation, precaution, and Carib-
communities (Lapointe et al. 2004). In the biplot, there bean reefs. Coral Reefs 25:441–450
were no coral species that aggregated in the chl a space, Aronson RB, Precht WF, Macintrye IG (1998) Extrinsic control of spe-
cies replacement on a Holocene reef in Belize: the role of coral
consistent with the Wndings of van Woesik et al. (1999) and disease. Coral Reefs 17(3):223–230
Tomascik and Sander (1985), who demonstrated the nega- Aronson RB, Precht WF, Macintyre IG, Murdoch TJT (2000) Coral
tive correlation between chl a concentration and coral bleach-out in Belize. Nature 405:36
cover, growth and diversity. However, it is diYcult to sepa- Aronson RB, Precht WF, Toscano MA, Koltes KH (2002) The 1998
bleaching event and its aftermath on a coral reef in Belize. Mar
rate the eVects of chl a on the structuring of coral communi- Biol 141(3):435–447
ties because it is collinear with a variety of other water Bellwood DR, Hughes TP, Folke C, Nystrom M (2004) Confronting
quality variables. the coral reef crisis. Nature 429:827–833
In summary, results of our analyses revealed a strong Boyer JN, Jones RD (2002) A view from the bridge: external and inter-
nal forces aVecting the ambient water quality of the Florida Keys
correspondence between key predictor variables (e.g., mac- National Marine Sanctuary (FKNMS). In: Porter JW, Porter KG
roalgae, sponge and chl a) and characteristics of the hard (eds) The Everglades, Florida Bay, and coral reefs of the Florida
coral community structure in the Florida Keys. Water qual- Keys. CRC Press, Florida, pp 609–628
ity variables divide the reef tract of the Florida Keys into Carpenter RC (1990) Mass mortality of Diadema antillarum I. Long-
term eVects on sea urchin population-dynamics and coral reef
two major groups: the Upper–Middle Keys reefs, character- algal communities. Mar Biol 104(1):67–77
ized by lower values of water quality variables and the Chiappone M, Sullivan KM (1994) Patterns of coral abundance deWn-
Lower Keys, characterized by higher values of water qual- ing nearshore hardbottom communities of the Florida Keys. Fla
ity variables. The negative relationship between hard coral Sci 57:108–125
Chiappone M, Sullivan KM (1996) Distribution, abundance and spe-
cover and macroalgal/sponge cover suggests a phase-shift cies composition of juvenile Scleractinian corals in the Florida
in the coral reef structure in the Florida Keys from coral- reef tract. Bull Mar Sci 58(2):555–569
dominated to macroalgae and sponge-dominated communi- Clarke KR, Ainsworth M (1993) A method of linking multivariate
ties. However, the correlative nature of our analyses community structure to environmental variables. Mar Ecol Prog
Ser 92:205–219
precludes the identiWcation of precise causal mechanisms Clarke KR, Gorley RN (2006) PRIMER v6: User manual/tutorial.
that underlie the structuring of the hard coral community in PRIMER-E Ltd, Plymouth
the Florida Keys. It is also probable that the concurrent Connell JH (1997) Disturbance and recovery of coral assemblages.
increase of macroalgal and sponge cover in the Florida Coral Reefs 16:101–113
Done TJ (1992) Phase shifts in coral reef communities and their
Keys reXects an alternative state as a result, rather than the ecological signiWcance. Hydrobiologia 247:121–132
cause of coral demise. We recognize that other factors such Dustan P (1977) Vitality of reef coral populations oV Key Largo,
as resource availability, herbivory intensity, supply-side Florida: recruitment and mortality. Environ Geol 2:51–58
123
852 Mar Biol (2008) 154:841–853
Dustan P, Halas JC (1987) Changes in the reef-coral community of Keller BD, Donahue S (eds) (2006) 2002–03 sanctuary science report:
Carysfort Reef, Key Largo, Florida: 1974–1982. Coral Reefs an ecosystem report card after Wve years of marine zoning. US
6:91–106 Department of Commerce, National Oceanic and Atmospheric
Edmunds PJ, Elahi R (2007) The demographics of a 15-year decline in Administration, National Ocean Service, OYce of National
cover of the Caribbean reef coral Montastraea annularis. Ecol Marine Sanctuaries, Florida Keys National Marine Sanctuary,
Monogr 77:3–18 Marathon
Engel S, Pawlik JR (2005) Interactions among Florida sponges. I. Reef KuVner IB, Walters LJ, Becerro MA, Paul VJ, Ritson-Williams R,
habitats. Mar Ecol Prog Ser 303:133–144 Beach KS (2006) Inhibition of coral recruitment by macroalgae
Galindo HM, Olson DB, Palumbi SR (2006) Seascape genetics: a cou- and cyanobacteria. Mar Ecol Prog Ser 323:107–117
pled oceanographic-genetic model predicts population structure Lapointe BE (1999) Simultaneous top-down and bottom-up forces
of Caribbean corals. Curr Biol 16(16):1622–1626 control macroalgal blooms on coral reefs. Limnol Oceanogr
Gardner TA, Gill JA, Grant A, Watkinson AR (2003) Long-term 44:1586–1592
region-wide declines in Caribbean corals. Science 301:958–960 Lapointe BE, Matzie WR (1996) EVects of stormwater nutrient dis-
Gardner TA, Cote IM, Gill IA, Grant A, Watkinson AR (2005) Hurri- charges on eutrophication processes in nearshore waters of the
canes and Caribbean coral reefs: impacts, recovery, patterns, and Florida Keys. Estuaries 19:422–435
role in long-term decline. Ecology 86(1):174–184 Lapointe BE, Barile PJ, Matzie WR (2004) Anthropogenic nutrient
Ginsburg RN (ed) (1994) Proceedings of the colloquium on global as- enrichment of seagrass and coral reef communities in the Lower
pects of coral reefs: health, hazards and history. Rosenstiel School Florida Keys: discrimination of local versus regional nitrogen
of Marine and Atmospheric Science, University of Miami, Miami sources. J Exp Mar Biol Ecol 308:23–58
Ginsburg, RN, Shinn EA (1994) Preferential distribution of reefs in the Leichter JJ, Stewart HL, Miller SL (2003) Episodic nutrient transport
Florida reef tract: the past is the key to the present. In: Ginsburg to Florida coral reefs. Limnol Oceanogr 48:1394–1407
RN (ed) Proceedings of the colloquium on global aspects of coral López-Victoria M, Zea S (2005) Current trends of space monopoliza-
reefs: health, hazards and history. Rosenstiel School of Marine tion by encrusting excavating sponges on Colombian coral reefs.
and Atmospheric Science, University of Miami, pp 21–26 Mar Ecol 26:33–41
GraVelman J (2001) Quality statistics in canonical correspondence Lopez-Victoria M, Zea S, Weil E (2006) Competition for space
analysis. Environmetrics 12:485–497 between encrusting excavating Caribbean sponges and other coral
Green EP, Bruckner AW (2000) The signiWcance of coral disease epizo- reef organisms. Mar Ecol Prog Ser 312:113–121
otiology for coral reef conservation. Biol Conserv 96(3):347–361 McClanahan TR, Muthiga NA (1998) An ecological shift in a remote
Greenstein BJ, PandolW JM (1997) Preservation of community struc- coral atoll of Belize over 25 years. Environ Conserv 25(2):122–130
ture in modern reef coral life and death assemblages of the Florida McCook LJ (1999) Macroalgae, nutrients and phase shifts on coral
Keys: implications for the quaternary fossil record of coral reefs. reefs: scientiWc issues and management consequences for the
Bull Mar Sci 61:431–452 Great Barrier Reef. Coral Reefs 18:357–367
Hill MS (1998) Spongivory on Caribbean reefs releases corals from McCune B (1997) InXuence of noisy environmental data on canonical
competition with sponges. Oecologia 117:143–150 correspondence analysis. Ecology 78(8):2617–2623
Hughes TP (1994) Catastrophes, phase shifts, and large-scale degrada- McCune B, MeVord MJ (1999) PC-ORD ver. 5. Multivariate analysis
tion of a Caribbean coral reef. Science 265(5178):1547–1551 of ecological data. MjM Software, Gleneden Beach, Oregon
Hughes TP, Tanner JE (2000) Recruitment failure, life histories, and McCune B, Grace JB (2002) Analysis of ecological communities.
long-term decline of Caribbean corals. Ecology 81:2250–2263 MJM Software Design, Gleneden Beach, Oregon
Hughes TP, Baird AH, Bellwood DR, Card M, Connolly SR, Folke C, McManus JW, Polsenberg JF (2004) Coral-algal phase shifts on coral
Grosberg R, Hoegh-Guldberg O, Jackson JBC, Kleypas J, Lough reefs: ecological and environmental aspects. Prog Oceanogr
JM, Marshall P, Nystroem M, Palumbi SR, PandolW JM, Rosen B, 60:263–279
Roughgarden J (2003) Climate change, human impacts, and the McManus JW, Menez LAB, Kesner-Reyes KN, Vergara SG, Ablan
resilience of coral reefs. Science 301(5635):929–933 MC (2000) Coral reef Wshing and coral-algal phase shifts: impli-
Idjadi JA, Lee SC, Bruno JF, Precht WF, Allen-Requa L, Edmunds PJ cations for global reef status. ICES J Mar Sci 57:572
(2006) Rapid phase-shift reversal on a Jamaican coral reef. Coral Miller MW, Hay ME (1998) EVects of Wsh predation and seaweed
reefs 25:209–211 competition on the survival and growth of corals. Oecologia
Jackson JBC (2001) What was natural in the coastal oceans? Proc Natl 113:231–238
Acad Sci USA 98:5411–5418 Moulding AL (2005) Coral recruitment patterns in the Florida Keys.
Jackson JBC, Kirby MX, Wolfgang HB, Bjorndal KA, Botsford LW, Rev Biol Trop 25(Suppl 1):75–82
Bourque BJ, Bradbury RH, Cooke R, Erlandson J, Estes JA, Murdoch TJT, Aronson RB (1999) Scale-dependent spatial variability
Hughes TP, Kidwell S, Lange LB, Lenizan HS, PandolW JM, Pett- of coral assemblages along the Florida Reef Tract. Coral Reefs
erson CH, Steneck RS, Tegner MJ, Warner RR (2001) Historical 18:341–351
overWshing and the recent collapse of coastal ecosystems. Science Nugues MM, Smith GW, Hooidonk RJ, Seabra MI, Bak RPM (2004)
293:629–638 Algal contact as a trigger for coral disease. Ecol Lett 7:919–923
Jompa J, McCook LJ (2003a) Coral-algal competition: macroalgae Ogden JC, Porter JW, Smith NP, Szmant AM, Jaap WC, Forcucci D
with diVerent properties have diVerent eVects on corals. Mar Ecol (1994) A long-term interdisciplinary study of the Florida Keys
Prog Ser 258:87–95 seascape. Bull Mar Sci 54(3):1059–1071
Jompa J, McCook LJ (2003b) Contrasting eVects of turf algae on cor- Palmer MW (1993) Putting things in even better order: the advantages
als: massive Porites spp. are unaVected by mixed-species turfs, of canonical correspondence analysis. Ecology 74(8):2215–2230
but killed by the red alga Anotrichium tenue. Mar Ecol Prog Ser PandolW JM, Bradbury RH, Sala E, Hughes TP, Bjorndal KA, Cooke
258:79–86 RG, McArdle D, McClenachan L, Newman MJH, Paredes G,
Jones RD, Boyer JN (2002) FY2002 annual report of the water quality Warner RR, Jackson JBC (2003) Global trajectories of the long-
monitoring project for the water quality protection program in the term decline of coral reef ecosystems. Science 301(5635):955–
Florida Keys National Marine Sanctuary. Technical Report # 958
T192 of the Southeast Environmental Research Center, Florida Pawlik JR (1998) Coral reef sponges: do predatory Wshes aVect their
International University, Miami distribution? Limnol Oceanogr 43:1396–1399
123
Mar Biol (2008) 154:841–853 853
Pawlik JR, Steindler L, Henkel TP, Beer S, Ilan M (2007) Chemical Szmant AM, Forrester A (1996) Water column and sediment nitrogen
warfare on coral reefs: sponge metabolites diVerentially aVect and phosphorus distribution patterns in the Florida Keys, USA.
coral symbiosis in situ. Limnol Oceanogr 52(2):907–911 Coral Reefs 15:21–41
Petraitis PS, Dudgeon SR (2004) Detection of alternative stable states ter Braak CJF (1986) Canonical correspondence analysis: a new eigen-
in marine communities. J Exp Mar Biol Ecol 300:343–371 vector technique for multivariate direct gradient analysis. Ecol-
Porter JW, Meier OW (1992) QuantiWcation of loss and change of Flo- ogy 67:1167–1179
ridian reef coral populations. Am Zool 32:625–640 ter Braak CJF, Looman CWN (1994) Biplots in reduced-rank regres-
Porter JW, Dustan P, Jaap WC, Patterson KL, Kosmynin V, Meier sion. Biometrical J 36(8):983–1003
OW, Patterson ME, Parsons M (2001) Patterns of spread of coral ter Braak CFJ, Verdonschot PFM (1995) Canonical correspondence
disease in the Florida Keys. Hydrobiologia 460:1–24 analysis and related multivariate methods in aquatic ecology.
Porter JW, Kosmynin V, Patterson KL, Porter KG, Jaap WC, Wheaton Aquat Sci 57(3):255–289
JL, Hackett K, Lybolt M, Tsokos CP, Yanev G, Marcinek DM, Tomascik T, Sander F (1985) EVects of eutrophication on reef building
Dotten J, Eaken D, Patterson M, Meier OW, Brill M, Dustan P corals. I. EVects of eutrophication on the reef building coral Mon-
(2002) Detection of coral reef change by the Florida Keys coral tastraea annularis. Mar Biol 87:143–155
reef monitoring project. In: Porter JW, Porter KG (eds) The Ever- van Woesik R, Tomascik T, Blake S (1999) Coral assemblages and
glades, Florida Bay, and coral reefs of the Florida Keys: an eco- physico-chemical characteristics of the Whitsunday Islands: evi-
system sourcebook. CRC, Boca Raton, pp 749–769 dence of recent community changes. Mar Freshw Res 50:427–440
Quinn GP, Keough MJ (2006) Experimental design and data analysis Walther GR, Post E, Convey P, Menzel A, Parmesan C, Beebee TJC,
for biologists. Cambridge University Press, Cambridge Fromentin JM, Hoegh-Guldberg O, Bairlein F (2002) Ecological
Rakocinski CF, Lyczkowski-Shultz J, Richardson SL (1996) Ichtyoplank- responses to recent climate change. Nature 416:389–395
ton assemblage structure in Mississippi Sound as revealed by canon- Ward-Paige CA, Risk MJ, Sherwood OA, Jaap WC (2005) Clionid
ical correspondence analysis. Estuar Coast Shelf Sci 43:237–257 sponge surveys on the Florida reef tract suggest land-based nutri-
Rogers CS, Miller J (2006) Permanent ‘phase shifts’ or reversible ent inputs. Mar Pollut Bull 51:570–579
declines in coral cover? Lack of recovery of two coral reefs in Wheaton J, Jaap WC, Porter JW, Kosminyn V, Hackett K, Lybolt M,
St John, US Virgin Islands. Mar Ecol Prog Ser 306:103–114 Callahan MK, Kidney J, Kupfner S, Tsokos C, Yanev G (2001)
Shulman MJ, Robertson DR (1996) Changes in the coral reefs of San EPA/FKNMS Coral reef monitoring project: executive summary.
Blas, Caribbean Panama: 1983 to 1990. Coral Reefs 15:231–236 FKNMS symposium: an ecosystem report card, Washington DC
Smith NP, Lee TN (2003) Volume transports through tidal channels in
the middle Florida Keys. J Coast Res 19:254–260
123