Global Change Biology Global Change Biology (2015) 21, 1153–1164, doi: 10.1111/gcb.12792

Regional-scale dominance of non-framework building corals on Caribbean reefs affects carbonate production and future reef growth CHRIS T. PERRY1, ROBERT S. STENECK2, GARY N. MURPHY1, PAUL S. KENCH3, EVAN N. E D I N G E R 4 , S C O T T G . S M I T H E R S 5 and P E T E R J . M U M B Y 6 1 Geography, College of Life and Environmental Sciences, University of Exeter, Exeter EX11 1RP, UK, 2School of Marine Sciences, University of Maine, Darling Marine Center, Walpole, ME 04573, USA, 3School of Environment, The University of Auckland, Private Bag 92019, Auckland, New Zealand, 4Department of Geography and Department of Biology, Memorial University, St. John’s, NL A1B 3X9, Canada, 5College of Marine and Environmental Sciences, James Cook University, Townsville, Qld 4810, Australia, 6Marine Spatial Ecology Lab, School of Biological Sciences, University of Queensland, Brisbane, Qld 4072, Australia

Abstract Coral cover on Caribbean reefs has declined rapidly since the early 1980’s. Diseases have been a major driver, decimating communities of framework building Acropora and Orbicella coral species, and reportedly leading to the emergence of novel coral assemblages often dominated by domed and plating species of the genera Agaricia, Porites and Siderastrea. These corals were not historically important Caribbean framework builders, and typically have much smaller stature and lower calcification rates, fuelling concerns over reef carbonate production and growth potential. Using data from 75 reefs from across the Caribbean we quantify: (i) the magnitude of non-framework building coral dominance throughout the region and (ii) the contribution of these corals to contemporary carbonate production. Our data show that live coral cover averages 18.2% across our sites and coral carbonate production 4.1 kg CaCO3 m 2 yr 1. However, non-framework building coral species dominate and are major carbonate producers at a high proportion of sites; they are more abundant than Acropora and Orbicella at 73% of sites; contribute an average 68% of the carbonate produced; and produce more than half the carbonate at 79% of sites. Coral cover and carbonate production rate are strongly correlated but, as relative abundance of non-framework building corals increases, average carbonate production rates decline. Consequently, the use of coral cover as a predictor of carbonate budget status, without species level production rate data, needs to be treated with caution. Our findings provide compelling evidence for the Caribbean-wide dominance of non-framework building coral taxa, and that these species are now major regional carbonate producers. However, because these species typically have lower calcification rates, continued transitions to states dominated by non-framework building coral species will further reduce carbonate production rates below ‘predecline’ levels, resulting in shifts towards negative carbonate budget states and reducing reef growth potential. Keywords: Acropora, carbonate budgets, caribbean, coral reefs, Orbicella, reef growth Received 11 August 2014; revised version received 6 October 2014 and accepted 14 October 2014

Introduction Coral reefs globally have experienced major declines in coral cover over the past few decades (Bruno & Selig, 2007; Schutte et al., 2010) and, at many sites, this decline has been accompanied by a marked reduction in architectural complexity (Alvarez-Filip et al., 2009). Collectively, these changes are radically altering coral reef ecology and threatening future ecosystem functioning. These changes are increasingly observed globally, but are nowhere more evident than within the Caribbean basin (Alvarez-Filip et al., 2009; Schutte et al., 2010). Correspondence: Chris T. Perry, +44 (0)1332 723334, fax +44 (0) 1392 723342, e-mail: [email protected]

© 2014 John Wiley & Sons Ltd

Average coral cover on Caribbean reefs has declined from ~35% in the 1970’s to around 15% today (Gardner et al., 2003; Schutte et al., 2010; Jackson et al., 2014). Coral diseases have caused much of this decline, destroying communities of both branched Acropora species (Aronson & Precht, 2001) and massive coral taxa (Bruckner & Bruckner, 2006), including those belonging to the species complex Orbicella (= Montastraea, sensu Budd et al., 2012). These species were not only historically important space-occupiers in Caribbean fore-reef habitats to depths of about 15 m (Fig. 1a, b), but have also been major reef framework builders in the Caribbean since at least the Pleistocene (Pandolfi & Jackson, 2007). Chronic, long-term overfishing (Bellwood et al., 2004; Paddack et al., 2009), the disease-induced die-off 1153

1154 C . T . P E R R Y et al. (a)

(b)

(c)

(d)

(e)

(f)

Fig. 1 Common historical and contemporary coral species on Caribbean coral reefs. (a) Expansive stands of the Elkhorn coral Acropora palmata at ~4 m depth in St. Croix (1973); (b) Fore-reef site at ~15 m depth at Discovery Bay, Jamaica, dominated by expansive stands of the Staghorn coral Acropora cervicornis (1968); (c–f) Common non-framework building coral species that have become dominant on modern Caribbean reefs,(c) Millepora complanata, Grand Cayman (November 2013); (d) Agaricia agaricites (arrowed) colonising A. cervicornis rubble, Belize (November 2011); (e) Porites astreoides, Belize (November 2010); (f) Agaricia tenuifolia, Belize (November 2010). Image (A) by Robert Steneck, and image (B) by Eileen Graham and courtesy of Ken Johnson, Natural History Museum, London.

and lack of recovery of the key herbivorous sea urchin Diadema antillarum (Beck et al., 2014), together with local impacts from hurricanes, coral bleaching events and eutrophication, have exacerbated the rate and magnitude of these changes (Jackson et al., 2014). In addition to an overall loss of coral cover, an associated consequence, both in the Caribbean and at many sites globally, has been a reshaping of shallow water coral reef communities (Green et al., 2008; Alvarez-Filip et al., 2011a), with ‘novel’ assemblages (sensu Yakob & Mumby, 2011) emerging on many reefs. These are typically dominated by taxa that have not historically been important as major reef framework builders within fore-reef habitats. In the Caribbean, these include slowgrowing domed, plating and encrusting species of Agaricia, Porites astreoides and Siderastrea (Fig. 1c–f)

(Green et al., 2008; C^ ote & Darling, 2010), some of which have been described as “weedy” or opportunistic taxa (C^ ote & Darling, 2010; Darling et al., 2012), and digitate species of Porites and Madracis. These recent ecological changes are of interest for several reasons. Firstly, transitions of this type that have occurred synchronously over large spatial scales appear to be without precedent in the Holocene record of reef-building in the region (Aronson et al., 1998). Thus, the transitions to states dominated by non-framework building taxa observed at some sites over the past few decades (Green et al., 2008) may herald the start of a period of unique ecological change. Secondly, it has recently been argued that these taxa may have a higher probability of persisting through future disturbance regimes (C^ ote & Darling, 2010), and thus that they may become © 2014 John Wiley & Sons Ltd, Global Change Biology, 21, 1153–1164

C O R A L C O M M U N I T Y S H I F T S A N D F U T U R E R E E F G R O W T H 1155 palmata in depths less than ~6 m, and Acropora cervicornis at depths of between ~6–18 m), as well as declines in the abundance of the massive coral Orbicella spp. which, predisturbance, were collectively major space occupiers and high rate carbonate producers (Vecsei, 2001). In light of these findings, and in the context of reports of shifts to states of non-framework building coral taxa dominance at sites in the Caribbean (Green et al., 2008; Alvarez-Filip et al., 2011a,b; Yakob & Mumby, 2011), and other reefs globally (McClanahan et al., 2007; Rachello-Dolmen & Cleary, 2007; Graham et al., 2014), we use a newly assembled Caribbean-wide dataset to quantify the magnitude of non-framework building coral species dominance across the region and to examine levels of carbonate production by these species. Specifically, we address three main research questions: (i) How abundant are non-framework building coral taxa on modern shallow water Caribbean coral reefs; (ii) How much carbonate are these coral taxa producing compared to historically important Caribbean framework building species and (iii) What contribution are different non-framework building coral taxa now making to overall carbonate production rates on the regions shallow-water reefs? To address these questions we used quantitative survey data collected from 75 individual reefs located in 17 countries/islands (Fig. 2) that span the major Caribbean marine ecoregions (Spalding et al., 2007). We then combined the survey data with an existing database (see Perry et al., 2012) on coral linear extension rates and skeletal density to estimate whole reef coral carbonate production rates (kg CaCO3 m 2 yr 1), based on the absolute coverage of each species and their known calcification rates

increasingly important components of Caribbean coral reef communities. Thus the emergence of reefs dominated by such taxa may increasingly represent the new ecological norm on many Caribbean reefs (Yakob & Mumby, 2011). Thirdly, many of these non-framework building species also have low calcification rates (Alvarez-Filip et al., 2013) and this, in combination with an overall coral cover decline, has increased concerns that reef carbonate budgets will be reduced, negatively impacting future reef growth and habitat maintenance (Kennedy et al., 2013; Perry et al., 2013). Reef carbonate budgets are determined by the rate at which a reef produces and accumulates calcium carbonate, which is mostly a function of the amount of carbonate produced by corals and coralline algae, less that lost through physical removal or biological erosion (bioerosion) (Perry et al., 2008). The balance between these will strongly determine the carbonate budgetary status of a reef; where positive there is the potential for reef growth but, where net negative, erosion of the reef structure may occur (Eakin, 1996; Edinger et al., 2000; Mallela & Perry, 2007; Perry et al., 2008). From an ecological change perspective, it is pertinent to note that many of these production and erosion processes are themselves biologically driven, and thus are inherently susceptible to environmental perturbations. Indeed, a recent study of reef carbonate budget states at a range of sites across the Caribbean, reported significantly reduced rates of average net carbonate production compared to historical estimates, and substantial declines in rates of long-term reef accretion compared to Holocene averages from the region (Perry et al., 2013). These changes were in large part attributed to a decline in the abundance of fast-growing species of Acropora (Acropora

W 85o

Gd. Cayman Belize

Eleuthera, Bahamas St. Croix, US Virgin Islands Anguilla St. Maarten Barbuda Antigua

Caribbean Sea

40 km

Bonaire

N 15o

Canouan

St Lucia St Vincent

Union Tobago Cays

Bequia Mustique St. Vincent and the Grenadines

Petit St. Vincent Carriacou 10 km

Fig. 2 Location of study sites used in this study. Insets show details of southern Antilles sites and of sites in the St. Vincent and Grenadines islands. © 2014 John Wiley & Sons Ltd, Global Change Biology, 21, 1153–1164

1156 C . T . P E R R Y et al. (calcification rate being a function of linear grown and skeletal density).

Materials and methods Replicate benthic surveys (n ranging from 3 to 6) were conducted at each study site, with data collected from a total of 340 survey transects across a range of shallow fore-reef sites to depths of 20 m, although most data were collected at a depth of 10 m (see Table S1 for site survey details). Each site within a region or island was separated by at least 1 km. Data were collected over the following periods: Bonaire (2010), Belize (2011), Eleuthera (2012 and 2014), Grand Cayman (2012), and for the Eastern Caribbean sites in 2014. The range of countries included in our study inevitably occur within different regions with respect to prevailing wave energy/hurricane frequency (Chollett et al., 2012) and other disturbances, and thus some inherent variability in their background ecological condition, as a function of recent disturbance history, must be assumed. However, the ecological condition of most of the sites is considered representative of the spectrum of reef ecological states presently observed in shallow fore-reef habitats across much of the region (Jackson et al., 2014). At each site, quantitative data were collected to determine both benthic community composition and to measure biologically driven carbonate production rates. Data were also collected on substrate rugosity (as a function of total reef surface relative to linear distance) so that the distance covered by each benthic species could be determined in relation to the true 3-dimensional surface of the reef. This approach ensures that any ‘canopy effect’ bias (e.g., Goatley & Bellwood, 2011) from planar assessments is avoided. Along each transect we recorded the abundance of all corals (to species or morphological group level) along with all other space-occupying organisms. For all sites in The Bahamas, Belize, Bonaire and Grand Cayman benthic cover was measured as the surface area cover for each colony/substrate type beneath a 10 m linear transect line (following the ReefBudget methodology of Perry et al., 2012), whilst for the Eastern Caribbean sites benthic data were collected using an adapted version of the AGRRA method based on a point intercept methodology. The former method calculates species cover as a function of colony and substrate rugosity, whilst under the latter method we used transect averaged rugosity measures to quantify species cover. Other benthic cover categories included CCA and other calcareous encrusters (we also investigated areas of macroalgal cover to determine where living CCA occurred under the algal canopy); turf algae; macroalgae; non-encrusting calcareous algae (Halimeda spp. etc.) and sediment/rubble. We distinguish between four main groups of corals: (i) the historically important reef framework building Acropora and Orbicella species; (ii) domed, plating and encrusting forms of the genera Agaricia, Siderastrea and Porites astreoides (some species of which have been defined as ‘weedy’ or opportunistic; C^ ote & Darling, 2010; Darling et al., 2012); (iii) branching species of the genera Porites (mainly P. furcata and P. porites) and Madracis, which again some authors define as ‘weedy’(Darling et al., 2012) and which, with rare exceptions from lagoonal setttings

(Aronson et al., 2005), are also not historically important framework builders across the Caribbean and (iv) the hydrozoan Millepora spp., which is an important space occupier at some sites and which is commonly found encrusting and overgrowing other coral species in disturbed habitats (Lewis, 1989). We refer to the corals in groups 2–4 collectively as ‘nonframework builders’ in the text. All other corals e.g., of the genera Diploria, Colpophyllia etc. are grouped as ‘other’. To calculate rates of coral carbonate production by each coral species we used the ReefBudget methodology of Perry et al. (2012), integrating mean percent cover of each coral species with species-specific, or nearest equivalent species (see Perry et al., 2012) measures of skeletal density (g cm 3) and linear growth (cm yr 1) as derived from published sources. Regional variations in calcification rates at the species level are relatively poorly constrained at present and thus we acknowledge that variations around the mean calcification rates that we use may result for some species due to regional differences caused by, for example, sea-surface temperature variations (e.g., Carricart-Ganivet and Merino 2001). However, our approach allows us to use, where possible, same species (or at least same coral morphological group) data from comparable water depths in our calculations. Census data are combined with rugosity measures to yield a value for coral carbonate production (kg CaCO3 m 2 yr 1) relative to actual transect surface area. Details of supporting coral datasets are available on the ReefBudget website at: http://geography.exeter.ac.uk/reefbudget/.

Results Coral cover across our sites averaged 18.2%, with individual site averages ranging from 2.6% to 54.4% (Fig. 3a). Lowest country averages were in Eleuthera, Bahamas (4.9%) and the highest in St. Vincent (54.4%) and in the Tobago Cays (32.9%) (Fig. 3a). Cover of the historically important framework building species of Acropora averaged only 0.2% (range 0–5.5%) and of Orbicella only 4.4% (range 0–19.9%) (Fig. 3b, c). Acropora only occurred on transects at 10 of our 75 sites and only comprised >20% of the coral community at two sites, but was the single most abundant coral taxa at neither. Orbicella was present at 62 sites and the single most abundant coral group at 18 of the 75 sites. In contrast, cover of the domed and plating/encrusting non-framework building ‘group’ (comprising Agaricia spp., Porites astreoides and Siderastrea spp.) averaged 5.8% across all sites (range 0.2–29.6%). Corals of this group were present at all sites and comprised the most abundant components of the coral community at 42 of the 75 sites. Cover of the branching group of non-framework builders (the branched Porites and Madracis taxa) averaged 2.4% (range 0–27.6%). These coral species occurred on transects at 58 sites, but were the most abundant taxa at only 4 sites. Cover of Millepora spp. averaged 1.5% (range: 0–15.6%) (Fig. 3d, e, f), and although present at © 2014 John Wiley & Sons Ltd, Global Change Biology, 21, 1153–1164

1156 C . T . P E R R Y et al. (calcification rate being a function of linear grown and skeletal density).

Materials and methods Replicate benthic surveys (n ranging from 3 to 6) were conducted at each study site, with data collected from a total of 340 survey transects across a range of shallow fore-reef sites to depths of 20 m, although most data were collected at a depth of 10 m (see Table S1 for site survey details). Each site within a region or island was separated by at least 1 km. Data were collected over the following periods: Bonaire (2010), Belize (2011), Eleuthera (2012 and 2014), Grand Cayman (2012), and for the Eastern Caribbean sites in 2014. The range of countries included in our study inevitably occur within different regions with respect to prevailing wave energy/hurricane frequency (Chollett et al., 2012) and other disturbances, and thus some inherent variability in their background ecological condition, as a function of recent disturbance history, must be assumed. However, the ecological condition of most of the sites is considered representative of the spectrum of reef ecological states presently observed in shallow fore-reef habitats across much of the region (Jackson et al., 2014). At each site, quantitative data were collected to determine both benthic community composition and to measure biologically driven carbonate production rates. Data were also collected on substrate rugosity (as a function of total reef surface relative to linear distance) so that the distance covered by each benthic species could be determined in relation to the true 3-dimensional surface of the reef. This approach ensures that any ‘canopy effect’ bias (e.g., Goatley & Bellwood, 2011) from planar assessments is avoided. Along each transect we recorded the abundance of all corals (to species or morphological group level) along with all other space-occupying organisms. For all sites in The Bahamas, Belize, Bonaire and Grand Cayman benthic cover was measured as the surface area cover for each colony/substrate type beneath a 10 m linear transect line (following the ReefBudget methodology of Perry et al., 2012), whilst for the Eastern Caribbean sites benthic data were collected using an adapted version of the AGRRA method based on a point intercept methodology. The former method calculates species cover as a function of colony and substrate rugosity, whilst under the latter method we used transect averaged rugosity measures to quantify species cover. Other benthic cover categories included CCA and other calcareous encrusters (we also investigated areas of macroalgal cover to determine where living CCA occurred under the algal canopy); turf algae; macroalgae; non-encrusting calcareous algae (Halimeda spp. etc.) and sediment/rubble. We distinguish between four main groups of corals: (i) the historically important reef framework building Acropora and Orbicella species; (ii) domed, plating and encrusting forms of the genera Agaricia, Siderastrea and Porites astreoides (some species of which have been defined as ‘weedy’ or opportunistic; C^ ote & Darling, 2010; Darling et al., 2012); (iii) branching species of the genera Porites (mainly P. furcata and P. porites) and Madracis, which again some authors define as ‘weedy’(Darling et al., 2012) and which, with rare exceptions from lagoonal setttings

(Aronson et al., 2005), are also not historically important framework builders across the Caribbean and (iv) the hydrozoan Millepora spp., which is an important space occupier at some sites and which is commonly found encrusting and overgrowing other coral species in disturbed habitats (Lewis, 1989). We refer to the corals in groups 2–4 collectively as ‘nonframework builders’ in the text. All other corals e.g., of the genera Diploria, Colpophyllia etc. are grouped as ‘other’. To calculate rates of coral carbonate production by each coral species we used the ReefBudget methodology of Perry et al. (2012), integrating mean percent cover of each coral species with species-specific, or nearest equivalent species (see Perry et al., 2012) measures of skeletal density (g cm 3) and linear growth (cm yr 1) as derived from published sources. Regional variations in calcification rates at the species level are relatively poorly constrained at present and thus we acknowledge that variations around the mean calcification rates that we use may result for some species due to regional differences caused by, for example, sea-surface temperature variations (e.g., Carricart-Ganivet and Merino 2001). However, our approach allows us to use, where possible, same species (or at least same coral morphological group) data from comparable water depths in our calculations. Census data are combined with rugosity measures to yield a value for coral carbonate production (kg CaCO3 m 2 yr 1) relative to actual transect surface area. Details of supporting coral datasets are available on the ReefBudget website at: http://geography.exeter.ac.uk/reefbudget/.

Results Coral cover across our sites averaged 18.2%, with individual site averages ranging from 2.6% to 54.4% (Fig. 3a). Lowest country averages were in Eleuthera, Bahamas (4.9%) and the highest in St. Vincent (54.4%) and in the Tobago Cays (32.9%) (Fig. 3a). Cover of the historically important framework building species of Acropora averaged only 0.2% (range 0–5.5%) and of Orbicella only 4.4% (range 0–19.9%) (Fig. 3b, c). Acropora only occurred on transects at 10 of our 75 sites and only comprised >20% of the coral community at two sites, but was the single most abundant coral taxa at neither. Orbicella was present at 62 sites and the single most abundant coral group at 18 of the 75 sites. In contrast, cover of the domed and plating/encrusting non-framework building ‘group’ (comprising Agaricia spp., Porites astreoides and Siderastrea spp.) averaged 5.8% across all sites (range 0.2–29.6%). Corals of this group were present at all sites and comprised the most abundant components of the coral community at 42 of the 75 sites. Cover of the branching group of non-framework builders (the branched Porites and Madracis taxa) averaged 2.4% (range 0–27.6%). These coral species occurred on transects at 58 sites, but were the most abundant taxa at only 4 sites. Cover of Millepora spp. averaged 1.5% (range: 0–15.6%) (Fig. 3d, e, f), and although present at © 2014 John Wiley & Sons Ltd, Global Change Biology, 21, 1153–1164

Coral carbonate production (G)

1158 C . T . P E R R Y et al. Gd Cayman

Eleuthera, Bahamas

10 m

5m

Other

5.0

5.0

4.0

4.0

3.0

3.0

3.0

2.0

2.0

2.0 5 m

1.0

1.0

1.0

0

10.0

Coral carbonate production (G)

10 m

9.0 Belize

9.0

SW5 TR5 CR10 CR5 TR10 SW10

0

5.0

DF5

0

BASP PA

Acropora spp.

SD

DF10

CP

BT CR HW SS

Anguilla

St. Maarten 10.0 Barbuda

5m

2.5 m

Non-framebuilding domed and plating corals Orbicella spp.

15-20 m 10 m

10 m

St.Croix

Millepora spp. Non-framebuilding branched corals

4.0

5m

8.0

All 10 m

All 10 m

9.0 Antigua All 10 m

8.0

7.0

10 m

7.0

7.0

7.0

7.0

6.0

6.0

6.0

6.0

6.0

5.0

5.0

5.0

5.0

5.0

4.0

4.0

4.0

4.0

4.0

3.0

3.0

3.0 5 m

3.0

3.0

2.0

2.0

2.0

2.0

2.0

1.0

1.0

1.0

1.0

1.0

0

0

WF2.5 BN5 WF5 WF10 BS10 BS2.5 BS5 BN10 CV

SR

20.0

St. Lucia 19.0

0 FRPRMZ

0 GI1 GR1 SPR PL2 GR2 PL1

12.0

12.0

11.0

11.0

10.0

10.0

10.0

9.0

9.0

9.0

8.0

8.0

8.0

7.0

7.0

7.0

6.0

6.0

6.0

5.0

5.0

5.0

4.0

4.0

4.0

3.0

3.0

3.0

2.0

2.0

2.0

1.0

1.0

1.0

0

0 TU PP

AR CG1 SU CG2 AC

10 m 5m

11.0

FY

WW LB GI3 SC

17.0

12.0

MA

GI2

Bonaire All 10 m

13.0

Coral carbonate production (G)

NS

St. Vincent and the Grenadines

14.0

All 10 m

0

SAN SAS

0 BL

FR

DT

DR

ST

LA PL

CS PIL

St. Vincent Bequia Mustique Canoan

PR

HSN HSR MY

NR

MO

WI WB

CI

WH ND10 ND5 CA

Tobago Union Carriacou Cays & PSV

Fig. 4 Rates of coral carbonate production (G, where G = kg CaCO3 m2 yr 1) by Acopora spp, Orbicella spp., and by different groups of non-framework building corals: domed, plating and encrusting forms of the genera Agaricia, Porites astreoides and Siderastrea; branching species of the genera Porites and Madracis; the hydrozoan Millepora spp.; and all ‘other’ corals at reef sites across the Caribbean. See Fig. 3 caption for site codes. At the three sites where ‘other’ species make a significant relative contribution to carbonate production, the relevant species were: Diploria at site BS2.5 (St. Croix), Diploria and Colpophyllia at site BL in St. Vincent, and Diploria at site CI in Bonaire.

58 sites Millepora spp. were the most abundant species at only three sites. Collectively, the non-framework building corals represented the most abundant components of the coral communities at 55 of our 75 study sites (or 73% of sites). Furthermore, they contribute to >50% of the coral cover at 44 of 75 sites (59%). However, analysis of these data also indicates the overall abundance of different non-framework building taxa varies between sites across the region. In decreasing

order of abundance as regional averages, these are: P. astreoides 3.5% (range 0–18.1%), Agaricia spp. 1.8% (range 0–12.9%); Siderastrea spp. (mainly S. siderea) 1.8% (range 0–14.6%); branched Porites spp. 1.7% (range 0–26.9%), Millepora spp. 1.4% (range 0–15.6%) and branched Madracis spp. 0.6% (range 0–7.8%). These data provide compelling evidence from sites across the entire Caribbean region, and at different shallow water fore-reef sites, for the dominance of coral © 2014 John Wiley & Sons Ltd, Global Change Biology, 21, 1153–1164

C O R A L C O M M U N I T Y S H I F T S A N D F U T U R E R E E F G R O W T H 1159 taxa that were not historically important Caribbean reef framework builders, and especially of the domed and plating/encrusting species of Agaricia, Siderastrea and Porites astreoides. Our data also show that given their relative dominance on the regions reefs, this same group of corals now have a dominant influence on contemporary rates of coral carbonate production. This is despite the fact that calcification rates vary markedly among taxa. Carbonate production across our entire dataset averaged 4.1 kg CaCO3 m 2 yr 1, ranging from individual reef site averages of 0.2 to 16.2 kg CaCO3 m 2 yr 1 (Fig. 4). The lowest rates were at sites in Eleuthera and Barbuda, the highest from individual localities in St. Lucia, Carriacou and Bonaire (Fig. 4). These rates are a function of both the abundance and the calcification rates of different coral taxa, but our data show that the collective group of non-framework building corals produce more carbonate than species of Acropora and Orbicella at 62 of the 75 sites (or 83%). The domed and plating/encrusting group of non-framework builders produce more carbonate alone than Acropora and Orbicella at 38 of our 75 study sites (51%). Furthermore, at a significant number of sites (17 of 75 sites, or 23%) species of branched Porites and Madracis also produce more carbonate than species of Acropora or Orbicella, as does Millepora spp. at 31 of 75 sites (41%). Rates of carbonate production across all sites average 0.8 kg CaCO3 m 2 yr 1 for the domed and plating non-framework building group (range 0.02– 3.0 kg CaCO3 m 2 yr 1), and for the branched group of non-framework builders 1.43 kg CaCO3 m 2 yr 1 (range 0–15.1 kg CaCO3 m 2 yr 1), compared to regional averages of 0.05 kg CaCO3 m 2 yr 1 for species of Acropora and 1.05 kg CaCO3 m 2 yr 1 for Orbicella spp. Millepora production averages 0.6 kg CaCO3 m 2 yr 1 (range 0–6.4 kg CaCO3 m 2 yr 1). We note that species of Orbicella are the most important carbonate producers on only 20 of our 75 reefs (26%), whilst species of Acropora contribute to >20% of total coral carbonate production at only three sites [Seal Reef, Anguilla (A. palmata contributes to 29% of carbonate production), Tobacco Reef 5 m site, Belize, (A. palmata contributes to 34% of carbonate production) and Prostelyte Reef, St Maarten (A. cervicornis and A. palmata contribute to 21% of carbonate production)]. We also note considerable variation in the overall contributions made by different non-framework building coral taxa to carbonate production rates and some apparent geographic variations in terms of dominant producers. In decreasing order of abundance of average carbonate production rates at the site level, these are: branched Porites 1.4 kg CaCO3 m 2 yr 1 (range 0– 14.7), Millepora spp. 0.6 kg CaCO3 m 2 yr 1 (range 0– 6.4); P. astreoides 0.4 kg CaCO3 m 2 yr 1 (range 0–1.9); © 2014 John Wiley & Sons Ltd, Global Change Biology, 21, 1153–1164

branched Madracis spp. 0.3 kg CaCO3 m 2 yr 1 (range 0–4.8), Agaricia spp. 0.2 kg CaCO3 m 2 yr 1 (range 0–1.0) and Siderastrea spp. 0.2 kg CaCO3 m 2 yr 1 (range 0–2.6). Species of Agaricia and Porites astreoides are important contributors to carbonate production at most sites (reflecting their high abundances at most sites –see above), but in relative terms are especially important at sites located in western, central and north-eastern areas of the Caribbean (Belize, Grand Cayman, Eleuthera, St. Croix), whilst Millepora make an especially significant contribution at sites in the northern part of the Antilles (St. Maarten, Barbuda and Antigua). In contrast, at sites in the southern part of the Antilles, species of branched Porites and Madracis are commonly important carbonate producers (Fig. 5).

Discussion The ecological and physical structure of Caribbean coral reefs has markedly changed since the early 1980’s. Our benthic ecological data, based on surveys from 75 sites spanning a wide geographic area of the Caribbean, confirms that widespread transition to states of low live coral cover have occurred at most sites, consistent with reported regional trends (Schutte et al., 2010; Jackson et al., 2014). Indeed, average coral cover across our sites, which is 18.2%, closely corresponds to other recent estimates of 16.0% (Schutte et al., 2010) and 16.8% (Jackson et al., 2014). The majority of sites with average coral cover >30% were in the southern Antilles and southern Caribbean (Bonaire). Most significantly, our data also show that non-framework building coral taxa, comprising species of the genera Agaricia, Madracis, Millepora, Porites and Siderastrea, now collectively represent the dominant components of the coral communities i.e., collectively account for the highest proportion of coral cover, at a very high proportion of sites (73%; see Fig. 3). In contrast, both elkhorn (A. palmata) and staghorn (A. cervicornis) corals, and species of Orbicella, which were major reef framework builders in Caribbean forereef environments throughout the Holocene (Macintyre & Glynn, 1976; Macintyre et al., 1977; Gischler & Hudson, 2004; Hubbard et al., 2005), are now relatively minor components of most shallow fore-reef communities. Orbicella spp. remain important components of coral assemblages at a number of sites, especially at sites in the southern Antilles and southern Caribbean (Fig. 3), but contribute to >20% of coral cover at only 33 sites, and are the single most important group of corals at only 18 of the 75 sites. Acropora spp. were present on transects at only nine sites, but only comprise an average 7.6% of the coral cover within these sites. These

Coral carbonate production (G)

1158 C . T . P E R R Y et al. Gd Cayman

Eleuthera, Bahamas

10 m

5m

Other

5.0

5.0

4.0

4.0

3.0

3.0

3.0

2.0

2.0

2.0 5 m

1.0

1.0

1.0

0

10.0

Coral carbonate production (G)

10 m

9.0 Belize

9.0

SW5 TR5 CR10 CR5 TR10 SW10

0

5.0

DF5

0

BASP PA

Acropora spp.

SD

DF10

CP

BT CR HW SS

Anguilla

St. Maarten 10.0 Barbuda

5m

2.5 m

Non-framebuilding domed and plating corals Orbicella spp.

15-20 m 10 m

10 m

St.Croix

Millepora spp. Non-framebuilding branched corals

4.0

5m

8.0

All 10 m

All 10 m

9.0 Antigua All 10 m

8.0

7.0

10 m

7.0

7.0

7.0

7.0

6.0

6.0

6.0

6.0

6.0

5.0

5.0

5.0

5.0

5.0

4.0

4.0

4.0

4.0

4.0

3.0

3.0

3.0 5 m

3.0

3.0

2.0

2.0

2.0

2.0

2.0

1.0

1.0

1.0

1.0

1.0

0

0

WF2.5 BN5 WF5 WF10 BS10 BS2.5 BS5 BN10 CV

SR

20.0

St. Lucia 19.0

0 FRPRMZ

0 GI1 GR1 SPR PL2 GR2 PL1

12.0

12.0

11.0

11.0

10.0

10.0

10.0

9.0

9.0

9.0

8.0

8.0

8.0

7.0

7.0

7.0

6.0

6.0

6.0

5.0

5.0

5.0

4.0

4.0

4.0

3.0

3.0

3.0

2.0

2.0

2.0

1.0

1.0

1.0

0

0 TU PP

AR CG1 SU CG2 AC

10 m 5m

11.0

FY

WW LB GI3 SC

17.0

12.0

MA

GI2

Bonaire All 10 m

13.0

Coral carbonate production (G)

NS

St. Vincent and the Grenadines

14.0

All 10 m

0

SAN SAS

0 BL

FR

DT

DR

ST

LA PL

CS PIL

St. Vincent Bequia Mustique Canoan

PR

HSN HSR MY

NR

MO

WI WB

CI

WH ND10 ND5 CA

Tobago Union Carriacou Cays & PSV

Fig. 4 Rates of coral carbonate production (G, where G = kg CaCO3 m2 yr 1) by Acopora spp, Orbicella spp., and by different groups of non-framework building corals: domed, plating and encrusting forms of the genera Agaricia, Porites astreoides and Siderastrea; branching species of the genera Porites and Madracis; the hydrozoan Millepora spp.; and all ‘other’ corals at reef sites across the Caribbean. See Fig. 3 caption for site codes. At the three sites where ‘other’ species make a significant relative contribution to carbonate production, the relevant species were: Diploria at site BS2.5 (St. Croix), Diploria and Colpophyllia at site BL in St. Vincent, and Diploria at site CI in Bonaire.

58 sites Millepora spp. were the most abundant species at only three sites. Collectively, the non-framework building corals represented the most abundant components of the coral communities at 55 of our 75 study sites (or 73% of sites). Furthermore, they contribute to >50% of the coral cover at 44 of 75 sites (59%). However, analysis of these data also indicates the overall abundance of different non-framework building taxa varies between sites across the region. In decreasing

order of abundance as regional averages, these are: P. astreoides 3.5% (range 0–18.1%), Agaricia spp. 1.8% (range 0–12.9%); Siderastrea spp. (mainly S. siderea) 1.8% (range 0–14.6%); branched Porites spp. 1.7% (range 0–26.9%), Millepora spp. 1.4% (range 0–15.6%) and branched Madracis spp. 0.6% (range 0–7.8%). These data provide compelling evidence from sites across the entire Caribbean region, and at different shallow water fore-reef sites, for the dominance of coral © 2014 John Wiley & Sons Ltd, Global Change Biology, 21, 1153–1164

C O R A L C O M M U N I T Y S H I F T S A N D F U T U R E R E E F G R O W T H 1161 Our findings thus strongly support the hypothesis that the major coral species driving carbonate production on Caribbean reefs have changed dramatically since the early 1980’s, when species of Acropora (A. palmata and A. cervicornis), as well as species of Orbicella, were not only major space occupiers on Caribbean reefs, but were also the major carbonate producers (Vecsei, 2001). In terms of reef carbonate production and biological carbonate budgets, the loss of Acropora species has been especially significant. Extension rates for these branched corals are up to 9–10 cm per year (Bak, 1976; Gladfelter et al., 1978; Gladfelter, 1984) and, where abundant, these species contribute to high rates of carbonate production (probably in excess of 10 kg CaCO3 m 2 yr 1; Kinsey, 1981; Vecsei, 2001), and historically were capable of driving high rates of vertical reef accretion through the Holocene (Macintyre et al., 1977). In previous net carbonate budget assessments from the Caribbean we identified a critical live coral cover threshold of ~10%, below which overall reef budget states typically become net negative, threatening reef growth potential (Perry et al., 2013). Here we also identify, using an expanded dataset encompassing a far greater number of reefs from across the region, a strong correlation between coral carbonate production rate and live coral cover (Fig. 6a), regardless of which groups of corals are most dominant on the reefs. The low coral cover states we report from many reefs are thus likely a strong influence on the relatively low coral carbonate production rates derived from our data (average 4.1 kg CaCO3 m 2 yr 1). However, because different coral species have different calcification rates, shifts in relative coral species abundance and thus in the assemblages of corals that are most prevalent at a site, will also dictate the magnitude of coral carbonate production decline. This is reflected in the different linear regressions generated as a function of whether sites were dominated either by species of Orbicella, or by the domed and encrusting group of species of Agaricia, Porites astreoides and Siderastrea (the two ‘groups’ that collectively dominated the coral assemblages at 66 of the 75 sites) (Fig. 6b). Across the range of sites we examined, a clear trend of declining overall rates of coral carbonate production was also observed as the relative abundance of the non-framework building coral taxa increased (Fig. 6c). Thus, with some confidence, we draw the conclusion that as non-framework building coral taxa become increasingly important on Caribbean reefs so average rates of carbonate production will also generally decline. However, perhaps the central issue here is to what extent are the low carbonate production rates revealed in our dataset simply a function of the © 2014 John Wiley & Sons Ltd, Global Change Biology, 21, 1153–1164

cumulative effects of an overall decline in live coral cover and a relative increase in abundance of nonframework building coral taxa? As noted above, there is compelling evidence in this and other site-specific studies that non-framework building species have clearly increased in relative abundance on Caribbean reefs (e.g., Green et al., 2008) and, that based on analysis of event deposits (Scoffin & Hendry, 1984), core records (Aronson et al., 1998) and Holocene (Greer et al., 2009) and Pleistocene (Perry, 2001) reef outcrop studies, this is an apparently novel state within the Caribbean. Although we cannot determine whether non-framework building coral abundance has increased in absolute terms on a site-by-site basis, our data do indicate a strong positive correlation between live coral cover and absolute cover of these taxa (Fig. 6d), and this observation has two important implications. First, that non-framework building coral taxa dominance is not simply a function of the generally low coral cover states now observed on many Caribbean reefs and, secondly, that the dominance of these coral taxa does not inherently limit the occurrence of relatively high live coral cover states. Coral cover reached 55% in our study, with the highest coral cover states occurring on reefs dominated by the domed and plating group of low relief non-framework building coral taxa (Fig. 6a). However, because these taxa typically have lower rates of calcification in shallow water than species of Acropora and Orbicella (Huston, 1985) this means that even sites with relatively high cover of living corals can still have low overall production rates, where that cover is dominated by these low relief non-framework building species. This finding has two further key implications. Firstly, that the use of total coral cover alone, as a predictor of reef functionality from a carbonate production perspective, needs to be treated with caution. Threshold levels will be as much a function of coral community composition as of overall coral cover. Data on how these two factors interact in a budgetary context are largely absent, but our findings point to a critical need to better understand the implications of different coral cover and coral community states to determine desirable carbonate production regimes. Secondly, that even transitions to states of relatively high non-framework building coral cover will not ensure that reefs maintain positive budgetary states. This suggestion is supported by recent conceptual modelling approaches (AlvarezFilip et al., 2013) that predicted a general lowering of carbonate production rates as non-framework building species dominance increased. Our data, collected from a wide range of Caribbean reef sites and habitats, not

1162 C . T . P E R R Y et al. (b) 18 16 14 12 10 8 6 4 2 0

Reefs dominated by:

Coral carbonate production (kg CaCO3 m2 yr–1)

Coral carbonate production (kg CaCO3 m2 yr–1)

(a) Orbicella sp. Domed/plating species of Agaricia, Porites and Siderastrea Other corals

0

10

20

30

40

50

60

18 16 14 12 10 8 6 4 2 0

Reefs dominated by: Orbicella sp. Domed/plating species of Agaricia, Porites and Siderastrea

0

10

Live coral cover (%)

(c)

30

40

50

60

50

60

(d) 18 16 14 12 10 8 6 4 2 0

35

Weedy coral cover (%)

Coral carbonate production (kg CaCO3 m2 yr–1)

20

Live coral cover (%)

0

20

40

60

80

100

Weedy coral cover as proportion of total coral cover (%)

30 25 20 15 10 5 0

0

10

20

30

40

Live coral cover (%)

Fig. 6 (a) The linear regression and 95% confidence interval for the relationship between coral cover and coral carbonate production at 75 sites across the Caribbean (R² = 0.5853, y = 0.2289x, P < 0.01). Sites are differentiated into those dominated by either Orbicella spp., by the domed and encrusting group of species of Agaricia, Porites and Siderastrea, or by ‘other’ corals, mainly the branched species of Porites and Madracis, or by Diploria spp. Acropora spp. did not dominate at any site; (b) The linear regression and 95% confidence intervals for the relationship between coral cover and coral carbonate production at sites dominated by the two most common corals or group of corals; Orbicella spp. (R² = 0.5376, y = 0.282x, P < 0.01), and the domed and encrusting group of species of Agaricia, Porites and Siderastrea (R² = 0.7509, y = 0.2107x, P < 0.01); (c) The linear regression and 95% confidence interval for the relationship between relative % of domed and plating non-framework building corals and coral carbonate production (R² = 0.2321, y = 0.0539x + 7.2579, P < 0.05); (d) The linear regression and 95% confidence interval for the relationship between coral cover and % cover of non-framework building corals (R² = 0.4665, y = 0.3969x + 1.4322, P < 0.01).

only confirms the widespread dominance of such coral taxa in most fore-reef habitats and on most reefs across the region, but also that these non-framework builders are becoming increasingly important as reef carbonate producers. The findings of this study also have direct relevance to wider concerns about future reef functionality. These relate to the integrity and maintenance of habitat architectural complexity, and to future reef growth potential. The domed, plating and encrusting non-framework builders that are becoming so important on Caribbean reefs today also have the lowest levels of morphological rugosity (Alvarez-Filip et al., 2011b). Thus their increasing dominance has implications both for carbonate production rates and for habitat complexity (see also Alvarez-Filip et al., 2013 for modelled scenarios of such changes), with combined knock-on effects on ecological processes and thus for many key reef ecosystem services. This is especially the case where key reef-associated species depend upon the intricate structures that more morphologically complex coral species (i.e., branched corals) provide (e.g., Rogers et al., 2014). Per-

sistence of the presently observed coral cover and coral community states we report are thus likely to have important implications not only for future reef carbonate budgets but also for multiple aspects of reef functionality that depend on the maintenance of reef framework structures. The general and significant lowering of carbonate production rates we report, that are linked to fundamental changes in coral species type and abundance, also have major implications for future reef growth potential. Converting carbonate production rates to long-term reef growth (accretion) rates is not a simple issue (Hubbard et al., 1990). In previous work, and using a small subset of the data we present here, we estimated that current rates of vertical reef growth potential on reefs in the Caribbean averaged only ~ 0.6 mm yr 1 for sites at 5 m depth, and 2.1 mm yr 1 at 10 m depth (Perry et al., 2013). These were derived from sites with average carbonate production rates of 4.6 and 3.1 kg CaCO3 m 2 yr 1 at the respective water depths, and were considered highly optimistic estimates of accretion given that no account could be taken for epi© 2014 John Wiley & Sons Ltd, Global Change Biology, 21, 1153–1164

1160 C . T . P E R R Y et al. Coral carbonate production (G)

Belize

Gd Cayman

5m

Eleuthera, Bahamas

10 m

3 2

5m 10 m

15-20 m

5m

1 0

Branched Madracis spp. (M. aurentenra/M. decactis) Branched Porites spp. (P. porites/P. furcata) Millepora spp. Siderastrea spp. Porites astreoides Agaricia spp.

10 m

SW5 TR5 CR10 CR5 TR10 SW10

DF5

BA PA

DF10

SD

SP

BT CP

CR HW SS

St. Maarten St. Croix

All 10 m

Coral carbonate production (G)

6 5m

5 4

2.5 m

3

10 m

2

Coral carbonate production (G)

Antigua All 10 m

10 m

1 0

Barbuda All 10 m

Anguilla

5m WF2.5 BN5 WF5 WF10 BS10 BS2.5 BS5 BN10 CV

SR

SAN

FR PR MZ

SAS

GR1 SPR PL2 GR2 PL1

8

St. Lucia

St. Vincent and the Grenadines

7

All 10 m

All 10 m

GI1 NS

GI2 WW LB GI3 SC

Bonaire 10 m

6 5 4 5m

3 2 1 0

MA

FY

TU

PP

AR

CG1 SU CG2 AC

St. Vincent

BL

FR

ST

Bequia

DT

DR

PL

LA

PIL

CS

Mustique Canoan

PR

HSR

HSN MY

Tobago Cays

NR

MO

WI WB

CI

WH ND10 ND5 CA

Union Carriacou & PSV

Fig. 5 Rates of coral carbonate production (G, where G = kg CaCO3 m2 yr 1) by individual species of non-framework building coral taxa, Agaricia spp. Porites astreoides, Siderastrea spp., branched Porites spp. and branched Madracis spp., and the hydrozoan Millepora spp., at reef sites across the Caribbean. See Fig. 3 caption for site codes.

findings not only confirm localized reports of increases in the relative abundance of non-framework building coral taxa, mainly of the genera Agaricia, Porites and Siderastrea, measured at several Caribbean sites (Green et al., 2008; Alvarez-Filip et al., 2011a, 2011b), but importantly provide clear evidence that these transitions have occurred across the entire region. It is difficult to quantitatively determine whether the abundance of such non-framework building coral taxa has changed in absolute terms across our sites due to a paucity of baseline data. However, our data provides strong evidence that at least their relative abundance has markedly increased in many areas and that this state appears to be without precedent in the recent geological history of reef-building in the region (Aronson et al., 1998; Perry, 2001). A pressing question with both ecological and geologic relevance arising from these observations, is what impact are these changes having on rates of carbonate production on Caribbean reefs? Three key factors are important in determining rates of coral carbonate production at the reef scale. The first is live coral cover, the second is the abundance of different

coral species, and the third is the rate of coral calcification by each of those coral species present at any given site. It follows that any overall lowering of coral carbonate production rates may occur in response to changes in any one of these parameters. As discussed above, coral cover has certainly declined precipitously on Caribbean reefs since the early 1980’s (Jackson et al., 2014), and the rise of macroalgae resulting from both the loss of dominant corals and fisheries-induced declines of herbivores, contributes to locking coral reefs into these low coral cover conditions (e.g., Steneck et al., 2014). These changes, which also affect reef habitat complexity (Alvarez-Filip et al., 2013), have profound implications for a wide range of key ecological functions on reefs, including fisheries (Rogers et al., 2014), sediment generation (Perry et al., 2011) and habitat maintenance (Saunders et al., 2014). Carbonate production rate data presented here demonstrate that such changes in functionality are likely to encompass major changes in rates and patterns of biologically driven reef framework carbonate production, with implications for the future maintenance of reef framework structures. © 2014 John Wiley & Sons Ltd, Global Change Biology, 21, 1153–1164

1162 C . T . P E R R Y et al. (b) 18 16 14 12 10 8 6 4 2 0

Reefs dominated by:

Coral carbonate production (kg CaCO3 m2 yr–1)

Coral carbonate production (kg CaCO3 m2 yr–1)

(a) Orbicella sp. Domed/plating species of Agaricia, Porites and Siderastrea Other corals

0

10

20

30

40

50

60

18 16 14 12 10 8 6 4 2 0

Reefs dominated by: Orbicella sp. Domed/plating species of Agaricia, Porites and Siderastrea

0

10

Live coral cover (%)

(c)

30

40

50

60

50

60

(d) 18 16 14 12 10 8 6 4 2 0

35

Weedy coral cover (%)

Coral carbonate production (kg CaCO3 m2 yr–1)

20

Live coral cover (%)

0

20

40

60

80

100

Weedy coral cover as proportion of total coral cover (%)

30 25 20 15 10 5 0

0

10

20

30

40

Live coral cover (%)

Fig. 6 (a) The linear regression and 95% confidence interval for the relationship between coral cover and coral carbonate production at 75 sites across the Caribbean (R² = 0.5853, y = 0.2289x, P < 0.01). Sites are differentiated into those dominated by either Orbicella spp., by the domed and encrusting group of species of Agaricia, Porites and Siderastrea, or by ‘other’ corals, mainly the branched species of Porites and Madracis, or by Diploria spp. Acropora spp. did not dominate at any site; (b) The linear regression and 95% confidence intervals for the relationship between coral cover and coral carbonate production at sites dominated by the two most common corals or group of corals; Orbicella spp. (R² = 0.5376, y = 0.282x, P < 0.01), and the domed and encrusting group of species of Agaricia, Porites and Siderastrea (R² = 0.7509, y = 0.2107x, P < 0.01); (c) The linear regression and 95% confidence interval for the relationship between relative % of domed and plating non-framework building corals and coral carbonate production (R² = 0.2321, y = 0.0539x + 7.2579, P < 0.05); (d) The linear regression and 95% confidence interval for the relationship between coral cover and % cover of non-framework building corals (R² = 0.4665, y = 0.3969x + 1.4322, P < 0.01).

only confirms the widespread dominance of such coral taxa in most fore-reef habitats and on most reefs across the region, but also that these non-framework builders are becoming increasingly important as reef carbonate producers. The findings of this study also have direct relevance to wider concerns about future reef functionality. These relate to the integrity and maintenance of habitat architectural complexity, and to future reef growth potential. The domed, plating and encrusting non-framework builders that are becoming so important on Caribbean reefs today also have the lowest levels of morphological rugosity (Alvarez-Filip et al., 2011b). Thus their increasing dominance has implications both for carbonate production rates and for habitat complexity (see also Alvarez-Filip et al., 2013 for modelled scenarios of such changes), with combined knock-on effects on ecological processes and thus for many key reef ecosystem services. This is especially the case where key reef-associated species depend upon the intricate structures that more morphologically complex coral species (i.e., branched corals) provide (e.g., Rogers et al., 2014). Per-

sistence of the presently observed coral cover and coral community states we report are thus likely to have important implications not only for future reef carbonate budgets but also for multiple aspects of reef functionality that depend on the maintenance of reef framework structures. The general and significant lowering of carbonate production rates we report, that are linked to fundamental changes in coral species type and abundance, also have major implications for future reef growth potential. Converting carbonate production rates to long-term reef growth (accretion) rates is not a simple issue (Hubbard et al., 1990). In previous work, and using a small subset of the data we present here, we estimated that current rates of vertical reef growth potential on reefs in the Caribbean averaged only ~ 0.6 mm yr 1 for sites at 5 m depth, and 2.1 mm yr 1 at 10 m depth (Perry et al., 2013). These were derived from sites with average carbonate production rates of 4.6 and 3.1 kg CaCO3 m 2 yr 1 at the respective water depths, and were considered highly optimistic estimates of accretion given that no account could be taken for epi© 2014 John Wiley & Sons Ltd, Global Change Biology, 21, 1153–1164