Introduction

Tropical coral reefs represent the most diverse marine ecosystem, the recent global decline of which has been documented repeatedly (e.g., Pandolfi et al. 2003). One of the threats is anthropogenic carbon dioxide (CO2) emission, which lowers carbonate saturation state in ocean surface waters, hence the potential of corals to secrete their calcium carbonate (CaCO3) skeleton, and eventually diminishes reef accretion-rate (Kleypas et al. 1999). But exactly how and where do reefs accrete? It is a widely held concept among geoscientists and biologists that tropical coral reefs in shallow water should accrete faster than those in deep water (Schlager 1981, compilations on his Figs. 4 and 5) (Table 1). Tropical reef-building scleractinian corals harbor unicellular algae (dinoflagellates) in their tissue as photosymbionts, which are both crucial for the secretion of the aragonitic CaCO3 skeleton by consuming CO2 and for providing nutrition through photosynthesis (Muller-Parker and D’Elia 1997). For example, growth-rate, i.e., skeletal extension-rate, of massive corals of the Montastraea group (average rate of 10 mm/yr), major builders of Caribbean reefs, decreases exponentially with increasing water depth (Bosscher and Schlager 1992). Growth-rates of the branched Acropora palmata (average rate of 90 mm/yr), which forms thickets in 0–5 m of water and which is also a major builder of Caribbean reefs, exceeds growth-rates of Montastraea sp. by an order of magnitude (Dullo 2005). Hence, reefs dominated by fast-growing branched acroporid corals should build up faster than those with massive corals, as indicated by Adey (1975, St. Croix), Davies and Hopley (1983, Great Barrier Reef), Macintyre et al. (1985, Antigua), or Hubbard et al. (1998, Caribbean) (Table 1). According to these studies, accretion-rates of branched coral reefs were highest, however, rates of massive and branched coral reefs overlapped. Also, differences in accretion-rate between massive and branched coral reef sections were not checked statistically, and paleo-waterdepth data were not considered. This study analyses accretion data of branched and massive core sections, including paleo-waterdepth data, from Holocene barrier and atoll reefs of Belize, the largest reef system in the Atlanic Ocean.

Table 1 Some accretion-rates of Holocene reefs discussed in the text
Full size table

Materials and methods

Twenty-five drill cores were taken on Belize barrier and atoll reefs with a portable rotary drill and wireline system during four expeditions, which took place between October 1995 and July 2002 (Fig. 1). Individual cores are as long as 20 m, and the total core length is 275 m. All cores investigated here were drilled on marginal reef pavements (sensu James et al. 1976) close to reef crests. Cores were investigated with regard to sedimentology and 14C radiometric age (Gischler and Hudson 1998, 2004; Gischler and Lomando 2000). For the present study, additional taphonomic studies were conducted in order to check whether or not dated corals are in situ. Autochthonous corals were distinguished from transported corals in the core by checking (qualitatively) for upcore-oriented corallites, and degrees of marine encrustation, cementation, and bioerosion. Paleo-waterdepth of 33 dated Acropora palmata and Montastraea sp. corals was determined in the core material by measuring the distance from each calibrated datum to the Holocene Belize sea-level curve, which is based on 69 dates from corals and peats (Gischler and Hudson 2004; Gischler 2006). Reef accretion-rate was measured between radiometric dates (Table 2). Rate of Holocene sea-level rise was measured in 500 year increments on the Belize sea-level curve. A caveat refers to the fact that paleo-waterdepths were calculated using the sea-level curve, which was based on some of the same age data. However, as postglacial sea-level curves for the Caribbean realm exhibit very similar trends, i.e., steep rises until ca. 6 kyrs BP followed by a gradual rise to modern level (e.g., Gischler 2006), problems of circular reasoning can be largely excluded.

Fig. 1
figure 1

Map of the study area offshore Belize including barrier reef and three offshore atolls and 25 corehole locations (open circles)

Full size image
Table 2 Basic data from Belize barrier and atoll reef core material
Full size table

Results

Holocene reef composition

In Belize barrier and atoll reefs, the branched Acropora palmata and members of the massive Montastraea group are by far the most abundant corals in Holocene reefs (Gischler and Hudson 1998, 2004) (Figs. 2c, d; 3). Similarly, at modern barrier and atoll reef margins, A. palmata thickets are found in shallow water (0–4 m) whereas abundant Montastraea sp. occur in deeper water (>4 to almost 40 m) on the reef slope (Fig. 2a, b). Other common corals in the core include brain corals (Diploria sp., Colpophyllia sp.) (Fig. 2e), massive corals such as Siderastrea siderea, branched Porites sp. and Acropora cervicornis (Fig. 2h), and the hydrocoral Millepora sp. The latter coral is quite common in well-cemented coral grainstone (Fig. 2g). Unconsolidated sand and rubble makes up 30% of the core. Further constituents include coralline algae, Halimeda sp., some mollusk shell and echinoderm fragments, and benthic foraminifera (Gischler and Hudson 1998, 2004). Microbialite crusts occur occasionally along the core. They grew in former reef cavities, together with rare brachiopods, small coralline sponges, encrusting forams, and stylasterids (Fig. 2d, k).

Fig. 2
figure 2

Photos from modern Belize reefs and from core samples. a A. palmata zone in shallow water (0–2 m). Modern reef margin near barrier reef core holes 6 and 7. b M. annularis zone in 6 m of water near Glovers Reef core 2. c Branches of A. palmata with vertical corallites. Branches are cemented together largely by high-magnesium-calcite cements (dark grey). Note bioerosion by boring mollusks. Barrier reef core 4, −6 m. d Piece of M. annularis with corallites oriented upcore. Note microbialite crust. Barrier reef core 1, −6 m. e Thick crust of coralline algae and forams Homotrema rubrum (red dots) on Colpophyllia. Lighthouse Reef core 7, −3 m. f A. palmata fragment, which is bored, encrusted by coralline algae (white) and microbialite (dark grey) in upper left part. Barrier reef core 11, −4 m. g Millepora blades cemented together by high-magnesium-calcite cement and microbialite (lower right part). Barrier reef core 7, −5 m. h Branched coral cemented by thick cement crust. Barrier reef core 4, −7 m. i Thin-section micrograph of Montastraea coral with aragonite needle cement and microcrystalline and peloidal high-magnesium-calcite cements. Barrier reef core 7, −6 m. k Cavity in core with coralline sponge (Ceratoporella nicholsoni) and tecideinid brachiopod (Argyrotheca sp.). Barrier reef core 1, −10 m

Full size image
Fig. 3
figure 3

Taphonomic signatures in the rotary cores from Belize. (a) Barrier Reef. (b) Offshore atolls. Cores TR 6, 8, 9 are not shown because Holocene recovery was below 10%. Holocene age data from 14C-dating are presented as calibrated ages; Pleistocene age data from U/Th-dating

Full size image

Even though re-deposition can never totally be ruled out in a 5 cm diameter core, the taphonomic signatures individually and collectively indicate that the large majority of dated corals are at or close to their original habitat (Figs. 2, 3). Vertical corallites are observed in A. palmata and Montastraea fragments throughout the core (Fig. 2c, d). Marine encrustations, mainly by coralline algae, are also quite common (Fig. 2e, f). Bioerosion by bivalves, sponges, and polychaetes may also be seen in many cores (Fig. 2c–h; 3). Cementation is exclusively marine-phreatic with aragonite needle crusts and microcrystalline and peloidal high-magnesium-calcites (Fig. 2i). Cementation is more common in the uppermost sections of the Holocene core (Fig. 3).

Accretion-rates

Accretion-rate as calculated from the 33 reef core sections ranges from 0.46 to 7.50 m/kyr, and averages 3.03 m/kyr (Table 2). Surprisingly, average accretion-rates of reef sections dominated by the massive, slow growing Montastraea (3.31 m/kyr) are higher than those of reef sections with the branched, fast growing Acropora palmata (2.56 m/kyr). According to the t test, these differences in accretion-rate are statistically not significant (P < 0.187). Curiously, accretion-rate was higher in deeper water as it was in shallower water (Fig. 4a). The correlation is statistically significant (r = +0.44, P < 0.01). The same trend is seen when plotting reef sections dominated by Acropora and Montastraea separately (Acropora: r = +0.51, P < 0.03; Montastraea: r = +0.49, P < 0.08). The encountered paleo-depth range is 10 m. In general, the large majority of Acropora data points plots above 3 m paleo-waterdepth, whereas most of the Montastraea data points are found below that depth (Fig. 4a). Between 4 and 10 m paleo-waterdepth, there are only two Acropora points left. The relatively clear bathymetric separation of Acropora and Montastraea data points further supports the contention that most of the corals dated were not significantly transported away from their original habitat.

Fig. 4
figure 4

(a) Holocene reef accretion-rate in Belize plotted against paleo-waterdepth. (b) Holocene reef-accretion-rate in Belize versus rate of sealevel-rise

Full size image

Discussion

Interpretation of accretion-rates

Reef accretion results from the complex interplay of construction by reef builders, destruction by biological and physical impacts (i.e., bioerosion and storms), detrital sediment accumulation and consolidation (i.e., diagenesis). In this context, there are three possible explanations for the results, which apparently contradict the fact that growth of reef-building corals decreases with water depth and that massive reef corals grow slower than branched ones.

First, there is evidence from reefs of Australia (Lizard Island, One Tree Island) and the Bahamas (Lee Stocking Island) for a decrease in the rate of bioerosion by sponges, echinoderms, mollusks, and microbes with increasing water depth (Kiene and Hutchings 1994; Vogel et al. 2000). Qualitative data obtained during this study shows no clear trends in that massive and branched corals exhibit similar degrees of bioerosion. More quantitative studies will be necessary in order to validate local observations and draw a more general picture and patterns of bioerosion-rates in various reef settings.

Second, massive, deeper water Montastraea apparently accrete in more undisturbed settings as compared to shallow water acroporid thickets, which repeatedly get abraded, broken, and leveled out during tropical storms and cyclones. Hughes and Connell (1999) discussed the selective damage during hurricanes, i.e., how branched corals are most storm-susceptible and massive corals largely storm-resistant. In Belize, for example, hurricanes hit every 6 years in the twentieth century on average (NOAA National Hurricane Center, http://www.nhc.noaa.gov) Also, the accomodation space in deeper water is higher as compared to shallow water. Indeed, Holocene reef accretion-rates increase with the rate of sealevel-rise (r = +0.41, P < 0.019), i.e., with increasing accomodation space (Fig. 4b). In order to draw more far-reaching conclusions, it would be desirable to include reefs outside of the cyclone belt in the analysis. Following the line of argument used here, high accretion-rates among branched coral reefs are to be expected in cyclone-free reefs. Suitable study examples would include reefs of the Maldives (Indian Ocean) and the Gilbert, Phoenix, or Line Islands (Pacific Ocean), which are located close to the equator.

Third, detrital sediment accumulation and infill within the coral framework is more likely in deeper as compared to shallower water because of lower depositional energy. This process is also of importance for reef accretion, and it has been shown by James et al. (1976) how internal reef sediment may be preferentially cemented, thereby consolidating Holocene reefs. No significant difference in the degree of reef cementation was seen in branched versus massive coral reef sections during this study though. In general, reefs were better cemented in the upper core sections as compared to the lower parts. Due to the decreasing rate of sea-level rise, late Holocene reefs have been exposed for a comparably long time to carbonate-rich waters, which are constantly being pumped through the reef framework.

The following aspects of reef accretion were not included in this study. There exists carbonate production by other reef-building organisms such as, calcareous algae, which was only qualitatively analyzed in this study (Fig. 3), and by mollusks, foraminifera, and echinoderms. Detailed quantitative studies in shelf-edge reefs of St. Croix (Caribbean Sea) have shown, however, that the large majority (>90%) of reefal carbonate production is by corals whereas production by other organisms is usually < 10% (Hubbard et al. 1990). Likewise, core sections of unconsolidated rubble and sand, which make up about 30% of the core material studied here (Fig. 3) were not included in the analysis. This is because radiometric dating within this facies is problematic due to the effect of time-averaging in these largely detrital deposits. There are three sections within this facies, however, based on which accretion-rate can be estimated though. Corals directly above and below rubble and sand sections were age-dated in two sections on the barrier reef (accretion-rates 1.71 and 1.82 m/kyr) and one section on the atolls (1.14 m/kyr). Average accretion-rate (1.56 m/kyr) in this facies are apparently much lower as compared to the coral facies analyzed. Finally, it would be desirable for future studies to also inlcude quantitative data on carbonate production per area and time units in the comparison of massive and branched coral facies. However, this kind of data are still very rare and usually based on estimates (Chave et al. 1972), or on extensive and long-term observations and data collection (e.g., Hubbard et al. 1990).

Comparison with other data

There is only a limited number of published data from other locations available for comparison, which contain paleo-waterdepth and accretion data of branched versus massive coral reefs, and which can be analyzed in the above fashion (Fig. 5). For Belize barrier and atolls reefs, there are two other core studies, which were performed near Carrie Bow Cay (between our core location BBR 1 and BBR 6/7, Fig. 1). Macintyre et al. (1981) drilled three core holes in the fore reef at water depths of 12–16 m. They found Holocene successions of massive and branched corals. Accretion-rates ranged from 1.8 to 8.3 m/kyr. Shinn et al. (1982) drilled four cores in the shallow fore reef, in the back reef, and on Carrie Bow Cay. Accumulation-rates ranged from 7.2 to 7.9 m/kyr. When plotting accretion-rates of reef sections between radiometric dates, which are mostly dominated by massive corals, versus paleo-waterdepth (range from 13.9 to 23 m), a positive correlation (r = +0.22, P < 0.678) may be seen as well, however, it is statistically not significant (n = 6).

Fig. 5
figure 5

Reef accretion and paleo-waterdepth data for some other locations discussed in the text

Full size image

For the wider Caribbean, Hubbard et al. (2005) compared accretion-rates versus paleo-waterdepth in several locations (Florida, Panama, Bahamas, West Indies) and found no clear trend (r 2 =  0.003). Also, reefs with branched corals did not significantly accrete faster (average 3.83 m/kyr) as compared to those with massive corals (average 3.07 m/kyr). The depth range observed in the study was 15 m.

Further afield, for the Great Barrier Reef, Davies et al. (1985) reported that over a depth range of 20 m, deep reefs grew much faster during initiation as compared to shallower reefs. When plotting the data of their Tables 1 and 2, accretion-rate shows a decrease with increasing paleo-waterdepth, however, there is no statistically significant correlation (r = -0.26, P < 0.401) (Fig. 5). In fringing reefs of New Caledonia, over a depth range of 15 m, there is an increase of accretion-rate with depth visible (r = +0.17, P < 0.407), like in Belize, however, it is not statistically significant (Cabioch et al. 1995, their Table 1). Paleo-water-depth was calculated to the mean relative sea-level curve of New Caedonia for this comparison. When plotting massive and branched reef sections separately, branched corals exhibited the same trend as described above, whereas accretion-rates of massive coral reef sections decreased with paleo-waterdepth. Again, no significant correlations could be detected. In the western Indian Ocean, there is no trend visible when comparing accretion-rate and paleo-waterdepth of keep-up and catch-up reefs in Mayotte, Mauritius, La Réunion, and the Seychelles (r = +0.059, P < 0.839). In this case, paleo-waterdepth was estimated by measuring from the middle of each accretionary increment to the sealevel-curve (Camoin et al. 2004, their Fig.4). The depth range is 20 m. Montaggioni (2005) recently compiled available data on Holocene reef growth in the Indo-Pacific realm, and found the highest accretion-rates among arborescent and tabular coral reef sections. Even so, accretion-rates were highly variable and not dictated by coral growth form. For example, in the Holocene of Tahiti, accretion-rates of massive Porites reef sections were in the same range and sometimes even higher as rates of branched Acropora reef sections (Montaggioni et al. 1997).

Conclusions

Even though different trends in accretion-rate versus water-depth exist in different reef locations worldwide, there were no statistically significant correlations observed that would support the concept that shallower reefs grew faster than deeper reefs, and that reefs with branched corals grew faster than those dominated by massive corals. Massive coral reefs in slightly deeper water (5–10 m) appear to be of paramount importance because they accrete as fast as or even faster than shallow acroporid coral reef systems, as demonstrated for the Belize barrier and atoll reefs. Certainly, more analyses of other Holocene reef locations—including paleo-waterdepth and reef accretion data, which compare branched versus massive coral reefs—are desirable in order to validate the conclusions presented here.