Coral reefs form some of the planet’s most biologically diverse ecosystems, providing numerous ecosystem goods and services1. Much of this functionality is linked to the structure of the reefs themselves, that provide both complex 3-dimensional habitats, and breakwater structures that modify wave energy regimes and act as protective breakwaters for adjacent shorelines. However, at the global scale, coral reefs have been severely impacted over recent decades by multiple human disturbances2. Coral cover is estimated to be declining by 1–2% per annum across the Indo-Pacific3, and has already declined by an average of ~80% in the Caribbean since the mid-1970s (4). Commensurate with these declines has been a loss of reef architectural complexity5. Climate change is an additional threat. Elevated sea-surface temperatures have caused widespread coral bleaching6, and increasing atmospheric CO2 concentrations are projected to drive further warming and ocean acidification7. These changes have important implications for coral reef ecosystems generally, but it has also been suggested that such changes will result in lower rates of reef carbonate production8, which will limit the potential for coral reef growth in the future and, potentially, lead to a collapse of reef structures7. Quantitative data to support these ideas are essentially absent, but clearly any such loss of vertical growth capacity will profoundly inhibit the ability of reefs to keep pace with projected increases in sea-level, and severely impede many of the ecosystem functions and services that are underpinned by reef structures and their associated topographic complexity. The geomorphic state of reefs, as measured by the development and maintenance of their topographically complex carbonate structures, is dependent upon the net accumulation of calcium carbonate. This is a function of the balance between constructional (for example, coral and coralline algal production) and erosional (biological and physical erosion) processes8. Where the balance is positive, net accumulation (and thus reef growth) is typical, but where the system switches to a net negative state, such as may happen under conditions of high biological erosion, net erosion of reef structures can occur8. Short-term transitions of this type have been documented at individual sites following local disturbances9. Key questions that arise, however, are: what impacts have regional scale changes in coral reef ecology had on the carbonate production states of shallow-water reef habitats? how do carbonate production rates calculated for contemporary ecosystems compare with those established over mid- to late Holocene timescales, that is, how do they compare with rates calculated for the period pre- major human pressure in the region? and what implications do these changes have for reef growth potential in the future? Here, we report contemporary rates of reef carbonate production and bioerosion measured from 101 transects on 19 coral reefs in 4 countries (Bahamas, Belize, Bonaire and Grand Cayman) from across the Caribbean (Fig. 1). We then use these data to determine net rates of biologically-driven carbonate production (kg CaCO3 m−2 year−1) and resultant accretion rates (mm year−1) (Methods). Within these countries data were collected from a range of common Caribbean reef habitats: nearshore hardgrounds, Acropora palmata habitats, Montastraea spur-and-groove zones, fore-reef slopes, and deeper (18–20 m) shelf-edge Montastraea reefs. The countries examined occur in different regions with respect to prevailing wave energy/hurricane frequency10, and thus some degree of inherent variability in their background ecological conditions, as a function of recent disturbance history, must be assumed. However, the general ecological condition of most of the sites examined was remarkably consistent, and typified the spectrum of reef ecological states presently observed in shallow-water habitats across much of the region4 11: on most of the reefs live coral cover was less than ~25–30% (often markedly so); most shallow water sites ( 5 G (Fig. 1; see Supplementary Table S1). The most productive reefs were inside the ‘no dive reserve’ in Bonaire, where average net production was +3.63 G (5 m depth) and +9.53 G at 10 m depth (Fig. 1). At the transect scale 37 of the 101 transects had negative budgets and 22 had rates between 0–1 G. Only nine transects had rates >5 G and just 5 rates >10 G. The remainder were between 1 and 5 G (Fig. 2a). Net carbonate production rates vary between and within habitats. Montastraea spur-and-groove habitats had the highest G values (mean 3.0 G; range −0.47 to 16.68 G; Fig. 2b): all other habitats had mean G values 50% lower than rates calculated for mid- and late-Holocene periods and in many cases are markedly lower (especially in the habitats previously dominated by A. palmata). High gross production rates close to high historical values (>10 G) were calculated only for the few sites with high live coral cover dominated by healthy Montastraea populations. In contrast to earlier states of Caribbean reef ecology, where high carbonate production rates were driven by corals of the genera Acropora and Montastraea, there is thus now an essentially monospecies dependency on corals of Montastraea complex to maintain positive production states at the sites where rates remain high. These remaining high productivity sites may thus quickly transition to very low net production states if regional declines in Montastraea populations continue18. With regard to net carbonate production rates our data indicate that while ~65% of transects surveyed exhibited positive net carbonate production states, 58% had rates 10 m palaeodepths)23 and which thus equate to the range of depth intervals we examined across the Caribbean. We converted our production rate estimates to potential accretion rates (mm year−1) using established approaches based on carbonate density and the average porosity of reef framework23 24, but also accounted for the re-incorporation of a proportion of bioerosion-derived sediment into the framework (Methods). The resultant average accretion rate across our sites is 1.36 mm year−1 but is highly variable (range: −1.17 to 11.93 mm year−1). The very high end rates at our ‘healthiest’ sites were thus comparable with those calculated for some Holocene high productivity Acropora dominated reefs in the Caribbean, for example, Alacran reef, Mexico: 12 mm year−1 (ref. 25). However, for transects 10 m) reef environments, the relative difference between the two may be changing as shallow water habitats experience relatively greater rates of ecological decline. Many shallow water reefs across the Caribbean are therefore probably at an accretionary threshold. Although there is no solid evidence for significant regional scale loss (erosion) of the underlying framework structure of reefs at present, the geomorphological complexity of many reef surfaces is clearly declining5, and our data suggest that under present trends this may ultimately extend to loss of the underlying (Holocene) structure. The potential for production states to revert to those typical of the past is likely to be highly site specific, but also strongly influenced by external factors. For example, if communities of branched Acropora recover, as observed at a few select sites in the region29, then relatively rapid shifts back to higher net production states may occur. However, where communities persist in altered coral community states, dominated by ‘weedy’ taxa such as Porites astreoides and Agaricia agaricites 3 30, then more persistent low net (or negative) erosional regimes will likely endure. Differential exposure to high magnitude physical disturbance events across the region, which may increase as oceans warm31 (but see also ref. 32), and uncertain responses to projected more frequent bleaching episodes, will also inhibit positive budget transitions. That many of the reefs that retain high carbonate production rates (>5 G) have an essentially monospecific dependency on corals of the Montastraea complex is an additional pressing concern. Loss of these corals would have major consequences for reef growth potential in the future. Finally, given that coral cover is also on a general downward trajectory on reefs throughout the Indo-Pacific region3, our findings raise important questions about contemporary reef growth potential globally, and about how resilient the geomorphic structure of reefs will be if coral cover continues to decline in the face of changing environmental and climatic regimes. Methods Quantifying gross and net carbonate production and erosion rates We collected data on gross carbonate production and erosion rates to determine net carbonate production (G, where G=kg CaCO3 m−2 year−1)15 and accretion rates (mm year−1) from four countries across the Caribbean (Bahamas, Belize, Bonaire and Grand Cayman) between November 2010 and March 2012. As we were interested in determining the relative importance of different biological carbonate producers and eroders in different environments, and examining these in the context of ecological change, we adopted the ReefBudget census-based methodology to determine rates of G 14. To derive measures of benthic carbonate production we used census approaches to determine abundance of benthic carbonate producers and calculated carbonate production rates using published linear extension and density metrics for individual species (or nearest equivalent species), following ref. 14. Estimates of substrate erosion rates were based on a census of bioeroding sponge tissue cover, available substrate for microendolithic bioerosion, metrics on species and size class of bioeroding urchins, and metrics on species-size-life phases of bioeroding parrotfish locally calibrated with bite rate data. Comparing net production rates within and between reef sites A one-way ANOVA, with Bonferroni multiple pairwise comparisons, was used to test for differences in G between countries. However, the spatial nature of the data meant that transects within reefs were likely to be more similar to each other than to transects on other reefs. The data were also unbalanced as the number of transects surveyed at each reef was not constant. Consequently, linear mixed effects models were chosen to examine the relationships between net carbonate production (G) and potential controls including the percentage cover of hard corals and macroalgae33 34 (Supplementary Tables S2 and S3). All modelling was performed in SPSS 19. Initially a saturated model was chosen, such that all sensible fixed and random effects of interest were included. The data were nested within reef and country and restricted maximum likelihood estimation was used to run the model. Thereafter, Akaike Information Criterion was used to select a suitable covariance structure and subsequently the best combination of random effects. This model was assessed graphically by examining a histogram of the standardized residuals to check for normality and by plotting the residuals versus the predicted values. Heterogeneity of variance was clear at this stage and the dependent variable (net carbonate production) was transformed: log (y+3). The model was run again with the transformed data as the dependent variable and the assumptions of normality and homogeneity of variance were confirmed. The fixed effects were then assessed by examining the significance of regression parameters33 34 (see also Supplementary Methods). Comparing contemporary and Holocene reef accretion rates To compare contemporary and Holocene reef accretion rates, rates of net carbonate production were converted to potential accretion rates (mm year−1) using established approaches based on carbonate density and framework porosity15 21 22. To account for the incorporation of bioerosion-derived sediment back into the reef structure, and thus as an addition to the reef accretion rate metric, we used a proportional sediment incorporation rate for the Caribbean of 50% based on data in ref. 21. In this calculation we assumed that all urchin and endolithic sponge-derived erosional products, and 50% of parrotfish-derived erosional products (as a mobile taxa that defecates both on areas of reef framework and into sand channels) were available for potential incorporation. Comparisons with Holocene reef accretion rates in the Caribbean were based on the depth and habitat-stratified reef core data sets published in Hubbard23. Author contributions C.P. initiated the study, collected and analysed the field data and wrote the manuscript. G.M., P.K., E.E., S.S. and R.S. helped development the underpinning methodology, collected and analysed the field data, and contributed to manuscript writing. P.M. helped develop the methodology, helped with site selection, analyse the data and contributed to the manuscript. Additional information How to cite this article: Perry, C. T. et al. Caribbean-wide decline in carbonate production threatens coral reef growth. Nat. Commun. 4:1402 doi: 10.1038/ncomms2409 (2013). Supplementary Material Supplementary Information Supplementary Tables S1-S3 and Supplementary Methods.
