In-situ incubation of a coral patch for community-scale assessment of metabolic and chemical processes on a reef slope

Introduction

The functionality of many reef systems is intrinsically linked to their structural habitat complexity (Newman et al., 2006; Graham & Nash, 2013; Kennedy et al., 2013). On tropical coral reefs, the three-dimensional habitat relies primarily on the ability of corals to deposit large quantities of calcium carbonate. Over recent decades, corals reefs have been under threat at a global scale by a large number of anthropogenic impacts such as ocean warming, overfishing, eutrophication, and ocean acidification (Hoegh-Guldberg, 1999; Gardner et al., 2003; Hoegh-Guldberg et al., 2007; De’ath et al., 2012; Anthony et al., 2008; Baker, Glynn & Riegl, 2008).

The consequential decline in coral cover and the reduction in historically dominant framework building coral species has already resulted in a substantial loss of 3D-complexity on many tropical reefs (Edinger & Risk, 2000; Alvarez-Filip et al., 2011a; Perry et al., 2015; De Bakker et al., 2016; Hughes, 1994; Hughes et al., 2007).

The impact of individual aspects of environmental change on coral reef health has been assessed in a number of laboratory experiments (Gilmour, 1999; Burkepile & Hay, 2009). However, in situ community-scale measurements of metabolic and geochemical processes would enable characterization of total reef metabolism. Net community calcification (NCC) is considered to reflect the overall response of the community to environmental change and is therefore monitored as a proxy for coral reefs’ status (Gattuso et al., 1993; Kleypas et al., 1999; Edinger et al., 2000). Field validation of coral accretion/decline is required to test whether observed experimental responses can be translated to whole ecosystems and in situ conditions. Coral reef decline or community compositional change can be estimated qualitatively from visual inspection of the same site over time (Aronson & Precht, 1997), or by digital comparison of photographs taken at intervals (Porter & Meier, 1992; Coles & Brown, 2007; De Bakker et al., 2016). The disadvantage of such visual assessments, however, is that results are confined to areas that have been visited previously and are not quantitative with respect to NCC. Still, the latter may be estimated from visual inspections using typical, species-specific calcification rates (Perry, Spencer & Kench, 2008). And although demonstrated to have fair accuracy (Porter & Meier, 1992; Alevizon & Porter, 2015; Chow et al., 2016), “carbonate budgeting” estimates do not allow estimating seasonal variability of NCC (Courtney et al., 2016), and are inherently insensitive to rapid environmental change. The same goes for elevation-change analyses using coral cores as alone do not relate alteration in seafloor structure to cause (Hubbard, Miller & Scaturo, 1990; Yates et al., 2017). Furthermore, the integrated effects due to organismal interactions cannot be assessed with such an approach. For example, ocean acidification may reduce coral NCC (Andersson & Gledhill, 2013) and at the same time increase erosion rates by sponges (Fang et al., 2013; Webb et al., 2017). Census approach has yet to include the role of several bioeroders (excavating sponges) (Murphy et al., 2016), which have been observed to become increasingly dominant on Caribbean reefs (Chaves-Fonnegra, Zea & Gómez, 2007). Moreover, ongoing ocean acidification appears to promote the contribution of chemical CaCO₃ dissolution to total bioerosion by sponges even further (Duckworth & Peterson, 2013; Wisshak et al., 2013). Lastly, Silbiger & Donahue (2015) suggest that, under future climate conditions of increased pCO₂ and ongoing warming, dissolution of existing reef carbonates is likely to be more affected than the growing of new reefs as such. Together this implies that there is an urgent need for in situ determination of NCC at the ecosystem level.

Direct approaches to accurately quantify NCC generally rely on determining the flux of alkalinity between water column and reef (Smith & Key, 1975). For reefs in environments characterized by a relatively linear flow of water over the reef, the upstream/downstream method (Odum & Hoskin, 1958; Gattuso et al., 1996) can be employed to determine NCC (Shaw et al., 2014; Koweek et al., 2015; Albright et al., 2016). For less unidirectional flow regimes, estimates based on overall residence time and knowledge of offshore conditions is needed (Courtney et al., 2016). In environments where low turbulence allows buildup of appreciable chemical vertical gradients, these gradients have been used to calculate net fluxes (McGillis et al., 2011; Takeshita et al., 2016). For fully exposed reefs, where no measurable accumulation may occur even in the boundary layer, the use of incubators is necessary. Several such incubation methods have been designed and applied. Most incubators cover a limited area (Patterson, Sebens & Olson, 1991; Haas et al., 2013; Camp et al., 2015), allowing single-species incubations. In most cases, numerous incubations are necessary to accurately capture variability between different locations on a reef in accretion/erosion and thus accurately estimate whole ecosystem NCC. When employing small-volume incubators, care must also be taken to maintain representative hydrodynamic conditions for the incubation species. Moreover, incubations must be terminated before NCC becomes depressed (for example by depletion of oxygen). Larger incubation structures (Yates & Halley, 2003) better capture variability on a community scale and convey environmental hydrodynamic conditions (surge) which, on the other hand, may cause inadvertent leakage of enclosed water. This potential exchange between ambient and enclosed water complicates the interpretation of observed chemical changes, particularly for signals that take relatively long to manifest themselves (e.g., alkalinity). The latter limitation restricts this method into hydrodynamically favorable (i.e., calm) conditions (McGillis et al., 2011). Additionally, due to obvious logistical challenges, most or all in situ incubations have been carried out on the reef flat.

Here, we aim to assess diurnal coral reef metabolic rates by investigating the in situ inorganic carbonate system over a reef slope coral reef patch offshore from the island of Saba, Dutch Caribbean. We use a tent-based incubation system in an environment with relatively strong currents and swell-induced near-bottom surge that caused modest exchange between the incubated reef and ambient water. Exchange between the enclosed and surrounding seawater is used to our advantage as this maintains oxygenated conditions during the incubation, thus allowing for increased incubation periods. Precise monitoring of temperature and salinity, both inside and outside the tent, allows for accurate determination of the amount of exchange across the enclosure. Explicitly accounting for the role of ambient variability, the benthic fluxes originating within the tent are inferred with high accuracy. Comprehensive monitoring of CO₂ system parameters (dissolved inorganic carbon, total alkalinity), dissolved oxygen, and nutrients (phosphate, nitrate, nitrite, and ammonium) allows for subsequent quantification of integrated whole ecosystem coral reef metabolic processes (NCC and net community production) in a highly hydrodynamic environment.

Methods

Site and substrate

Reef incubations were performed west of the island of Saba in Ladder Bay at the Ladder Labyrinth mooring site of the Saba National Marine Park (17.6261°N, 63.2602°W) (Field permit approved by the Dutch Ministry of Infrastructure and Environment: RWS-2015/38370), between October 26th and 29th 2015, at a depth of 21 m (Fig. 1). The reef in this area has a distinct spur-and-groove morphology, and is located on a steep incline from the heights of Saba toward the ∼500 m deep stretch between the island and the Saba Bank carbonate platform. Coral reefs around Saba harbor a relatively rich diversity of marine species in the context of the wider Caribbean (Etnoyer, Wirshing & Sanchez, 2010). The location at which the tent was placed, was chosen such that a patch of coral reef with a community representative of the wider area was fully enclosed (Fig. 2). Within the 2 × 2 m enclosure, one larger and several smaller carbonate structures were present, acting as main substratum for benthic biota, together resulting in a in a total hard surface area ∼4.4 m² and a 14% surface enlargement (rugosity). Abiotic components (sand and bare rock) accounted for 61% of the total surface area within the enclosure. Algae (algal turf, Lobophora spp., Dictyota spp.) covered 22%, sponges (among others Agelas sp. and Callyspongia plicifera) covered 7% and calcifying species such as corals (including Orbicella faveolata, Meandrina meandrites, and Diploria clivosa) and crustose coralline algae covered 4.2, and 6.6% of the total surface area, respectively (Fig. 2). No macro-bioeroding organisms were visible. A small number of heterotrophic animals, including small fish, crustaceans, and nudibranch, were present during the time of the incubation.

Results

Application of Eq. (1) to data collected during incubations four and five yields a leak rate of the enclosure f of ∼0.007 min⁻¹. This indicates that ∼25 kg of seawater (i.e., 0.007 × 3,000 kg) is exchanged every minute between the incubation enclosure and the environment. Although f ∼0.007 accurately relates interior and ambient salinity observations made during incubations 4 and 5, we have no direct means of ascertaining that it also applies to preceding incubations. However, given the sparse ambient salinity data, we assume this water exchange rate to be constant throughout all incubations. Because the sealing of the tent to the substrate remained unchanged from the second incubation onward, the time history of ambient salinity S_out^calc was derived under the assumption of f being ∼0.007. During incubations the error between S_out^meas and S_out^calc was −0.023 ± 0.19 (range: 34.7–36.0). The good match indicates validity of this simple adjective exchange model.

The ambient hydrographic conditions reflect variable admixture of a deeper, colder, and more saline component into the warmer, and fresher waterbody that is more commonly encountered at the incubation site (see Fig. 3A). In these ambient waters, C_T and A_T increase with S as is expected. Oxygen, too, is observed to increase with increasing S, due to the higher solubility in the more saline and—crucially—colder water. Regressions between S and O₂/C_T/A_T are presented in Fig. 3, panels B, D, and F. The time histories of these properties, derived using S_ambient^calc, are presented in Fig. 3, panels C, E, and G. For nutrients, no regressions were performed since ambient concentrations were essentially invariant at zero compared to the in-tent changes during incubations (Fig. S2).

Figure 4 illustrates the results of our approach for incubation 4. For an overview of all results from all incubations, please refer to Table S4 and/or Fig. S3. The model employed fits the measurements for C_T and A_T relatively well: the RMSE of the fit of C_T is 3.5 µmol kg⁻¹—identical to the measurement uncertainty of C_T itself. For A_T the fit (5.3 µmol/kg) is slightly worse than instrument precision (3.5 µmol kg⁻¹).

For the five incubations, consumption of oxygen from the incubated seawater ranged from −10 to −30 µmol kgSW⁻¹ h⁻¹ (Table S4). Concomitant increase of C_T, PO₄, and NO₂₊₃ strongly suggests respiration to be the dominant process throughout these incubations. Respiration should decrease A_T slightly due to the release of nutrients, but an even larger decrease is inferred by the full model suggesting a significant role for net calcification during incubation 2, 3, and 4. For the incubation 1 and 5 on the other hand, slight CaCO₃ dissolution is inferred. Prior to inferring rates, values of A_T were adjusted to account for the release of nutrients during respiration. Therefore, by definition the respiration A_T rate is zero for all incubations (Table 2).

Results obtained from incubations using the smaller, secondary tent suggests at most a very limited role for sedimentary processes (see Fig. S4): although exterior concentrations of A_T and C_T increased by as much as 60 µmol kg⁻¹ in the second half of the incubation, interior A_T and C_T during those 4 h did not increase at rates higher than 2.5 µmol kg⁻¹ h⁻¹. This limited response suggests a leak rate between 0.2% and 1% min⁻¹. Appreciable accumulation may therefore be expected if fluxes are present, also because of the small tent’s favorable volume-to-surface ratio (275 vs 750 l m⁻² for the dome tent). Over the first 165 min of the incubation, however, interior A_T and C_T increase only by 1 and 2 µmol kg⁻¹ compared to initial conditions (commensurate flux: ∼0.3 and 0.6 mmol m⁻² h⁻¹, that is, one or two orders of magnitude lower than in the main tent; note that appreciable errors apply both to measurements and conversion to flux). These increases may be linked to exchange with ambient waters. No changes in pyramid tent nutrient concentrations were observed, suggesting low rates of productivity or respiratory processes. For subsequent calculations, we consider the contribution by sedimentary processes to be negligible.

Subsequently, the concentration changes in oxygen, inorganic carbon, etc. in the dome-tent were used to calculate the fluxes during each of the five main incubations (Table 2). No trends were observed in TOC concentrations within the tent during incubations, despite clear trends being observed for C_T.

Deviations of the tent leak rate from the nominal 0.7 % min⁻¹ (or “breach events”) are observed during all incubations. These deviations from our assumption may negatively affect the inferred rates of change in O₂ concentrations, as evident from the differing slopes of trace and fit near the starts of incubations. Considering only the first 30 min of the oxygen traces (preceding tentative breach events in all incubations), appreciably higher oxygen consumption is inferred than when considering the full traces (Table S5). At constant ambient S, O₂, C_T, and A_T, leakage at a rate higher than the 0.7% min⁻¹ that we assume would result in underestimation of the true fluxes of O₂, C_T, and A_T. No change would be observed in the O-flux/C-flux ratio that we infer. Conversely, at times of sudden high ambient S, O₂, C_T, and A_T, leakage at a rate higher than the 0.7% min⁻¹ that we assume would result in overestimation of C_T and A_T fluxes, and underestimation of O₂ fluxes. This would change the O-flux/C-flux ratio that we infer. However, as we do not know if the O₂ trace is asymptotic, or that the tent did indeed leak at higher or lower rates than 0.7% min⁻¹, results are likely still valid (i.e., when no breach events occurred, and O₂ deviations resulted from sensor artefacts or true biological activity). We therefore maintained the conceivably affected incubations in the paper.

The NCC inferred from our incubations ranges from −0.2 to 1.2 kg CaCO₃ m⁻² year⁻¹ which is on average higher but still in range than the NCC estimated from the ReefBudget method (0.36 kg CaCO₃ m⁻² year⁻¹, Table 3).

Discussion

Our results indicate that the tent incubation is an effective tool for in situ quantification of reef fluxes in reef-overlaying water. Quantification of fluxes was achieved despite strong variability in ambient conditions and in the presence of appreciable swell-induced seawater exchange. To this end we applied a comprehensive conceptual framework for the interpretation of the measured concentration differences. This method allows for a volume exchange between the environment and the incubation thereby replenishing the latter and keeping the O₂ levels within the tent near ambient conditions resulting in minimized unrepresentative reef community metabolism. By continuously monitoring the inside environment and assuming constant exchange rate, fluxes within our incubation can be treated as if acquired by a flow through system. Nonetheless, future application of this or similar incubation methods could be further improved by continuous monitoring of the exchange rate, rather than assuming it to be constant throughout the incubation. This could be obtained for instance by running a second thermosalinograph outside the tent. The application of a conceptually simpler “asymptotic” model yields different and less well-constrained results. Particularly, the direction of the CaCO₃ dissolution flux may be seen to be reversed in the simpler method (see also Fig. 4). In all incubations, for both A_T and C_T, the uncertainty in measured concentration differences and the variability between results may be greater for the simpler model (Table S4; Fig. S3). In the case of A_T (Fig. 4F), the assumed-to-be-constant input of A_T inferred by the model applied here is of opposite sign to the simpler asymptotic model result (−0.6 vs +7.1 µmol kg⁻¹ h⁻¹). This reversal of sign of the A_T rates observed between the two models (asymptotic and full) is caused by the inability of the asymptotic model to account for occasional intrusion of high-A_T ambient water into the tent during sudden changes in ambient hydrography. The asymptotic model (panels C and D) shows a relatively good fit of the observations of O₂ and NO₂₊₃ around the fitted curve, which is due to the invariant ambient concentrations of these parameters.

In contrast with previous studies carried out at shallower depths using either Lagrangian drifts or incubations, all net diurnal rates from this study are strongly skewed toward respiration suggesting net heterotrophy in all incubations. Studies performed at shallower depths shift between net autotroph and net heterotrophy over the course of a day (Yates & Halley, 2003; Albright, Langdon & Anthony, 2013; Albright et al., 2015). However, previously reported average net respiration rates occurring at night on shallower reefs are comparable to results from this study (14.5–35.5 mmol C m⁻² h⁻¹). Results are also comparable to previously reported values at depth. For example, Middelburg, Duarte & Gattuso (2005) compiled a global mean coral reef respiration rate of 131 ± 46 mol C m⁻² year⁻¹. Specifically for their categories “outer reef slopes” and “high activity areas” they report values of 140 ± 70 and 413 ± 187 mol C m⁻² year⁻¹, respectively. This range compares well to the rates reported here (105–298 mol C m⁻² year⁻¹ for the full planar surface). A stoichiometric comparison of the inferred fluxes of C_T, oxygen and nutrients is combined with (i) the canonical “Redfield ratio” (Redfield, 1963) of the elemental composition of open ocean phytoplankton and (ii) the median elemental composition of benthic macroalga (Atkinson & Smith, 1983), likely resembling the composition of the labile fraction of the locally present organic carbon (Table 4). This shows that the community incubated in this experiment respires carbon and nutrients in a ratio that resembles the composition of benthic macroalgae (Atkinson & Smith, 1983) which indicates that the observed signal is indeed originating from the sedimentary, benthic, macrofaunal, and/or bacterial constituents of the enclosed community.

A strong correlation between NCC and net community productivity in reef environments is well documented (Gattuso et al., 1996; Shaw, McNeil & Tilbrook, 2012; Shaw et al., 2015; McMahon et al., 2013; Albright et al., 2015), however, no correlation was found in our incubations. The NCC inferred from our incubations ranged from −0.2 to 1.2 kg CaCO₃ m⁻² year⁻¹ which is on average higher than the mean recorded rate of 0.2 kg CaCO₃ m⁻² year⁻¹ associated with a Floridian reef patch (10% coral cover) in shallower waters (Yates & Halley, 2003) using a similar method.

Applying the ReefBudget approach to our benthic census data for comparison, we obtain a NCC for the full incubated substrate surface (i.e., sand and hard substrate) of ∼0.36 kg CaCO₃ m⁻² year⁻¹ (see Table 3). This estimate is slightly lower but comparable to the average NCC inferred from our incubations (∼0.42 kg CaCO₃ m⁻² year⁻¹, Table 3). The range of NCC estimates inferred from our results indicates how sensitive metabolic and chemical processes on coral reefs are to their environment. The chemical flux-based method as presented here is appreciably sensitive to the effects of the surrounding hydrological conditions on the substrate and this may be the source for a slight discrepancy compared to the ReefBudget approach. NCC rates acquired by Perry et al. (2013) associated with a coral cover ranging between 4% and 5% and at similar depth (17–20 m) in the Bahamas are all negative (ranging from −0.01 kg CaCO₃ m⁻² year⁻¹ to −0.23 kg CaCO₃ m⁻² year⁻¹). This may be explained by varying community composition such as the absence of macro-bioeroders within our tent. Furthermore, the flux-based method does not assess the mechanical component of bioerosion (caused by parrot fish or sponges for instance) which is important to the process of reef accretion.

The ReefBudget method offers a fast and convenient tool for estimating reef biogenic carbonate production states both on a remarkable temporal and spatial scale. Although the incubated flux-based approach, may be more sensitive to unstable and varying reef states, it cannot offer such a large spectrum of study. However, it provides an assessment of the full community without having to determine calcification/dissolution rates as a function of surface area. This can be very useful to assess the effect of endolithic species or determine the impact that some understudied organism may have on the chemical conditions. For instance, benthic cyanobacterial mats have been shown to proliferate around the islands of Curacao and Bonaire since 2003 (De Bakker et al., 2017) and are described to effect pH on a local scale (Hallock, 2005; Paerl & Paul, 2012). However, close to no records on how these mats may alter reef chemical conditions and subsequently impact the calcifying/bioeroding community are available. Currently, the ReefBudget approach relies on various assumptions regarding the calculation of each biological component. As such, the flux-based approach described here should not be regarded as a substitute for survey methods such as the ReefBudget, but rather as a complementary tool. Using the flux-based approach, it will become easier to determine missing components and variations in chemical dissolution/calcification on a spatial (e.g., depth) and also smaller temporal scale (i.e., diurnal cycle, seasonality), therewith improving survey based carbonate budget assessments.

To determine if the respiration signal might be an artefact of the incubation treatment, we identify potential causes that may perturb the signal. The estimated contribution by macrofauna such as fish, crustaceans, and nudibranchs to the observed respiration signal is deemed to be negligible: considering a fish mass-specific O₂ consumption rate of ∼100 mgO₂ h⁻¹ kg⁻¹ (Roche et al., 2013), and assuming 100 grams of fish to be present in the tent (which is likely a strong overestimate as only very few small fish were observed during incubations), we calculate a contribution to C_T in the incubation of ∼0.1 µmol kg⁻¹ h⁻¹, which is two orders of magnitude smaller than the observed respiration rates of 10–30 µmol kg⁻¹. Additionally, we rule out that “free floating” TOC (e.g., coral exudates) is the material that is respired. While clear fluxes are inferred for C_T, no trends were observed in TOC concentrations within the tent during incubations. Alternatively, no significant depression was observed of average interior TOC values (84 ± 9 µmol kg⁻¹, n = 39) relative to exterior TOC (86 ± 7 µmol kg⁻¹, n = 8). Although the tentative drop in TOC resembles the small drop observed in dedicated dissolved organic carbon (DOC) depletion experiments (De Goeij & Van Duyl, 2007), the lack of volume flow through the tent means TOC cannot be more than a very minor source of respirable carbon. Absence of depletion of suspended labile TOC notwithstanding, TOC may still play a role in the form of mucus if that is adhered to substrate or incubator, out of reach of sampling but available for bacterial respiration. However, the respiring biomass required for the observed C_T increases is unlikely to be present in the form of bacteria, especially shortly after incubation start. Indeed, Wild et al. (2004) show from small-scale incubations that the bacterial degradation of coral mucus, introduced into their incubators (containing only sediment and water column) at high concentrations, occurs at rates of 0.7–2.1 mmol m⁻² h⁻¹. That compares to rates around 60–175 mmol m⁻² h⁻¹ observed in our experiment, suggesting remineralization of adhered mucus plays at best a minor role in our incubations, further suggesting the observed fluxes to originate from the macroscopic biotic substrate.

In that category, sponges are the most likely organisms respiring, having appreciable biomass and containing ample energy stores to maintain respiration during the incubation periods, in which only limited amounts of organic carbon are available for filter feeding. Hadas, Ilan & Shpigel (2008) report (Red Sea) sponge basal oxygen consumption to be ∼50% of consumption featured during full water pumping activity, which means that sponge respiration largely continues even when filter feeding ceases. These authors report a rate of 2.4 µmol O₂ h⁻¹ g⁻¹ (wet weight). Similarly, Ludeman, Reidenbach & Leys (2017) report sponge oxygen consumption (standardized to sponge volume) ranging from 0.3 to 3 µmol h⁻¹ ml⁻¹, with strong species dependence. Assuming the higher end of this range applies to the sponges incubated in our experiment (mostly Agelas sp., C. plicifera), and assuming as much as five kg wet weight of sponge to have been present in the enclosure, we account for ∼5 µmol kg⁻¹ h⁻¹ of the observed rates of ∼30 µmol kg⁻¹ h⁻¹. Maintained respiration by sponges throughout the series of incubations could be fueled by filter feeding during the ∼50% of the time in which the incubator was open to the ambient water. Recent research by McMurray et al. (2018) showed that species hosting abundant symbiotic microbes (i.e., high microbial abundance, HMA) primarily consumed DOC, while the diet of species with low microbial abundances (LMA) primarily consisted of detritus and picoplankton. They further pointed out that it remained unknown if DOC released by LMA species could be a source of food for HMA species. The main sponges incubated in our experiment are Agelas sp. and C. plicifera and represent respectively a HMA and a LMA sponge. We tentatively infer that this may explain partly the observed high respiration rate. Nevertheless, we cannot infer if sponges are able to maintain metabolic balance throughout the incubation period, or that they deplete their stores. Further analyses such as bacterial counts would be needed to answer such questions. From Table 2 we conclude oxygen consumption rate during daytime (incubations 2 and 4) to be lower by ∼5 µmol kg⁻¹ h⁻¹ than night time rates (incubations 3 and 5), hinting at a role of primary producers (corals, CCA, macro and microalgae).

The lack of accumulation of C_T in the secondary, small incubation places our sediment at the very low end of literature values regarding respiration. For example, Middelburg, Duarte & Gattuso (2005) report a global mean sediment respiration value of ∼8.5 ± 7 mmol C m⁻² h⁻¹ (as approximated from their Figure 11.3). Our observed low values may be reasonable considering the highly hydrodynamic nature of the incubation environment which likely hampers settlement of substantial amounts of organic matter onto and into the sediment. In addition, the volcanic sandy composition of the sediment around Saba may be less prone to dissolution than coralline sediment and could explain the insignificant increase in A_T in the small tent. However, Eyre et al. (2018) shows similar results for sediment around Cook Islands which is mostly composed of calcareous fragments (Wood, 1967). Eyre et al. (2018) shows that dissolution in reef sediment across different locations around the world is negatively correlated with the aragonite saturation state (Ω_ar). Average Ω_ar of ambient water around the tent incubation throughout the experiment is calculated to be 3.85 which is more comparable to islands (Bermuda and Tetiaroa) showing accretion in reef sediment. The combined effects of hydrodynamics and sand composition are likely to explain why our results present neither accretion nor dissolution in our tent’s sediment.

Conclusions

Flux-based carbonate budget studies, as presented here, provide quantitative data on the functional state of reefs in terms of biologically driven carbonate production which is particularly sensitive to ambient environmental conditions. As such, they can be particularly useful for temporal studies, especially to reveal not only diurnal and seasonal patterns but also to capture shifts in functionality of reef systems. We incubated a coral reef patch situated in a high-energy environment which caused a limited amount of seawater exchange. Monitoring of conditions within and outside the tent allowed for determination of the exchange rate and thereby allowed for correcting the respiration and calcification rates. Application of this procedure shows that this reef patch is characterized by NCC inside the tent at a rate within range but on average higher than fluxes reported in previous studies for shallower reef systems indicating coherence in our results. However, the range of NCC estimates inferred from our results accounts for the sensitivity of this reef patch to the surrounding environment. Furthermore, the net heterotrophy reported here both during the day and the night differs from studies performed at shallower depths where shifts between net autotrophy and net heterotrophy are observed. Future research may include various types of substrates and comparison between regions with varying water quality.

Supplemental Information