Introduction

Ocean acidification and climate change are large-scale threats to the viability of coral reefs (e.g. Albright et al. 2016, 2018; Bellwood et al. 2004; Cornwall et al. 2021; Eyre et al. 2018; Hoegh-Guldberg et al. 2007; Veron et al. 2009). The potential impacts of acidification are highly heterogeneous on the reef, and algal ridges may be particularly vulnerable. Algal ridges are an essential structural component of many coral reef systems. They develop as raised rims on the exposed windward margins of coral atolls and reefs (Fig. 1) (Adey 1975; Ladd et al. 1950) and play a critical role in protecting leeward reef flat, back reef and lagoon structures by dissipating energy associated with destructive waves (Harris and Vila-Concejo 2013; Harris et al. 2011; Johansen 2018). Algal ridges frequently acquire coral rubble created and transported during heavy storms and cyclones, which subsequently becomes cemented and incorporated into the ridge structure, increasing the height and protective potential (e.g. Macintyre et al. 2001; Perry 1999; Rasser and Riegl 2002; Talavera et al. 2021; Teichert et al. 2020).

Fig. 1
figure 1

A. One Tree Island (23°30’S; 152°05’E) study area B. site locations are denoted by numbers 1 to 5, respectively and the algal pool investigated during this study. The prevailing wind direction is generally from the southeast. North is upward

Full size image

Calcium carbonate production on the surface of algal ridges is dominated by crustose coralline algae (CCA), which precipitate metastable high-Mg calcite (HMC; Mg0.14–0.20Ca0.80–0.86CO3). Calcification rate estimates obtained for algal rims and the adjacent reef crest environments range between 1.5 and 10.3 kg CaCO3 m−2 y−1 (Chisholm 2000).

Internally, the diagenetic precipitation of HMC minerals is important for cementing and stabilising rubble and sediments within the reef framework (Macintyre et al. 2001; Perry 1999; Rasser and Riegl 2002). Like CCA HMC, submarine HMC cements generally contain between 4 and 24 mol% MgCO3 (Frank and Lohmann 1996; MacIntyre and Marshall 1988b; Mitchell et al. 1987). The exact formation mechanism of diagenetic HMC is unclear, but the prevailing paradigm is that cement formation within the algal ridge and reef crest results from the pumping of oxic seawater supersaturated with respect to carbonate minerals through the reef framework, combined with CO2 degassing driven by changes in hydrostatic pressure due to wave action (MacIntyre and Marshall 1988b; Perry and Hepburn 2008; Rasser and Riegl 2002). This mechanism was hypothesised to account for the extensive development of submarine cement on the high-energy windward margins of reefs with ample wave energy to drive the circulation of oxygenated seawater within the reef subsurface (James et al. 1976; Marshall 1986). Consequently, coral reef systems where HMC dominates may be particularly vulnerable to ocean acidification (Macintyre and Marshall 1988a; Marshall 1986) as HMC is the most soluble calcium carbonate polymorph and is predicted to become undersaturated in surface ocean waters by the end of the century (Andersson et al. 2008; Biscaye et al. 1988).

However, submarine cements and associated diagenesis and lithification products have also been observed in low-energy reef systems (Land and Moore 1980; Marshall 1986), and some lagoon settings (Macintyre and Aronson 2006), with the latter tending to be less consolidated. The occurrence of mineral carbonate precipitation in low-energy environments suggests that there may be an alternative chemical pathway for diagenetic precipitation within the algal ridge framework, as high-energy oxic pumping fails to explain low-energy analogues (Macintyre 1985; Perry and Hepburn 2008).

Possible factors influencing mineral carbonate cement formation include photosynthesis, respiration, and subsurface diagenetic processes under low oxygen conditions. Photosynthesis can only occur at the surface of algal ridges where light is available. Photosynthesis influences carbonate water chemistry through the consumption of carbon dioxide (CO2) and the production of organic matter, which increases seawater pH and carbonate ion concentration, thereby promoting calcium carbonate precipitation.

To fuel primary production, new nutrients are required. Background nitrate concentrations are generally low in reef settings, with a significant amount of the new nitrogen needed to fuel photosynthesis coming from nitrogen fixation associated with the benthic community (El-Khaled et al. 2021; Hatcher and Frith 1985; Roth et al. 2020). Nitrogen fixation does not change total alkalinity (TA) (see Equation S2 for the chemical definition of TA). Only after the release of ammonia from organic matter respiration (Eq. 1) and its oxidation to nitrate under oxic conditions does TA change (Eq. 2) (Wolf-Gladrow et al. 2007).

Organic matter respiration under oxic conditions has the opposite effect to that of photosynthesis, where seawater pH and carbonate ion concentration decrease due to CO2 production (Eq. 1). Oxic respiration also affects TA through the release of organic nitrogen as ammonia such that the change in alkalinity relative to dissolved inorganic carbon (DIC) is via the N:C (y:x) stoichiometry of organic matter (Tribble 1993). The nitrification (Eq. 2) of ammonia to nitrate tends to occur under oxic conditions but in the dark (Glaze et al. 2022; Lu et al. 2020).

$$({\text{CH}}_{2} {\text{O}})_{x} ({\text{NH}}_{3} )_{y} ({\text{H}}_{3} {\text{PO}}_{4} )_{z} + x{\text{O}}_{2} \rightleftharpoons x{\text{CO}}_{2} + x{\text{H}}_{2} {\text{O}} + y{\text{NH}}_{3} + z{\text{H}}_{3} {\text{PO}}_{4}$$
(1)
$${\text{NH}}_{3} + 2{\text{O}}_{2} \rightleftharpoons {\text{NO}}_{3}^{ - } + {\text{H}}^{ + } + {\text{H}}_{2} {\text{O}}$$
(2)

Bacterially mediated oxidation of organic matter under low oxygen conditions can substantially impact inorganic carbon system equilibria (Coleman et al. 1985, 1993; Lin et al. 2018; Tribble 1993). Under low oxygen conditions, denitrification can occur via Eq. (3) (Boudreau 1996), however, the amount occurring relative to oxic respiration tends to be low due to low water column nitrate concentrations (Erler et al. 2013). The decomposition of organic matter can also be facilitated by sulphate-reducing bacteria, resulting in TA and hydrogen sulphide production (Eq. 4).

$$\begin{aligned} & 5({\text{CH}}_{2} {\text{O}})_{x} ({\text{NH}}_{3} )_{y} ({\text{H}}_{3} {\text{PO}}_{4} )_{z} + 4x{\text{NO}}_{3}^{ - } \rightleftharpoons \\ & x{\text{CO}}_{2} + 4x{\text{HCO}}_{3}^{ - } + 2x{\text{N}}_{2} + 5y{\text{NH}}_{3} + 5z{\text{H}}_{3} {\text{PO}}_{4} \\ \end{aligned}$$
(3)
$$\begin{aligned} & 2({\text{CH}}_{2} {\text{O}})_{x} ({\text{NH}}_{3} )_{y} ({\text{H}}_{3} {\text{PO}}_{4} )_{z} + {\text{xSO}}_{4}^{{2 - }} \rightleftharpoons \\ & 2x{\text{HCO}}_{3}^{ - } + x{\text{HS}}^{ - } + x{\text{H}}^{ + } + 2{\text{yNH}}_{3} + 2{\text{zH}}_{3} {\text{PO}}_{4} \\ \end{aligned}$$
(4)

Organic matter oxidation under sulphate-reducing conditions produces TA and DIC in a 1:1 ratio, plus an additional amount of TA associated with the ammonia released via the N:C (y:x) stoichiometry of organic matter. A Redfield stoichiometric of 550:30:1 (C:N:P) or similar is usually applied to reef systems (Atkinson and Smith 1983; Tribble 1993).

Organic matter oxidation can proceed before sulphate reduction via Eqs. (5 and 6) if manganese and iron are present.

$$\begin{aligned} & ({\text{CH}}_{2} {\text{O}})_{x} ({\text{NH}}_{3} )_{y} ({\text{H}}_{3} {\text{PO}}_{4} )_{z} + 2x{\text{MnO}}_{2} + 4x{\text{H}}^{ + } \rightleftharpoons \\ & x{\text{CO}}_{2} + 2x{\text{Mn}}^{{2 + }} + y{\text{NH}}_{3} + z{\text{H}}_{3} {\text{PO}}_{4} \\ \end{aligned}$$
(5)
$$\begin{aligned} & 9({\text{CH}}_{2} {\text{O}})_{x} ({\text{NH}}_{3} )_{y} ({\text{H}}_{3} {\text{PO}}_{4} )_{z} + 4x{\text{FeOOH}} + 4x{\text{SO}}_{4}^{{2 - }} \rightleftharpoons \\ & 9x{\text{HCO}}_{3}^{ - } + 4x{\text{HS}}^{ - } + x{\text{H}}^{ + } + 4x{\text{Fe}}^{{2 + }} + 6x{\text{H}}_{2} {\text{O}} + 9y{\text{NH}}_{3} + 9z{\text{H}}_{3} {\text{PO}}_{4} \\ \end{aligned}$$
(6)

While the manganese and iron reduction pathways are important in many sediment porewater systems, for reef systems where carbonate-rich sediments dominate, the manganese and iron reduction pathways can be ignored due to low manganese and iron concentrations (Chambers et al. 2001; Werner et al. 2006).

Reaction pathways 1 to 4 reduce pH and saturation state through the release of DIC. For reaction pathway 4, the oxidation of hydrogen sulphide back to sulphate would further lower pH and saturation state. The lowering of pH will initially induce the dissolution of metastable carbonate minerals, e.g. HMC, until the build-up of TA in interstitial waters leads to oversaturation with respect to a more stable calcium and magnesium carbonate polymorph. In the presence of HMC, dissolution is likely to buffer the carbonate system such that HMC dissolution and the precipitation of a more stable carbonate mineral phase—via incongruent dissolution—are likely to occur simultaneously.

Hydraulic conductivity within the reef framework plays a significant role in water movement and the respiration of organic matter. Bulk water movement through high hydraulic conductivity/high permeability zones can have water velocities on the order of 10 m d−1 in the vertical direction and residence times on the order of days (e.g. Buddemeier and Oberdorfer 1986; Oberdorfer and Buddemeier 1986). In contrast, water movement through frameworks with low hydraulic conductivity has much lower velocities, translating into long residence times. For reef frameworks with high hydraulic conductivity and low residence times, substantial water exchange can facilitate the transport of organic matter, potentially leading to increased respiration and HMC dissolution. In frameworks with low hydraulic conductivity, the transport of organic matter is limited, thus reducing respiration, which influences carbonate precipitation and dissolution (Oberdorfer and Buddemeier 1986).

For Davies Reef, Buddemeier and Oberdorfer (1986) noted high hydraulic conductivity and water velocities and low residence times for unconsolidated sediments. The geochemistry for porewater irrigating these sediments indicated that the system was anoxic with measurable sulphide concentrations but with most alkalinity concentrations lower than seawater, indicating net precipitation of calcium carbonate.

This work investigates whether diagenetic calcite precipitation occurs under anoxic conditions along the windward margin of One Tree Island, Queensland, Australia. We show the dissolution and precipitation of calcium carbonate polymorphs can occur within reef frameworks characterised by low hydraulic conductivity (Buddemeier and Oberdorfer 1986). Our results come from the low permeability algal ride pavement at One Tree Island. We also compare our results to those reported from permeable carbonate sands of the Checker Reef system, Hawaii (Tribble 1990; Tribble et al. 1990).

Methods

Sampling site selection

Sampling was conducted in January 2015 and May 2016 at One Tree Island, which is located in the Capricorn-Bunker group of islands at the southern end of the Great Barrier Reef (Fig. 1). The One Tree Island reef platform consists of a relatively thin layer of Holocene reef growth on top of older Pleistocene reefs that developed during interglacial high stands between approximately 450 and 125 ka (Davies and Kinsey 1977; Dechnik et al. 2015). Reef growth at One Tree Island recommenced following the inundation of the Pleistocene platform, which occurred between 9 and 10 ka (Davies and Kinsey 1977; Dechnik et al. 2015). Maximum accretion rates are typically observed in areas of high hydrodynamic energy (typically exposed windward margins), with exposed reefs tending to accrete lagoon-ward (Dechnik et al. 2016; Marshall 1986). The southern and eastern facing flanks at One Tree Island have well-developed algal rims and leeward reef flats ranging from 200 to 600 m in width between the reef crest and lagoon (Fig. 1). The age of the One Tree Island algal ridge ranges between 2 and 7 kyr (Dechnik et al. 2015; Harris et al. 2015; Marshall and Davies 1982).

The algal ridge at One Tree Island is characterised by CCA pavements that host a variety of algal turfs in addition to rubble deposited by destructive events (typically storms). Five sites were chosen for the installation of shallow boreholes to sample the upper 50 cm of the subsurface algal ridge framework and interstitial porewaters (sites 1 to 5). Carbon system dynamics on the surface of the algal ridge were determined by sampling water from algal pools at low tide (Fig. 1). Sites 1 to 4 form a transect across the algal ridge from the reef crest to the rubble zone, and site 5 was selected as a comparison for the transect sites. Cores were collected at site 1, located near the point where waves broke at low tide (and near to the isolated algal pool that was investigated), site 3, located 100 m further towards the lagoon and near the centre of the algal ridge, and site 5, located 200 m to the east near the centre and apex of the algal ridge.

Pool and dome experiments

A shallow pool located on the algal ridge near sites 1 and 2 was sampled periodically at low tide between 6 January 2015 and 8 January 2015 (Fig. 2). As the tide recedes, various pools receive water draining across the algal rim from the direction of the lagoon towards the open ocean. Some of these pools become isolated at low tide. The average salinity for the algal pool we sampled during the study period was 35.5 ± 0.1. The composition of the benthic community within the algal ridge area was surveyed at approximately 20 m intervals and was categorised as follows: algae, foraminifera, sand and rubble and other invertebrates (Fig. 3). The ‘turf’ itself was made up of algae, which blanketed the substrate and were usually exposed at low tide. The other dominant groups were benthic foraminifera and other invertebrates.

Fig. 2
figure 2

Sampling the algal ridge benthic community at One Tree Island reef. A Isolated algal pool sampled at low tide. B Benthic dome setup used for experiments. C Isolated algal pool with benthic dome setup installed. D Carbonate core drilled from a site adjacent to borehole site S2. E Close-up of the upper section of the core. Note the consolidated nature of the core and the algal turf on top of the core

Full size image
Fig. 3
figure 3

The reef flat benthic community for our study area at the One Tree Island reef

Full size image

Two perspex dome experiments were also set up to quantify carbonate production and dissolution on the algal ridge (Fig. 2B and C). The domes were placed within an algal pool and filled with surrounding algal ridge surface water, and all air was removed from each dome to prevent further gas exchange. A flexible extrusion of closed-cell foam (Moroday Handy-Seal™) was applied to the base of each dome to improve the seal between the dome and the benthos. Domes were placed on the algal ridge turf environments, effectively isolating a section of benthos and allowing the photosynthesis, respiration, calcification and dissolution processes within the domes to be recorded as variations in seawater TA and DIC chemistry. A temperature data logger (Aquatic 2, TG-4100) was placed in each dome, and seawater was circulated within each dome by pumping water (8 L min−1) using a 12 V battery-powered submersible pump (Topsflo, China) out through a side port back into the dome through a second port located on top of the dome. A tee was placed inline so that water circulating within the dome could be sampled periodically for dissolved oxygen (DO), pH, and TA at approximately 20 to 30 min intervals for about 3 h. For the dome experiments, two 60 mL samples were typically collected for pH and TA analysis.

Borehole preparation and sampling

Small boreholes (24 mm diameter) were drilled to four depths (10 cm (site 2 only), 20 cm, 30 cm and 50 cm) at multiple locations (Sites 1–5) across the algal ridge using a portable hammer drill (Ramset, Australia) fitted with a tungsten carbide bit. Triplicate boreholes were drilled to each depth at each site. The sediment produced during drilling was flushed from the boreholes using a portable water pump to avoid the potential for dissolution of the fine calcium carbonate particulates produced by drilling.

Borehole porewater samplers were constructed from 19 mm diameter pipe end caps and were perforated with 3 mm holes around the cylindrical face and connected to the surface using 4 mm (internal diameter) low-density polyethylene tubing (Fig. 4). Each cap was covered with a 1 μm nylon mesh membrane to prefilter extracted porewater. The tube and filter were connected with custom-made reinforced rubber washers that produced a tight fit with the sides of the borehole. Each micro borehole was sealed with a threaded cap to prevent seawater from entering the well during submersion, which could be removed for sampling when required. After installation, each borehole cavity was backfilled with a fast-setting cement-based grout (SikaGrout-UW, Australia).

Fig. 4
figure 4

A Borehole sampler for interstitial water sampler and B Installation of a sample in a borehole

Full size image

The boreholes were left undisturbed for at least four days after installation in January 2015 to allow the porewaters to re-equilibrate. Boreholes were sampled during the day at low tide. Due to the consolidated algal ridge's low permeability (Fig. 2E), sampling the porewaters was challenging. We used a 60 mL syringe to extract porewaters. The low hydraulic conductivity required a significant force to extract water from many bores, thus, porewater sample volumes were restricted to 60 mL or less.

At each site, water was sampled for total sulphide, DO, TA, and pH analysis. Samples for TA (~ 25 mL) were 0.2 μm syringe filtered (30 mm diameter, PES membrane) immediately after collection. There was some uncertainty as to whether the boreholes would be fully re-equilibrated only four days after the disturbance of creating them. To account for this uncertainty, the boreholes were resampled in May 2016 and subject to the full suite of measurements as in 2015. These sets of measurements showed similar patterns, and the majority of the data considered here were from the 2016 sampling after a full year of equilibration (Table S3).

Pore-water analysis

Dissolved oxygen was measured to a precision of ± 0.01 mg L−1 using a portable meter (A323, Orion Star). The probe was calibrated according to the manufacturers' instructions. Total sulphide concentration was measured using the Methylene Blue Method (Method 8131, Hatch) on a portable spectrophotometer (DR2800 Hach). The calibration of the spectrometer followed the manufacturers' instructions. The method had a precision ± 8 μg L−1 based repeat sample analysis. Samples were measured immediately following collection to limit the oxidation and loss of any reduced sulphide species.

Unfiltered pH samples were stored immediately after collection in dark freezer bags containing ice blocks to limit biological and chemical reactions. pH samples were prioritised for analysis upon returning to the One Tree Island laboratory, which usually occurred within 1–4 h after collection. Total alkalinity samples were refrigerated and kept in the dark until analysis, usually within 2–6 h of the collection, with only a few stored for up to 24 h. Before sample measurement, sample bottles were warmed to 25 °C. pH measurements followed the m-cresol purple (mCP) method outlined by Liu et al. (2011) and were made with an Ocean Optics 4000+USB spectrophotometer fitted with a 5 cm cell with a temperature sensor and readback (Nand and Ellwood 2018). This method had a precision of ± 0.007 pH units for this study. Total alkalinity was determined using the one-point titration method of Nand and Ellwood (2018) (see supplementary online materials). Briefly, this involved accurately weighing the sample, adding hydrochloric acid containing the bromophenyl blue (BPB) dye and then reweighing the sample. The pH of the titrated sample was then determined by measuring the BPB absorption characteristics using the Ocean Optics 4000+USB spectrophotometer. Total alkalinity was calculated using Equations S1 to S5 (see supplementary online materials). The acid used for TA measurements was calibrated by titration against a Dickson-certified reference material (CRM). The analytical precision of multiple titrations of the Dickson CRM in the field laboratory at volumes similar to porewater samples was ± 16 μmol kg−1 (n = 47). This precision was lower than the “home” laboratory precision of < 2 μmol kg−1 (Nand and Ellwood 2018). Dissolved inorganic carbon, pCO2 and carbonate ion concentrations were calculated from measured pH and TA results using CBsyst, a Python module for calculating seawater carbon system chemistry (https://github.com/oscarbranson/cbsyst, https://doi.org/10.5281/zenodo.1402261),

Geochemical modelling

We developed equilibrium and kinetic expressions for the key reactions involved in organic matter oxidation and calcium carbonate dissolution. For this work, we considered Eqs. 1–6 and for mineral carbonate dissolution and precipitation via Eq. (7)

$${\text{Ca}}_{x} {\text{Mg}}_{{\left( {1 - x} \right)}} {\text{CO}}_{3} + {\text{CO}}_{2} + {\text{H}}_{2} {\text{O}} \rightleftharpoons 2{\text{HCO}}_{3}^{ - } + x{\text{Ca}}^{2 + } + \left( {1 - x} \right){\text{Mg}}^{2 + }$$
(7)

Organic matter oxidation via denitrification, manganese reduction and iron reduction processes (Eqs. 3, 5 and 6) were minor because seawater nitrate (Hatcher and Frith 1985; Silverman et al. 2012), dissolved manganese and dissolved iron concentrations (Fe < 10 nM, Ellwood unpublished data) at One Tree Island were generally low compared to available sulphate for reductive processing of organic matter.

Equilibrium modelling approach: Our geochemical equilibrium modelling of organic matter respiration under oxic and sulphate-reducing conditions followed an approach similar to that of Tribble (1993). For this model, discrete steps for organic matter oxidation occurred. Within each step, the speciation of the carbon, sulphur, nitrogen, phosphorus and boron species (HCO3−, CO32−, HS−, S2−, NH3, H2PO4−, HPO42−, B(OH)4−) were determined by solving a series of speciation equations simultaneously (see supplementary material and Equations S6 to S23; (Tribble 1990)). Mass balance equations were used to track changes in DIC, DO, sulphate, ammonia, phosphate and borate species associated with the incremental oxidation of organic matter under oxic followed by anoxic conditions. In situ conditions were applied to different solutions undergoing sulphate reduction in a closed system, i.e. no atmospheric oxygen or CO2 was exchanged, and no porewater was exchanged. We assumed a C:N:P: ratio of 550:30:1 for organic respiration (Atkinson and Smith 1983). Mineral carbonate dissolution and precipitation were also computed within lower and upper bounds and applied to each increment depending on the carbonate saturation state (Ω) (see supplementary material text, Equation S27 and Figure S1). An apparent mineral carbonate Ksp of 10−6.1 is used to calculate the carbonate saturation state (Figure S3). This value represents the apparent solubility and is based on the minimum carbonate ion concentrations calculated for bore samples based on their TA and pH values (Figure S3). Note that this value is larger than the aragonite solubility constant of 10−6.189 for standard seawater conditions but smaller than that for HMC, estimated to be ≈10−5.6 for 15% Mg in HMC (Drupp et al. 2016).

Kinetic modelling approach: Rate equations were developed for the oxidation of organic matter via Eqs. 1–6. Like the equilibrium mode, in the kinetic model, during each time step (dt), the speciation of the carbon, sulphur, nitrogen, phosphorus and boron species (HCO3−, CO32−, HS−, S2−, NH3, H2PO4−, HPO42−, B(OH)4−) were determined. The saturation state for carbonate mineral formation was calculated during each time step, and precipitation or dissolution of mineral carbonate was allowed to proceed via the following rate equations:

$$R_{{{\text{cal}}}} = k_{{{\text{cal}}}} \cdot {\text{An}} \cdot \frac{{\left( {{\Omega } - 1} \right)}}{{\left( {K_{{{\text{max}}}} + \left( {{\Omega } - 1} \right)} \right)}}$$
(8)
$$R_{{{\text{diss}}}} = k_{{{\text{diss}}}} \cdot {\text{An}} \cdot \left( {{\Omega } - 1} \right)^{n}$$
(9)

where Rcal and Rdiss represent the rates for precipitation and dissolution, respectively, An is the specific mineral reactive area in 1 kg water, Kmax is the half-saturation constant for precipitation (Gehlen et al. 2007), set to 0.4, and n is the reaction order, set to 2.3 (Gehlen et al. 2007). kcal and kdiss are kinetic rate constants for calcium carbonate precipitation and dissolution, respectively, and set to values of 0.036 × 10−3 mol m−2 h−1 (Nassar et al. 2018) and 0.2 × 10−3 mol m−2 h−1 (Andersson et al. 2007; Langdon et al. 2000), respectively. To calculate An, we assume an active reef framework area of 44 m2 with an average particle diameter of 1 mm and 10% voids (Table 1). Using this approach, the total reactive area equals 154 m2 kg−1. This value is less than 292 m2 kg−1 used by Barkouki et al. (2011) and Nassar et al. (2018) but is reflected in our percentage void volume and particle size. We set the model area to 200 m3, assume water only percolates vertically to a depth of 1 m and use variable hydraulic conductivities ranging from 0.02 m h−1 (low hydraulic conductivity) to 2 m hr−1 (high hydraulic conductivity) to simulate variable sediment permeabilities (Buddemeier and Oberdorfer 1986; Oberdorfer and Buddemeier 1986). From this, we obtained water residence times ranging from 5 h for a vertical velocity of 5 m d−1 for unconsolidated sediment to 200+hr for a vertical velocity of < 0.2 m d−1 for consolidated sediment. Water residence times for bores are likely to be on the order of 200+hours as they are consolidated, and they have low hydraulic conductivity. For kinetic model runs, we show a residence time equal to 40 h, so the model trajectories can be easily tracked (Figure S2).

Table 1 Composition of cores collected from the algal ridge at sites 1, 3 and 5
Full size table

We undertook two main experiments with the kinetic model. For these experiments, we initially set the dissolved organic carbon concentration (DOC) for water percolating within the framework to 6 and 21 μmol L−1, which represents the range of labile DOC concentrations obtained by Lønborg et al. (2018) for tropical coastal waters of the Great Barrier region. We also experimented with DOC concentrations up to 200 μmol L−1. Note that we only considered DOC and not particulate organic matter in the model as the consolidated nature of our sampling area likely restricts large particle movement within the framework. Rate constants for the various organic matter oxidation processes (Eqs. 1–6) were taken from Wang and Van Cappellen (1996). In the first experiment, we set TA, DIC, and DO concentrations to 2197 μmol kg−1, 1919 μmol kg−1 and 250 μmol L−1, respectively. This experiment simulates the evolution of TA, DIC, and DO for algal ridge water after it has been flooded with open seawater at high tide. In the second experiment, we set TA, DIC and DO concentrations to 2150 μmol kg−1, 1650 μmol kg−1 and 450 μmol L−1, respectively. This experiment simulates the evolution of TA, DIC, and DO to represent waters that could potentially infiltrate the reef framework during the day when photosynthesis is dominant (Figure S5).

Drill Core collection and dating

Drill cores were collected for mineralogical and petrological analysis at each site using a 25 mm by 500 mm diamond-tipped coring shaft. Core recovery was between 50 and 70% for the 4 cores collected.

Cores were photographed and sectioned along the vertical axis. A number of polished thin sections were prepared from selected core sections for mineralogical and petrological identification. Thin sections were imaged using a Leica DM600_M automated optical microscope (Berry et al. 2008). A Java-based scientific imaging software was used to process images for quantitative area analysis. Colour thresholds were adjusted, and masks were generated for calcite/aragonite, HMC, and voids. The particles were then analysed for the total area covered by the separate masks. The area of in situ CCA HMC was measured manually for each photograph, and the area was then subtracted from the total area of HMC in the thin section. This allowed us to differentiate the micritic HMC sedimentary infill from in situ CCA HMC.

Cores were also uranium-series dated by Laser Ablation Multicollector Inductively Coupled Plasma Mass Spectrometry (LA-MC-ICPMS) using the procedure outlined by Eggins et al. (2005). Uranium-series dating was complemented with radiocarbon dating of site 1 (see supplementary online materials).

Results

Mineralogy and petrography

Thin section analysis of drill cores confirmed that the investigated algal ridge at One Tree Island is well-cemented, dense and highly heterogeneous (Fig. 5). The two major mineral carbonate phases found were aragonite and HMC. Aragonite was present as skeletal fragments from corals, echinoids, molluscs, and gastropod shells. HMC was assigned to one of two groups based on whether it was in situ CCA material or micritic cement, which formed an inorganic matrix. The top of each core showed evidence of extensive boring, and the surface material was characterised as HMC rubble (Fig. 5). For nine of the twelve thin sections, HMC was the most abundant mineral phase, whereas, for the other three, aragonite dominated due to the presence of large coral fragments (Table 1; Fig. 5).

Fig. 5
figure 5

A Thin section of core 1 top shows extensive boring in the top few millimetres and large coral fragments (white material); B Thin section of site 5 core material also shows extensive boring in the top few millimetres from autotrophic and heterotrophic microborers, such as euendoliths, and macro borers, such as polychaetes, molluscs, and holothurians (Hutchings 2011; Schneider 2011). C Thin section of site 5 centre (32–34 cm depth) composed largely of coralline algae (brown layers) and a matrix of lithified coralline algae sand (brown matrix surrounding the white fragments)

Full size image

Uranium-series and radiocarbon ages for cores collected at sites 1, 3 and 5 ranged between 4 and 10 kyr, with most ages falling between 5 and 7 kyr. The ages for CCA and coral components were generally consistent except for site 3, where CCA ages were 2 to 3 ky younger than coral component ages (Table S2, Figure S4). The ages presented here are consistent with previously obtained ages for One Tree Island (Dechnik et al. 2015, 2016; Harris et al. 2015; Marshall and Davies 1982).

Algal ridge surface water chemistry

Changes in the water chemistry of tidally isolated pools on the surface of the algal ridge have been used to assess and quantify the geochemical and biological processes that contribute to algal ridge growth (e.g. Shaw et al. 2014; Silverman et al. 2012). Systematic temporal variation was observed throughout this study in all water chemistry parameters measured within the algal pools during the day and night (Fig. 6). Water temperature varied from a maximum of 29.2 °C to a minimum of 23.7 °C during the night between 6 and 8 January 2015. The average salinity of the algal pool across both day and night samplings was 35.5 ± 0.1. The TA for samples ranged between 2024 and 2494 μmol kg−1, pH ranged between 7.570 and 8.657, and DO ranged between 70 and 470 μmol L−1. During the day, DO, pH and carbonate ion concentrations all increased with time during the isolation of the pool (Figs. 2, 6 and S5). These changes were accompanied by decreases in TA, pCO2 and dissolved inorganic carbon (DIC). During the night, these trends were reversed (Fig. 6). The variations seen for the 2015 campaign are consistent with subsequent, more detailed measurements in 2017 and 2019 for algal ridge waters (Branson et al. 2023).

Fig. 6
figure 6

Contour plot of carbonate ion concentration as a function of DIC and TA. Overlayed are TA and DIC day and night values for the domes and algal pool. Also presented are values for the open seawater (n = 3) and borehole porewaters from 2015 and 2016. Contour lines represent constant carbonate ion concentrations, and vectors show photosynthesis/respiration using a Redfield stoichiometry of 550:30:1 (C:N:P) and CaCO3 precipitation/dissolution

Full size image

Borehole results

Porewater samples were obtained from 40 of the 42 micro boreholes inserted into the algal ridge. The majority of borehole samples contained measurable quantities of total sulphides (> 0.03 μmol L−1) (Table S3), with concentrations up to 67 μmol L−1. However, most also contained measurable oxygen (range 9 to 118 μmol L−1). As mentioned, for a number of bores, extracting the required volume of water required to make geochemical measurements was challenging. Thus, there was the possibility of exposure to oxygen during the extraction process. The presence of hydrogen sulphide is incompatible with DO, so we interpret the presence of oxygen in the samples as being introduced when drawing water out of the borehole or representing a mix of waters extracted from anoxic and oxic pores where waters have undergone differing degrees of organic diagenesis within the algal ridge framework (Buddemeier and Oberdorfer 1986). The sulphide concentrations reported in this study should not be considered exact but rather as the minimum possible concentration. That said, the presence of sulphide indicates anoxic conditions are present with the majority of bores sampled.

The TA of porewaters sampled in 2015 and 2016 ranged between 1614 and 2318 μmol kg−1 (Fig. 6). There was no correlation between TA and total sulphide concentration (Figure S6). Porewater pH ranged between 7.575 and 8.239. The low pH measurements obtained from specific boreholes are consistent with oxic and anoxic respiration of organic matter producing CO2 and the precipitation of mineral carbonate (Eqs. 1–4 and S29)—both respiration and precipitation reactions lower pH. A plot of TA versus calculated DIC highlighted a strong relationship (r2 = 0.89) with a slope of 0.90 ± 0.04 and an intercept of − 153 ± 8 μmol kg−1 over a ∆DIC and ∆TA range of 815 and 736 μmol kg−1, respectively (Fig. 7). The majority of porewater DIC and TA concentrations were below that measured for open seawater and surface algal pool water (Fig. 7) and fell close to a carbonate ion concentration horizon that ranged between 80 and 100 μmol kg−1 (Figs. 6 and S3).

Fig. 7
figure 7

The difference in TA and DIC for porewater samples relative to that of open seawater near One Tree Island with TA and DIC concentrations of 2197 μmol kg−1 and 1919 μmol kg−1, respectively. The black line is the best-fit line with a slope of 0.9 ± 0.04 and an intercept of − 153 ± 8 μmol kg−1, n = 79 and r2 = 0.89. The dark and light blue shaded areas represent 95% confidence and prediction levels, respectively

Full size image

Modelling

The modelling results are split into two parts. We first consider organic matter respiration and its influence on DIC and TA associated with the dome experiments, and then we examine how respiration might proceed in porewaters under anoxic conditions.

Algal ridge surface water modelling

Respiration processes in algal ridge surface waters are modelled as oxic processes (Eq. 1) until DO is depleted, then switch to anoxic processes (Eqs. 3 and 4). While at no stage did we observe anoxic conditions in the algal pool or in the domes (Figure S5), this modelling approach provides a tool for us to evaluate the trajectory of TA and DIC change as DO is consumed (Fig. 8). Starting at the open seawater TA and DIC concentrations when DO is in excess (i.e. DO = 250 μmol L−1), dome and algal pool water accumulate large amounts of metabolic CO2, as highlighted by a 150 μmol kg−1 increase in DIC before the onset of carbonate mineral dissolution (Model 1, organic carbon equals 200 μmol L−1, Fig. 8A). This CO2 release reduces solution pH and hence the carbonate ion concentration in the water until the carbonate saturation concentration is reached and dissolution begins, which increases both DIC and TA increase (Fig. 8A). Once DO was entirely consumed, DIC and TA increased in approximately equimolar amounts, with the addition of TA associated with ammonia released from organic matter respiration responsible for deviations from a 1:1 ratio.

Fig. 8
figure 8

Kinetic model results for the evolution of surface water to porewater transition. Contour plot of carbonate ion concentration as a function of TA and DIC for dome, algal pool and porewaters (2015 and 2016) (as in Fig. 6). The results are overlayed with two model curves. Model experiment 1 (panels A and C) tracks changes in pH (A) and sulphide concentration (C) for a parcel of water as it percolates through the carbonate framework. The DOC values represent three concentrations (μmol L−1). The starting TA and DIC are 2197 μmol kg−1 and 1919 μmol kg−1, respectively, and DO was 250 μmol L−1. In experiment 2 (panels B and D), the starting TA and DIC are 2150 μmol kg−1 and 1650 μmol kg−1, respectively, and DO is 450 μmol L−1. Note the residence time was set to 40 h, hence the termination of the curve at a carbonate ion concentration of ~100 µmol kg-1  for the lowest DOC concentration

Full size image

Porewater modelling

Most of the TA and DIC results for porewater samples fall below that of open seawater (Fig. 6). Porewater bores were sampled during the day at low tide when photosynthesis and calcium carbonate production dominate on the algal ridge. The question is whether our daytime sampling of the bores biases our results to lower TA and DIC values as a result of mineral carbonate formation. For model experiment 1, we used starting open water TA and DIC values and compared this model experiment 2, where DIC was lowered, and DO was raised to represent algal ridge daytime waters (Fig. 8). The trajectory for experiment 2 is similar to that of experiment 1; however, DO is higher and DIC lower due to its DO production and DIC consumption associated with photosynthesis (Figure S5). At the lowest DOC concentration, we can see that calcium carbonate formation dominates the change in TA and DIC in both models, such that they both decrease until they hit the saturation horizon at a carbonate ion concentration of ~ 75 μmol kg−1 (Fig. 8). After that, both DIC and TA increase due to respiration and carbonate dissolution. The kinetic mode predicts sulphate reduction, such that sulphide production should occur even while calcium carbonate formation occurs (Fig. 8C and D). Experiments 1 and 2 spanned a duration of 40 h (Figure S2). This duration encompassed several day-night cycles and multiple tidal flooding events on the algal ridge. As a result, it is unlikely that the bore results are significantly biased by sampling conducted during the daytime.

At a higher DOC concentration of 21 μmol L−1, respiration proceeds, and DIC increases while TA decreases as a result of calcium carbonate formation. Again, once TA and DIC hit the carbonate saturation horizon, their concentrations increase. At a DOC concentration of 200 μmol L−1, respiration dominates where the increase in DIC is coupled with a slight decrease in TA associated with ammonia release, nitrification (if DO is present), denitrification, and then sulphate reduction (Fig. 8). When TA and DIC reach the carbonate saturation horizon, their concentrations increase as a result of mineral carbonate dissolution and ammonia release associated with anoxic respiration.

We used the equilibrium mode to constrain the boundaries for measured pore water pH and sulphide concentration evolution across four scenarios (Fig. 9). In two scenarios, carbonate mineral precipitation and dissolution were maintained at a Ω of 1, and the starting DO concentration was either 0 or 250 µmol L−1. Under these conditions, pH was buffered by mineral carbonate dissolution and precipitation. In the other two scenarios, carbonate mineral precipitation and dissolution were not constrained, i.e. the saturation states were allowed to differ from 1 with no carbonate precipitation or dissolution, and the starting DO concentration was either 0 or 250 µmol L−1. In the equilibrium model simulations where porewater pH was buffered by mineral carbonate dissolution and precipitation, the initial porewater pH is 7.63 even at low respired DIC concentrations (Fig. 9) and remained unchanged until the respired DIC concentration reached ~ 80 μmol kg−1, after which pH declined to values below 7.2 as respired DIC concentrations exceed 1000 μmol kg−1. In model simulations without carbonate dissolution and precipitation, the pH declined from a value of ~ 8.00 as respiration proceeded and did so faster when DO was initially set to 250 μmol L−1. For the unbuffered carbonate system, pH dropped below 7 with (250 μmol L−1) and without (0 μmol L−1) DO at total sulphide concentrations of 223 µmol L−1 and 736 µmol L−1, respectively (Fig. 9).

Fig. 9
figure 9

Porewater pH versus sulphide concentration for samples collected in 2015 and 2016 along with equilibrium model curves. Also plotted are the data from Tribble (1993) for comparison. Model curves are for equilibrium experiments conducted with (Ω = 0.99, and 1.01) and without (Ω = 0, and ∞) carbonate mineral dissolution and precipitation and with and without DO present (units in μmol L−1). Starting DIC and TA were set at 1919 ± 3 μmol kg−1 and 2197 ± 2 μmol kg−1 (n = 3), the value of surface ocean waters near One Tree Island (OTI). The curves provide a boundary for pH changes with sulphide production from organic matter oxidation

Full size image

The measured porewater pH and total sulphide concentrations mostly fall within the bounds of the carbonate buffered and unbuffered simulations (Fig. 9). The majority of the data lie close to the carbonate buffered simulation with DO set to 250 μmol L−1. This suggests that organic matter degradation and evolution of DIC and TA within porewaters proceeds initially via oxic and then anoxic respiration processes (i.e. Equations 1–4). The low sulphide concentrations also suggest the easily oxidisable organic carbon content of bore waters is low—lower than porewater concentrations for the Tribble (1993) study.

The majority of the borehole DIC and TA concentrations fall below those of surface ocean water near One Tree Island (Fig. 6). This requires net carbonate precipitation from seawater that enters the reef framework beneath the algal rim and is consistent with the evolution of modelled porewater DIC and TA concentrations, where mineral carbonate saturation is close to 1 (Fig. 8). Under this constraint, mineral carbonate precipitation lowers both DIC and TA until the carbonate dissolution horizon is reached. Both our kinectic and equilibrium model runs were able to mimic the large drop in porewater DIC and TA concentrations relative to the open seawater starting conditions, which is consistent with mineral carbonate formation within the reef framework beneath the algal ridge. Both models also mimic the expected pH decline with decreasing carbonate ion concentration.

Discussion

The consolidated lithological and mineralogical nature of the algal ridge where the cores were recovered and where the boreholes were placed is consistent with a dominance of algal turf on top of carbonate bindstone (Dechnik et al. 2015). The consolidated nature of the algal ridge results in low hydraulic connectivity, with hydraulic connectivity values likely to be similar to the reef plate section at Davies Reef (Oberdorfer and Buddemeier 1986). Standing water also remains during low tide for a number of pools on the algal ridge, which is consistent with this idea of low hydraulic connectivity (Fig. 2A and C).

Most measured porewater TA and DIC values are lower than surface ocean water at One Tree Island (Figs. 6, 7 and 8), which is consistent with porewater evolution involving the precipitation of carbonate minerals within the subsurface framework of the algal ridge. The prevalence of sulphate reduction at shallow depths (> 10 cm) beneath the algal ridge surface indicates cementation of the internal ridge framework is likely occurring under anoxic conditions (Buddemeier and Oberdorfer 1986). The presence of dissolved sulphide also indicates carbonate precipitation is not just the result of wave-driven pumping of well-oxygenated water into the framework interior as previously proposed (Ginsberg et al. 1971; Macintyre 1985; Matthews 1974; Perry and Hepburn 2008).

Our modelling shows that mineral carbonate precipitation is needed to generate porewater DIC and TA concentrations well below that of seawater that enters the subsurface reef framework (Figs. 7 and 8). However, our kinetic model, especially model run 1 lacks the ability to simulate incongruent carbonate dissolution and precipitation, which is likely to be needed to reach the lowest measured TA and DIC porewater concentrations (Fig. 8). Model run 2 does reach these low concentrations, however, the starting DIC and DO concentrations are different from the measured open water values, which resets the water chemistry over the reef flat at each high tide. Thus, the initial chemical composition of the water entering the reef framework is likely a blend of various water chemistries. These chemistries could include those from different times of day (day and night) and tidal states (low and high tide), all of which are influenced by multiple daily tidal cycles that wash over the algal ridge.

Mineral carbonate precipitation can lower the carbonate ion concentration to the point where porewater concentrations lie near the saturating carbonate ion concentration of approximately 75 µmol kg−1 (Figs. 6 and S3). The precipitation of mineral carbonate also lowers pH, such that pH alone cannot be used as the sole diagnostic tracer of diagenesis within porewaters. However, the presence of measurable amounts of sulphide within porewaters indicates diagenetic processes linked to sulphate reduction are occurring within the reef framework (Fig. 9), consistent with measurements for Davies Reef (Buddemeier and Oberdorfer 1986). Sulphate-reducing bacteria have been implicated in the production of cryptic carbonate production that occurs within the reef framework (Heindel et al. 2012, 2010; Pigott and Land 1986; Webb et al. 1998). Our results are consistent with this idea.

The low pH values of porewaters from beneath the algal ridge at One Tree Island have also been observed in other reef systems, e.g. Davies Reef (7.74 to 7.76), Oahu (7.48 to 7.60), Marathon Key (7.61–7.87), Checker Reef (7.68 to 7.78), and Shiraho Reef (Buddemeier and Oberdorfer 1986; Drupp et al. 2016; Sansone et al. 1990; Tribble et al. 1990; Yamamoto et al. 2015). However, in these other cases, TA and DIC values are generally higher than for open ocean waters. This requires mineral carbonate dissolution to evolve TA and DIC to concentrations higher than for open ocean waters. This is particularly well documented for sediments where chemical diagenesis and dissolution of calcium carbonate occur under anoxic conditions with sulphate reduction (e.g. Krumins et al. 2013; Pigott and Land 1986; Tribble 1993; Yamamoto et al. 2015).

Our equilibrium model (Fig. 9) allows us to estimate changes in DIC and TA concentrations associated with both oxic and anoxic respiration and changes associated with carbonate dissolution and precipitation. For example, when assuming a starting seawater DO concentration of 250 μmol L−1 and taking the sample with the highest measured sulphide concentration (66 µmol L−1), and an estimated DIC change of + 382 μmol kg−1 produced by associated respiration processes, the model predicts a concurrent change in TA of + 155 μmol kg−1. Adding these concentrations to open water values produces DIC and TA values of 2301 μmol kg−1 and 2352 μmol kg−1, respectively, for a system without carbonate buffering and 2367 μmol kg−1 and 2484 μmol kg−1, respectively, for a system buffered by carbonate mineral dissolution (Fig. 10). These model results are higher than the measured in situ DIC and TA concentrations of 2001 μmol kg−1 and 2088 μmol kg−1, which indicates a reduction of DIC and TA through mineral carbonate formation.

Fig. 10
figure 10

A TA versus DIC for porewater samples collected in 2015 and 2016. The symbol colours indicate the porewater sulphide concentration, while the contour lines represent the carbonate ion concentrations. B TA versus DIC for a measured porewater water sample containing 66 µmol L−1 of sulphide. Also plotted are porewater model results, assuming oxic followed by anoxic respiration for open seawater to produce 66 µmol L−1 of sulphide. The two theoretical results are for a system with and without carbonate buffering, i.e. Ω carbonate saturation state boundaries are between 0.99 and 1.01. The triangle sample represents open seawater where precipitation of mineral carbonate has occurred until Ω is between 0.99 and 1.01, but with no respiration. Panels C and D are similar to panels A and B, except the data are taken from Tribble (1990, 1993), and calculations are for a water sample containing 484 µmol L−1 of sulphide. Note the change in scale for DIC, TA and H2S concentrations between panels A, B and C, D

Full size image

Applying our equilibrium model to the data reported for Checker Reef, Hawaii, we find for Tribble's highest measured sulphide concentration of 484 µmol L−1 (Tribble 1993), a predicted DIC change of +1215 μmol kg−1 and TA change of + 1034 μmol kg−1, which corresponds to buffered (and unbuffered) DIC and TA values of 3188 μmol kg−1 (3134 μmol kg−1) and 3339 μmol kg−1 (3231 μmol kg−1). These values closely match Tribble’s measured DIC and TA values of 3108 μmol kg−1 and 3141 μmol kg−1, respectively, and are clearly higher than open seawater. This indicates mineral carbonate dissolution is associated with oxic and anoxic respiration processes within the Checker Reef system, with little or no net carbonate precipitation occurring (Fig. 10).

A major difference between the One Tree Island and Checker Reef systems and, indeed, other permeable reef systems (e.g. Cyronak et al. 2013; Drupp et al. 2016) is the nature of the substrate through which the seawater percolates. The depth where porewaters were collected on the Checker Reef system allowed large volumes of water to be collected at flow rates > 500 ml min−1 from well samplers (Sansone et al. 1988; Tribble 1993). The residence time of water within the Checker reef framework is between 1 and 7 days (Tribble 1990). In contrast, the One Tree Island algal ridge pavement and subsurface are densely cemented, comprising fragments of branching corals and mollusc shells encrusted and bound by coralline algae (Fig. 5) (Dechnik et al. 2015; Marshall 1983; Marshall and Davies 1982). Water movement beneath the One Tree Island algal ridge on the windward side of the island is restricted, with residence time likely to be similar to those measured by Oberdorfer and Buddemeier (1986) for the consolidated reef plate at Davies Reef. This limited water exchange, a saturation index higher than 1 relative to less soluble mineral carbonate polymorphs and the presence of sub-oxic and anoxic bacteria, e.g. sulphate-reducing bacteria, provides an environment where mineral carbonates can form. We argue that this provides a mechanism for the precipitation of calcium carbonate, thereby reducing TA and DIC values below that of slowly ingressing water. This is consistent with what Buddemeier and Oberdorfer (1986) observed for Daiveis Reef, and the work by Marshall (1983) where he observed ubiquitous and abundant mineral carbonate cement formation occurring near where we collected our cores at One Tree Island. These cements include magnesium carbonate cement in the form of bladed spar, micrite, peloids and aragonite cement. In contrast, for other more permeable reef systems (e.g. Cyronak et al. 2013; Drupp et al. 2016), where there is active water movement through the permeable system, extensive organic matter degradation by sub-oxic and anoxic bacteria leading to mineral carbonate dissolution exceeding that of cement formation, raising TA and DIC values above that of ingressing seawater.

Recent work by Kessler et al. (2020) highlighted elevated saturation state values relative to the theoretical dissolution values for the dissolution of permeable calcium carbonate reef sands using a flow-through reactor. They also found that microbial fermentation reactions result in the breakdown of organic material and the production of hydrogen gas and DIC (Dong et al. 2023). In their kinetic experiments, the fermentation reactions were found to consume alkalinity through the production of organic acids, which lower TA. If this were to occur in a confined porewater environment, such as in the algal ridge framework at One Tree Island, then TA concentration would conceivably decrease below that of the ingressing seawater. At the same time, the DIC concentration would increase, resulting in a lower carbonate ion concentration and saturation state. While we do not have residence time measurements for porewater samples from the algal ridge, we expect them to be long on the order of days to weeks. Under these conditions, we might expect to reach quasi-equilibrium between mineral carbonate dissolution and carbonate formation at a carbonate ion concentration of approximately 75 μmol kg−1 (Figure S3).

Agal ridge surface water

The carbonate chemistry changes of surface water on the One Tree Island algal ridge between day and night are consistent with photosynthesis/respiration and calcium carbonate production/dissolution. The observed TA versus DIC slope of 0.42 ± 0.03 (n = 24) is consistent with organic matter production combined with calcium carbonate formation. Net community production contributes 20% to the change in DIC, with the remainder associated with calcification (Fig. 6). The changes in DIC associated with organic matter production and calcification also influence the pCO2 concentration. Calcification is expected to raise pCO2, while organic matter production will lower pCO2. Measured pCO2 values during the day dropped as low as 54 µatm (range 54 to 196 µatm), thus favouring HMC preservation through an increase in carbonate ion concentration but potentially limiting photosynthetic production (Ordoñez et al. 2019).

Nighttime changes in DIC and TA are coupled and increase with time following partial isolation from open seawater at low tide. The nighttime TA:DIC slope is 0.61 ± 0.06 and is significantly different to that measured during the daytime (Fig. 6). This nighttime TA:DIC slope is consistent with net respiration of organic matter and dissolution of carbonate minerals promoted by the accumulation of respired CO2. The pCO2 concentration reached as high as 1522 µatm (range 775 to 1522 µatm), with an estimated respired CO2 contribution of 70% to the change in DIC, and 30% from carbonate mineral dissolution.

The nocturnal changes in TA and DIC associated with the respiration of organic matter and dissolution of carbonate minerals for the dome and algal ridge experiments contrast porewater TA and DIC. While TA and DIC increased relative to open seawater values for the dome and algal ridge samples, porewater TA and DIC values decreased (Fig. 6). The carbonate ion concentrations for algal ridge samples and porewaters samples are similar, ranging between about 80 and 100 μmol kg−1, and close to the HMC saturation horizon, however, the contrasting TA and DIC behaviour between the dome and algal ridge samples and porewaters highlights the impac of sulphate-reducing bacteria and their influence on cryptic carbonate production under anoxic conditions.

These results have implications for how algal ridges may be affected by future ocean acidification. Incongruent dissolution represents a pathway through which metastable magnesium carbonate polymorphs are replaced by more stable carbonate minerals, potentially increasing the resilience of algal ridges to future changes. However, this process requires both the production of TA by sulphate-reducing bacteria under anoxic conditions and a low connectivity internal ridge framework that allows the products of dissolution to build up and buffer the system. While ocean acidification is not likely to directly alter the rate of sulphate reduction, it could change conditions within the algal ridge framework. If ocean acidification were to result in increased dissolution on top of and within the algal ridge framework, it may increase the hydraulic conductivity of the algal cement and allow oxic conditions to occur to greater depths, inhibiting both sulphate reduction and the buffering of interstitial waters, and subsequently preventing the cementation of the interior algal ridge occurring. Given the buffering of interstitial waters by dissolved mineral phases, the prevalence of anoxic conditions in close proximity to the ridge surface, and the tightly consolidated nature of the algal ridge presently observed, this appears unlikely to occur in the near future.

Conclusions

The precipitation of carbonate minerals within the interior of the algal ridge framework at One Tree Island occurs under largely anoxic conditions and is likely driven by TA produced by sulphate-reducing bacteria. Porewater TA and DIC results indicate that cryptic carbonate production removes TA and DIC from interstitial waters under thermodynamically favourable conditions created by sulphate-reducing bacteria. Kinetic and equilibrium modelling of porewater conditions indicates that TA and DIC evolution, involving carbonate precipitation, occurs within the subsurface framework of the algal ridge under anoxic conditions and is not a result of wave-driven pumping of well-oxygenated water into the framework interior. Using hydrogen sulphide concentrations, we estimate up to 153 μmol kg−1 of TA is produced from respiration processes within the algal ridge framework. Interestingly, we estimate that significantly more TA is removed through carbonate mineral production than is produced through sulphate reduction, such that the majority of porewater TA concentrations are below the open seawater concentration. The precipitation of mineral carbonate also lowers interstitial water pH, so that pH alone cannot be used as a definitive tracer for organic carbon diagenesis. The simultaneous precipitation and dissolution of carbonate minerals suggest that diagenetic stabilisation occurring within reefs may make them less susceptible to ocean acidification than CCA-driven surface consolidation, as interstitial precipitation occurs at pHs far lower than surface waters as a result of the sulphate reduction pathway. That said, if ocean acidification results in increased dissolution on top of the algal ridge framework, oxic conditions penetrate to greater depths, preventing the cementation of the algal ridge framework.