Sediment from the central section of the Great Barrier Reef (Australia) were collected from a fringing near-shore reef off Magnetic Island (Geoffrey Bay) in May 2014 and from the reef flat of Davies Reef, a mid-shelf coral reef in May 2015. Sediments were transported to the facilities of the National Sea Simulator (SeaSim) at the Australian Institute of Marine Science (Townsville, Queensland, Australia) on the day of collection. Prior to the start of experiments, sediment were maintained in flow-through aquaria receiving 2 L min⁻¹ filtered seawater of ambient pCO₂ under low light conditions (~20 μmol photons m⁻² s⁻¹). The sediments were sieved (< 2 mm) to remove most macrofauna, homogenized and, inundated with filtered water (< 0.2 μm), added into chambers and flumes of the respective experiments.

Magnetic Island sediments were studied under diffusive conditions for 14 days. Sediments were placed into flumes with a working section (20 × 25 cm) to a height of 5 cm, covered by a 5 cm high water column. Each flume received seawater of either target pCO₂ of ~400 (“Control”), ~600 (“Medium”), and ~800 μatm (“High”) at an inflow rate of 1.2 L min⁻¹ (3 flumes per pCO₂ treatment). The unidirectional flow had a velocity of ~1 cm s⁻¹ and was passed through a diffuser to create laminar flow above the sediment. A light:dark cycle of 12:12 h with a ramp time of 1 h was established by white and blue LED modules (Sol White, Aqua Illumination, USA). The photosynthetically active radiation (PAR) at the sediment surface was ~350 μmol photons m⁻² s⁻¹. Davies Reef sediments were studied under advective conditions for 9 days. A 10 cm thick sediment layer was incubated under a 16 cm high water column within acrylic cylindrical chambers (diameter: 19 cm) that were sealed with a light-permeable lid. Each chamber received seawater at an inflow rate of 0.15 L min⁻¹ with three chambers each under a pCO₂ treatment of ~400 (“Control”) and ~1300 μatm (“High”). The water column was stirred by a transparent rotating disc (15 cm diameter, 1 cm thick) mounted 10 cm above the sediment surface. The disk rotated with 40 rpm, which was shown to create pressure gradients of ~2.5 Pa between the center and the edges of the chamber (Janssen et al., 2005). This is similar to the pressure gradients occurring at current velocities between 10 and 20 cm s⁻¹ (Huettel et al., 1996), which are frequently encountered on Davies Reef flat (Pickard, 1986). The light conditions were the same as for the diffusive experiment.

The seawater used for experiments was collected 500 m offshore from SeaSim, filtered and brought to the experimental temperature (Table 1). The respective pCO₂ levels were achieved using pure CO₂ as described by Ow (2016). pH was measured daily using a multi-parameter meter (HQ40d, Hach) with a glass electrode (IntelliCAL™ pH, PHC201) calibrated with pH_NBS buffers. The pH measurements were recalculated on the total scale (pH_T) by offset-correction using photometric pH measurements with Tris-buffered saline or m-Cresol (Dickson et al., 2007). On 3 days 250 mL of the overlying water was sampled and poisoned by addition of 125 μL of a saturated mercury chloride solution for analysis of total alkalinity (TA) by acid titration (Vindta 3C, Marianda, Germany).

The carbonate chemistry in the water column (Table 1) and porewater was calculated using the software package “seacarb” for R (Lavigne et al., 2011; R Core Team, 2015). Using the “carb” function we calculated pCO₂, dissolved inorganic carbon (DIC), and the saturation states of calcite (Ω_calc) and aragonite (Ω_arag) from TA, pH_T, temperature and salinity. To calculate the saturation states of the high-magnesium calcite (Ω_HMC) we utilized the function “Om” that computes Ω_HMC using the Mg-content vs. stability curve for minimally prepared biogenic Mg-calcites of Plummer and Mackenzie (1974).

To estimate porewater Ω, we used the measured porewater pH and a wide range of TA concentrations ranging between water column concentrations of 2,320–2,360 μmol kg⁻¹ and 4,000 μmol kg⁻¹.

Porosity of the upper 0–1 cm was determined as the weight loss of a known volume of sediment after freeze-drying. Freeze-dried samples were used to determine total carbon (TC), total nitrogen (TN), and (after decalcification) total organic carbon (TOC) using an elemental analyzer (Euro EA 3000, EuroVector). Total inorganic carbon (TIC) was calculated as the difference between TC and TOC.

Grain size distributions obtained by dry-sieving (60°C, until constant weight) using a calibrated sieve stack were used to graphically determine the median grain size with the program GSSTAT; Poppe et al., 2004). For sediments incubated under advective conditions, the permeability of the upper 4 cm was determined with the falling-head method (Klute and Dirksen, 1986).

The mineralogical composition of the sediments was measured by X-ray diffraction. Dried bulk samples were pulverized (< 20 μm particle size) and prepared with the Philips backloading system. The X-ray diffraction was measured on a Philips X'Pert Pro multipurpose diffractometer. Identification of the minerals was performed using the Philips software X'Pert HighScore™. Semi-quantification of each mineral was based on Relative Intensity Ratio values calculated after Chung (1974). Mg-content of Mg-rich calcites was derived from reference minerals provided by the International Centre for Diffraction Data database.

The pH and oxygen microsensors were built, calibrated and used as described in de Beer et al. (1997) and Revsbech (1989), respectively. The tip size of the pH microsensors was 150 μm and a response time (t₉₀) of < 10 s. The oxygen microsensors had a tip size of 150 μm and a response time (t₉₀) of < 5 s. Details on the microsensor setup are described in Polerecky et al. (2007). In the diffusive experiment, parallel porewater profiles of O₂ and pH (step size: 400 μm) were obtained in the dark and light beginning after 4 h at the respective condition. Due to the measurement duration of each profile (10–15 min) and in order to increase replication, the data of 2–4 days were pooled. In the advective experiment, porewater pH profiles in the dark were obtained using a vertical steps size of 500 μm close to the wall (position A), between center and wall (position B) and in the center (position C) of each chamber on day 9. For this, the flow-through system was interrupted, the chamber lids and stirring discs were removed and the overlying water was decreased to a height of 10 cm. During the measurements, the water was gently mixed by a submersible pump at a velocity 1 cm s⁻¹.

Hyperspectral imaging was used to non-destructively estimate the abundance of MPB biomass from the chlorophyll content at the sediment surface over the duration of the advective experiment, following the methodologies described by Polerecky et al. (2009) and Chennu et al. (2013, 2015). The hyperspectral imager (Pika II; Resonon, United States) was mounted on a moveable rig and, during the scan of the chambers, captured back-scattered light off the sediment in 480 spectral bands (~2 nm resolution) over the range of 430–900 nm. During scanning, the sediments were illuminated by broadband light delivered by a light chain (GS-TZ-10, Taizhoe Tianze Lamp Industries Co Ltd., China) radially arranged around the chamber walls. The sediment's chlorophyll content was estimated from the logarithmic difference of the reflectance at the wavelengths of in-vivo chlorophyll absorption maximum (666 nm) and edge (703 nm). This calculation was performed at each pixel, and the values averaged over the imaged area of the sediment scanned. To calibrate the hyperspectral imaging signal for actual chlorophyll a content, three sediment cores (diameter: 2.8 cm) were inserted into each of the six chambers at the end of the experiment. From each core, the sediment surface was imaged and subsequently the top 5 mm of the sediment was sliced for chlorophyll a determination, and regression against the spectral index for chlorophyll. The MPB biomass was also estimated at the end of the diffusive experiment. For this, the upper 3 mm of the sediments on three randomly selected positions in each flume were collected for photopigment analysis using core liners (diameter: 1 cm). Photopigment samples from both experiments were immediately frozen (−80°C), photopigments were extracted as described by Al-Najjar et al. (2012) and measured using the HPLC method after Wright et al. (1991).

Net photosynthesis in the light and oxygen consumption in the dark was measured in both experiments. In the diffusive experiment, oxygen fluxes (JO₂) were calculated from the oxygen profiles according to Fick's first law of diffusion

where DO₂ is the diffusion coefficient of oxygen, [O₂] is the oxygen concentration and z is the depth. The JO₂ was calculated for the upper 0–1.5 mm of the sediment taking into account porosity (ϕ) and the diffusion coefficient was corrected for tortuosity (θ, D'O₂ = DO₂/θ²) adjusted for sands (θ² = ϕ^1−m, m = 2; Boudreau, 1997). Oxygen flux measurements in the advective experiment were performed on day 8. The water flow-through was interrupted and water column oxygen concentrations were measured before and after an incubation of 2–3 h. Oxygen fluxes were calculated as the concentration difference between the start and the end of an incubation taking into account the volume of the overlying water, the sediment surface area and the incubation time. To investigate the short-term response of low pH on oxygen consumption in sediments used for the advective experiment, we measured volumetric oxygen consumption rates by the percolation method described in de Beer et al. (2005). Homogenized sediment was filled into a core liner (diameter: 5 cm), which was closed from below with a rubber stopper containing a valve that allowed the percolation of water. The sediment was constantly covered by a 5 cm high column of air-saturated water with a pH_T of either 8.03 or 7.54, which was continuously percolated downward through the sediment at a rate of 30 mL min⁻¹. Using oxygen and pH microsensors, the values at 1 cm below the sediment surface were monitored. Once the values were constant, the percolation was stopped and the dynamics of oxygen and pH were monitored. The same procedure was then repeated with pH_T 7.54 water (n = 1 for each pH condition).

Sulfate reduction rates (SRR) were measured in sediments of the diffusive experiment using the whole core injection method according to Jørgensen (1978). On the last experimental day, one sediment core (26 mm diameter) was retrieved from each flume. Using a syringe with a hypodermic needle, 50 μL ³⁵SO42− (~1000 kBq) was vertically injected into each core, distributing an equal amount of tracer into each depth layer. After incubation for 6–6.5 h in the dark in a water bath (25.5°C), sediment cores were frozen (−20°C), sliced into depth intervals of 0–0.5, 0.5–1.5, 1.5–3, 3–5 cm and fixed in 10 mL of 20% (w/v) zinc acetate. Volumetric SRR were determined by the single-step chromium distillation technique after Kallmeyer et al. (2004) as modified by Røy et al. (2014). Porewater sulfate concentrations were determined by non-suppressed anion exchange chromatography (Metrohm 761 Compact IC, Herisau, Switzerland). Areal SRR were calculated by integrating the volumetric SRR over the measured sediment column.

Fluxes of dissolved organic carbon (DOC), nutrients and total alkalinity were measured in the advective experiment from water samples used for oxygen flux measurements. Samples for DOC were fixed using 250 μL HCl and stored at 4°C until analysis using a TOC analyzer (Shimadzu TOC-5000A, Japan). Nutrient samples (Si, PO43−, NO2−+NO3−, NO2−) were filtered through 0.2 μm cellulose-acetate filters, frozen (−20°C) and analyzed using standard methods (Ryle et al., 1981). Total alkalinity (TA) samples of 50 mL were poisoned with 100 μL of a saturated HgCl₂ solution and kept with no headspace until measurement by automatic gran titration using a Metrohm 855 titrosampler (Metrohm, Switzerland) as described by Vogel et al. (2016). The TA anomaly technique (Chisholm and Gattuso, 1991) was used to calculate net CaCO₃ fluxes, assuming that the ratio in fluxes between TA and CO32− is 2:1.

All statistical tests were performed using the statistical software R (R Core Team, 2015). To test for differences between means of pCO₂ treatments we used the analysis of variance (ANOVA) for the diffusive experiment (three treatment levels) and t-test for the advective experiment (two treatment levels) at a significance level of p < 0.05. All data was first tested for normal distribution and homogeneity of variances.

Within each experiment the sediment characteristics were similar in all pCO₂ treatments (Supplementary Table 1). Sediments from both Magnetic Island and Davies Reefs were carbonate sands (TIC content: 11.2–11.6% dry weight) with a similar median grain size (~0.49 mm) and the carbonate minerals were dominated by aragonite (83–85%), followed by 10–11% HMC (13%-Mg-calcite). Sediments used for the advective experiment were highly permeable (1.3–1.4 × 10⁻¹¹ m²).

The pCO₂ treatments had no differential effects on the growth of MPB in both the diffusion- (Supplementary Table 2) and advective experiment (Figure 1). The horizontal distribution of MPB biomass in sediments of the advective experiment was heterogeneous, with a local maximum in the center of all chambers (Figure 1A). The spectral signal vs. the actual chlorophyll a concentration showed a linear relationship with R² = 0.62 (data not shown). In both experiments, the MPB was dominated by diatoms as indicated by high concentrations of fucoxanthin (a marker pigment for diatoms; Supplementary Table 2).

Net photosynthesis in the light and oxygen consumption in the dark were not influenced by the pCO₂ level in the diffusive (Figure 2) and advective experiment (Figure 3). Over 24 h the sediments were net autotrophic, releasing 3–18 and 37–47 mmol O₂ m⁻² d⁻¹, respectively. Volumetric oxygen consumption rates in sediments used for the advective experiment were not affected by the pCO₂ treatment on the short-term (4.5 μmol O₂ L⁻¹ min⁻¹, Figure 4). This was measured at a depth of 1 cm using the percolation method. During the period when oxygen was still present, the porewater pH decreased to 7.95 when the initial porewater pH was 8.03 (Figure 4A), and remained stable when the initial porewater pH was 7.54 (Figure 4B).

Average areal sulfate reduction rates in the diffusive experiment were not affected by the pCO₂ treatment (Table 2) and contributed 4–12% to the total carbon mineralization.

Under advective conditions, the net precipitation of CaCO₃ in the light at high pCO₂ was 5.8 times lower than under control pCO₂ (Figure 3B). In the dark, the CaCO₃ dissolution was increased 2.7-fold under high pCO₂. On a daily basis the sediments were net precipitating CaCO₃ under control pCO₂, while they were net dissolving under high pCO₂. The nutrient and DOC fluxes were not influenced by the pCO₂ treatment nor light conditions (Supplementary Table 3).

Under diffusive conditions, the oxygen profiles in the light and in the dark were similar between all pCO₂ treatments (Figure 5A), showing that the pCO₂ treatment had no effects on oxygen production and consumption. When illuminated by light, the porewater pH in the upper 5 mm was influenced by the water column pH (Figure 5B). Namely, the peak in porewater pH at ~2 mm caused by photosynthetic activity of the MPB decreased with decreasing water column pH. In the dark, the influence of reduced water column pH was less pronounced, but resulted in a lower porewater pH close to the sediment surface (4–6 mm). At 1 cm depth the porewater pH approached 7.4–7.6 with increasing depth both in the dark and the light. In the advective experiment, the patterns in porewater pH followed the porewater flow path in the stirred chambers (Figure 6). The water column pH had the largest influence on the porewater pH in the inflow region at the edge of the chamber (position A) and in the upper 3–5 mm at the other positions. The porewater pH decreased along the porewater flow path from the edge to the center of the chamber (position C) with both pCO₂ treatments reaching a final pH of ~7.5 at depths >5 mm. The high pCO₂ treatment showed this pH already at position B, while the pH was 0.1–0.2 units higher in the control pCO₂ treatment.

Our study demonstrates in two independent experiments that medium-term exposure to elevated pCO₂ does not significantly influence biotic processes associated with photosynthesis and OM remineralization in coral reef sediments. A lack of pH effects could be explained by the pH conditions typically experienced by sediment microorganisms. Close to the sediment surface, microorganisms carrying out photosynthesis and aerobic respiration may be adapted to wide pH ranges due to diel pH variations up to 1 unit caused by their own activity (Werner et al., 2008; Schoon et al., 2010; Rao et al., 2012). In contrast, the similar sulfate reduction rates at all pCO₂ levels in the diffusive experiment can be explained by the buffered porewater pH in deeper sediment layers, where a major part of sulfate reduction occurs.

We expected a stimulation of photosynthesis and MPB growth by elevated pCO₂, reducing a possible CO₂ limitation that can occur when high photosynthetic activity shifts the carbonate system away from CO₂ (Cook and Røy, 2006). The absence of any CO₂ limitation could be due to the following reasons. First, the MPB community in our experiments was dominated by diatoms, which are known to possess very efficient carbon concentrating mechanisms that convert HCO3− to CO₂ (Reinfelder, 2011). Second, porewater advection can relieve MPB from CO₂ limitation by a rapid resupply of free CO₂ (Cook and Røy, 2006). Third, MPB may have been limited by inorganic nutrients rather than CO₂ (Uthicke and Klumpp, 1998). This is supported by the observation of the highest MPB biomass in the center of the advective chambers, where the longest flow path of porewater emerge at the surface, thus supplying the highest nutrient load (Huettel and Gust, 1992). Similar to our study neither photosynthesis nor OM remineralization were affected by elevated pCO₂ in coral reef sediments in Heron Island (GBR, Australia, Anthony et al., 2013; Cyronak et al., 2013a; Cyronak and Eyre, 2016; Trnovsky et al., 2016) and Moorea (French Polynesia, Comeau et al., 2014, 2015). However, in reef sediments from Sesoko Island (Okianwa, Japan) high pCO₂ stimulated photosynthesis and bacterial growth, while oxygen consumption was reduced (Sultana et al., 2016). Thus, biotic responses toward OA in reef sediments may differ between coral reefs suggesting the need for further study.

OA conditions drastically increased CaCO₃ dissolution in the advective experiment on Davies Reef flat sediments by changing the carbonate chemistry in the porewater. High pCO₂ turned the sediments from net precipitating (−6.0 mmol CaCO₃ m⁻² d⁻¹) to net dissolving (9.7 mmol CaCO₃ m⁻² d⁻¹) on a daily basis. Even though we made no measurements of CaCO₃ dissolution in the diffusive experiment, the lower porewater pH at elevated pCO₂ in the light and in the dark indicates a lower porewater Ω. This likely resulted in lower photosynthetic CaCO₃ precipitation in the light and increased dissolution in the dark that has also been described for other reef sediments under diffusive conditions (Cyronak et al., 2013a).

Although the two experiments are not directly comparable, generally higher dissolution rates can be expected under advective than diffusive conditions under similar pCO₂ conditions due to a deeper penetration of low Ω water into the sediments under advective conditions (Cyronak et al., 2013a). Indeed, in the advective experiment the water column pH affected deeper sediment layers, particularly at the inflow region of the stirred chambers. In contrast, in the diffusive experiment differences in porewater pH were limited to the upper sediment layers. Irrespective of the pCO₂ level and solute transport regime, an increase in sediment dissolution has been found by all studies under elevated pCO₂ (Anthony et al., 2013; Cyronak et al., 2013a; Comeau et al., 2014, 2015; Cyronak and Eyre, 2016; Trnovsky et al., 2016), suggesting that this OA response is common of coral reef sediments.

It has been suggested that sediment dissolution occurs faster under high pCO₂ because less metabolic CO₂ is required to reach the CCT, so that more metabolic CO₂ can drive dissolution (Andersson, 2015). This is supported by the temporal and spatial patterns in porewater pH in sediments studied under advective conditions. In the percolation experiment, an initial pH of 8.0 was reduced by metabolic CO₂ from aerobic respiration, suggesting that the CCT was not reached. In contrast, the CCT was likely reached at an initial porewater pH of 7.5 so that metabolic CO₂ addition resulted in dissolution that buffered the pH. This can also explain the patterns in porewater pH in the stirred chambers. Along the flow path from the chamber edge toward the center, the porewater accumulates metabolic CO₂. At high pCO₂ the CCT was reached already in the middle of the porewater flow path, because the pH of 7.5–7.6 did not decrease further along the path. Even though such low pH also occurred at control pCO₂, the pH gradually decreased along the flow path, indicating that the CCT was reached later and only close to the outflow. As the CCT is reached earlier at high pCO₂, more metabolic CO₂ is available to drive dissolution beyond the CCT. At similar CO₂ production rates this will result in increased sediment dissolution under high pCO₂, which we observed.

Our calculations of porewater Ω indicate that high-magnesium calcites (HMC) are the predominantly dissolving minerals. The porewater pH in sediments studied under both diffusion- and advective-dominated conditions never decreased below 7.4–7.6, demonstrating that the pH is buffered by the dissolution of the carbonate sediments. At decreased pH they start dissolving, thus stabilizing the pH. The studied sediments were both enriched in HMC (13%-Mg-calcite), which was the most soluble carbonate mineral present. In all elevated pCO₂ treatments the water column was undersaturated with respect to HMC, but not to aragonite. We calculated the porewater Ω using the minimum porewater pH_T of 7.4–7.6 and a wide range of TA concentrations ranging between water column concentrations (2,320–2,360 μmol kg⁻¹) and 4,000 μmol kg⁻¹ that are rarely exceeded in coral reef sediments (e.g., Andersson, 2015; Yamamoto et al., 2015; Drupp et al., 2016). Indeed, this resulted in undersaturating conditions with respect to HMC (Ω_HMC = 0.33–0.89), but not to aragonite (Ω_arag ≥ 1). This suggests that HMC is the dissolving phase in both studied sediments rather than the much more abundant aragonite. While the preferential dissolution of HMC is in line with findings in natural reef sediments (Andersson et al., 2007; Drupp et al., 2016), conditions of Ω_arag < 1 cannot be ruled out entirely as they still could occur in microenvironments (e.g., in hot spots of particulate organic matter, in grain crevices).

Comparing the oxygen fluxes and dissolution rates in the Davis Reef sediments with results from in-situ studies on Heron Island shows that our rates are realistic for reef settings. Oxygen fluxes measured in our study are in the range of in-situ measurements on the Heron Island reef flat during daytime (−167 to −16 mmol O₂ m⁻² d⁻¹), night (14 to 114 mmol O₂ m⁻² d⁻¹), and on a daily basis (−26 to 25 mmol O₂ m⁻² d⁻¹, Rao et al., 2012; Cyronak et al., 2013b; Cyronak and Eyre, 2016). Also our CaCO₃ dissolution rates in the control pCO₂ treatment are similar to in-situ estimates of dissolution during daytime (−86 to −2 mmol CaCO₃ m⁻² d⁻¹), night (4 to 57 mmol CaCO₃ m⁻² d⁻¹), and on a daily basis (−15 to 16 mmol CaCO₃ m⁻² d⁻¹, Rao et al., 2012; Cyronak et al., 2013a,b; Cyronak and Eyre, 2016). Furthermore, the increase in daily dissolution by 16 mmol CaCO₃ m⁻² d⁻¹ is only slightly higher than the 7–11 mmol CaCO₃ m⁻² d⁻¹ found in in-situ manipulation experiments that used the same stirring rate, but applied lower pCO₂ (800–950 μatm, Cyronak et al., 2013a; Cyronak and Eyre, 2016). This indicates that our rate estimates are reasonable and can be used to assess the ecological consequences of increased dissolution under high pCO₂ for Davis Reef.

The increase in sediment dissolution caused by OA will have only a small effect on NEC on Davies Reef flat. The current average NEC of the reef flat of 120 mmol CaCO₃ m⁻² d⁻¹ (Albright et al., 2013) are 7.5 times higher than the increase in dissolution rates under high pCO₂. Since sediments cover less than 30% of the flat area (Klumpp and McKinnon, 1989; Albright et al., 2013), the increase of dissolution rates will decrease NEC on Davies Reef flat by < 4%. However, sediment dissolution effects on NEC will be higher on reefs or reef areas with lower NEC and higher sediment cover (Comeau et al., 2015; Cyronak and Eyre, 2016). Therefore, future declines in calcification (e.g., Andersson et al., 2009; Dove et al., 2013) and increased bioerosion rates (e.g., Wisshak et al., 2012; Reyes-Nivia et al., 2013) could significantly increase the relative importance sediment dissolution for NEC.

Sediment dissolution rates will likely be higher in the Davis Reef lagoon, because it is deeper (5–27 m, average 16 m, Pickard, 1986) than the reef flat and there is less light for photosynthesis. Sediment dissolution increases when the ratio between daily gross photosynthesis and respiration (P/R) decreases (Rao et al., 2012; Cyronak and Eyre, 2016). While our estimates of oxygen fluxes are similar to results of in-situ studies on the Davies Reef lagoon (Table 3), the P/R of 2–2.4 in our study is much higher than the 0.4–1.2 found in-situ (Hansen et al., 1987, 1992). Thus, our dissolution estimates will likely underestimate the dissolution occurring in the lagoon sediments. However, benthic photosynthesis and respiration in reef sediments were shown to have a strong and linear control over dissolution (Cyronak et al., 2013b). We therefore estimated present day and future dissolution rates using in-situ oxygen fluxes from Hansen et al. (1987, 1992) and the linear relation of dissolution vs. oxygen flux under control and high pCO₂ in our study (Figure 7). This results in present daily dissolution of −0.5 to 4.0 mmol CaCO₃ m⁻² d⁻¹ and future dissolution under high pCO₂ of 14.5 to 18.7 mmol CaCO₃ m⁻² d⁻¹ (Table 3). These dissolution rates can be regarded as maximum because they do not account for possibly lower porewater advection rates in the finer lagoon sands (median grain size 1.2–2.9 mm, Hansen et al., 1987). The increase in sediment dissolution due to porewater advection using the same stirring velocities as in our study (40 rpm) was shown to be 2.5 higher than under diffusive conditions (Cyronak et al., 2013a). Considering this, our estimates for future daily dissolution under diffusive conditions are 5.5 to 9.9 mmol CaCO₃ m⁻² d⁻¹, which can be considered minimum.

We estimated how such increases in dissolution rates relate to the accumulation rate of sediments into the lagoon and how it could affect the mineralogy. Maximum sediment accumulation rates on the Davies Reef lagoon during the last 4,000 years range between 1,800 and 3,600 g CaCO₃ m⁻² y⁻¹ or 1.3–2.5 mm y⁻¹ (Tudhope, 1983). Considering a sediment porosity of 0.5 and the density of calcite (2.71 g cm⁻³), the increase in dissolution under high pCO₂ could reduce sediment accumulation by 0.2–0.4 mm y⁻¹, thus by 6–31%. This could slow down reef accretion and make it harder for lagoons to keep up with future sea level rise.

Furthermore, the mineral composition of the lagoon sediments may drastically change due to the preferential dissolution of HMC. The accumulation of HMC in the lagoon can be calculated as 10% of the total sediment accumulation, thus 180–360 g CaCO₃ m⁻² y⁻¹ or 4.9–9.8 mmol CaCO₃ m⁻² d⁻¹. Since HMC are likely the predominantly dissolving minerals, even our minimum dissolution estimates under high pCO₂ could reduce the accumulation rates of HMC by at least 66%. This indicates that Davies Reef lagoon sediments could experience a fast decline in HMC, which has also been suggest for other coral reefs (Haese et al., 2014). This could lead toward a higher relative abundance of more stable minerals like aragonite (Andersson et al., 2007). Compared to HMC, more CO₂ is needed to reach undersaturating conditions with respect to aragonite so that dissolution will then occur at a lower porewater pH that may affect sediment biota.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.