Threshold of carbonate saturation state determined by a CO 2 control experiment

Threshold of carbonate saturation state determined by a CO2 control experiment S. Yamamoto, H. Kayanne, M. Terai, A. Watanabe, K. Kato, A. Negishi, and K. Nozaki The University of Tokyo, Hongo 7-3-1-S1-737, Bunkyo-ku, Tokyo 113-0033, Japan Tokyo Institute of Technology, O-okayama 2-12-1-W8-13, Meguro-ku, Tokyo 152-8552, Japan National Institute of Advanced Industrial Science and Technology, Umezono 1-1-1, Tsukuba, Ibaraki 305-8568, Japan now at: Air Water Inc. 2-6-40, chikkoshinmachi, Nishi-ku, Osaka, 592-8331, Japan


Introduction
The oceans are a large carbon reservoir that absorbs atmospheric CO 2 , which then equilibrates to bicarbonate (HCO − 3 ) and carbonate (CO 2− 3 ) ions.More than 30 % of the CO 2 emitted into the atmosphere by human activities is taken up by the oceans (Sabine et al., 2004), lowering the pH of surface water and decreasing the saturation state (Kleypas et al., 2006).Future uptake of CO 2 by the oceans is predicted to reduce seawater pH by 0.3 to 0.5 units over the next few decades (Caldeira and Wickett, 2003).
The saturation state of seawater for a given mineral is a measure of the thermodynamic potential for the mineral to form or dissolve.The saturation state of calcium carbonate ( ) is defined as follows: As a result, will rise by carbonate dissolution.The carbonate-bicarbonate system acts as a buffer to atmospheric CO 2 levels and ocean acidification.Hence, carbonate dissolution results in an increase to the buffer capacity.If we are to predict future changes in the carbon balance between the atmosphere and the ocean accurately, and the degree of consequent ocean acidification, it is crucial that we understand this buffer system.
Previous studies have mainly investigated the effects of elevated pCO 2 on the net production and calcification of marine organisms by laboratory experiments.For example, Langdon and Atkinson (2005) showed that by 2065 the coral calcification rate will decrease by 40 % to 83 % compared with pre-industrial levels.Data from Gattuso et al. (1998), Leclercq et al. (2000), Marubini et al. (2001Marubini et al. ( , 2002)), and Reynaud et al. (2003), meanwhile, predict a decline of only 1 % to 18 %.Although there are differences between low-and high-sensitivity data sets, the calcification rates of marine organisms are expected to decrease.
The value of varies with carbonate mineralogy, and the solubility of Mg-calcite varies according to the magnesium content.In general, minerals with a higher Mg content show higher solubility, but the solubility of synthetic and biogenic Mg-calcite varies under laboratory conditions, and there is variability in the solubility of biogenic Mg-calcite from different organisms (Morse and Mackenzie, 1990).
At present, a (aragonite saturation state) values are 1 to 2 in high-latitude regions, and 3 to 4.5 in low-latitude regions.
Although carbonate dissolution at low latitudes (e.g.coral reefs) was not predicted, net dissolution under conditions of a > 1 has been observed at several coral reef sites (Boucher et al., 1998;Leclercq et al., 2002;Kayanne et al., 2005;Yates and Halley, 2006;Nakamura and Nakamori, 2009).These studies have shown that dissolution is possible even when a > 1, due to the high solubility of Mg-calcite.When the Mg content is >8 % to 12 %, Mg-calcite dissolves more readily than aragonite (Plummer and Mackenzie, 1974;Bischoff et al., 1987;Morse et al., 2006).Morse and Mackenzie (1990) divided the solubilities of Mg-calcite into three major categories: (1) Plummer and Mackenzie (1974) solubility; (2) the "best-fit" biogenic Mgcalcite solubilities (Bischoff et al., 1987;Walter and Morse, 1984); and (3) synthetic Mg-calcite solubilities (Bischoff et al., 1987;Mucci and Morse, 1984).Of these, category 1 has the highest values for Mg-calcite solubility, with the solubility of 12 % to 15 % mol Mg-calcite exceeding aragonite solubilities by a factor of five.Category 2 shows significantly higher solubilities than category 3 due to the instability of biogenic factors, although the overall trend in category 2 is similar to that in category 3.While categories 2 and 3 reflect the thermodynamic solubilities of biotic and abiotic Mg-calcite, it is likely that category 1 reflects kinetic (rather than thermodynamic) factors, including the retention of reactive surface particles due to minimal sample cleaning and lack of annealing (Bischoff et al., 1993;Morse et al., 2007).However, category 1 does reflect reactivity in nature.
The relationship between and the kinetic dissolution rate of synthetic and biogenic calcium carbonates in the laboratory has been described by Keir (1980).Hales and Emerson (1997) reviewed the result of Keir (1980) and made corrections to their original measurements.However, these dissolution rates are not consistent with those obtained from field observations.Moreover, no Mg-calcite dissolution experiments have been performed using natural seawater and used total alkalinity or dissolved inorganic carbon in estimating a values and dissolution rates.Indeed, it is pCO 2 that alters a ; consequently, laboratory studies need to set a values using pCO 2 to attain consistency with the environment.It is essential to determine the relationship between and the kinetic dissolution rate of biogenic Mg-calcite in order to understand naturally buffering systems.
The aim of this work is to clarify the relationship between a and the rate of Mg-calcite dissolution.We measured the dissolution rate of coralline aragonite and Mg-calcite excreted by several organisms under conditions of a > 1, using an experiment system that controlled pCO 2 in seawater.We measured the a threshold of biogenic Mg-calcite, calculated by total alkalinity (A T ) and dissolved inorganic carbon (C T ).

Experimental design
We designed an experimental system with conditions matching those of a natural coral reef.Carbonate sediments were collected from Shiraho reef and prepared with minimal treatment (ultrasonic cleaning and drying).Shiraho reef (24 • 22 N, 124 • 15 E) is situated at the southeast coast of Ishigaki Island.Natural seawater was used for the dissolution experiment and the a value of the seawater was controlled by CO 2 , rather than by HCl or NaOH.
The experimental system (Fig. 1) consists of four components: a seawater tank, a dissolution chamber, a CO 2 gas unit, and a flow-through analyzer for A T and C T .The experimental procedure is described below.
First, CO 2 gas (420 to 2210 ppm) levels were prepared using the CO 2 gas unit, imitating CO 2 conditions similar to those close to the present day (420 ppm), 2 × pre-industrial levels (∼560 ppm), 4 × pre-industrial levels (∼1120 ppm), and 8 × pre-industrial levels (∼2240 ppm).Seawater in the tank was circulated through the seawater line (bypass line) for 10 to 12 hours to allow the seawater and introduced CO 2 to equilibrate.The pCO 2 was then checked with a nondispersive infrared gas analyzer (NDIR: LI-820, LI-COR).After the pCO 2 had stabilized, A T and C T were determined using the flow-through analyzer before the experiment began.Samples were placed in the dissolution chamber, and seawater was circulated through the seawater line (dissolution line).A T and C T were measured again after several hours, both in the middle and at the end of the experiment.The experiment was performed as a time series, and A T and C T were analyzed periodically through its duration.
The conditions for each experiment are listed in Table 1.Temperature in the incubator was maintained at 26 • C and the dissolution experiment was performed at seven different pCO 2 levels for bulk sediment, four levels for coralline algae and foraminiferans, and three levels for coral.

System components
Seawater tank: CO 2 gas was introduced to the seawater by using a 3.5 m coiled tube with a single small hole to allow for bubbling and equilibration.The coiled tube is made of fluorocarbon polymers and is a closed system.Seawater pCO 2 in the tank was monitored continuously by using a membrane tube and NDIR similar to the system used by Saito et al. (1995).
Dissolution chamber: By using a pump, seawater was introduced to the dissolution chamber (600 ml) from the seawater tank.Carbonate samples for dissolution experiments were placed in this chamber.Mesh filters (200 µm) were affixed to both sides of the chamber to prevent loss of samples.
CO 2 gas unit: Gas mixtures of CO 2 at concentrations between 420 and 2210 ppm were prepared by mixing CO 2 free gas (passed through soda lime traps to remove trace CO 2 ) with pure CO 2 using two mass flow controllers, and then dissolved in seawater.The gas flow rate was 400 ml min −1 in all experiments.
Flow-through analyzer: The sample seawater was introduced directly to the flow-through analyzer (Kimoto Electric Company Limited) by switching the flow line (Kimoto et al., 2001;Watanabe et al., 2004).Certified reference materials (A.Dickson, University of California) were used to calibrate the system.Sodium carbonate solutions were used for the C T calibration.The analytical accuracies of A T and C T were within 3 µmol kg −1 and standard deviations of A T and C T were 1.1 µmol kg −1 and 2.1 µmol kg −1 respectively (Kimoto et al., 2002;Watanabe et al., 2004).

Samples
For the dissolution experiment, we used commercially available seawater (Nihon-Aquarium-Service Co., Ltd.) collected from a depth of 500 m at 34.7 • N, 139.4 • W (near Izu Islands, Japan) and sterilized by UV rays.The seawater was filtered using a 0.45 µm capsule filter.The carbonate samples used were coral, coralline algae, and sediment sampled from Shiraho reef at Ishigaki Island, Japan.We collected these samples from the surface sediment layer using a scoop.The sampling point was 600 m from the shoreline, in the center of a shallow lagoon in the Shiraho reef, where typical  The major size fraction of the coral reef sediment was from −1.0 to 0.0 φ (1 to 2 mm), from which 800 particles were separated for each dissolution experiment.We consider that our dissolution experiment represents the actual reef environment, because this size fraction (1 to 2 mm) represents 40 % of the size spectrum within the reef sediments.The mineralogy of each sample was identified by X-ray diffraction (XRD) analysis.The samples consisted of coral, foraminiferans, and coralline algae (see Fig. 2).The MgCO 3 content was estimated by the position of the peak X-ray strength (Goldsmith and Graf, 1958).The magnesium contents of the coralline algae and foraminiferans were 16.5 ± 0.4 and 13.3 ± 0.4 mol %, respectively, and the coral was pure aragonite.

Calculation
The A T of seawater increases by 2 moles for every 1 mole of calcium carbonate dissolution.The carbonate dissolution rate is measured by analyzing the change in A T .Salinity changes also affect the A T .Hence, small salinity changes that occurred due to evaporation over the course of the experiment were corrected for by using pre-determined relationships between gas flow and salinity change rate.Dry gas  is required to produce the exact pCO 2 values, so the bubbling of CO 2 into the seawater causes evaporation, and the salinity rises during each experiment.Table 2 shows a series of CO 2 volume flows with the resultant salinities after 7 h.Salinity increases in direct proportion to the gas mass flow, allowing it to be calibrated during the experiment according to the post-experiment salinity value.Figures 3 and 4 show time series of A T without and with salinity calibration.A T increased even in the experiment without a carbonate sample; however, after salinity calibration, the change in nA T (normalized total alkalinity) was about 2 to 3 µmol kg −1 , a value within the error range of the measurement.Using Eq. (3), A T was standardized to a constant salinity, and the dissolution rate was then calculated:

Gas Flow Salinity
where nA T is the normalized total alkalinity (µmol kg −1 ), representing the total alkalinity standardized to the salinity; S average is the average salinity during all the experiments; and S sample is the calibrated salinity according to Table 2. C T was also standardized to salinity, and nC T (normalized dissolved inorganic carbon, µmol kg −1 ) was obtained.Salinity was measured using a salinometer (PORTASAL 8410A, Guildline Instruments Limited).IAPSO (International Association for Physical Sciences of the Ocean) standard seawater was used for calibration.The precision of salinity analysis is ±0.003.Dissolution rates were calculated as follows: where nA T is the difference in nA T over the course of the experiment, m w is the weight of seawater, M is the molecular weight of calcium carbonate (=100), m s is the average weight Biogeosciences, 9, 1441-1450, 2012  of the carbonate sample over the course of the experiment, and t is the duration of the experiment.Due to the design of the system, a small amount of seawater remains in the pump and tubes at the end of each run.However, we can confirm that only a small amount of seawater (10 ml = less than 0.5 % of total volume) remained in the pump at the end of each experiment, and this was corrected for when determining the mass balance of A T .Seawater f CO 2 and a values were calculated from A T , C T , seawater temperature, and salinity, using the calculation program CO2sys (http://cdiac.esd.ornl.gov/oceans/co2rprt.html;DOE, 1994).The total scale for pH was used in all calculations, employing the equilibrium constants (K 1 and K 2 ) reported by Mehrbach et al. 1973;refit by Dickson and Millero, 1987).
The conditions for each experiment are listed in Table 1 and the results of all experiments are listed in Supplementary Material Table 1.The results of the bulk dissolution experiment are shown in Fig. 5.During the experiments, nA T varied between 6 and 25 µmol kg −1 .No increase in nA T was observed when pCO 2 was 420 ppm, and the change in nA T was highest when pCO 2 was 2030 ppm.
Figure 6 shows the average dissolution rates, calculated from nA T changes, plotted against averaged a .The average dissolution rate was fastest (4.0 × 10 −3 % h −1 ) when a was 1.3 and slowest when a was 3.7.
Similarly, Fig. 7 shows the results of dissolution experiments on coralline algae, foraminiferans, and coral, together with those of the bulk sediment.In all cases, the dissolution rate increases as a decreases.The threshold value of foraminiferan and coralline algal dissolution is 3.0 < a < 3.2, and that of the bulk sediment is 3.7 < a < 3.8.The dissolution rate of the coral shows no significant change when 1.5 < a < 2.0.

Discussion
At any given value of a , the relative dissolution rate is typically in the order of coralline algae > foraminiferans = bulk sediment > coral (Fig. 7).As in earlier studies (Bischoff et al., 1993;Morse et al., 2006Morse et al., , 2007)), the differences among samples presumably resulted from the solubility differences of minerals with varying Mg content.

Relationship between solubility and a
We compared the solubilities obtained in this study with those measured previously (Morse and Mackenzie, 1990) using the a threshold of coralline algae and foraminifera.The solubility of biogenic Mg-calcite is calculated using two different methods: Plummer-Mackenzie solubility (Plummer and Mackenzie, 1974) and biogenic best-fit solubility (Walter and Morse, 1984;Bischoff et al., 1987).The value of Mg-calcite (16 mole % Mg) is 0.2 by the former method and 0.8 by the latter, when a = 1.0.This discrepancy originates from differences in pretreatment and the experimental method (Bischoff et al., 1993;Morse et al., 2006).For example, the Plummer-Mackenzie method carbonate samples were washed in an ultrasonic bath and then dried, whereas for the biogenic best-fit solubility they were not only washed in an ultrasonic bath, but were chemically treated with H 2 O 2 to remove organic matter.
In the present work, the −log(K * sp ) values of coralline algae (16.5 mol ± 0.4 mole % Mg) and foraminiferans (13.3 mol ± 0.4 mole % Mg) are 7.80 and 7.82, respectively, and according to Plummer and Mackenzie (1974), −log(K * sp ) value of 12.7 mole % Mg is 7.82.Our samples were cleaned only in an ultrasonic bath and dried at 40 • C for about 12 h, following Plummer and Mackenzie (1974).The solubilities calculated in this study are similar to those reported by Plummer and Mackenzie (1974), and are comparable to those obtained from field data, because the minimal processing employed in this study resulted in a state similar to that occurring in the sediment.Moreover, a significant dissolution rate of coral was obtained even when a = 1.1.Since a (calculated from A T and C T ) has an error of 0.05, biogenic aragonite starts to be dissolved where  1.0 < a < 1.1.Alternatively, biogenic aragonite may be slightly more soluble than synthetic aragonite because of its heterogeneity and instability.

Evaluation of Mg-calcite dissolution
The relationship between dissolution rate and a , as obtained from the present experiments, is as follows:  5) to ( 7), the net dissolution of bulk sediment was zero at 3.7 < a < 3.8 and 3.0 < a < 3.2 for foraminiferans and coralline algae.
Figure 8 shows the total dissolution rate (i.e. the dissolution rate of Foraminifera + coralline algae) plotted against the bulk dissolution rate.The value of the y-axis is given as follows: Total calculated dissolution rate The composition ratio is derived from Fig. 2; in the case of a negative dissolution rate, a value of zero was used, assuming no precipitation of calcium carbonate.In Fig. 8, DR bulk should be equal to the total calculated dissolution rate (DR) if foraminiferans and coralline algae are the only dissolving grains in the bulk sediment; however, DR is smaller than DR bulk , especially for a > 3.0.
The sediment samples consist of 75 % foraminiferans, coralline algae, coral, and other aragonite; 3 % echinoids, and 22 % "other minerals"; the latter two components were not included in the calculations.The echinoid test is composed of Mg-calcite with a low MgCO 3 content (8.0 mole % Mg), corresponding to a small contribution to the bulk dissolution rate.The "other minerals" are considered to be calcite or Mg-calcite, because they did not acquire coloration when dyed with Co(NO 3 ) 2 (aragonite colors redpurple), and over 95 % of the sediment in the Shiraho reef is carbonate.Any minerals in the minor fraction appear to be much more soluble than the foraminiferans and coralline algae, meaning that we may have underestimated the dissolution rate.However, at least ∼70 % of the dissolution can be explained from foraminiferan and coralline algal dissolution, and this finding is applicable to reef environments where sediment grains consist mainly of these biogenic carbonates.

Comparison of the present results with field observations
We compared the results of bulk dissolution rate vs. Calcification of corals and other calcifiers is observed even at night, and results in a decrease of total alkalinity (Gattuso et al., 1999).Hence, a total alkalinity increase by carbonate dissolution (net dissolution) can be observed when dissolution is greater than gross calcification.Table 3 lists the net dissolution rate and a values.When a decreased from 3.1 to 1.0, net carbonate dissolution was observed.Andersson et al. (2009) calculated average rate of total dissolution by using long-term experimental data and estimated values of 1.6 mmol m −2 h −1 in the control (average a = 2) and 3.74 mmol m −2 h −1 in a treatment mesocosm (average a = 1).From Eq. ( 8), we can estimate that dissolution rates for our study are 3.2 and 5.0 mmol m −2 h −1 for a values of 2.0 and 1.0, respectively.This result is consistent with previous studies.

Future impact of ocean acidification on Mg-calcite dissolution
The burning of fossil fuels and future uptake of CO 2 by the oceans are predicted to reduce a to 2.0 to 3.0 by 2100 (Kleypas et al., 2006).Because high Mg-calcite is more soluble than both calcite and aragonite, it is the first responder to the decreasing saturation state of seawater (Andersson et al., 2003(Andersson et al., , 2007(Andersson et al., , 2009;;Morse et al., 2006).The present results show that the threshold a of biogenic Mg-calcite in taxa such as coralline algae and foraminifera is 3.0 to 3.2.Mg-calcite is dissolved only nocturnally at present, but will eventually begin to dissolve diurnally also.Kayanne et al. (2005) estimated the community calcification rate to be 70 to 127 mmol m −2 day −1 for Shiraho Reef.If the average daytime a value becomes 3.0 and the night-time value becomes 2.0, the carbonate dissolution rate will be 1.4 mmol m −2 h −1 diurnally and 3.2 mmol m −2 h −1 nocturnally (see Eq. ( 9)), which equals at least 55.2 mmol m −2 day −1 .Because the sandy regions are about three times larger than the coral habitats at Shiraho reef (Kayannne et al., 2005), Mg-calcite dissolution in these sands will have a major influence on the Shiraho reef ecosystem.Moreover, the decreased calcification caused by the decline of a would result in negative net calcification of carbonate ecosystems at Shiraho reef in the future.
coral zonation could be seen.The sediment was collected in August 2007, and the coral and coralline algae in October 2007.Samples were cleaned in an ultrasonic bath sonicator and dried at 40 • C for about 12 h.

Fig. 5 .
Fig. 5.The result of bulk dissolution experiment (nA T and duration).

Fig. 6 .
Fig. 6.The relationship between average dissolution rate and a (bulk).

Table 1 .
Condition for each experiment.

Table 2 .
Relationship between change of gas flow, salinity and A T .

Table 3 .
Carbonate dissolution rates reported from carbonate environments and mesocosms.