Hydrogeochemistry and Acidic Property of Submarine Groundwater Discharge Around Two Coral Islands in the Northern South China Sea

Submarine groundwater discharge (SGD) is an important source of nutrients in many coastal regions, yet little information is available on its carbonate chemistry and controlling factors. This study examined the processes and factors controlling the hydrogeochemistry and acidic property of the groundwaters and SGD waters of two isolated coral islands, Liuqiu Island (13 km off southwestern Taiwan) and Dongsha Island (located in the northern South China Sea, 420 km away from Liuqiu Island). Our results showed that the total alkalinity and dissolved inorganic carbon (DIC) of the fresh SGD waters were controlled mainly by the chemical weathering of carbonate minerals. Part of the DIC came from the organic matter decomposition or soil CO2, reducing the pH and CO3 2− concentration. Distributions of the carbonate chemistry and nutrients of the SGD waters were controlled mainly by physical mixing between the groundwater and the ambient seawater under the seabed, the so-called subterranean estuary. The Ca2+ released through weathering significantly increased the saturation state of aragonite or calcite, reducing the corrosiveness of the SGD waters on the carbonate rocks. This study is likely the first to examine the effects of the acidic property of SGD waters on the biogenic carbonate spine of a sea urchin and a pteropod shell. The spring water with similar carbonate chemistry to that of the freshwater SGD endmember from Liuqiu Island with a saturation state of aragonite of 0.96 caused observable dissolution on the spine of a sea urchin and a pteropod shell, but the spine dissolved more readily. This was because the spine is made of high-Mg calcite, which has higher solubility than that of aragonite or calcite. Such a result implies that some marine organisms with carbonate skeletons or shells containing high Mg:Ca ratios may suffer the impact of ocean acidification earlier. Although the SGD may contribute less than 10% of freshwater discharge by rivers to the coastal area, its impact on coastal biogeochemical cycles and ecosystems due to its acidic property and continual effect on the coast all year round deserves further investigation.

Submarine groundwater discharge (SGD) is an important source of nutrients in many coastal regions, yet little information is available on its carbonate chemistry and controlling factors. This study examined the processes and factors controlling the hydrogeochemistry and acidic property of the groundwaters and SGD waters of two isolated coral islands, Liuqiu Island (13 km off southwestern Taiwan) and Dongsha Island (located in the northern South China Sea, 420 km away from Liuqiu Island). Our results showed that the total alkalinity and dissolved inorganic carbon (DIC) of the fresh SGD waters were controlled mainly by the chemical weathering of carbonate minerals. Part of the DIC came from the organic matter decomposition or soil CO 2 , reducing the pH and CO 3 2− concentration.
Distributions of the carbonate chemistry and nutrients of the SGD waters were controlled mainly by physical mixing between the groundwater and the ambient seawater under the seabed, the so-called subterranean estuary. The Ca 2+ released through weathering significantly increased the saturation state of aragonite or calcite, reducing the corrosiveness of the SGD waters on the carbonate rocks. This study is likely the first to examine the effects of the acidic property of SGD waters on the biogenic carbonate spine of a sea urchin and a pteropod shell. The spring water with similar carbonate chemistry to that of the freshwater SGD endmember from Liuqiu Island with a saturation state of aragonite of 0.96 caused observable dissolution on the spine of a sea urchin and a pteropod shell, but the spine dissolved more readily. This was because the spine is made of high-Mg calcite, which has higher solubility than that of aragonite or calcite. Such a result implies that some marine organisms with carbonate skeletons or shells containing high Mg: Ca ratios may suffer the impact of ocean acidification earlier. Although the SGD may contribute less than 10% of freshwater discharge by rivers to the coastal area, its impact on coastal biogeochemical cycles and ecosystems due to its acidic property and continual effect on the coast all year round deserves further investigation.

INTRODUCTION
River inputs play an essential role in global water cycles and biogeochemical cycles. They transport water and large amounts of nutrients, such as nitrate, phosphate, and silicate, to support the estuaries' high productivity and biodiversity. In addition to the river waters, there is also fresh groundwater discharging into the ocean through the seabed directly. This phenomenon is called submarine groundwater discharge (SGD) (Burnett et al., 2003;Bishop et al., 2017;Taniguchi et al., 2019). Taniguchi et al. (2019) defined SGD as "the flow of water through continental and insular margins from the seabed to the coastal ocean, regardless of fluid composition or driving force", stating that islands typically have higher SGD, in terms of fluxes per unit area of landmass, than continents. According to this definition, the SGD includes both the freshwater and recirculated (or recharged) seawater. Primarily due to difficulties in observations and measurements, there are not as many studies on SGD as there are on riverine systems. The amount of water fluxes of SGD is considerably high, accounting for 1-10% of the total riverine discharge (Burnett et al., 2001;Santos et al., 2021). However, the proportion of the nutrient flux of the SGD could be higher. This is because SGD water generally has a longer residual time than that of river water before entering the oceans, accumulating a more significant effect of organic matter decomposition; chemical weathering; and human activities, such as sewage inputs and fertilizer leakage. (Wang et al., 2018;Chen et al., 2019;Santos et al., 2021). Consequently, SGD water is usually high in nutrients but low in dissolved oxygen (DO) and pH (Wang et al., 2018). Coupling with the dissolution of minerals, such as CaCO 3 , SGD water could also be high in dissolved inorganic carbon (DIC) and total alkalinity (TA) (Wang et al., 2018). Since the pre-industrial period, the oceans have been absorbing an increasing amount of human-released CO 2 from the atmosphere. Consequently, the pH, saturation state of calcite (Ω Cal ), and aragonite (Ω Ara ) are gradually decreasing with time. Such a process is called ocean acidification. Generally speaking, the pH of the surface ocean is around 8.25 on the NBS scale (or about 8.1 on the total scale), and the value is expected to have a 0.3 pH unit drop at the end of this century due to the increasing atmospheric CO 2 (Orr et al., 2005;Hoegh-Guldberg et al., 2007;Dore et al., 2009;Bates et al., 2014). It is well known that ocean acidification reduces the calcification rates of many marine organisms that use CaCO 3 as their skeleton. Indeed, SGD could be an acidity source for coastal oceans.
Previous studies have shown that the SGD is a vital source of nutrients in many coastal regions, but little information is available on its carbonate chemistry and the controlling factors. Using Taiwan as an example, a previous study showed that in the coastal region, there is a larger amount of inorganic carbon (such as carbon dioxide) brought in by the SGD than organic carbon produced by phytoplankton from the nutrients brought in by the SGD (Wang et al., 2018). The extra-inorganic carbon may eventually move to the atmosphere in the form of CO 2 . That is, SGD is a source of CO 2 that ends up in the atmosphere. Studies have pointed out that the lower the Ω Cal or Ω Ara values, the lower the calcification rate of some marine organisms with calcium carbonate shells (Hoegh-Guldberg et al., 2007). Of note is that the Ω Cal and Ω Ara of SGD water could be lower than one (undersaturation) (Wang et al., 2018). In that case, CaCO 3 starts to dissolve. The Ω ara value of the discharged submarine groundwater is anticipated to decrease from oversaturation to undersaturation with the increasing atmospheric CO 2 in Sanya Bay in the northern South China Sea . Understanding the hydrogeochemistry and carbonate chemistry of the SGD water is critical for examining and predicting the impact of human activities on the chemistry of SGD and on the coastal ecosystem, especially on coral reefs where the highest marine biodiversity exists.
This study aims to investigate the processes and factors controlling the hydrogeochemistry and the acidic property of the SGD water and groundwater of two isolated coral islands, Liuqiu Island (13 km off southwestern Taiwan) and Dongsha Island (located in the northern South China Sea, 420 km away from Liuqiu Island) ( Figure 1). Liuqiu Island is a 4 km long and 2 km wide NW-SE elongated coral-reef island with an area of 6.8 km 2 . The island has been raised at a rate of about 2 mm/year, mainly due to mud diapirism (Shih et al., 1991;Chen et al., 2014). Dongsha Island is a 2.8 km long and 0.9 km wide E-W coral reef island with an area of 1.8 km 2 . It is the only reef island among a circular coral reef with a diameter of about 25 km, namely, the Dongsha Atoll. We collected water samples around Liuqiu Island and Dongsha Island. The carbonate chemistry and major ions of rainwaters, spring waters, and SGD waters were analyzed. Factors controlling the chemical weathering and the carbonate chemistry are discussed. This study is likely the first to examine the effect of the acidic property of SGD water on the biogenic carbonate spine of a sea urchin and a pteropod shell.
A hollow stainless steel needle with a total length of 0.5 m was inserted from the water into the seabed 0.2-1 m below the sea surface around the coasts of Liuqiu Island and Dongsha Island. The SGD waters were drawn using the needle and stored immediately in a plastic-made sampling bag to minimize the gases exchange between the water samples and air. Field water temperature, salinity (or conductivity), and DO were measured using a Hach HQ40 portable multimeter. DO was calibrated using H 2 O-saturated air, and salinity was calibrated with the OSIL-produced International Association for the Physical Sciences of the Oceans (IAPSO) salinity standard seawater. Water samples were collected and poisoned with 0.05% saturated HgCl 2 . Waters for pH, TA, DIC, and CH 4 measurements were stored in borosilicate glass bottles. Waters for nutrient measurements were filtered with a 0.45 μm Millex-HA syringe filter (MF-Millipore mixed cellulose ester membrane) and stored in PP bottles. TA was measured by Gran titration with an open-cell system using the Apollo-manufactured total alkalinity titrator (model: AS-ALK2). DIC was measured with an infrared detector using an Apollo-manufactured DIC analyzer (model AS-C6L). Both TA and DIC were calibrated against certified reference material for ocean CO 2 measurements provided by Dr. A.G. Dickson at the Scripps Institution of Oceanography Marine Physical Laboratory, University of California, San Diego. For the brackish water and seawater samples, pH on the total scale was measured using a spectrophotometer with meta-cresol purple at 25°C. For the freshwater samples, the pH on the NBS scale was measured with a glass electrode pH meter at 25°C. To compare the pH data on the same scale, we converted the pH on the total scale to the NBS scale. The differences in pH values of our samples between two scales (pH on the NBS minus pH on the total scale), determined using the CO 2 System Calculations Program version 2.3 (Pierrot et al., 2006), is as high as 0.16 at a salinity of 1.4 and as low as 0.14 at a salinity of 36.4. pH values reported in this study are on the NBS scale at 25°C.
Two SGD samples collected from Dongsha Island formed black precipitation when HgCl 2 was added for preservation. This could be due to the formation of HgS when HgCl 2 reacted with the H 2 S, which is one of the gases that commonly exist in groundwater. The chemical reaction is as follows: Of note is that this reaction generates hydrochloric acid, further reducing the pH and TA of the water samples. Assuming all the black precipitation was HgS, using its weight, we corrected the changes in TA and pH due to the formation of HCl.
Nitrate plus nitrite, phosphate, and silicate were calibrated with the OSIL nutrient standards, and detailed information was given by Chen and Wang (2006). CH 4 was measured with the head-space equilibration technique (Weiss, 1981) using a gas chromatograph (Agilent 7,890) fitted with a flame ionization detector (FID). The primary standard was 1.16 ppmV CH 4 (MESA Specialty Gas, United States). The precision of repeated analysis of water samples was about ±3% in routine sample analysis. The CO 3 2− , HCO 3 − , and partial pressure of CO 2 (pCO 2 ) were calculated using the CO 2 System Calculations Program version 2.3 (Pierrot et al., 2006). The dissociation constants for the carbonate chemistry were taken from Millero (2010). Ω Cal and Ω Ara were calculated using the following equation: where K sp , the stoichiometric solubility product, was taken from Mucci (1983). Major ions were analyzed for the spring waters collected at stations S1-S4 on 1987/9/26, the spring water collected at station S5 on 2020/ 12/3, and the rainwater collected at R1 on 2020/8/26. Cations of samples collected in 1987 were measured using the Dionex ion chromatography system (model: 2000i). Cl − was determined using the mercury thiocyanate method (Florence and Farrar, 1971), SO 4 2− was determined using turbidimetric method (American Public Health Association, 1981), and HCO 3 − was calculated using the measured TA and pH. Major ions of water samples collected in 2020 were analyzed using ion chromatography at the Ultra Trace Industrial Safety Hygiene Laboratory of the SGS Taiwan Ltd. The results are shown in Table 1. For the remaining water samples, the Ca 2+ used to calculated the Ω Cal and Ω Ara values was estimated using the linear correlation with the spring water at S5 (Ca 2+ 1931 μmol/kg at a salinity of 0.42; see Table 1) mixed with the standard seawater (Ca 2+ 10,282 μmol/kg at a salinity of 35 (Millero et al., 2008)) as endmembers.
A sea urchin, Hemicentrotus pulcherrimus, was collected from Liuqiu Island. A 10% KOH solution was used for tissue digestion. The shell of a pteropod, Creseis virgula, was collected around the coast of southwestern Taiwan. Spring water collected at station S5 on 2021/ 12/3, and surface seawater collected at the coast of southwest Taiwan on 2021/1/12 were poisoned with 0.05% saturated HgCl 2 and sealed for the incubation experiments. The spine and shell were incubated in a 250 ml borosilicate glass bottle filled with the spring water for 60 days at 25°C. The seawater incubated with the spine was used as the control. On the 30th day, we refreshed the spring water and seawater, as well as taking images under a microscope (model: MSHOT MZ88) using a Canon EOS 600D digital camera on the 1st, 30th and 60th days.

RESULTS AND DISCUSSION
Major Ions and Their Implied Material Source From Chemical Weathering Table 1 shows the conductivity, salinity, and the major ions of the spring waters at stations S1-S5 and the rainwater at station R1.
Generally speaking, atmospheric depositions and the chemical weathering of rocks are two significant sources of ions for natural waters. Human activities, such as sewage input, could also contribute major ions. Of note is that the low CH 4 concentrations of the spring water (Table 2) suggested that the influence of human activities on the major ions of the spring water was minimal, as local sewage discharge usually has several orders of magnitude higher in CH 4 concentration (Borges et al., 2018). Major ions in the spring waters of Liuqiu Island are contributed mainly from the chemical weathering of rocks, as the major cations (Na + , K + , Ca 2+ , Mg 2+ ) and anions (Cl − , SO 4 2− , HCO 3 − ) in rainwater collected on 2020/8/26 were negligibly small when compared with the spring water in Liuqiu Island collected on 2020/12/3 ( Table 1). Taking Na + , K + , Ca 2+ , Mg 2+ as examples, the rainwater was able to contribute only 0.2, 1.1, 0.1, and 0.2%, respectively, to that of the spring water. Such a result implies that various chemical weathering processes contribute the most to the spring water's major ions. The positive correlations between the major ions and salinity of the spring waters of Liuqiu Island and the rainwater provided further evidence for this suggestion ( Figure 2). Several chemical weathering reactions are listed as follows: The dissolution of halite: The dissolution of a silicate mineral (using sodium feldspar as an example): The dissolution of CaCO 3 or MgCO 3 (using CaCO 3 as an example): The dissolution of dolomite: The oxidation of pyrite: Na + was the most or second most abundant among the cations, followed by Ca 2+ in the rainwater and spring waters, whereas HCO 3 − was the most abundant anion followed by Cl − (Table 1). Generally, Cl − ion mainly came from halite dissolution or marine spacy. Seawater has a Na + : Cl − ratio of approximately 0.86 (Millero et al., 2008). In contrast, the freshwater samples shown in Table 1 had excess Na + rather than Cl − with ratios between 1.03 and 1.63. The almost 1:1 correlation between the Na and Cl ions (Figure 3) suggested that the Cl − and the equivalent of Na + came from the halite dissolution or marine spacy. The remaining Na + (Na + minus Cl − , corrected for halite dissolution contribution) and K + were from Na-or K-feldspar weathering, which was expected to contribute the same amount of TA in the form of bicarbonate ion as shown in Eq. 2. Such a result is in accordance with previous study that in Liuqiu Island siliceous soils cover carbonate rocks (Holocene coral reef or Pleistocene limestone) (Cheng et al., 2011). The carbonate rocks' dissolution mainly contributes to the Ca 2+ and Mg 2+ ions (see Eqs 3, 4. The [Ca 2+ ]/[Mg 2+ ] ratio of 3.8 of the spring water is in accordance with that of the soil pedons in Liuqiu Island, ranging between 1.4 and 7.1 (average 3.7 ± 1.9, n 21) (Cheng et al., 2011). The SO 4 2could be contributed by the CaCO 3 or MgCO 3 dissolution with the participation of H 2 SO 4 instead of H 2 CO 3 . But this is unlikely to explain the amount of SO 4 found in the spring waters of Liuqiu Island, as the rainwater contained a negligibly small amount of SO 4 . Pyrite oxidation (see Eq. 5) might contribute the SO 4 to the spring water and the surface freshwaters in Liuqiu Island. Under pyrite oxidation, 1 mol of SO 4 formation would lead to 2 mol of TA deduction due to sulfuric acid formation.

Carbonate Chemistry and the Acidic Property of Spring Water
Rainwaters sampled contained only a small amount of TA (-19-94 μmol/kg, n 9) and DIC 38-89 μmol/kg, n 9), whereas our time-series data at station S5 showed that the spring water at S5 had significantly higher TA (6,624-6,704 μmol/kg) and DIC (7,309-7,931 μmol/kg) values ( Table 2). As mentioned above, the dissolution of feldspar or carbonate rocks contributed TA and DIC. Based on Eqs 1-3, regarding the spring water sampled on 2020/12/3, we estimated (Na + − Cl − ) + K + + 2 (Ca 2+ + Mg 2+ ) 5787 μmol/kg for TA and (Ca 2+ + Mg 2+ ) 2439 for DIC. Note that CaCO 3 or MgCO 3 weathering consumed CO 2 . The organic matter decomposition could contribute to this CO 2 in the groundwater. Soil air has a significantly higher CO 2 concentration than that of the atmosphere and could also contribute to the CO 2 in the groundwater. Using the TA value, we calculated that the DIC value to be 6,252 μmol/kg when the pCO 2 was the same as that in the atmosphere (about 400 μatm; 100% humidity). That is, we estimated that there was (7,848-6,252) 1,596 μmol/kg of excess DIC or CO 2 in the spring water with respect to the atmospheric CO 2 level. Figure 4 shows the estimations. Again, this amount of CO 2 could be contributed by the organic matter decomposition or by soil air. Organic matter decomposition increased the DIC value but slightly decreased the TA value. In comparison, the CO 2 gas exchange between soil air and groundwater would not change TA. However, we could not quantify the amount of TA change with this 1,596 μmol/kg excess CO 2 , as we could not quantify the amount of CO 2 in the soil air-groundwater CO 2 exchange or in the groundwater's organic matter decomposition. Based on Eqs 1-3, we estimated the TA value of the spring water to be 5,787 μmol/kg, explaining 86% of the measured value of 6,701 μmol/kg. The excess CO 2 of 1,596 μmol/kg accounted for 20% of the measured DIC of 7,848 μmol/kg (Figure 4). Such results suggest that the chemical weathering mainly contributed to the TA and DIC values, and part of the DIC was contributed by the increase in CO 2 through organic matter decomposition or soil air-groundwater CO 2 exchange.
Surface seawaters around southwest Taiwan and in the northern South China Sea generally had pH values of about 8.20 (about 8.05 on the total scale) or higher (Chen et al., 2010;Lui and Chen, 2015;. In contrast, the spring waters collected between 2017 and 2021 had a relatively low pH value, ranging 7.02-7.25 and averaging 7.15 ± 0.09 (n 5). The dissolution of carbonate rock consumes dissolved CO 2 and generates TA and DIC in a ratio of 2:1. Feldspar dissolution consumes CO 2 and generates TA as well. That is, increasing TA and DIC in a ratio equal to or higher than 2:1 increases the pH, Ω Ara , and Ω Cal values but decreases the pCO 2 value. Such a change also increases the carbonate buffering capacity of a water body by reducing the extent by which the pH, Ω Ara , and Ω Cal values are reduced under additional CO 2 dissolution (Chou et al., 2013). Indeed, these water bodies' relatively low pH values were mainly due to the dissolution of CO 2 through the soil air-groundwater CO 2 exchange and organic matter decomposition. Although the dissolution of carbonate rocks generated a considerable amount of CO 3 2− and Ca 2+ , the addition of a high amount of excess CO 2 reduced the CO 3 2− concentrations of the spring waters to just 6.3-9.4 μmol/kg due to thermodynamics. Consequently, the Ω Ara and Ω Cal were 0.92-1.40 and 1.50-2.28, respectively. With a considerable amount of excess CO 2 , the pCO 2 of the spring waters could be as high as 37,316 μatm ( Table 2).

Physical Mixing Behaviors in the Subterranean Estuaries
In Liuqiu Island, the values of temperature, DO, pH, TA, DIC, Ω Ara , and SiO 2 in the spring water did not considerably change across seasons ( Figure 5). The temperatures of the SGD samples and seawaters showed strong seasonal variations. The strong negative linearities between TA, DIC, NO 3 − + NO 2 − , SiO 2 , and the salinity of the SGD samples suggested that the carbonate chemistry and nutrients of the SGD samples were controlled mainly by physical mixing between the groundwater and ambient seawater within the seabed, the so-called subterranean estuary. The freshwater endmembers of the TA, DIC, and SiO 2 , determined using the SGD data and linear regressions methods, were similar to those of the spring water, suggesting that the carbonate chemistry of the discharged submarine groundwater at station A1 had similar carbonate chemistry to that of the spring water at station S5, which had no observable changes across seasons. The differences in TA and DIC between the linear-regression determined freshwater endmembers of the SGD samples and the spring waters were (7,640) 1,277 μmol/kg and (8,330-7,649) 681 μmol/kg, respectively ( Figure 5; Table 2). The differences in TA and DIC had a ratio of 1.88. Such a result reflects that the higher TA and DIC of the SGD freshwater endmembers were mainly due to the further dissolution of carbonate rocks, which has changes of TA and DIC in a ratio of two. Generally, water with a lower pH or Ω Ara value is more corrosive to CaCO 3 . Although the spring water contained a very low CO 3 2− concentration, it had a Ω Ara value that was sometimes slightly below one. This is because the spring water contained a high concentration of Ca 2+ that came from the weathering of the CaCO 3 rock, increasing the Ω ara value. That is, the dissolution of CaCO 3 minerals around the coral islands effectively reduced the corrosiveness of the groundwater and hence the SGD water, reducing the environmental stress that was due to the discharge of the acidic SGD. Meanwhile, the linearities of TA and DIC vs. salinity of the SGD samples in Liuqiu Island suggested that the residual time of the groundwater in the freshwater-seawater mixing zones, the so-called subterranean estuaries, was short. The increase in buffering effect of the SGD water due to an additional dissolution of carbonate rock in the subterranean estuary was insignificant, making SGD is an additional acidic source with similar carbonate chemistry of the groundwater for the coastal areas.
The SGD waters likely had a freshwater endmember with a lower NO 3 − concentration than that of the spring water. The PO 4 3− was scattered across the salinity gradient largely because the groundwater and the ambient seawater contained similar but low concentrations of PO 4 3− . The use of chemical fertilizer could be an important source of nutrients for groundwater. According to the Liuqiu Township Office, only 0.01 km 2 (0.15% of the total area) is farmland in Liuqiu Island. Most important is that nearly all of the farmland is dryland. Without sufficient freshwater (such as river water) to support a considerable scale of agriculture, the influence of the use of fertilizer on the chemistry of the groundwater in Liuqiu Island or Dongsha Island is expected to be insignificant. Sewage input could be another critical factor influencing the chemistry of groundwater in Liuqiu Island or Dongsha Island. In Liuqiu Island, the CH 4 concentrations were low for both the spring water and SGD samples, showing no observable influence of anthropogenic activities in our sampling sites, such as local sewage discharge, which is usually several orders of magnitude higher in CH 4 concentration (Borges et al., 2018). Noting, fresh SGD can be observed in Liuqiu Island in both the wet and dry seasons, showing that it has a continual effect on FIGURE 3 | Na − vs. Cl − of the rainwater and spring waters on Liuqiu Island. Table 1 shows the information in detail. The solid line shows the 1:1 ratio.
FIGURE 4 | Estimated TA and DIC of the spring water at station S5 using major ion concentrations and the excess CO 2 calculated with measured TA and pCO 2 400 μatm, assuming the air-water CO 2 equilibrium.
Frontiers in Earth Science | www.frontiersin.org September 2021 | Volume 9 | Article 697388 the coast of Liuqiu Island all year round. In contrast, the surface seawaters had high DO, pH, and Ω ara values but low nutrient concentrations. Such a result was likely due to the enhanced productivity achieved via the support of nutrients from the SGD or from the Gaoping River, which is about 13 km from Liuqiu Island. Local discharges may also affect this and warrant further investigation.
In Dongsha Island, which is about 420 km from Liuqiu Island, the DO and pH of the groundwater, SGD waters, and seawaters had positive correlations with salinity, whereas the TA, DIC NO 3 − + NO 2 − , PO 4 3− , SiO 2 and CH 4 showed negative correlations with salinity ( Figure 6). Such results suggest that the carbonate chemistry and nutrients of the SGD waters were controlled mainly by physical mixing between the groundwater and seawaters. Furthermore, the carbonate chemistry and nutrients of water samples collected over the wet and dry seasons in Dongsha Island were consistent with each other, showing stable chemical properties of the groundwater endmember for the SGD waters. Interestingly, the trends of the pH, TA and DIC vs. salinity plots suggested that the freshwater endmembers of the SGD waters in Dongsha Island were similar to those in Liuqiu Island. Such a result suggests that the factors controlling the carbonate chemistry of the groundwaters and SGD waters in Dongsha Island might be Island. The rainwater, spring water, SGD water, and seawater are in light blue, black, red, and blue, respectively. The red lines extended to the axes and equations show the linear regressions for the conservative parameters of the SGD data.
Frontiers in Earth Science | www.frontiersin.org September 2021 | Volume 9 | Article 697388 similar to those in Liuqiu Island. In the seagrass meadows in the inner lagoon, the TA and DIC released from the sediment, as well as the high productivity increased the TA:DIC ratio, making the seawater around the seagrass meadows high in DO, TA, pH and Ω Ara but low in DIC and pCO 2 values (Chou et al., 2018). Our water samples were likely not affected by the seawater with relatively high TA but low DIC from the seagrass meadows, maintaining values similar to that of the offshore surface waters.
Notably, in Dongsha Island, the high CH 4 and PO 4 3concentrations of the groundwaters, SGD waters, and seawaters suggest that sewage discharges might have affected these waters. Similar to the case in Liuqiu Island, fresh SGD waters could be found around Dongsha Island in the dry season when the rainfall was low, suggesting that the SGD in Dongsha Island maintains a continual effect around the coast all year round and not to mention in the wet season when the rainfall is high.

Effect of Acidic Property and Corrosiveness of SGD on CaCO 3 Skeleton and Shells
Notably, the pH could be as low as 7.02 for the spring water in Liuqiu Island and 7.07 for the groundwater in Dongsha Island (Figures 5, 6). Most pH values of the SGD samples in Liuqiu Island and Dongsha Island, however, were far below 7.95, reaching as low as 7.30 for Liuqiu Island and 7.34 for , SiO 2 , and CH 4 vs. salinity of groundwater, SGD water, and seawater in Dongsha Island. The groundwater, SGD, and seawater are in green, red, and blue, respectively. The red lines extended to the axes and equations show the linear regressions for the conservative parameters of the SGD data.
Frontiers in Earth Science | www.frontiersin.org September 2021 | Volume 9 | Article 697388 Dongsha Island. Such results reflect the acidic properties of the groundwaters and SGD waters in these two coral reef Islands. As mentioned above, the spring water at station S5 in Liuqiu Island had similar carbonate chemistry to that of the fresh SGD endmembers in Liuqiu and Dongsha Islands. Although the spring water and biogenic carbonate minerals are spatially separated, interactions between the spring water and biogenic carbonate minerals could provide insights into the acidic effects of the SGD on carbonate skeletons and shells. On this basis, the spring water of Liuqiu Island collected on 2020/ 12/3 with a Ω Ara of 0.96 and preserved with saturated HgCl 2 was used to conduct an incubation experiment. A surface seawater sample collected at the coast of southwestern Taiwan on 2021/1/12 with a Ω Ara value of 2.73 was used for comparison. The spines of sea urchins (H. pulcherrimus) were incubated in the water samples and observed after 30 and 60 days. The shell of a pteropod (C. virgula) was also incubated in the spring water for observation. The incubation results showed that the spine of H. pulcherrimus had no observable difference after incubation in seawater for 60 days (Figure 7). In contrast, the spine showed significant dissolution after 30 days and 60 days incubations in the spring water of Liuqiu Island. The base of the spine showed a reduction in size, and the shaft of the spine was etched, becoming rough (Figure 7). The shell of C. virgula also showed an indication of dissolution, turning from transparent to milky white. Since HgCl 2 was added to inhibit microbial activity, we expected that the dissolution of the spine or shell caused by microbial activities to be insignificant. Our observation showed that the spine of the sea urchin dissolved more readily than the shell of the pteropod. This was because the spine of the sea urchin is composed of highmagnesium calcite, which has higher solubility than that of aragonite (Grossmann and Nebelsick, 2013;Haese et al., 2014). Furthermore, even though the shells of pteropods are composed of aragonite, the shapes, microstructures, and MgCO 3 :CaCO 3 ratio of the shells are different, causing the shells to have different solubility and dissolution rates.
It is worth mentioning that our incubation experiments indicated the acidic property of SGD on biogenic carbonate minerals. However, the effects of acidic SGD on live marine organisms are expected to be more complicated. This is because Ω Ara and other factors, such as pH, temperature, pCO 2 and dissolved oxygen, also influence the physiology of marine organisms (Glover and Kidwell, 1993;Shirayama and Thornton, 2005;Kroeker et al., 2013;Van de Waal et al., 2013;Peck et al., 2016;Mekkes et al., 2021). Moreover, the outermost organic layer, periostracum, or other organic matters of many calcifying organisms can effectively counter the CaCO 3 dissolution under an acidic environment (Glover and Kidwell, 1993;Peck et al., 2016). Of note is that both incubation experiments and field samples showed that acidified conditions enhanced carbonate shell dissolution and reduced calcification rates of many live calcifying organisms (Glover and Kidwell, 1993;Orr et al., 2005;Shirayama and Thornton, 2005;Hall-Spencer et al., 2008;Talmage and Gobler, 2010;Lischka et al., 2011;Moya et al., 2016;Peck et al., 2016;Zhai, 2018;Li and Zhai, 2019;Zhai et al., 2019;Mekkes et al., 2021). Indeed, the reported acidified conditions FIGURE 7 | Images of (bottom left) the spine of H. pulcherrimus and (right) shell of C. virgula before (0 days) and after incubation in the spring water at station S5 with Ω Ara 0.96 for 30 and 60 days. Seawater with Ω Ara 2.73 incubated with (top left) the spine of H. pulcherrimus was used as the control.
Frontiers in Earth Science | www.frontiersin.org September 2021 | Volume 9 | Article 697388 mostly had higher pH and Ω ara values but lower pCO 2 values than that of the SGD samples in Liuqiu Island or Dongsha Island. For instance, the shell of Limacina helicina significantly degraded after 29 days of incubation when pH on the total scale and pCO 2 were 7.66 and 1,150 μatm, respectively (Lischka et al., 2011). In the North Yellow Sea, the community net calcification rate was nearly zero when the Ω ara value reached 1.5-1.6 (Zhai, 2018;Li and Zhai, 2019;Zhai et al., 2019). Additionally, the CaCO 3 sediments would be dissolved when the Ω ara value of the seawater reached 2.92 ± 0.16 or lower (Eyre et al., 2018). This is largely due to the additional acidic effect as a result of the CO 2 produced by the decomposition of organic matter in sediment. The Liuqiu Island and Dongsha Island examples reveal that SGD could be an important acidic input to coastal oceans.

CONCLUSION
The hydrogeochemistry and carbonate chemistry of the groundwaters, SGD waters, and the seawaters in two coral islands were examined in this study. Our results show that although the dissolution of carbonate rocks increased pH, Ω Ara , Ω Cal and the buffering capacity of the groundwater, the CO 2 from the organic matter decomposition or soil CO 2 enhanced the corrosiveness of the groundwater, shifting Ω Ara from oversaturation to undersaturation. The acidic property of the SGD waters was verified using biogenic carbonate minerals. Although SGD amounts to only several percent of the riverine discharge, it is a common natural phenomenon that could affect the coast all year round, especially in winter when there is minimal rainfall. We suggest that the impacts of SGD on coastal biogeochemical cycles and ecosystems due to its acidic property and continual effect on the coast all year round deserve consideration in further investigations.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusion of this article will be made available by the authors.