Seasonal Controls of Seawater CO₂ Systems in Subtropical Coral Reefs: A Case Study from the Eastern Coast of Shenzhen, China

Abstract

In situ field investigations coupled with coral culture experiments were carried out in the coral reef waters of the eastern coast of Shenzhen, Da’ao Bay (DAB), Dalu Bay (DLB), and Yangmeikeng Sea Area (YMKSA) to study the dynamics of the carbon dioxide (CO₂) system in seawater and its controlling factors. The results indicated that the CO₂ parameters were highly variable over a range of spatiotemporal scales, forced by various physical and biochemical processes. Comprehensively, DAB acted as a sink for atmospheric CO₂ with exchange flux of –1.51 ± 0.31 to 0.27 ± 0.50 mmol C m⁻² d⁻¹, while DLB and YMKSA acted as a CO₂ source with exchange fluxes of –0.42 ± 0.36 to 1.69 ± 0.74 mmol C m⁻² d⁻¹ and –0.58 ± 0.48 to 1.69 ± 0.41 mmol C m⁻² d⁻¹, respectively. The biological process and mixing effect could be the most important factor for the seasonal variation in total alkalinity (TA). In terms of dissolved inorganic carbon (DIC), in addition to biological process and mixing, its seasonal variation was affected by air–sea exchange and coral metabolism to some extent. Different from the former, the other CO₂ parameters, total scale pH (pH_T), partial pressure of CO₂ (pCO₂), and aragonite saturation state (Ω_A), were mainly controlled by a combination of the temperature change, biochemical processes, air–sea exchange, and coral metabolism, while water mixing has little effect on them. In addition, our results indicated that coral communities could significantly increase the DIC/TA ratio by reducing the TA concentration and increasing the DIC in the reef waters, which may promote the acidification of local seawater and need attention.

1. Introduction

Affected by human activities and global changes, the concentration of carbon dioxide (CO₂) in the atmosphere increased from 280 ppm before the Industrial Revolution to 420 ppm in 2023 [1]. If it continues to grow at this rate, atmospheric CO₂ concentrations are expected to double by the end of the 21st century, which will pose a serious threat to normal human activity [1]. How to effectively alleviate and control the continuous increase in atmospheric CO₂ is a major issue of concern for scientists and the relevant authorities.

The ocean accounts for 70% of the global surface area and can absorb about 30% of the CO₂ emitted by humans each year, so it plays an important role in adjusting the atmospheric CO₂ concentration [2]. However, this capture of atmospheric CO₂ significantly affects the chemical properties of the marine CO₂ system, and the result is a phenomenon called ocean acidification (OA) [3]. The consequence of OA is an increase in the partial pressure of seawater CO₂ (pCO₂) and a reduction in the pH, aragonite saturation state (Ω_A), and CO₂ buffering capacity by increasing the dissolved inorganic carbon (DIC) and hydrogen ion concentration in seawater [4], which has been found to potential impacts on marine calcifiers, e.g., hermatypic corals and calcifying algae [5,6,7,8,9,10].

In general, the dynamics of CO₂ systems in the ocean are mainly controlled by biogeochemical processes, e.g., photosynthesis, respiration and calcification, and temperature changes [11,12,13]. For instance, the photosynthesis of phytoplankton can absorb CO₂ from the atmosphere, thereby reducing the DIC concentration in seawater, while biological respiration and calcification processes can release CO₂ into the water column and increase the DIC concentration in seawater. Furthermore, temperature changes can significantly affect the dynamic of the CO₂ system through thermodynamic effects and the regulation of biological metabolic activities, such as photosynthesis and respiration [14,15]. In recent years, many studies have confirmed the dominant role that temperature changes play in the CO₂ system in seawater, especially regarding the control of pH and pCO₂ [16,17].

Compared with the ocean, the geochemical behavior of the CO₂ system in coastal water is more complex [18,19]. In addition to the impact of climate change, the dynamic of the coastal CO₂ system is also significantly affected by terrestrial input (such as nutrients and organic matter) and human activities, resulting in large spatial and seasonal variations in the carbonate system [20,21,22]. For example, previous studies have shown a potential link between eutrophication and OA, e.g., the North Yellow Sea [23] and Northern South China Sea [24]. This is because excess nutrient input can promote the reproduction of phytoplankton and subsequent aerobic decomposition of algae-derived organic matter in the water column, which may be the main reason for OA [25]. Thus, coastal OA and hypoxia usually occur simultaneously [26].

In addition, the focus on coastal coral reef habitats, the metabolic activities of corals, and reef organisms may also have an important effect on the seawater CO₂ system because of the formation of biological calcium carbonate and respiration [27,28]. Generally, coral calcification absorbs total alkalinity (TA) and DIC at a ratio of 2:1 from seawater, resulting in an increase in the pCO₂ and a decrease in the pH and Ω_A of the seawater, and vice versa [29]. Although extensive studies have been carried out, the dynamics and driving factors of the CO₂ system are currently uncertain and vary between different reefs, because the above processes are affected by the superposition of various factors, such as the coral and reef species, temperature, salinity, and dissolved oxygen (DO) [9,30,31]. Hence, there is a need to understand when and where the seawater CO₂ system will reach certain conditions within different coral reef habitats, as well as the specific mechanisms leading to these conditions.

The eastern coast of Shenzhen (ECSZ) is a typical subtropical coral habitat in the South China Sea. In recent years, the coral reef system in this region has undergone significant degradation due to global warming and coastal human activities, such as aquaculture activities, port construction, and coastal tourism activities. Previous studies have shown that corals covering in the ECSZ and adjacent Daya Bay declined from an average of >70% in the 1980s to below 30% after 2000 [32,33]. OA is likely to be a potentially important cause of coral reef degradation. Therefore, it is necessary to accurately evaluate the mechanism of OA by understanding the dynamics of the seawater CO₂ system in detail. However, it remains tremendously challenging to make satisfactory progress towards this goal. For that, the biogeochemical process of the seawater CO₂ system (TA, DIC, pCO₂, Ω_A, and pH) in the ECSZ, including Da’ao Bay (DAB), Dalu Bay (DLB), and the Yangmeikeng sea area (YMKSA), were investigated through a field investigation and indoor culture experiments of coral metabolism in November 2022 (autumn), February 2023 (winter), May 2023 (spring), and August 2023 (summer). The main purpose of this study was to (1) determine the spatiotemporal variations in the seawater CO₂ system and its main control processes in subtropical coral reefs; and (2) quantify the contribution of each process (temperature change, water mixing, air–sea exchange of CO₂, coral metabolism, and other biological processes) to the seasonal variation in the CO₂ system.

2. Materials and Methods

2.1. Study Area

The ECSZ, located in the northern part of the South China Sea, is dominated by typical subtropical monsoons, with an annual rainfall of 1900–2214 mm. Affected by the monsoons, there are obvious dry and wet seasons in the survey area. The dry season is from November to March of the following year, and the wet season (accounts for 90% of the annual rainfall) is from April to October. The water depths in most areas are <15 m, averaging around 10 m (Figure 1). The tide in the study area is an irregular, semi-diurnal tide, with the highest tide level being 1.72 m and the lowest tide level being −0.58 m. There are no large rivers flowing into the sea along the coast, only a stream (Yangmei River, YMR) discharging into YMKSA (Figure 1). Thus, terrestrial materials such as nutrients and organic matter from inland are mainly transported to the study area through tidal action, coastal currents, and groundwater, rather than runoff input.

In addition, fringing coral reefs develop in the study area, and the coral community is dominated by Acropora pruinose (A. pruinose), Porites lutea (P. lutea), Favia favus (F. favus), Acropora digitifera (A. digitifera), and Platygyra carnosus (P. carnosus), respectively. The coral coverage is generally low and patchy, varying annually between 25 and 40% (mean, 35%). In this study, we selected three contrasting coral reefs around the Dapeng Peninsula coastline, i.e., DAB, DLB, and YMKSA, with the reef areas of 19.3, 6.3, and 92.6 hectares, for investigation (Figure 1). Among them, DAB and YMKSA are significantly affected by human activities, e.g., coastal tourism activities, harbor construction, and nuclear power plants [34,35]. In comparison, DLB (located on the southern edge of the Dapeng Peninsula) is less affected by human activities, and the hydrodynamic conditions there are strong due to the influence of the coastal current, especially in the autumn and winter.

2.2. Sampling and Analyses

Four cruises were conducted in the DAB, DLB, and YMKSA in November 2022 (autumn), February 2023 (winter), May 2023 (spring), and August 2023 (summer), respectively. A total of 51 stations were set up in the study area, and the same station positions were visited each season, where both surface and bottom water samples were collected using a Niskin sampler. At the same time, we collected the dominant coral species and brought them back to the laboratory for culture experiments to explore the effects of coral metabolism on seawater carbonate in different seasons.

Temperature, salinity, DO, and chlorophyll a (Chl a) were measured in the field by a YSI EXO2 multiparameter (YSI company, Yellow Springs, OH, USA), and the corresponding precisions were ±0.01 °C, ±0.01, ±0.1 μmol L⁻¹, and ±0.01 μg L⁻¹, respectively. The preservation and determination of carbonate samples were carried out according to the ‘Guide to Best Practices for Ocean CO₂ Measurements’ [36]. Briefly, the total scale pH (pH_T) was measured using a pH meter (Thermo Electron Co., Waltham, MA, USA), against 2-amino-2-hydroxy-1,3-propanediol as the pH_T buffer, at a precision of ±0.01 pH unit. The TA was determined by a METTLER TOLEDO Easyplus Titration device (ET18, Zürich, Switzerland) with the precision of ±3 µmol kg⁻¹.

The remaining other CO₂ parameters were calculated using the CO2SYS program [26,37]. Specifically, the DIC was calculated from the data of TA, pH_T, temperature, and salinity. In order to further verify the reliability of the calculated data, some seawater samples (n = 30) were selected during the investigation period to measure the DIC concentration (DIC_Mea), using the Apollo SciTech DIC analyzer (AS-C3, Apollo SciTech Inc., Newark, DE, USA; precision of ±2 µmol kg⁻¹), and compared with the calculated DIC (DIC_Cal). The results showed that the calculated results were basically consistent with the measured results of the DIC (<10 μmol L⁻¹; Supplementary Figure S1), indicating that the calculated DIC data in this study were reliable. In addition, the pCO₂ and Ω_A at in situ (pCO_{2@in situ}, and Ω_{A@in situ}) and average temperature of 24.8 °C (pCO_{[email protected]°C} and Ω_{[email protected]°C}) were calculated from TA, DIC, temperature, and salinity, using CO2SYS program.

2.3. Incubation Experiments of Coral Metabolism

During the survey period of each season, the dominant species of coral (A. pruinose, P. lutea, F. favus, A. digitifera, and P. carnosus) were collected in the study area and immediately transported to the running-seawater aquarium facilities. After an acclimatization period of 10 days, the scleractinian corals were used for the incubation experiment. Specifically, the target corals were transferred to incubators containing 8 L of filtered seawater (experimental group). In addition, one incubator with only filtered seawater was selected as the control group. Three parallel experiments were set up in each group. All incubators were sealed and cultured without headspace for 48 h (with 12 h light–12 h dark cycle). The temperature (24, 18, 25, and 30 °C in four seasons) and light conditions (110, 105, 115, and 110 μmol m⁻² s⁻¹) of the incubation experiments were consistent with the site conditions to the maximum extent. The salinity of the seawater in the four culture experiments was 32.6 (autumn), 32.6 (winter), 33.0 (spring), and 32.3 (summer), respectively. Furthermore, the initial content of DO in all culture experiments was 250 μmol L⁻¹. The TA and DIC concentration in the seawaters were determined at 0, 24, and 48 h during the experiments. The release fluxes (RFs) of the TA and DIC by coral metabolism in different seasons were calculated as follows:

RF (i) = (k_i − k₀) × 24 × V × ρ/S

where i is the TA or DIC; k_i and k₀ are the linear regression slopes of TA or DIC (μmol kg⁻¹) vs. time (h) in the experimental sample and control groups, respectively; V (L) is the experimental seawater volume; ρ (kg m⁻³) is the density of seawater; and S (cm⁻²) is the surface area of the experimental coral and was obtained via the aluminum foil method [38].

2.4. Estimation of CO₂ Fluxes

The air–sea flux of CO₂, FCO₂ (mmol m⁻² d⁻¹), was obtained by the difference between the partial pressure of CO₂ on the sea surface (pCO_2S) and atmosphere (pCO_2A):

FCO₂ = k × K₀ × (pCO_2S − pCO_2A)

where K₀ and k (cm h⁻¹) are the solubility coefficient and gas transfer velocity of CO₂, respectively. In this study, 417.4 µatm (November 2022), 420.3 µatm (February 2023), 424.0 µatm (May 2023), and 419.7 µatm (August 2023) (ftp://aftp.cmdl.noaa.gov/products/trends/co2/co2_annmean_gl.txt, accessed on 10 September 2023) [1] were used as pCO_2A. Moreover, k was obtained according to the revised equation of Wanninkhof [39] in Sweeney et al. [40] as follows:

k = 0.27 × u₁₀² × (Sc/660)^−0.5

where Sc is the Schmidt number [39], and u₁₀ is the wind speed at a height of 10 m height (monthly averaged wind speed there), as obtained from the South China Sea and Adjacent Seas Data Center [41].

2.5. Budget Model to Distinguish Controlling Processes of CO₂ System

Based on the mass budget model proposed by Xue et al. [42,43], the contribution of each process (temperature change, water mixing, air–sea exchange of CO₂, coral metabolism, and other biological processes) to the seasonal net variation in the seawater CO₂ system was calculated. The specific calculation process is shown in Supplementary Material Section S1.1.

2.6. Statistical Analysis

Seasonal and spatial differences (between the DAB, DLB and YMKSA) in biogeochemical characteristics were analyzed using a one–way analysis of variance test (ANOVA) with SPSS Statistics software (version 22). The threshold value for statistical significance was taken as p < 0.05.

3. Results

3.1. Hydrological Features

Figure 2 shows the seasonal variation characteristics, namely seawater temperature, salinity, DO, and Chl a content, of the study areas (DAB, DLB, and YMKSA) for the four periods surveyed. Temperature presented a large seasonal variation, with the lowest value of 18.0 ± 0.1 °C (mean ± SD) in February 2023 (winter) and maximum value of 34.5 °C in August 2023 (summer) (Figure 2a). Regionally, there was no significant difference between DAB, DLB, and YMKSA (p > 0.05). As for salinity, its values varied from 31.8 ± 0.8 to 33.1 ± 0.1 and exhibited small fluctuations (Figure 2b). Significant differences in salinity between DAB, DLB, and YMKSA were observed during most of the survey periods (p < 0.05). Spatially, the horizontal distribution of temperature and salinity in the DAB, DLB, and YMKSA was not obvious in the dry season (Supplementary Figures S2–S4). In contrast, temperature (salinity) showed a significant gradient decrease (increase) from the nearshore to offshore areas in the rainy season, especially in August 2023 (Supplementary Figures S1–S4), thus indicating the input of terrestrial fresh water.

During the study period, the DO ranged from 92.4 ± 25.6 μmol L⁻¹ to 287.2 ± 11.4 μmol L⁻¹ (Figure 1), with the AOU values from 120.1 ± 25.9 μmol kg⁻¹ to −62.3 ± 11.1 μmol kg⁻¹. From November 2022 to August 2023, the DO in seawater decreased gradually (Figure 2c). In the summertime (August 2023), the bottom DO of the study area, especially in the DAB, was located at the edge of low oxygen, i.e., 93.8 μmol L⁻¹ [44], which posed a potential threat to the normal metabolism of benthic organisms. As can be seen in Supplementary Figure S5, there was no statistically significant difference between the sites in DO in the DAB; however, its values in the DLB increased from the nearshore area to the offshore area, except for in November 2022 (Supplementary Figure S6). In the YMKSA, there was no statistically significant difference in DO among the sites in the dry season (November and February), while the DO increased from the nearshore zone to the offshore area in May 2023, which was opposite to the distribution of surface DO in August 2023. Furthermore, it is worth noting that the relatively low DO concentrations in August 2023 appeared right in the coral reef area, especially in the YMKSA, indicating a higher biological oxygen consumption process in the reef area.

The surface Chl a exhibited a marked variation, 0.46 ± 0.23 μg L⁻¹ to 9.45 ± 3.44 μg L⁻¹, with the highest (May 2023) and lowest values (February 2023) all observed in the YMKSA (Figure 2d). The seasonal variations in Chl a in the DAB, DLB, and YMKSA were different. For instance, the Chl a in the DLB gradually increased from November 2022 to August 2023, while its highest values in the DAB and YMKSA were all observed in May 2023 (Figure 2d). Spatially, the Chl a was generally higher in the nearshore zone than it was in the offshore zone in the DAB, DLB, and YMKSA in May 2023, which was opposite to the distribution in other seasons (Supplementary Figures S5–S7). In addition, the Chl a was >10 μg L⁻¹ in most of the coastal stations in May 2023 of the YMKSA, indicting the occurrence of a bloom [42].

3.2. Spatiotemporal Variations in Marine CO₂ System

3.2.1. TA and DIC

The concentration of TA and DIC in the study areas (DAB, DLB, and YMKSA) varied considerably among the seasons (p < 0.05). As shown in Figure 3, the highest TA and DIC values, 2225.3 ± 5.9 µmol kg⁻¹ and 2008.1 ± 9.6 µmol kg⁻¹, respectively, were all recorded in February 2023 for the DLB, while the lowest values of 2160.0 ± 39.1 µmol kg⁻¹ and 1984.8 ± 88.8 µmol kg⁻¹, respectively, were all found in August 2023 for the YMKSA (Figure 3a,b). There were significant differences (p < 0.05) in the concentrations of TA and DIC between different regions (DAB, DLB, and YMKSA) during the survey periods, except for in May 2023. For instance, the values of TA and surface DIC in the YMKSA were significantly lower than those in the DAB and DLB, especially in August 2023 (Figure 3a,b).

Spatially, during the autumn survey, the relatively low TA values were mainly distributed in the nearshore area of the DAB, which contrasted with the distribution of DIC. In other seasons, the spatial distribution of TA and DIC was not obvious; it was mainly has a patchy distribution (Supplementary Figure S8). In the DLB, the TA and DIC showed an increasing trend from the nearshore area to the offshore area in November 2022 and May 2023 (except for individual cases), while it was the reverse situation in August 2023; however, the TA and DIC showed a patchy distribution in February 2023 (Supplementary Figure S9). Similarly, the lower TA values mainly appeared in the inshore place in November 2022, May 2023, and August 2023 of YMKSA, which was contrary to the distribution pattern in February 2023 (Supplementary Figure S10). In terms of the DIC, its distributions in the surface water were higher in the nearshore zone than they were in the offshore zone, except in November 2022 (Supplementary Figure S10). On the contrary, the bottom DIC values in the inshore place were significantly lower than they were in the offshore zone in May 2023 and August 2023; meanwhile, no obvious distribution was found in other seasons (Supplementary Figure S10).

3.2.2. pH_{T@in situ} and Ω_{A@in situ}

As shown in Figure 3c, the average pH_{T@in situ} values of the DAB, DLB, and YMKSA sites exhibited clear seasonal variations, with the highest value, 8.19 ± 0.02, being observed in the surface water in February 2023 of the DLB and the lowest value, 7.95 ± 0.12, being observed in the bottom water in August 2023 of the YMKSA, which was opposite to the change in DIC (Figure 3). For different areas, the bottom pH_{T@in situ} of the YMKSA was significantly lower than that of the DAB and DLB in February 2022, May 2023, and August 2023 (p < 0.05). Contrary to the DIC, the bottom pH_{T@in situ} was generally lower than that of the surface water (p < 0.05); a difference of up to 0.22 pH units in the YMKSA occurred in August 2023 (Figure 3c).

The Ω_{A@in situ} values ranged from 2.14 ± 0.55 to 3.62 ± 0.16, with the highest and lowest values all observed in August 2023 for the DAB (surface water) and YMKSA (bottom water) (Figure 3c). Regionally, the bottom Ω_{A@in situ} in the YMKSA was generally lower than that in the DAB and DLB, especially in August 2023 (Figure 3c). Similarly, the Ω_{A@in situ} value for the bottom water was significantly lower than that for the surface water; a difference up to 1.21 Ω occurred in August 2023 in the YMKSA (Figure 2c).

The spatial distributions of pH_{T@in situ} and Ω_{A@in situ} are shown in Supplementary Figures S8–S13. During the investigation period, the lower values of pH_{T@in situ} and Ω_{A@in situ} were mainly distributed in the nearshore areas of the DAB, DLB, and YMKSA. In addition, during the survey period of November 2022 and August 2023, we clearly found that the values of pH_{T@in situ} and Ω_{A@in situ} in the coral reef area were significantly lower than those outside the reef area, especially in the bottom water (except for individual), thus indicating a potential link between coral reefs and OA to a certain extent. Similar results have been found in other coral reef areas, e.g., Dongsha Atoll (northern South China Sea) [45]. In addition, previous studies have shown that marine calcified organisms require the Ω_A value to be >1.5 for optimal growth [19]. Thus, ~40% of the YMKSA in the bottom water may not be suitable for the growth of calcified organisms.

3.2.3. pCO_{2@in situ} and FCO₂

In the surface water, the mean pCO_{2@in situ} values of the study area varied from 294.1 ± 30.1 µatm to 519.9 ± 45.0 µatm; its spatiotemporal variation was opposite to that of pH_{T@in situ} and Ω_{A@in situ} (Figure 3). The corresponding values of FCO₂, shown in Supplementary Figure S14, varied from –1.69 ± 0.41 mmol C m⁻² d⁻¹ to 1.70 ± 0.74 mmol C m⁻² d⁻¹. The DAB could absorb CO₂ from the atmosphere in November 2022, February 2023, and August 2023, with absorption fluxes of 1.51 ± 0.31 mmol C m⁻² d⁻¹, 0.48 ± 0.28 mmol C m⁻² d⁻¹, and 0.23 ± 0.99 mmol C m⁻² d⁻¹, respectively. However, during the survey of May 2023, DAB was a source of CO₂, with an average release flux of 0.27 ± 0.50 mmol C m⁻² d⁻¹. As for DLB, this region was a sink of CO₂ in November 2022 (−1.03 ± 0.08 mmol C m⁻² d⁻¹) and February 2023 (−0.49 ± 0.31 mmol C m⁻² d⁻¹), but it was a source of CO₂ in May 2022 (0.42 ± 0.36 mmol C m⁻² d⁻¹) and August 2023 (1.69 ± 0.74 mmol C m⁻² d⁻¹). However, the YMKSA was a source of CO₂ in February 2023 (0.29 ± 0.17 mmol C m⁻² d⁻¹) and May 2023 (0.58 ± 0.48 mmol C m⁻² d⁻¹) but a sink of CO₂ in November 2022 (−1.69 ± 0.41 mmol C m⁻² d⁻¹) and August 2023 (−0.21 ± 1.00 mmol C m⁻² d⁻¹) (Supplementary Figure S14).

As for the bottom pCO_{2@in situ}, its values ranged from 313.6 ± 25.9 µatm in November 2022 (DAB) to 767.6 ± 218.7 µatm in August 2023 (YMKSA). As shown in Figure 3d, the values for the bottom pCO_{2@in situ} in the DAB, DLB and YMKSA were significantly higher than those in the surface water in August 2023 (up to 360 µatm), which may significantly promote the source properties of CO₂ in the survey area, especially when the hydrodynamic conditions are strong [26]. As shown in Supplementary Figures S11–S13, the spatial distributions of pCO_{2@in situ} in the bottom water were opposite to those of the pH_{T@in situ} and Ω_{A@in situ} during the investigation period.

3.3. Coral Metabolic Process

During the culture experiment period, all the target corals did not show bleaching or death, and they were well adapted to the indoor system environment. The release/absorption fluxes of TA and DIC by dominant corals in the study area (i.e., A. pruinose, P. lutea, F. favus, A. digitifera, and P. carnosus) were obtained via indoor culture experiments. The results showed that the daily fluxes of TA absorbed by corals ranged from 0.06 ± 0.01 μmol cm⁻² to 0.77 ± 0.19 μmol cm⁻² during the investigation period, with the highest and lowest values occurring in F. favus (August 2023) and P. lutea (May 2023), respectively (Figure 4a). In addition, the daily flux of the DIC released by corals ranged from 2.08 ± 0.61 μmol cm⁻² in May 2023 (A. digitifera) to 10.86 ± 12.46 μmol cm⁻² in November 2022 (A. digitifera).

Compared with other studies, the TA absorption flux by corals in this study was almost the same as that of Favites complanata (F. complanata) (0.38 ± 1.26 μmol cm⁻²) by Guo et al. [46], but it was significantly lower than that of the A. digitifera (27.12 ± 2.33 μmol cm⁻²), Acropora formosa (A. formosa) (16.60 ± 0.70 μmol cm⁻²), and Siderastrea siderea (S. siderea) (54.0 μmol cm⁻²) by Kuffner et al. [47], Nishida et al. [48], and Chan et al. [49]. In addition, the DIC release flux of corals in this study was significantly higher than that of the Galaxea fascicularis (G. fascicularis) (0.53 μmol cm⁻²) and slightly lower than that of Acropora hyacinthus (A. hyacinthus) 13.82 μmol cm⁻² by Zhao et al. [50]. In fact, due to differences in the metabolism of corals under indoor and outdoor conditions, the calculated DIC release (and TA absorption) results may deviate from the actual situation, but this does not fundamentally affect its applicability in this study.

4. Discussion

4.1. Comparison of CO₂ Parameters with Other Reefs around the World

Table 1. Comparisons of the values of TA (μmol kg⁻¹), DIC (μmol kg⁻¹), pH_T, pCO₂ (μatm), and Ω_A from some coral reef waters in the world.

Location	Sampling Time	TA μmol kg−1	DIC μmol kg−1	pHT	pCO2  μatm	ΩA	Reference
Pedra da Risca do Meio Coral Reef	August and November 2020	2325 ± 19	2019 ± 16	7.980 ± 0.008	475 ± 28	n.d.	Cotovicz et al. [30]
Great Barrier Reef–Australia	September 2009 to August 2016	2288 ± 44	1989 ± 45	8.03 ± 0.05	404 ± 46	n.d.	Lønborg et al. [55]
Trawler Reef	August 2014	2289.3 ± 4.9	2003.0 ± 19.7	8.011 ± 0.020	n.d.	n.d.	Hannan et al. [56]
Big Vicki’s Reef	August 2014	2284.0 ± 9.8	1984.7 ± 14.7	8.044 ± 0.023	n.d.	n.d.	Hannan et al. [56]
Palfrey Reef	August 2014	2278.4 ± 7.4	1981.9 ± 22.7	8.048 ± 0.036	n.d.	n.d.	Hannan et al. [56]
Coral Reef Lagoon Kaneohe Bay	September 2003 to September 2004	2180 ± 36	1920 ± 16	n.d.	460 ± 52	n.d.	Fagan and Mackenzie [51]
The coast of Iriomote Island (Japan)	August 2017	2211 ± 44	1878 ± 103	8.009 ± 0.177	415 ± 81	n.d.	Akhand et al. [52]
Reef flat in Northeastern Brazil	July 2006	1857.6 ± 42.1	1623.0 ± 39.2	8.13 ± 0.12	371.6 ± 9.2	n.d.	Akhand et al. [52]
Luhuitou fringing reef, Sanya Bay, China	July 2010	2312.1 ± 15.3	1994.7 ± 40.9	8.007 ± 0.048	420.1 ± 62.4	n.d.	Zhang et al. [53]
Yongxing Island, China	July to August 2009	2421 ± 142	n.d.	8.23 ± 0.22	456 ± 249	n.d.	Yan et al. [54]
Fiery Cross Reef, China	July to August 2009	2240 ± 56	n.d.	8.22 ± 0.03	395 ± 25	n.d.	Yan et al. [54]
Great Barrier Reef, Australia	November 2011 to April 2012	2276 ± 16	1954 ± 25	8.03 ± 0.03	404 ± 40	n.d.	Albright et al. [57]
Gulf of California, Mexico	January to May 2014	n.d.	n.d.	n.d.	n.d.	3.03 ± 0.05	Norzagaray et al. [58]
Reef flats of Kiritimati Island	May and June 2018	2256.1 ± 22.2	1989.8 ± 59.6	7.95 ± 0.08	n.d.	3.1 ± 0.4	Knebel et al. [59]
Dongshan coral habitat	May 2019	2221 ± 0	1960 ± 4		402 ± 6	2.92 ± 0.05	Dong et al. [31]
	August 2019	2249 ± 1	2038 ± 0		619 ± 15	2.47 ± 0.01
	December 2019	2209 ± 1	2009 ± 4		436 ± 11	2.26 ± 0.03
Gray’s Reef	July 2006 to October 2007	n.d.	n.d.	8.08 ± 0.05	377 ± 10	3.47 ± 0.46	Xue et al. [42,43]
Eastern coast of Shenzhen, China	November 2022	2184.6 ± 24.6	1879.7 ± 30.5	8.12 ± 0.03	317.8 ± 26.2	3.40 ± 0.18	In this study
	February 2023	2219.6 ± 13.8	1998.4 ± 15.9	8.17 ± 0.03	405.2 ±35.8	2.50 ± 0.11
	May 2023	2178.8 ± 21.7	1938.3 ± 22.7	8.12 ± 0.03	458.5 ± 33.7	2.81 ± 0.15
	August 2023	2153.7 ± 37.0	1914.8 ± 58.5	8.06 ± 0.07	579.5 ± 119.8	2.89 ± 0.44

Note(s): n.d. = no data.

shows the concentration characteristics of seawater CO₂ parameters (TA, DIC, pH_T, pCO₂, and Ω_A) in the ECSZ (DAB, DLB, and YMKSA) and other coral reef areas in the world. In comparison, the results of CO₂ parameters in the study area were comparable to those of most other reefs listed in Table 1. For instance, the TA concentration there was comparable to that in the Coral Reef Lagoon Kaneohe Bay [51], coast of Iriomote Island [52], and Dongshan coral habitat [31], but it was significantly lower than that in the Pedra da Risca do Meio Coral Reef [30], Luhuitou fringing reef [53], and Yongxing Island [54]. As for the DIC, its concentration was comparable to that in the Great Barrier Reef [55], Big Vicki’s Reef [56], Palfrey Reef [56], Coral Reef Lagoon Kaneohe Bay [51], and Dongshan coral habitat [31], but it was significantly higher than that in the reef flat in Northeastern Brazil [52]. For other CO₂ parameters, e.g., pH_T, pCO₂, and Ω_A, their values were also comparable to most of the coral reef areas listed in Table 1. In addition, as mentioned above, the seawater CO₂ parameters in the ECSZ showed obvious seasonal variations and regional differences (DAB, DLB, and YMKSA), which were comprehensively forced by the physical and biogeochemical processes, such as temperature changes, water mixing, air–sea exchange, and biological metabolism, which are described in the following sections.

4.2. Factors Controlling the Variability of Seawater CO₂ Parameters

4.2.1. Temperature

Temperature is usually an important factor affecting the carbonate system in seawater. On the one hand, a temperature change can affect the solubility of CO₂ in seawater and the dissociation equilibrium of carbonate; that is, the solubility of CO₂ decreases with the increase in temperature. On the other hand, temperature can affect the metabolic activities of organisms, thereby regulating the inorganic carbon dynamic in seawater [17,60]). As can be seen in Figure 5, unlike for the surface pH_T and bottom Ω_A, temperature showed a significant linear relationship with pH_T, pCO₂, and Ω_A, with all data considered (n = 192), confirming the dominant role of seasonal changes in temperature on these CO₂ parameters (especially the pH_T and pCO₂). Similar results have been found in other survey sites, such as the Patos Lagoon [61], Gulf of Mexico [62], Aransas Ship Channel [63], East China Sea [64], Northern South China Sea [65], and Pearl River Estuary [66]. Furthermore, based on the above linear relationship (i.e., temperature vs. CO₂ parameters), the seasonal variation in temperature, i.e., 17.9–34.5 °C, can be estimated to explain 11.6% (43.4%), 17.9% (44.4%), and 55.6% (3.3%) of the changes in pH_T, pCO₂, and Ω_A of the surface (bottom) waters. Thus, temperature was an important factor controlling the seasonal variation in these CO₂ parameters.

For different regions (DAB, DLB, and YMKSA), the response of the CO₂ systems to seasonal temperature changes was different (Figure 5), which may be caused by the enhancement or canceling out of physical processes and biological processes [63]. As shown in Figure 5, temperature was significantly correlated with pH_T, pCO₂, and surface Ω_A (p < 0.001) in the DLB, and temperature changes can reveal 49.1–72.5% variations in those CO₂ parameters, suggesting the important control role of temperature. Different from that, there was no significant correlation between temperature and most CO₂ parameters in the DAB and YMKSA (many results deviate from their overall trajectory); temperature only showed a significant correlation with the bottom pH_T and pCO₂ and the surface Ω_A (p < 0.001 or p < 0.05), but the significance was relatively weak, especially in the YMKSA (p < 0.05), indicating the control effect of other factors besides temperature, e.g., biological processes and air–sea exchange of CO₂ [67].

4.2.2. Water Mixing

The water-mixing process can alter the concentration of TA and DIC in seawater, in turn, affecting the geochemical behavior of the CO₂ system [61,68,69]. Since salinity changes can effectively indicate the mixing process of different waters, the effect of water mixing on the carbonate system can be quantitatively (semi-quantitatively) revealed by fitting the relationship between the salinity and CO₂ parameters. As shown in Figure 6, taking all the data into account (n = 192), salinity showed a significant positive correlation with the surface TA (y = 44.6x + 724, p < 0.001, Figure 6a), surface DIC (y = 80.5x − 713.2 724, p < 0.001, Figure 6c), and bottom DIC (y = 33.4x + 858.0, p < 0.01, Figure 6c), indicating the significant effects of water mixing (dilution effect). According to the linear relationship between them, it can be estimated that the salinity changes in the surface (29.6–33.5) and bottom water (28.8–33.8) can explain 47.9%, 86.5%, and 58.1% of the changes in the surface TA, surface DIC, and surface DIC, respectively. However, no significant correlation between salinity and TA was observed in the bottom water, which may be related to biological calcification in the study area [27]. As mentioned above, the bottom TA concentration in the reef area was significantly higher than that in the outer area of the reef area, especially in the YMKSA (Supplementary Figure S10), thus indicating the significant effect of coral calcification, which was also confirmed in other studies [70,71,72]. For other CO₂ parameters (pH_T°C24.8°C, pCO_{[email protected]°C}, and Ω_{[email protected]°C}), similarly, they showed a significant linear relationship with salinity in the surface and bottom water (p < 0.001; Supplementary Figure S15), and the salinity change could reveal 29.2–70.5% of their variation. The above results indicated that the freshwater dilution may have an effect on the seasonal variation in the seawater CO₂ system. However, considering the low seasonal variation in salinity (amplitude of variation < 1.5), we believe that dilution may have a limited contribution to the change in the inorganic carbon system.

As for the different study areas (i.e., DAB, DLB, and YNKSA), however, we only found that the salinity in YMKSA showed a significant positive correlation with the carbonate parameters (except for the bottom TA) (p < 0.001; Figure 6 and Supplementary Figure S15), thus indicating the important influence of water mixing (dilution effect), which can explain 64.4–169.2% of the seasonal amplitude of carbonate parameters (TA, DIC, pH_T, pCO₂, and Ω_A) in water. However, focusing on DAB and DLB, no significant correlation was found except for the relationship between salinity and surface DIC, pH_T, and pCO₂ (weak significance, p < 0.05), thus indicating that there were other important factors besides water mixing, e.g., biological activity.

4.2.3. Photosynthesis and Respiration

Due to the extremely high biodiversity and biomass, the metabolic processes of organisms in coral reefs are very complex, including the primary production of phytoplankton and benthic algae; and the heterotrophic metabolism of corals, microorganisms, and reef organisms. These processes can all have a large impact on the CO₂ system in the water column [27,29].

Generally, the photosynthesis of algae can be described by the formula 106CO₂ + 122H₂O + 16HNO₃ + H₃PO₄ → (CH₂O)₁₀₆(NH₃)₁₆H₃PO₄ + 138O₂. Therefore, photosynthesis can reduce the DIC and increase the TA, thereby regulating the CO₂ system in seawater [25]. In this study, the surface Chl a showed a significant correlation with the NDIC, NpH_{[email protected]°C}, NpCO_{[email protected]°C}, and Ω_A on an annual basis, excluding the YMKSA data of May 2023 (n = 152), but relatively weak (p < 0.05), indicating that the primary production controlled the seasonal variation in the surface CO₂ system to some extent. However, as shown in Supplementary Figure S16, the data of the Chl a and CO₂ parameters in YMKSA for May 2023 significantly deviated from their overall relationship, which may be the result of multiple factors, e.g., biological calcification, respiration, and air–sea exchange [42,43].

In contrast to photosynthesis, the respiration process by microorganisms and reef organisms can increase DIC and reduce TA in seawater [25]. During the investigation period, we only analyzed the respiration and calcification processes of corals; we did not analyze the metabolic processes of other organisms (such as reef organisms and microorganisms). Therefore, it was difficult to directly quantify the contribution of the biological respiration to the seawater CO₂ system. Generally, the AOU is a key indicator of net metabolic activity, i.e., net process of photosynthesis and respiration [73]. Therefore, the relationship between the AOU and CO₂ parameters can effectively reveal the effect of biological metabolic process on the seawater CO₂ system. As demonstrated by Figure 7, taking all the data into account (n = 192), AOU showed a significant linear relationship with NDIC, NpH_{[email protected]°C}, NpCO_{[email protected]°C}, and NΩ_{[email protected]°C}, especially in the bottom seawater (p < 0.001), and its distribution pattern basically followed the trend of Redfield line [74], indicating that biological processes (photosynthesis and respiration process) could be the main processes controlling the seasonal variation in these CO₂ parameters [25].

Regionally, the AOU has a very significant linear relationship with the NDIC, NpH_{[email protected]°C}, NpCO_{[email protected]°C}, and NΩ_{[email protected]°C} in the bottom water of the DAB, DLB and YMKSA (p < 0.001, n = 192); however, their relationship was relatively weak in the surface water due to factors such as the air–sea exchange (Figure 7). In terms of the DIC, the slopes of the NDIC–AOU regression for the bottom water of the DAB and DLB, 0.69 and 0.63 (Figure 7b), were close to the Redfield stoichiometry, i.e., the C/O molar ratio of 0.77 (106/138) [74], which was similar to those observed in other coastal waters, e.g., Gulf of Trieste [75] and North Yellow Sea [26], reflecting the dominant role of phytoplankton photosynthesis and microbial respiration. However, the slope of NDIC–AOU regression in the YMKSA was 0.39, which was much lower than that in Redfield stoichiometry, which may be related to the metabolism of coral reef organisms, such as fish and corals [76,77,78,79].

Furthermore, according to the change in the AOU and the linear relationship between the AOU and CO₂ parameters, the variations in the DIC, pH_T, pCO₂, and Ω_A in the bottom water by biological processes were estimated to be 152.8 μmol kg⁻¹, 0.244, 310.1 μatm, and 1.33 of the DAB; 93.1 μmol kg⁻¹, 0.193, 256.5 μatm, and 1.05 of the DLB; and 87.1 μmol kg⁻¹, 0.335, 466.9 μatm, and 1.78 of the YMKSA, respectively. In comparison, the contribution of biological processes to the seasonal variation in the Ω_A was comparable to that of temperature; however, its contribution to pH_T and pCO₂ was relatively low, accounting for about 47–90% of the temperature. Of course, it is undeniable that there may be a large deviation between the results quantified by linear fitting and the actual values. The main reasons are as follows: (1) theoretically, the relationship between the AOU and pH_T, pCO₂, and Ω_A was not a simple first-order linear relationship; and (2) the distribution of the inorganic carbon parameter–AOU was scattered, especially in the summertime, which may be related to hypoxia.

4.2.4. Coral Metabolic Activity

Coral communities can change the concentration of TA and DIC in seawater through photosynthesis, respiration, and calcification to regulate the marine CO₂ system [27]. Based on the results of culture experiments and the distribution of corals in the study area, the contribution of coral metabolism to TA and DIC in seawater was preliminarily calculated. The results showed that coral metabolic processes may have little effect on TA and DIC in the seawater of the survey area. On average, coral metabolism only reduced TA in the water column of the DAB, DLB, and YMKSA by 0.16 ± 0.03 μmol kg⁻¹, 0.03 ± 0.01 μmol kg⁻¹, and 0.21 ± 0.04 μmol kg⁻¹ per month; correspondingly, this process increased the DIC by 3.57 ± 0.54 μmol kg⁻¹, 0.81 ± 0.12 μmol kg⁻¹, and 4.65 ± 0.70 μmol kg⁻¹, respectively. A previous study indicated that, in autumn, the microbial respiration of seawater in here can simultaneously reduce the concentration of TA by 14.4–22.8 μmol kg⁻¹ and increase the concentration of DIC by 90.0–168 μmol kg⁻¹ per month [28]. In comparison, the effect of coral metabolism on TA and DIC was much lower than that of microbial respiration. Similar results were also observed in other coral reefs, e.g., Kāne ‘ohe Bay [9].

However, during the survey period, we found the CO₂ parameters of the bottom water in the study area, especially in the YMKSA, showed significant differences between the reef area and the non–reef area (p < 0.05), which has also been found in other coral reef waters [54], reflecting the significant impact of coral communities on the reef area. It was assumed that coral metabolic activity only affected the CO₂ system within the reef area (ignoring the water exchange inside and outside the reef area). In this scenario, the coral metabolism can reduce the concentration of TA by 4.19 ± 1.74 μmol kg⁻¹ (DAB), 1.84 ± 0.34 μmol kg⁻¹ (DLB), and 2.71 ± 1.90 μmol kg⁻¹ (YMKSA) per month; correspondingly, it can increase the concentration of DIC by 88.14 ± 30.45 μmol kg⁻¹, 40.7 ± 6.12 μmol kg⁻¹, and 51.77 ± 17.40 μmol kg⁻¹, respectively, which was almost equivalent to the contribution of microbial respiration.

In addition, coral communities can affect the carbon cycle of their habitats. For instant, the metabolic process of corals can continuously release a large amount of active organic carbon (e.g., coral mucus) into seawater, especially under environmental stress, which can significantly affect the carbon metabolism process of water and sediment [80]. Furthermore, coral reefs can accumulate a large number of benthic organisms, and their metabolism also has an important impact on the CO₂ system. To sum up, we speculate that large–scale coral reef was a non–negligible factor leading to OA in coastal waters.

4.3. 1–D Model Reveals the Seasonal Driving Mechanism of CO₂ System

The contributions and relative importance of different controlling processes to the seasonal changes in the carbonate parameters (TA, DIC, pH_T, pCO₂, and Ω_A) are shown in Figure 8. The specific contributions of different processes (temperature change, water mixing, air–sea exchange of CO₂, coral metabolism, and other biological processes) were shown in Supplementary Material Section S1.2. Although the conclusions obtained from the 1–D model were generally consistent with those derived from the property regression, sometimes, they also illustrate how the various processes enhance or cancel each other.

As for TA, water mixing and biological processes (except coral metabolism) were generally the main driving forces for its seasonal variations (Figure 8a–c). However, in terms of different study areas (DAB, DLB and YMKSA), there were some differences in the effects of the above processes (i.e., water mixing and biological processes) on TA changes. For example, water mixing dominated the change in TA in the YMKSA from November 2022 to February 2023, and from May to August 2023, while the biological process controlled its change from February to May 2023. However, the seasonal variations in TA in the DAB and DLB were jointly controlled by water mixing and biological processes (except for DAB from November 2022 to February 2023).

In the DAB and YMKSA, DIC was controlled by biological processes, air–sea exchange, coral metabolism and water mixing (Figure 8d,f). In comparison, coral metabolism had little effect on the seasonal variation in DIC in the DLB, which was also true for other CO₂ parameters (pH_T, pCO₂ and Ω_A) (Figure 8). Among them, biological processes were the most important contributors to controlling the seasonal variation in DIC in the DAB and DLB from November 2022 to May 2023, while biological processes and water mixing jointly controlled their changes from May to August 2023 (Figure 8d,e). Different from the former, biological processes dominated the DIC changes in the YMKSA from February to May 2023. In other stages, biological processes, air–sea exchange, coral metabolism, and water mixing jointly controlled the seasonal variation in the DIC (Figure 8f).

As for other CO₂ parameters (pH_T, pCO₂, and Ω_A), biological processes, temperature changes, air–sea exchange, and coral metabolism (except DLB) jointly dominated their seasonal changes (Figure 8g–o). However, water mixing had little effect on the above parameters. As shown in Figure 8, overall, the seasonal driving factors of pH_T were opposite to that of pCO₂. Among them, biological processes, temperature changes, air–sea exchange, and coral metabolism were the main contributors to the changes in pH_T and pCO₂ of the DAB and YMKSA between November 2022 and February 2023 and between May 2023 and August 2023, while biological processes, temperature changes, and coral metabolism controlled their changes between February and May 2023. The seasonal driving factors of pH_T and pCO₂ in the DLB were almost consistent with those of DAB and YNKSA, except for the little effect on coral metabolism. Similarly, as for Ω_A, biological processes, temperature changes, air–sea exchange, and coral metabolism jointly dominated its seasonal changes in the DAB and YMKSA. In the DLB, biological processes, temperature changes, and air–sea exchange control its changes, while coral metabolism has little effect (Figure 8n).

5. Conclusions

In this study, we demonstrated the dynamics of the seawater CO₂ system and its controlling factors in the DAB, DLB, and YMKSA in the November 2022, February 2023, May 2023, and August 2023. Overall, DAB acted as a sink for atmospheric CO₂ with the exchange flux of −1.51 ± 0.31 to 0.27 ± 0.50 mmol C m⁻² d⁻¹, while DLB and YMKSA acted as a CO₂ source with the exchange fluxes of −0.42 ± 0.36 to 1.69 ± 0.74 mmol C m⁻² d⁻¹ and −0.58 ± 0.48 to 1.69 ± 0.41 mmol C m⁻² d⁻¹, respectively. In addition, ~40% of the YMKSA in the bottom water may pose a potential threat to coral growth due to Ω_A < 1.5.

The seasonal variation in the seawater CO₂ system was mainly driven by a variety of physical and biogeochemical factors, mainly including temperature change, mixing-effect biochemical processes, air–sea exchange, and coral metabolism. The biological processes and mixing effect could be the most important factors for the seasonal variation in TA. As for DIC, in addition to biological process and mixing, its seasonal variation was affected by the air–sea exchange and coral metabolism to some extent. Different from the former, the pH_T, pCO₂, and Ω_A were mainly controlled by the temperature change, biochemical processes, air–sea exchange, and coral metabolism, while water mixing had little effect on them. In addition, our results indicated that coral communities can significantly increase the DIC/TA ratio by reducing the TA concentration and increasing the DIC in the reef waters, which may promote the acidification of local seawater and need attention. Although this study focused on the subtropical coral reefs, it is reasonable to assume that the presence of calcified organisms in other tropical reefs will also accelerate coastal acidification.