JB, with an area of about 640 km², is the largest embayment in the southeastern part of the Korean Peninsula ( Figure 1 ). The tide in JB is semidiurnal, with a range between 0.45 m during neap tide and 1.8 m during spring tide (Kim et al., 2014). The current during spring tide is approximately 90 cm s^–1 and about 30 cm s^–1 during neap tide, while weak currents (< 10 cm s^–1) are observed in the inner bay (Kang, 1991). JB is surrounded by the cities of Jinhae, Masan, Goseong, Tongyeong, and Geoje with a combined population of more than a million people and including an enormous industrial complex (Kim et al., 2014). JB has been one of Korea’s major fisheries with high productivity and about 48 km² of the area is used for intensive large-scale aquaculture, including oyster, mussel, sea squirt, and blood cockle farming (Lee et al., 2018). In particular, the Jinhae-Tongyeong coast, including the JB, in southeastern Korea is a major hub for oyster aquaculture, representing about 80% of the nation’s total production (http://www.foc.re.kr). In addition, long water residence times and low tidal current velocities, because of the topographic features of the semi-enclosed bay, lead to deposition of biodeposits (feces and pseudo-feces) close to the aquacultural areas (Hyun et al., 2013).

Water and sediment core sampling, and other in situ experiments, were carried out in December 2022 in Dangdong Bay, which is one of the several inner bays located on the western side of JB ( Figure 1 ). Oyster (Grassostrea gigas) and ascidian (Styela clava) farming occur around the sampling area, and large quantities of oysters are harvested during November and February. The water temperature, salinity, pH, and dissolved oxygen (DO) in the water column were measured in situ using a calibrated multiparameter water quality meter (YSI-6600; YSI Inc., Yellow Springs, OH, USA). Water samples from the surface and bottom layer, for the measurement of dissolved inorganic nutrients, were collected using a Niskin water sampler (model 1010, General Oceanics Inc., Miami Gardens, FL, USA). The water samples for measuring dissolved inorganic nutrients were immediately filtered using a membrane syringe filter (pore size: 0.45 μm, Advantec, Japan) and then stored below –20°C pending analysis.

Sediment cores for geochemical analysis were collected by scuba divers in duplicate or triplicate using an acrylic core to minimize surface disturbance. They were stored onboard in coolers with ice. Any core containing visually significant bioturbation or shells along the walls, that could affect the vertical stratifications, was not retained. Sediment pore water samples were extracted using a Rhizon sampler (Rhizon CSS, Rhizosphere Research Products, Netherlands) at 1-cm depth intervals through predrilled and taped holes in the core (Seeberg-Elverfeldt et al., 2005). Pore water samples were kept frozen (−20°C) until processed in the laboratory. Sediment samples for measuring total organic carbon (TOC) and total nitrogen (TN) content were sectioned at 0–2-cm intervals and kept in a deep freezer (−20°C) until laboratory processing.

To estimate the in situ sediment total oxygen uptake (TOU) rate, diffusive oxygen uptake (DOU) rate, and benthic nutrient fluxes at the sediment–water interface, we deployed the autonomous in situ benthic lander on the sea floor with the assistance of a scuba diver. The benthic lander consists of the benthic chamber (BelcII), microprofiler (BelpII), and automatic water sampler (Lee et al., 2012; Cho et al., 2023). In brief, the BelcII contains an opaque rectangular chamber with a base area of 841 cm² (29 × 29 cm²), and an automatic syringe water sampler. After reaching the seafloor and waiting for 2 h, the chambers are enclosed by a lid with a motor-driven closing system. Once the lid is closed, four stirring bars on the lid rotate at a speed of ~30 rpm, creating a diffusion boundary with a layer thickness of about 300–700 μm at the sediment–water interface (Lee et al., 2012). Variations of O₂ concentration in the chamber are measured at 10-s intervals over the incubation period by an oxygen optode sensor (4330F, Aanderaa, Norway) mounted on the lid (Lee et al., 2012; Cho et al., 2023). Meanwhile, the BelpII consists of the four separate O₂ microoptode systems (OXR50-HS-SUB and FSO2-SUBPORT, PyroScience GmbH, Germany) and is equipped with a custom-built motor-driven linear stage, allowing for high spatial resolution (<100 μm) measurement of oxygen at the sediment–water interface. The O₂ microoptode consists of a retractable fiber of 230 μm with a tip diameter of 50–70 μm and a 90% response time of less than 0.8 s. The zero-reading for O₂ concentration was determined in the anoxic zone of the profiles, where oxygen has reached a constant value. Before deploying the benthic lander, the oxygen sensors for the chamber were calibrated at 100% air-saturation using air-bubbled water and 0% air-saturation using sodium dithionite-added water. After bringing the benthic chamber on board, water samples were collected by syringe sampler for nutrient analysis and filtered immediately using a membrane syringe filter (pore size 0.45 μm, Advantec, Tokyo, Japan) and then stored in a freezer (−20°C) until analysis.

Four acrylic cylindrical sediment traps with a diameter of 7 cm, a length of 60 cm (an aspect ratio of 8.6) were deployed for more than 24 h at each site to collect the vertical flux of particulate OC and TN (Lee et al., 2014; Kim et al., 2021; Cho et al., 2023). With the assistance of a scuba diver, these sediment traps were placed on the upper frame of the benthic lander. To ensure the preservation of particulate materials and prevent sample wash-out, the trap bottles filled with filtered saline water (more than 50) and attached to the bottom of each sediment trap after approximately 2 hours. A scuba diver subsequently recovered the sediment traps prior to retrieval of the benthic chamber. The overlying water was cautiously siphoned off within 3 h, and samples were stored in a refrigerator until laboratory processing.

Dissolved inorganic nutrient (NH₄ ⁺, NO_x, PO₄ ^3–, and Si(OH)₄) concentrations in the water column and sediment pore water were measured using an autoanalyzer (QuAAtro 39, SEAL Analytical, Mequon, WI, USA) within a week. To minimize potential artifacts for nutrient analysis, our samples were promptly thawed and filtered to remove particulate matter, ensuring that any silicate precipitate formed during freezing was excluded from analysis. Certified reference material (KANSO Ltd., Japan) was included in each batch of nutrient samples and were used to ensure the accuracy of the samples. Reproducibility was generally within ±7%. The porosity of the sediment was determined from the net weight difference between wet and dry sediment. The total organic carbon (TOC) and total nitrogen (TN) sediment contents were measured using a CHN analyzer (Thermo Finnigan Flash EA 1112) after acidification with 1 M HCl to remove CaCO₃. TOC and TN contents were calculated from raw data using calibration series obtained with sulfanilamide standards (C₆H₈N₂O₂S, Thermo Fisher Scientific Inc, USA). To estimate the vertical flux of total mass (TM), TOC, and TN, two aliquots of samples from the sediment trap were filtered through precombusted GF/F filters (6827-1315, Whatman, UK) and freeze-dried until reaching a constant weight. The filtered sample was acidified, and its TOC and TN contents were determined using the same process as for the sediment samples.

The ²¹⁰Pb activity in the core samples was measured using an alpha spectrometer equipped with low-background silicon-surface barrier detectors (PIPS Detector Canberra, USA) (Lee et al., 2012). This assumed that the total activity of ²¹⁰Po (daughter nuclide of ²¹⁰Pb) in the sediments had reached radioactive equilibrium with ²¹⁰Pb. For this measurement, about 0.5–1 g of powdered sediment was spiked with ²⁰⁹Po and then digested with a mixture of concentrated HNO₃ and HCl on a hot plate. After reducing the ferric ions with ascorbic acid in an acid solution, the Po isotopes (²⁰⁹Po and ²¹⁰Po) were spontaneously deposited onto a silver disk at 70°C for 6 h with stirring. The activity was counted over a day to obtain sufficient counts (> 1000). Excess ²¹⁰Pb (²¹⁰Pb_xs) was determined by subtracting average ²¹⁰Pb activity in the lower sediment layer from the measured ²¹⁰Pb activities.

The TOU and BNF (benthic nutrient flux) across the sediment–water interface (SWI) were calculated as:

where F is the TOU or the BNF (mmol m^–2 d^–1), dC/dt is the slope of the line derived from regressing the TOU or BNF concentration as a function of incubation time (mmol L^–1 d^–1), V/A is the chamber height [V: chamber volume (m³), A: chamber area (m²)].

The high resolution O₂ vertical profiles at the upper diffusive boundary layer (DBL) and SWI were determined to calculate the DOU rate (Jørgensen and Revsbech, 1985). The DOU rate was derived from the in situ diffusion coefficient of temperature and salinity and the oxygen gradient within the DBL using the following equation (Glud, 2008):

where D₀ is the molecular diffusion coefficient of oxygen corrected using in situ temperature and salinity (Broecker and Peng, 1974; Li and Gregory, 1974) and dO₂/dz is the oxygen concentration gradient (mmol cm^–4) within the DBL.

Using the vertical ²¹⁰Pb_xs profiles, and assuming a steady state, the apparent sedimentation rate (SR) was estimated using:

where A is the ²¹⁰Pb_xs activity in each sediment layer, A₀ is the ²¹⁰Pb_xs activity in the surface sediment (z=0), λ is the ²¹⁰Pb decay constant (0.031 y^–1), and SR is the sedimentation rate (cm y^–1). SR was estimated as the slope of the least-square regression between ln(²¹⁰Pb_xs) and sediment depth (z, cm) below the mixed surface layer of sediment.

We calculated the theoretical maximum sedimentation rate (Max_sed) to assess the influence of resuspended particles using the following equation (Lee et al., 2014):

where Max_sed is the maximum sedimentation rate (cm y^–1), the TM is the total mass flux collected by the sediment trap (g m^–2 d^–1), ϕ is the surface sediment porosity, and ρ is the dry bulk density (g cm^–3).

Analysis of variance (ANOVA) was performed using SPSS version 21 (SPSS for windows, SPSS INC., Chicago, USA). One-way ANOVA followed by Tukey’s multiple comparison or Kruskal-Wallis non-parametric multiple comparison tests were performed to assess spatial difference in the TOC and TN concentration of the sediment.

The water depth, temperature, salinity, DO, and nutrients of the water samples are listed in Table 1 . The water depth at the stations was approximately 15 m. Across the entire sampling site, water temperature, salinity, and pH ranged from 10.3 to 11.2°C (average, 10.8 ± 0.32°C), from 33.2 to 33.5 (average, 33.4 ± 0.09), and from 7.66 to 7.92 (average, 7.78 ± 0.10), respectively. Neither temperature nor salinity exhibited significant differences horizontally or vertically, suggesting a well-mixed water column in the study area. DO concentrations ranged from 270 to 302 μmol L^–1 (average, 285 ± 12.5 μmol L^–1) and were well saturated (96–108%). NH₄ ⁺, NO_x, PO₄ ^3–, and Si(OH)₄ concentrations ranged from 1.38 to 1.95 μmol L^–1 (average, 1.69 ± 0.22 μmol L^–1), from 1.17 to 2.66 μmol L^–1 (average, 1.78 ± 0.46 μmol L^–1), from 0.31 to 0.43 μmol L^–1 (average, 0.38 ± 0.04 μmol L^–1), and from 11.5 to 16.4 μmol L^–1 (average, 13.5 ± 1.54 μmol L^–1), respectively.

The depth profiles of TOC and TN contents in sediment at the study site are shown in Figure 2 . The TOC contents in the surface sediment (< 2 cm) at St.3 were 7.26 ± 0.28%, significantly higher than those measured at St.1 (4.47 ± 0.36%), St.2 (4.14 ± 0.03%), and St.4 (4.56 ± 0.14%) (ANOVA, p <0.05). Similarly, the TN content observed at St.3 (0.65 ± 0.11%) was high compared with St.1 (0.50 ± 0.001%), St.2 (0.40 ± 0.001%), and St.4 (0.52 ± 0.02%). Overall, the vertical profiles of the TOC and TN contents showed a slight decrease with depth, but the profiles at St.3 showed an exceptionally steep decrease with depth (ANOVA, p <0.05).

The vertical distributions of NH₄ ⁺, NO_x, PO₄ ^3–, and Si(OH)₄ concentrations in pore water are illustrated in Figure 3 . NH₄ ⁺ concentrations at St.1 remained low throughout the depth range, whereas they increased steeply in the range 0–6 cm but remained relatively constant at lower depths at St.3. NH₄ ⁺ concentration at St.2 slightly increased near the surface, and stayed constant below, and increased slightly with depth at St.4. The highest concentrations of NO_x were measured between 0 and 1 cm at all sites, except for St.3, and the concentration decreased below 1 μmol L^–1 at greater depths, indicative of denitrification and/or anammox processes. PO₄ ^3– and Si(OH)₄ in the pore water had similar vertical distribution patterns to the NH₄ ⁺ concentrations.

The temporal evolution of O₂ in the benthic chamber is illustrated in Figure 4 . Over time, there is a notable decrease in O₂ concentration, attributed to benthic mineralization of organic matter within the sediment. The TOU was estimated to be within the range of 38.4 to 49.6 mmol m^–2 d^–1 based on the gradient (dO₂/dt) of the relationship between O₂ concentration and time, as detailed in Table 2 .

The vertical distributions of O₂ in the pore water at St.2, St.3, and St.4 are presented in Figure 5 , measured using an in situ microprofiler (BelpII). Unfortunately, O₂ profiles at St.1 could not be obtained because the oxygen sensor’s height at the time of microprofiler installation was insufficient to penetrate from the DBL to the sediment. The mean oxygen penetration depths (n = 10) at St.2, St.3, and St.4 were 3.38 ± 1.33, 2.93 ± 1.70, and 8.49 ± 4.83 mm, respectively. DOU values, estimated using the oxygen slope within the DBL, ranged from 12.3 ± 1.8 to 15.1 ± 1.4 mmol m^–2 d^–1 ( Table 2 ).

The nutrient concentrations in the benthic chamber generally increased during the in situ incubation period, but decreases in NO_x at St.2, St.3, and St.4 imply that denitrification and anammox processes might significantly influence nitrogen cycling in JB sediments ( Figure 6 ). It remains uncertain if this phenomenon is related, but NH₄ ⁺ benthic flux at St.1 showed a minimum. BNF estimates were based on the linear relationships between nutrient concentration and time ( Figure 6 ). The BNF across the SWI ranged from 0.96 ± 0.04 to 2.28 ± 0.16 mmol m^–2 d^–1 for NH₄ ⁺, from –0.51 ± 0.28 to 0.18 ± 0.03 mmol m^–2 d^–1 for NO_x, from 0.02 ± 0.01 to 0.05 ± 0.04 mmol m^–2 d^–1 for PO₄ ^3–, and from 6.72 ± 0.14 to 9.11 ± 0.14 mmol m^–2 d^–1 for Si(OH)₄.

The vertical profiles of excess ²¹⁰Pb in the sediment showed intensive biological mixing to a depth of 20 cm ( Figure 2 ). The surface mixed layer at St.2 was significantly deeper than at other stations. In the vertical profiles of St.3 and St.4, the equilibrium state of ²¹⁰Pb was found at sediment depths deeper than 15 cm. We used average ²¹⁰Pb activities at equilibrium with ²²⁶Ra, which showed constant activity in deeper sediment layers, to calculate the excess ²¹⁰Pb activities. However, this equilibrium was not evident at St.1 and St.2 because the sediment cores collected at these stations were not long enough. Therefore, for St.1 and St.2, we used the ²¹⁰Pb activities at the deepest sediment layers available, at 23 cm for St.1 and 29 cm for St.2, where secular equilibrium with ²²⁶Ra was assumed. The sedimentation rates, calculated from the relationship between ²¹⁰Pb_xs and sediment depth, were estimated to be 0.33 ± 0.03 cm y^–1 at St.1, 0.13 ± 0.03 cm y^–1 at St.2, 0.27 ± 0.06 cm y^–1 at St.3, and 0.17 ± 0.07 cm y^–1 at St.4 ( Table 3 ).

The vertical flux of total mass (TM) ranged from 7.10 ± 0.80 to 27.4 ± 5.90 g m^–2 d^–1, with the highest values at St.2 ( Table 3 ). Vertical fluxes of OC (C_in) and nitrogen (N_in) were estimated to range from 45.5 ± 7.00 to 93.0 ± 25.3 mmol C m^–2 d^–1 and from 4.50 ± 0.50 to 9.30 ± 3.90 mmol N m^–2 d^–1, respectively, with the maximum values observed at St.2 ( Table 3 ).

TOU has been used as the proxy for the total carbon mineralization rate in marine sediment as it encompasses both aerobic respiration and the reoxidation of reduced inorganic constituents (such as NH₄ ⁺, Mn²⁺, Fe²⁺, and H₂S) from the anaerobically mediated OC degradation (Canfield et al., 2006; Glud, 2008). The estimation of OC mineralization through oxygen uptake relies on the premise that oxygen consumption directly correlates with the total carbon oxidized via aerobic and anaerobic pathways (Canfield et al., 1993). The OC oxidation rate (OC_ox) in the sediment was thus calculated from the TOU measurements, applying the Redfield stoichiometric ratio (C:O₂, 106:138, Redfield et al., 1963). The sedimentary OC_ox ranged from 29.5 ± 2.63 to 38.1 ± 4.59 mmol C m^–2 d^–1 (average; 34.3 mmol C m^–2 d^–1) and did not exhibit distinctive spatial differences ( Table 3 ). These ranges are lower than previous results of sea squirt and oyster farm (47–79 mmol C m^–2 d^–1, Lee et al., 2011), fish farm (52.2 mmol C m^–2 d^–1, Choi et al., 2020; 59.1–64.5 mmol C m^–2 d^–1, Kim et al., 2021), and those are higher than shellfish farm (20.6 mmol C m^–2 d^–1, Giles et al., 2006). In JB, the average ratio of OC_DIC, calculated from alkalinity and total dissolved inorganic carbon (DIC) flux, to OC_ox was 1.2, which suggests that anaerobic OC mineralization contributes to total OC oxidation (Lee et al., 2012). In addition, ratios > 1 in coastal sediments with higher organic contents indicate that a significant portion of the reduced substances are being released from the sediment layer into the bottom water or precipitated (e.g., FeS) with other reductants before undergoing complete oxidation in the oxic zone (Canfield et al., 2006; Ferrón et al., 2009). Considering that highly reduced substrates (e.g., H₂S, FeS, FeS₂) have been identified in the sediment (An, unpublished data), our estimations of OC oxidation, using TOU values and the Redfield ratio, are likely underestimated.

In general, TOU represents the sum of the DOU and benthic fauna-mediated O₂ uptake because DOU is a measure of the diffusive O₂ uptake via the sediment–water interface, excluding the impact of benthic fauna activities (Glud et al., 2003; Glud, 2008; Hicks et al., 2017; Kim et al., 2020b). In our results, the average ratio of TOU : DOU ranged from 3.12 to 3.28, which suggests significant benthic fauna activities in the JB sediments. The ratios estimated by the present study were higher than those reported in other hypoxic estuary (1.14–1.90, Bonaglia et al., 2014; 1.22–1.36; Seitaj et al., 2017), semi-enclosed bay (average 1.27, Glud et al., 2003), fish farm (0.9–1.7, Cathalot et al., 2012), shallow continental shelf (average 1.42, Lansard et al., 2008; 0.94–2.82, Kim et al., 2020b), and were comparable to those of a continental shelf and slope (0.78–4.25, Archer and Devol, 1992). Indeed, the deeper surface mixed layer in the sediment may indicate intensive biological activity ( Figure 2 ). Benthic fauna activities, including respiration, biological pumping, and bioirrigation, can significantly increase the TOU of coastal sediment (Glud et al., 2003; Glud, 2008). It can also result from increased reoxidation of reduced products from anaerobic respiration mediated by microbes, which can enhance the benthic O₂ uptake (Kristensen and Alongi, 2006; Kim et al., 2020b). Severe hypoxia has led to significant mass mortality and delayed recolonization of benthic species during the JB summer (Lim et al., 2006). However, the benthic community recovers from summer hypoxia and continues to do so until the following spring. Therefore, fauna activities associated with benthic O₂ uptake in the JB sediment during winter play an important role in the biogeochemical functioning of the benthic habitat.

To elaborate on the driving factors affecting particulate OC in the JB sediments, we estimated the OC mass budget at each station ( Figure 7 ). Assuming a steady state, the OC mass budget in the surface sediment layer equalizes input with outputs according to the following equation (Martens and Klump, 1984):

where OC_in is the input flux of OC into sediment from the water column, OC_ox is the flux of oxidation of OC in the sediment, and OC_burial is the burial flux of OC into the deep sediment layer. Lateral transport and biodeposition are the dominant input fluxes of OC in JB (Lee et al., 2012; Hyun et al., 2013). To facilitate the calculation of OC mass budget, the OC produced by primary production (PP) is assumed to be deposited onto surface sediment in shallow water. Therefore, the difference between OC_in and PP can be represented as the sum of lateral transport of OC and/or biodeposition from aquaculture activities (OC_lat+bio). Thus, the net of OC_in and PP may equate to the combined ambient lateral transport of OC and biodeposition (OC_lat+bio). The PP rate in the JB water column has been estimated to be 9.2 mmol C m^–2 d^–1 (NIFS, 2023). The OC_lat+bio ranged from 36.3 to 42.9 mmol C m^–2 d^–1 (average 40.0 ± 3.38 mmol C m^–2 d^–1), which is significantly higher than the PP in the water column. In JB, the aquaculture sector is mainly in the inner bay area and accounts for almost two-thirds of the commercial oyster and ascidian aquaculture in Korea (http://www.foc.re.kr). The aquaculture of filter-feeding shellfish (ex., oyster, mussel, and clam) and ascidian (S. clava) generates massive biodeposit volumes composed of pseudo-feces and feces that are enriched with organic content (Fiala-Médioni, 1974; Callier et al., 2006; Mitchell, 2006; Xia et al., 2019). In addition, the long-lines, net cages, and aquaculture infrastructure can change the hydrologic system and encourage the deposition of OC surrounding an aquaculture farm (Lee et al., 2011; Kim et al., 2020a). High contributions of biodeposits to OC_in mainly occur in and around sea squirt farms (Lee et al., 2012), mussel farms (Zúñiga et al., 2014; Lacoste et al., 2018; 2022), oyster–seaweed farms (Xia et al., 2019), and scallop farms (Zhou et al., 2006). Thus, massive biodeposits from oyster aquaculture seem to be the dominant source of OC_in in JB (Lee et al., 2011; Hyun et al., 2013).

The vertical fluxes collected by the sediment trap in coastal waters may be influenced by the resuspension of bottom sediments. The calculated Max_sed ranged from 0.42 cm y^–1 to 2.1 cm y^–1, which is approximately 2 to 3 times higher than the estimated SR from the excess ²¹⁰Pb profiles, except for St.2, where it is 16 times higher. Considering the uncertainties of the sediment trap (Reimers and Suess, 1983), we believe that our assessment of the influence of resuspension and/or biodeposition on OC_in is valid.

OC_burial into the deep sediment layer were calculated using the SR and the mean organic carbon content (largely constant) of the lower layer (> 30 cm) by applying the equation:

where SR is the sedimentation rate estimated from the excess ²¹⁰Pb activity (cm y^–1), ρ is the dry bulk density of the sediment, assumed to be 2.45 g cm^–3 (Kim et al., 2021), ϕ is the porosity, and OC_∞ is the mean organic carbon content below the mixed layer (mmol C g^–1). The burial fluxes of OC in the JB sediment ranged from 3.96 ± 1.00 to 7.17 ± 1.64 mmol C m^–2 d^–1 (average 5.91 ± 1.37 mmol C m^–2 d^–1) ( Table 3 ). Burial efficiencies, which were calculated as OC_burial/OC_in × 100, ranged from 4.25% to 15.8% ( Table 3 ), which is similar to, or higher than, previous results from a finfish cage farm (3.3%, Sim et al., 2023; 6–8%, Kim et al., 2021), a sea squirt farm (3.77%, Lee et al., 2012), and an oyster farm (6–10%, Kim et al., 2021). The findings indicate that a significant portion of the deposited OC was mineralized within the surface sediment layer.

The sum of OC_ox and OC_burial in JB was lower than the vertical flux of OC, which implies that OC is being transported through resuspension and deposited elsewhere. For example, OC released from aquaculture sediment can be transported and deposited over distances of tens to hundreds of meters through processes such as deposition and lateral transport (Hatcher et al., 1994; Kim et al., 2021). This laterally transported OC can also contribute to the vertical flux, influencing sedimentary OC dynamics in adjacent ecosystems (Kim et al., 2020a). The tidal current velocity in the study area was less than 10 cm s^–1, which is below the threshold current velocity (15–20 cm s^–1) required to lateral transport according to resuspended biodeposits (i.e., feces and pseudo-feces) (Widdows et al., 1998). However, a tentative threshold current velocity for reworking muds on the seabed in JB may be around 10 cm s^–1 at 1 m above the seabed (Lee et al., 2006). In addition, JB has numerous small islands, prominent coasts, and a swiftly shifting sea floor topography, and thus, topography and ocean currents combine to create small-scale eddy circulation (Kim et al., 2024). Therefore, we suggest that the source and physico-chemical processes of OC in the sediment are likely to have affected the calculated OC mass balance and should be further investigated to understand their potential consequences.

The fluxes of dissolved inorganic nitrogen (DIN, NH₄ ⁺ + NO_x; 1.11–2.03 mmol N m^–2 d^–1) and dissolved inorganic phosphate (DIP, PO₄ ^3–; 0.02–0.05 mmol P m^–2 d^–1) measured in JB were comparable with, or lower than, previous winter-time measurements at an oyster farm (5.74 mmol N m^–2 d^–1 and 0.32 mmol P m^–2 d^–1, Lee et al., 2011; 1.3 mmol N m^–2 d^–1 and0.05 mmol P m^–2 d^–1, Hyun et al., 2013; 8.95 mmol N m^–2 d^–1 and 0.51 mmol P m^–2 d^–1, Kim et al., 2021), a sea squirt farm (0.92 mmol N m^–2 d^–1; Lee et al., 2011), and a finfish farm (1.41 mmol N m^–2 d^–1 and 1.77 mmol P m^–2 d^–1; Choi et al., 2020; 5.45 mmol N m^–2 d^–1 and 1.67 mmol P m^–2 d^–1; Kim et al., 2021). In shallow coastal waters, DIN and DIP released from sediment can be an important source supporting primary production in the water column via benthic–pelagic coupling (Ferrón et al., 2009; Lee et al., 2012; Dixon et al., 2014; Choi et al., 2020). We calculated the potential contribution of benthic nutrients fluxes for primary production using the Redfield ratio (C:N:P = 106:16:1) and compared these with the apparent DIN and DIP demand for primary production of 1.39 mmol m^–2 d^–1 and 0.09 mmol m^–2 d^–1, respectively ( Table 4 ). In the present study, the DIN and DIP fluxes from sediment accounted for 82.1 to 149% and 23.1 to 57.6% of N and P primary production demand ( Table 4 ). In November, benthic DIN and DIP could support 29 to 152% and 0 to 137%, respectively, of primary production at sea squirt and oyster farms in JB and Tongyeong (Lee et al., 2011). In December, benthic DIN and DIP release 23 to 270% and 40 to 804%, respectively, of the primary production demand at finfish and oyster farms and reference sites in the inner part of Geoje-Tongyeong coastal area (Kim et al., 2021). These results indicate the importance of benthic nutrient regeneration in maintaining primary production in JB.