Jeju Island, a volcanic island situated off the southern coast of the Korean Peninsula, has an area of 1,847 km² and is predominantly composed of basalt, which is known for its high permeability (Kim et al., 2013; Kim et al., 2016) ( Figure 1 ). The island’s geological properties result in most of its rivers being classified as dry streams (Lee et al., 2022). The quality of the coastal water surrounding Jeju Island is influenced by various factors, including submarine groundwater discharge (SGD), land-based fish farm effluent, wastewater treatment plant (WWTP) effluent, and the surrounding water masses (Cho et al., 2021; Kwon et al., 2022; Lee et al., 2022; Ji et al., 2023). SGD acts as a pathway for the transport of DIN, DON, and trace metals (Jeong et al., 2012; Cho et al., 2019). The fluxes of DIN and DON due to SGD in Hwasun Bay, situated on Jeju Island between stations 2 and 3 in this study area (approximately 8 km apart), were found to exceed the averages observed for major rivers such as the Delaware, Colorado, and Stikine Rivers (Kim et al., 2013). In Hwasun Bay, the SGD-derived input of dissolved nutrients and organic matter accounted for over 90% of the total input flux (Cho et al., 2021). Furthermore, according to Kwon et al. (2022), fish farm effluent is a significant source of anthropogenic DOM and nutrients in coastal waters off Jeju Island. They found that fish farm effluent accounted for approximately 95% of NH₄ ⁺ (with an average concentration of 5.6 ± 4.1 μM) and 71% of humic-like fluorescent DOM (FDOM; with an average intensity of 0.86 ± 0.49 R.U.) in coastal waters of Jeju Island. WWTP effluent has also been highlighted as a significant source of nutrients in oligotrophic oceans (Lee et al., 2022).

We employed data from Korea Marine Environment Monitoring sampling stations, which have been established approximately 5 km from the shore for the management of the marine ecosystem in the coastal waters off Jeju Island. Water samples were collected from the coastal waters off Jeju Island, South Korea, during all four seasons in 2017 (February, May, August, and November for winter, spring, summer, and autumn, respectively) (Song, 2020). A total of 109 samples were obtained using a Niskin sampler from 12 sampling stations ( Figure 1 ), made up of surface samples (n = 46; excluding station 9 in May) and water column samples (n = 63; ranging from 10 to 90 m in depth, with an average depth of 36 ± 21 m). The selection of sampling stations in the water column was determined based on the water depth, with the number of stations (ranging from 2 to 3) fluctuating as the depth changed. Bottom water samples were obtained at a distance of approximately 5 m above the ocean floor.

Seawater samples for DOC, total dissolved nitrogen (TDN), dissolved inorganic nutrients (NO₃ ⁻, NO₂ ⁻, NH₄ ⁺, Si(OH)₄, and PO₄ ³⁻), and FDOM were filtered using pre-combusted glass fiber filter paper (GF/F; pore size = 0.7 µm; Whatman Inc., UK) subjected to 500°C for 5 h. Filter paper samples with suspended particles were used for the analysis of chlorophyll-a (chl-a); these were stored at −20°C in polypropylene conical tubes until analysis. FDOM samples were stored in pre-combusted amber glass vials and refrigerated (< 4°C) until analysis. The subsamples for DOC and TDN were transferred into pre-combusted glass ampules and acidified with 6 M HCl (pH ~2.0); the ampules were then flame-sealed. The subsamples for the dissolved inorganic nutrients were stored in high-density polyethylene bottles and frozen until analysis.

Salinity was measured in situ using a portable sensor (CyberScan PCD650; Thermo Fisher Scientific, MA, USA). DOC and TDN concentrations were analyzed using a TOC-V^CPH analyzer (Shimadzu, Japan), with measurements standardized based on the calibration curves for acetanilide (C:N = 8). The measured values of 44 μmol L⁻¹ for DOC (n = 6) and 32 μmol L⁻¹ for TDN (n = 6) were within 5% of the certified values for the deep-sea reference (DSR; University of Miami). The concentration of DON was determined by subtracting the DIN concentration from the TDN concentration, where DIN is the combined concentration of NO₃ ⁻, NO₂ ⁻, and NH₄ ⁺.

Three-dimensional (3D) fluorescence spectroscopy was conducted using a spectrofluorometer (FluoroMate FS-2; SCINCO, Korea). The excitation (Ex) and emission (Em) matrix (EEM) spectra were scanned within a wavelength range of 250–500 nm at 2 nm intervals and 250–600 nm at 5 nm intervals, respectively. Raman scattering by water was accounted for by subtracting daily fresh distilled water signals from the sample EEM data. The fluorescence intensity, measured in counts per second (cps), was normalized using quinine sulfate standards (i.e., the fluorescence spectrum of a quinine sulfate standard solution in 0.1 N H₂SO₄ at an Ex/Em of 350/450 nm) and expressed as parts per billion of quinine sulfate equivalent (ppb QSE). The EEM for all data with smoothing was obtained using MATLAB with Savitzky–Golay filters. The parallel factor analysis (PARAFAC) model was applied to the 3D EEM data and validated using split-half and core consistency analyses (Stedmon and Bro, 2008).

The fluorescence index (FI) was determined as the ratio of the fluorescence intensity at an Em wavelength of 450 nm to that at an Em wavelength of 500 nm when the excitation wavelength was 370 nm (McKnight et al., 2001). The autochthonous contribution was used as the basis of the biological index (BIX), which was measured as the ratio of the fluorescence intensity at an Em wavelength of 380 nm to that at an Em wavelength of 430 nm when the excitation wavelength was 310 nm (Huguet et al., 2009). The humification index (HIX) was calculated as the ratio of the average fluorescence intensity at an Em wavelength of 300–345 nm to that at an Em wavelength of 435–480 nm when the excitation wavelength was 255 nm (Zhang et al., 2010).

Inorganic nutrients were analyzed using a nutrient autoanalyzer (New QuAAtro39; SEAL Analytical, UK). Artificial seawater (salinity: 35) was used as the sample matrix for the blank and standard. According to the certified reference material (MOOS-1 from National Research Council, Canada), the analytical uncertainty was within 2% for DIN, Si(OH)₄, and PO₄ ³⁻.

The YSCW, KIW, KSW, and CDW were defined based on the uniformity of their physical properties and previous research (Pang and Hyun, 1998; Lee et al., 2016; Kang and Moon, 2022). Table S1 presents the representative characteristics of the four water masses. To implement a four-component end-member mixing model based on mixing ratios, it is important to establish representative values for the source water masses (Pang and Hyun, 1998; Kang and Moon, 2022). For this purpose, observational data from the YS, East China Sea, and Korean Strait surrounding Jeju Island from 2017 provided by the Korea Oceanographic Data Center were used to construct temperature–salinity (T-S) plots ( Figure 2 ).

In the T-S diagram, the fractions A, B, C, and D are the representative points of each source water mass when P is a mixture of four source water masses ( Figure 2C ). The mixing ratios of these water masses were denoted as ƒ_a, ƒ_b, ƒ_c, and ƒ_d, respectively, where ƒ_a + ƒ_b + ƒ_c + ƒ_d = 100%. When lines AB and DC have the ratio m:n, and lines AD and BC have the ratio k:l (where m + n = 1 and k + l = 1), point P satisfies the equation l(nƒ_a + mƒ_b) + k(mƒ_c + nƒ_d) = 1. The values of ƒ_a, ƒ_b, ƒ_c, and ƒ_d are thus l·n, l·m, k·m, and k·n, respectively (Pang and Hyun, 1998; Kang and Moon, 2022).

All statistical analyses, including principal component analysis (PCA), were performed using SPSS ver. 19 (IBM Corp., Armonk, NY, USA). Before the PCA, the variables were standardized by transforming them to have a zero mean and unit variance. This method is useful when the variables under analysis have different units of measurement or ranges of values.

The concentration of DOC in the coastal waters off Jeju Island ranged from 58 to 124 µM, with a significantly higher concentration observed during summer (mean: 82 ± 15 µM) than during the other seasons (winter: 63 ± 6.2 µM; spring: 71 ± 3.7 µM; and autumn: 69 ± 6.5 µM; p < 0.05) ( Table 1 ; Figure S1A ). These DOC concentrations were similar to those found in the East China Sea (60–120 µM), the coastal waters of Southeast Asia (50–110 µM), and major oceans such as the Pacific, Indian, and North Atlantic Oceans (60–80 µM), but slightly lower than those in China coastal waters (85–120 µM) (Carlson and Ducklow, 1995; Hansell and Carlson, 1998; Doval and Hansell, 2000; Hung et al., 2003; Kim et al., 2020; Sanwlani et al., 2022). Similarly, the concentration of DON also exhibited seasonal variation, with an overall range of 1.3 to 17.0 µM. The DON concentration during summer was significantly higher (mean: 6.8 ± 2.0 µM) than that during the other seasons (winter: 5.5 ± 1.5 µM, spring: 5.2 ± 2.8 µM, and autumn: 4.0 ± 0.8 µM; p < 0.05) ( Table 1 ; Figure S1B ). The DON concentration in the coastal waters off Jeju Island was comparable to the East China Sea (1.9–11 µM) and the East/Japan Sea (4–7 µM) and was slightly lower than that reported for the East China Sea in 2003 (6–9 µM) (Hung et al., 2003; Kim and Kim, 2013; Kim et al., 2020).

The concentrations of DOC and DON were comparable between all sampling stations during the individual seasons ( Figure S2 ). When comparing the difference in the concentration of DOC and DON between the surface and water column, in summer, the DOC and DON concentrations were significantly higher in surface water than those in the water column (p < 0.05; Figure S2B ). The average concentrations of TDN, DIN, PO₄ ³⁻, and Si(OH)₄ were 3.2 ± 0.9, 3.7 ± 1.9, 0.8 ± 0.7, 0.6 ± 0.2, and 7.8 ± 2.4 µM, respectively ( Table 1 ). In summer, the concentration of TDN (11 ± 1.9 µM) and PO₄ ³⁻ (0.8 ± 0.0 µM) was higher than during the other seasons. In contrast, the concentrations of DIN (4.5 ± 1.6 µM) and Si(OH)₄ (8.8 ± 2.2 µM) were the highest in winter ( Table 1 ; Figure S2 ). The mean concentration of chl-a was found to be low at 0.4 ± 0.1, 1.2 ± 1.1, 1.0 ± 0.8, and 0.8 ± 0.3 mg/m³ in winter, spring, summer, and autumn, respectively.

The variation in the three fluorescence indices according to the season is presented in Figure 3 . The FI ranged from 0.8 to 7.0, with a mean of 2.6 ± 1.1 over all seasons (winter: 2.7 ± 1.1; spring: 2.1 ± 0.7; summer: 2.2 ± 0.4; and autumn: 3.2 ± 1.3). An HIX value less than 1.5 and a BIX value larger than 1 indicate a biological or aquatic bacterial origin (Huguet et al., 2009; Zhang et al., 2010). In the present study, the HIX ranged from 0.2 to 1.4 (average: 0.6 ± 0.2), and the BIX ranged from 0.9 to 3.1 (average: 1.5 ± 0.4) across all seasons.

To investigate the underlying cause of the high DOC and DON concentrations, the mixing ratio of the water masses was analyzed. As previously documented, Jeju Island is influenced by the YSCW, KIW, KSW, and CDW (Pang and Hyun, 1998; Lee et al., 2016). The results of the four-component end-member mixing model revealed that the KIW and KSW were the primary contributors to the DOM levels, accounting for 59 ± 20 and 28 ± 13%, respectively, across all seasons, while the contributions of the YSCW and CDW were only 7.8 ± 2.2% and 5.2 ± 5.1%, respectively ( Table 2 ). However, the contributions of these water masses varied between the seasons. Figure 4 presents the PCA results for the ratio of the four water masses based on seasonal differences. The normalized values of the ratio of the four water masses in 103 of the 109 samples (i.e., those without missing temperature or salinity data) were used as input data for the PCA. The first two principal components (PC1 and PC2) accounted for 73% and 26% of the variance, respectively. The seasonal distribution fraction within the score plot could be explained by the distribution of the water mass in the loading plot ( Figure 4B ). According to the score plot, winter and spring samples exhibited a higher clustering within the KIW fraction (79% and 69%, respectively) of the loading plot. In contrast, the samples collected during summer were more strongly associated with the CDW and KSW fractions, suggesting that the CDW (12 ± 4.8%) and KSW (47 ± 16%) had a stronger impact during summer than during other seasons. The autumn samples were mostly clustered in the YSCW fraction of the loading plot, indicating that the YSCW (10 ± 4.3%) had a stronger impact during autumn than during other seasons. However, the samples were not clearly separated in the score plots by station ( Figure 4C ).

The concentrations of DOC and DON were comparable at all sampling stations during each season ( Figure S2A ), while the stations were not clearly separated in the PCA ( Figure 4C ), suggesting that the seawater was well mixed in this region. There was also no statistical difference (p < 0.05) in the concentration of DOM between the surface and water column samples, except during summer ( Figure S2B ). As illustrated in Figure 4B , the seasonal distribution was explained by the distribution of the water masses; however, the samples from the surface and water column were not clearly separated in the score plot except during summer, when the samples collected from surface water were orientated toward the CDW and KSW, while those from the water column were more strongly associated with the YSCW.

In this study, the four surrounding water masses (the YSCW, KIW, KSW, and CDW) were considered as sources of DOM in the coastal waters off Jeju Island. The contribution of each of these water masses can vary depending on the season, weather patterns, and ocean currents (Oh et al., 2013). Water mass analysis revealed that the KIW and KSW were the primary contributors to DOM across all seasons. These results are similar to those of a previous study that reported, based on salinity and ²²⁸Ra diagrams, that Kuroshio water was the highest contributor (up to 65%) to DOM in the vicinity of Jeju Island (Lee et al., 2014). However, the PCA of the ratio of the four water masses by season showed that the KIW had a greater impact on the coastal waters off Jeju Island during winter (79%) and spring (69%) than during the other seasons in 2017. In summer, the CDW (12%) and KSW (47%) had a stronger impact during summer than during the other seasons, while YSCW (10%) had a stronger impact during autumn. Ignoring the contribution of DOM production from biological processes due to the weak correlation between chl-a concentrations and DOC (r ^2 = 0.051, p > 0.01) and DON (r ^2 = 0.0006, p > 0.01) concentrations in this study area ( Figure S3 ), it was found that during winter, 49 ± 4.9 µM of DOC and 4.3 ± 1.2 µM of DON originated from the KIW. For spring, contributions were 49 ± 2.5 µM of DOC and 3.6 ± 1.9 µM of DON from the KIW. During summer, the contribution of the CDW resulted in DOC and DON concentrations of 10 ± 1.9 and 0.9 ± 0.3 µM, respectively, whereas the contribution of KSW resulted in DOC and DON concentrations of 38 ± 7.2 and 3.2 ± 0.9 µM, respectively.

Based on the results of the present study, it appears that the DOM distribution in the coastal waters off Jeju Island approximately 5 km from the shore is primarily influenced by the surrounding water masses rather than other sources such as SGD or fish farm and WWTP wastewater. While SGD is recognized as an important source of nutrients and trace metals around Jeju Island (Jeong et al., 2012; Cho et al., 2019), the DOC concentration in SGD is generally lower (26 ± 11 μM) than that in seawater, indicating a negligible contribution to the overall DOC inventory in seawater (Kim and Kim, 2017). Additionally, the TDN concentration (0.13 ± 0.03 mg/L) in the study area, which can act as a marker of aquaculture activity, was lower than that recorded from drainage samples (0.74 ± 0.46 mg/L) directly obtained at the discharge outlet in the aquafarm region and in the coastal water near the aquafarm outlet (0.25 ± 0.26 mg/L). However, it was similar to the TDN concentration observed at a control site (0.12 ± 0.05 mg/L) located more than 5 km away from the aquafarms (Kim et al., 2022).

Furthermore, in contrast to the findings reported by Oh et al. (2021) for areas influenced by fresh/saline groundwater and land-based fish farm wastewater within 0.5 km of the coast, we observed that the DIN concentration did not follow the expected pattern of a decrease with an increase in salinity ( Table S2 ), exhibiting an increase instead (Spearman’s r = 0.290, p < 0.01). This suggests that the study area was not directly influenced by SGD and fish farm effluent. Moreover, the ratio of reduced dissolved inorganic nitrogen (RDIN: NO₂ ⁻ and NH₄ ⁺) to total dissolved inorganic nitrogen (TDIN: NO₃ ⁻, NO₂ ⁻, and NH₄ ⁺) in the coastal water off Jeju Island (RDIN/TDIN: 0.4 ± 0.3) was lower than the ratio observed at eight WWTP effluent sites (0.9 ± 0.1) (Lee et al., 2022). The higher RDIN/TDIN ratio indicates that there is a larger proportion of nitrogen in its reduced state (RDIN) that has not undergone complete oxidation in the presence of WWTP effluent. However, certain specific patterns or variations in the distribution of DOM in the coastal seawater are not fully accounted for or explained by the conducted water mass analysis. These unexplained observations could stem from complex interplays of hydrodynamic or biochemical processes that extend beyond the scope of water mass analysis alone.

The relationship between DOC and DON concentrations and salinity during the four seasons in 2017 is summarized in Figure 5 . The DOC and DON concentrations exhibited a significant and weak negative correlation with salinity only during summer (r ^2 = 0.868, p < 0.01 and r ^2 = 0.416, p < 0.01, respectively), suggesting that the DOC and DON concentrations during the summer are primarily influenced by the CDW (Moon et al., 2009; Kim et al., 2020). Additionally, the outcomes of the water mass analysis in Section 4.1 corroborate that the concentrations of DOC and DON are impacted by the CDW. However, a DOC and DON pulse was observed in the salinity range of 29.8–33.6 ( Figure 5 ), and the extrapolated DOC and DON values (522 µM and 49 µM, respectively) at zero salinity were approximately three times higher than that previously reported at the Changjiang River mouth (DOC: 156 µM; DON: 14.5 µM) at zero salinity (Hung et al., 2003; Zhao and Gao, 2019). This suggests that the DOC and DON derived from the CDW did not undergo conservative mixing in this study area, implying that some mechanism led to the creation or alteration of DOC and DON rather than their simple preservation and transfer to Jeju waters. In general, DOM is mainly associated with phytoplankton production in the ocean (Duursma, 1961; Libby and Wheeler, 1997). However, the chl-a concentration was low in the present study (1.0 ± 0.8 µg/L), and there was a weak relationship between chl-a and DOC (r ^2 = 0.002, p > 0.01) and between chl-a and DON (r ^2 = 0.008, p > 0.01) ( Figure 6 ). In a previous study, a pulse of DOC was observed within a salinity range of approximately 24–35 in the East China Sea and YS adjacent to Jeju Island (Kim et al., 2020). It was suggested that the increase in DOC was related to microbial metabolism, indicated by the high percentage of particulate organic carbon in the suspended particulate matter of the estuarine mixing zone. Similarly, due to the activity of bacterial hydrolytic ectoenzymes, a portion of the settled organic matter may be continuously released in the form of DOM (Smith et al., 1992; Glibert, 1993).

The FI, HIX, and BIX results supported the conclusion that microbial activity was the source of the DOC and DON pulse. The FI is commonly used to determine the origin of FDOM, with values below 1.4 suggesting a terrestrial source, values ranging from 1.4 to 1.9 indicating terrestrially and microbially derived fulvic acids, and values above 1.9 suggesting a microbial origin (McKnight et al., 2001; Zhang et al., 2010). HIX values less than 1.5 and BIX values larger than 1 indicate a biological or aquatic bacterial origin (Huguet et al., 2009; Zhang et al., 2010). In the present study, the FI, HIX and BIX values all fell within the ranges associated with microbial activity across all seasons ( Figure 3 ).

According to Kim et al. (2020), particulate organic matter can be transformed into DOM through heterotrophic microbial activity, which is then converted into inorganic nutrients. During the summer season, this process was confirmed using PCA, with PC1 and PC2 accounting for 53% and 22% of the total variance, respectively, and using Spearman’s analysis ( Figure 7 ; Table S3 ). PC1 exhibited positive loadings for DOC and DON and negative loadings for salinity and DIN. The DOC concentration had a significant negative correlation with both salinity and DIN (Spearman’s r = -0.957 and -0.785, p < 0.01, respectively), indicating an inverse relationship. Similarly, the DON concentration had a significant negative correlation with both salinity and DIN (Spearman’s r = -0.540 and -0.716, p < 0.01, respectively). In the study region, the DON concentration was observed to decrease depending on the salinity range (29.8–33.6) ( Figure 8 ). An extrapolated value of approximately 5.1 µM was obtained using linear regression analysis (y = -1.34x + 49.2, r ^2 = 0.416). In contrast, the DIN concentration increased with a higher salinity within the range of 29.8–33.6. An extrapolated value of approximately 6.91 µM was derived using linear regression analysis (y = 1.84x + 54.5, r ^2 = 0.520). The DIN concentration can be influenced by the infiltration of land-based nitrogen fertilizers and their subsequent transport into the coastal ocean through SGD (Cho et al., 2019). However, when considering the relationship between DIN and salinity, the DIN concentration appeared to be influenced by sources other than SGD. The opposite patterns observed for the DON and DIN concentrations suggest that DON was converted into DIN through microbial activity. DON is the main source of reactive nitrogen in the surface ocean, with DON degradation leading to the release of inorganic nutrients when DOC is consumed by bacteria (Letscher et al., 2013).

These results could be explained by the observed organic matter cycle and the long journey of the CDW from the Changjiang River mouth to Jeju Island (~450 km). In summer, the CDW reaches the vicinity of Jeju Island within 20–35 days (Lee et al., 2014). During this period, the DIN concentration in the East China Sea rapidly decreases, particularly 100–200 km from the Changjiang River mouth (Kwon et al., 2018). This depletion is associated with the rapid DIN consumption by phytoplankton during long-range transportation. During transportation to Jeju Island, primary production is reduced due to the depletion of inorganic nutrients, and particulate organic matter derived from primary production can be produced. This particulate organic matter can be converted to DOM through microbial activity (Davis and Benner, 2005; Kaiser and Benner, 2009; Ji et al., 2023) and then to inorganic nutrients in the vicinity of Jeju Island. During the experimental period of 10 to 15 days, the average DON concentration decreased by 40-70%, and this is explained by increased microbial biomass and DIN concentration (Seitzinger and Sanders, 1997). The observed organic matter cycle, in which particulate organic matter is transformed into DOM and then converted into DIN through the influence of water masses in the study area, is a likely contributor as a source of DIN in oligotrophic regions such as Jeju Island.