CDOM spectral slope (S275–295) as tracers of water masses, CDOM heterogeneity, and Δ14C‐DOC in an oligotrophic marginal sea

The absorption spectral slope, S275–295, is an optical metric frequently employed for characterizing chromophoric dissolved organic matter (CDOM). We collected CDOM absorption (aCDOM) and fluorescence spectra from the oligotrophic offshore South China Sea to identify the major determinant of S275–295 and to explore the potential of S275–295 as physical and biogeochemical tracers. S275–295 linearly decreased with increasing humic‐like fluorescent DOM, revealing the latter as the primary control on S275–295. The variability of S275–295 consistently declined from the surface to deep water, making S275–295 a potential indicator of CDOM molecular heterogeneity. The S275–295 vs. aCDOM plot was found to be a useful tool for characterizing water masses. The 14C content of dissolved organic carbon (Δ14C‐DOC) shows a strong correlation to S275–295, demonstrating S275–295 to be a proxy of the 14C‐age of DOM. A provisional equation, ln(−Δ14C‐DOC) = (−0.85 ± 0.06) × ln(S275–295) + (8.78 ± 0.19), was proposed to estimate Δ14C‐DOC from S275–295 for global oligotrophic oceans.

Ubiquitous in the ocean, chromophoric dissolved organic matter (CDOM) is the fraction of dissolved organic matter (DOM) that absorbs solar ultraviolet (UV) and visible radiation (Coble 2007).A portion of CDOM, recognized as fluorescent DOM (FDOM), emits fluorescence upon absorption of light (Coble et al. 2014).FDOM is mainly composed of humic-like (FDOM H ) and protein-like (FDOM P ) materials based on their molecular structures and excitation-emission characteristics (Coble 1996).While CDOM and FDOM in river-impacted coastal waters are primarily of terrestrial origin (Raymond and Spencer 2015), CDOM and FDOM in open oceans are mainly produced in situ by microbial transformation of organic matter (Nelson and Siegel 2013;Jørgensen et al. 2014) or have mixed marine and terrestrial sources (Andrew et al. 2013;Cartisano et al. 2018).In the euphotic zone, phytoplankton exudation and zooplankton grazing also produce marine CDOM and FDOM (Carlson and Hansell 2015).
The absorption of solar UV and visible radiation by CDOM profoundly impacts the optics, microbial ecology, and primary production in the upper ocean (Arrigo and Brown 1996;Morel et al. 2010).It also leads to multifaceted photoreactions having far-reaching biogeochemical implications (Mopper et al. 2015).The resulting loss of chromophores (photobleaching) is the main sink of CDOM and FDOM H in the sunlit surface oceans (Coble et al. 1998;Stedmon and Markager 2005).Moreover, the absorption and fluorescence properties of CDOM and FDOM and their derived metrics have been widely employed as tracers and/or proxies for various purposes.For example, the absorption coefficients of CDOM (a CDOM ) and fluorescence intensities of FDOM have been frequently used to study the mixing behavior of DOM across the land-ocean continuum (e.g., De Souza Sierra et al. 1997;Matsuoka et al. 2012;Guo et al. 2014) and as proxies of the concentration of dissolved organic carbon (DOC) in estuarine and coastal waters (Fichot and Benner 2011;Osburn et al. 2015;Li et al. 2019a).The absorption coefficient of CDOM has also been employed for tracing meridional circulations in major ocean basins (Nelson et al. 2007(Nelson et al. , 2010;;Nelson and Siegel 2013) and for diagnosing water mass compositions in deep fjords (Xie et al. 2012).Fluorescence signatures of FDOM may serve as fingerprints of upwelling systems (Coble et al. 1998) and tracers of different biological sources of DOM (Quinlan et al. 2018).Plots of absorption spectral slopes against a CDOM have been adopted to distinguish terrigenous and autochthonous CDOM (Stedmon and Markager 2001;Stedmon et al. 2011) and characterize water mass mixing (D'Sa et al. 2014) in northern seas.In addition, the spectral slope of CDOM between 275 and 295 nm (S 275-295 ) has been demonstrated to be an indicator of the light exposure history, origin, and molecular weight of CDOM (Helms et al. 2008), and a tracer of terrigenous DOC in estuarine and coastal waters (Fichot and Benner 2012).
Here, we report a CDOM and FDOM dataset collected from the oligotrophic Northern South China Sea (NSCS) basin.We identified the major component of CDOM that controls S 275-295 and expanded the proxy role of S 275-295 for tracing water masses, molecular heterogeneity of CDOM, and the radiocarbon age of DOC.

Study site
The South China Sea (SCS) is a large semienclosed tropical marginal sea (area: $3.5 Â 10 6 km 2 ) in the western Pacific (Supporting Information Fig. S1), characterized by a three-layer circulation system: a cyclonic circulation in both the upper (< 750 m) and deep (> 1500 m) layers and an anticyclonic circulation in the intermediate layer (750-1500 m) (Gan et al. 2016).The Luzon Strait, with a maximal sill depth of 2400 m, is the only deep channel that connects the SCS with the western Pacific.Water flows into the SCS from the western Pacific in the upper and deep layers and exits from the SCS to the western Pacific in the intermediate layer via the Luzon Strait (Tian et al. 2006).While the shelf areas of the SCS are significantly influenced by riverine inputs, its slope and deep basin are of oligotrophic nature (Liu et al. 2002;Wu et al. 2003;Zhang et al. 2016;Du et al. 2017).Chlorophyll a (Chl a) concentrations in slope and basin waters range from 0.1 to 1.1 μg L À1 (Zhang et al. 2016;Xing et al. 2019) and show typical subsurface maxima at the depths of 40-100 m (Liu et al. 2002;Zhang et al. 2016).
Past studies on DOM in the NSCS were mainly focused on the distributions of DOC and its exchange with the western Pacific through the Luzon Strait (e.g., Dai et al. 2009;Wu et al. 2015;Li et al. 2021).Wang et al. (2017) reported the only study on CDOM and FDOM in the NSCS investigating the distributions of CDOM and FDOM in the surface and intermediate waters (< 1500 m), the influences of Kuroshio water intrusion and mesoscale eddies on CDOM and FDOM in the upper layer, and the export of CDOM and FDOM from the NSCS to the western Pacific.

Sample collection
Water sampling was conducted aboard the R.V. Dong-Fang-Hong 2 from May 20, 2016 to June 3, 2016 and from July 13, 2017 to August 8, 2017 in the deep basin of the NSCS; one station in the adjacent Kuroshio Current in the western Pacific (Sta.F2) was also sampled during the 2017 campaign (Supporting Information Fig. S1 and Table S1).Samples were collected at 8-19 depths from surface to bottom (depth range: 1609-4284 m) using acid-cleaned 12-L Niskin bottles mounted on a conductivity-temperature-depth (CTD) rosette.Upon collection, samples were filtered through 0.2-μm polyethersulfone filters (Pall Life Sciences) under low vacuum.The filtrates were transferred into acid-cleaned and precombusted clear-glass bottles with Teflon-lined screw caps for CDOM absorption and fluorescence spectra determinations.All samples were stored at 4 C in the dark until analyzed in a landbased laboratory.Seawater temperature and salinity were measured with a SeaBird 911 plus CTD profiler.

Sample analysis
The procedures for CDOM absorption and fluorescence spectra were previously reported by Yang et al. (2020).Briefly, samples were warmed to room temperature in the dark before analysis.CDOM absorbance spectra were scanned between 800 and 200 nm on a dual-beam UV-visible spectrophotometer (UV-2550, Shimadzu) fitted with 10-cm quartz cuvettes.A baseline correction was made according to Babin et al. (2003).The Napierian absorbance coefficient at wavelength λ (nm), a CDOM (λ) (m À1 ), was calculated as 2.303 times the measured absorbance at wavelength λ divided by the pathlength of the cuvette in meters.S 275-295 was derived using nonlinear fitting adapted from the method of Helms et al. (2008).
Fluorescence excitation-emission matrices (EEMs) were acquired using a Hitachi F-4600 fluorescence spectrophotometer fitted with a 1-cm quartz cell.PARAFAC analysis of the obtained EEMs was performed to characterize the FDOM (Stedmon and Bro 2008).The detailed procedures for PARAFAC modeling were reported by Li et al. (2019a) and Yang et al. (2020).
The lower detection limit of a CDOM , defined as three times the standard deviation of five MilliQ water blank measurements, was 0.015 m À1 at wavelengths of 250-400 nm.The mean relative standard deviations (RSD) of the measurements of 10 sets of duplicate samples were < 4% for a CDOM (330), 3% for S 275-295 , 2.5% for humic-like FDOM components, and 6% for protein-like FDOM components.

Vertical distributions of CDOM and FDOM
Two humic-like and three protein-like FDOM components were identified based on PARAFAC modeling.Detailed information on the five components can be found in Yang et al. (2020).To simplify the discussion, C H and C P were used to represent the sums of the humic-like and protein-like components, respectively.
Vertical distributions of the CDOM and FDOM variables at individual stations in the NSCS were similar and thus averaged (Fig. 1).The absorption coefficient at 330 nm and C H were depleted within the upper 25 m likely due to photobleaching (Fig. 1a,b) (Nelson and Siegel 2013).Conspicuous   330) and C H (Fig. 1d).Below 1000 m, a CDOM (330), C H , and S 275-295 kept relatively constant (Fig. 1a,b,  d).C P varied little throughout the water column but the data at individual depths were noisier (Fig. 1c).
The overall vertical distribution patterns of the CDOM and FDOM variables at Sta. F2 in the Kuroshio Current were similar to those in the NSCS.Notably, at depth > 1500 m, all four variables at Sta. F2 were comparable to those in the NSCS, consistent with the Pacific origin of the DW in the SCS.The rather homogeneous vertical distributions of CDOM and FDOM in the deep NSCS suggested that strong vertical mixing (Qu et al. 2006) played an important role in re-distributing CDOM and FDOM after Pacific water entered the NSCS through the Luzon Strait at $2400 m depth (Wang et al. 2011).

Discussion
Control on S 275-295 and heterogeneity of CDOM S 275-295 linearly decreased with increasing C H across different water masses in the water column (Fig. 2a).This inverse relationship could be largely ascribed to photobleaching of low-S 275-295 FDOM H in the sunlit surface ocean (Omori et al. 2020;Yang et al. 2020), which increases S 275-295 (Helms et al. 2008;Yamashita et al. 2013), and microbial production of low-S 275-295 FDOM H in the dark ocean (Yamashita and Tanoue 2008;Yang et al. 2020).As C H accounts for 85% of the S 275-295 variance (Fig. 2a), FDOM H appears to be the dominant control on S 275-295 .In contrast, no significant correlation exists between S 275-295 and C P (p > 0.05) (Supporting Information Fig. S4), implying little impact of FDOM P on S 275-295 .
The ranges and RSD of a CDOM (330) and S 275-295 were substantially smaller in the IW and DW than in the shallower water masses (Fig. 3; Table 1).As S 275-295 is an indicator of the molecular weight of CDOM (Helms et al. 2008;Martínez-Pérez et al. 2017;Cao et al. 2020), these CDOM vertical trends suggest that both the abundance and molecular composition of CDOM were more uniform in the IW and DW, which is consistent with rapid mixing within the interior of the NSCS (Qu et al. 2006).Notably, the RSDs of S 275-295 are 54% and 67% lower than those of a CDOM (330) in the IW and DW (Table 1), respectively.The molecular composition of CDOM is thus more homogenous than its abundance.It has been demonstrated that the ranges of the molecular weight and atomic ratios of O/C and H/C of DOM in the Pacific deep water are narrower than those in the Atlantic deep water, supporting the notion that microbial processing can reduce the molecular heterogeneity of DOM during global overturning circulation (Chen et al. 2014;Bercovici et al. 2018;Wang et al. 2022).We argue that the low CDOM heterogeneity in the NSCS DW inherits the signatures of Pacific deep water, which is the sole source of the DW in the SCS (Qu et al. 2006;Tian et al. 2006).Given the short basin-wide renewal time of the SCS DW (30-100 yr, Qu et al. 2006;Chang et al. 2010), which must be even shorter in the NSCS sector, the abatement of CDOM heterogeneity caused by local microbial processing is plausibly less consequential.Likewise, the effect of possible in situ production and local terrestrial input of CDOM into the DW, which tends to increase the CDOM heterogeneity, should be secondary as well.

S 275-295 : Tracer of water masses
No significant correlation between S 275-295 and a CDOM (330) exists for the composite data covering all water masses in the entire water column ( p > 0.05).However, the SW, SSW, and aggregate IW and DW are separated in the S 275-295 vs. a CDOM (330) plot, with the deeper water masses successively lying beneath the shallower ones and S 275-295 for each specific water mass decreasing with increasing a CDOM (330) (Fig. 2b).At a given a CDOM (330), S 275-295 decreases from the shallower to deeper water masses, further demonstrating that the relative content of the high-molecular-weight component in CDOM augments downward in the water column.Notably, the DW overlaps with the lower portion of the IW (Fig. 2b), suggesting that part of the CDOM in the IW originates from the DW, consistent with the strong vertical mixing in the NSCS (Qu et al. 2006).
The water mass-specific feature of the S 275-295 -a CDOM (330) relationships renders it a potentially powerful tool for identifying and characterizing water masses in the SCS, analogous to the function of the temperature-salinity plot (Supporting Information Fig. S2).This water mass-identifying functionality lies in the fact that the dominant physical and biogeochemical processes controlling the S 275-295 -a CDOM (330) relationship differ across water masses.S 275-295 decreases exponentially with increasing a CDOM (330) with a high degree of correlation in the SW (R 2 = 0.90, p < 0.0001) mainly due to the opposing effects of photobleaching on a CDOM (330) and S 275-295 .In contrast, the S 275-295 -a CDOM (330) relationship is insignificant in the IW and DW (R 2 = 0.11, p > 0.05; Fig. 2b), which may be explained by the much smaller variation of S 275-295 relative to that of a CDOM (330) in these two water masses (Fig. 3; Table 1).In the SSW, the competing influences of downward mixing of the SW and upward mixing of the IW and DW lead to a significant but relatively weaker S 275-295a CDOM (330) relationship (R 2 = 0.49, p < 0.001; Fig. 2b).
S 275-295-a CDOM relationships have previously been used to characterize water masses at the horizontal dimension in the shallower eastern Bering Sea (< 150 m) (D'Sa et al. 2014).To our knowledge, the present study is the first successful application of this approach to an oligotrophic deep basin composed of various water masses at the vertical dimension.S2) for the NSCS basin, we tested the relationship between S 275-295 and Δ 14 C-DOC.Details of reorganizing the Δ 14 C-DOC and S 275-295 data can be found in Text S1.In the NSCS, Δ 14 C-DOC was strongly correlated with S 275-295 in the entire water column (Fig. 4a, Supporting Information Fig. S5).Δ 14 C-DOC values predicted from the fitted equation agree with the measured ones within 4.1 AE 2.1% (range: À6.6-8.8%;Fig. 4b), comparable to the relative analytical uncertainty (AE 1-3%) of the Δ 14 C-DOC measurement (Druffel et al. 2016;Shan et al. 2020).S 275-295 can thus serve as a proxy of the DOC radiocarbon age, offering a simple, fast, economical, and yet robust alternative to the Δ 14 C-DOC technique, which requires extensive sample pretreatment and complex and costly analytical instrumentation (Druffel et al. 1989;Beaupré et al. 2007;McNichol and Aluwihare 2007).
The strong correlation between Δ 14 C-DOC and S 275-295 suggests that the major physical and biogeochemical processes dictating the two variables are tightly coupled.DOC is enriched with 14 C in the surface ocean (Druffel et al. 1992) where 14 C-enriched modern atmospheric CO 2 enters and is  utilized for sunlight-driven photosynthesis of particulate organic carbon from which DOC is derived (Carlson and Hansell 2015).S 275-295 is higher in the surface ocean due to sunlight-initiated photobleaching, which also enriches 14 C in DOC (Beaupré and Druffel 2012).Therefore, the correspondence between Δ 14 C-DOC and S 275-295 in the surface ocean reflects the fact that both biological production of DOC and photobleaching of CDOM are restricted within the euphotic zone and that both processes undergo similar temporal cycles (Siegel et al. 2002).It should be noted, though, that the penetration of the UV radiation into the water column, which mainly drives photobleaching, is substantially shallower than that of the visible radiation for photosynthesis.This partial mis-match of the light penetration depth may partly explain the nonlinearity of the relationship between Δ 14 C-DOC and S 275-295 (Fig. 4a).
Below the surface ocean, as the water mass ages with depth, DOC becomes increasingly depleted with 14 C due to radiocarbon decay and preferential microbial uptake of 14 Cenriched young DOC (Cherrier et al. 1999;Raymond and Bauer 2001a).In the meantime, microbial processes decrease S 275-295 during water mass aging in the dark ocean (Nelson et al. 2007;Catal a et al. 2015), leading to a positive correspondence between Δ 14 C-DOC and S 275-295 in the entire water column.The positive Δ 14 C-DOC-S 275-295 correlation is consistent with the inverse relationship between water mass ventilation age and CDOM spectral slope between 320 and 650 nm observed in the interior of the North Atlantic (Nelson et al. 2007).

Potential broad implications
Despite being a marginal sea, the SCS is of oligotrophic nature excluding the shelves (Liu et al. 2002), bearing similar biological characteristics to major oligotrophic oceans worldwide (Williams et al. 2013).Moreover, the DW in the SCS is sourced from the northwest Pacific and thus carries its physical and biogeochemical signatures (Tian et al. 2006;Dai et al. 2009;Wang et al. 2017).Past studies have shown that vertical profiles of a CDOM , FDOM, S 275-295 , DOC, and Δ 14 C-DOC collected from the SCS resemble those in the major ocean basins (Druffel et al. 2016(Druffel et al. , 2019;;Cartisano et al. 2018;Cao et al. 2020;Ding et al. 2020).Here, we take the Δ 14 C-DOC-S 275-295 relationship as an example to highlight the broad implications of the present study.Quasi-paired S 275-295 and Δ 14 C-DOC data were compiled for global oligotrophic oceans (Text S1); the dataset obtained, though limited in  number (n = 32), covers both surface and deep waters of the Atlantic and Pacific basins, including the well-studied BATS and HOT sites (Fig. 4c).The Δ 14 C-DOC vs. S 275-295 plot reveals a strong correlation between the two variables (p < 0.0001), with the regression line overlapping with that for the NSCS within the 95% confidential intervals (Fig. 4c).This high concordance not only confirms the broader applicability of the NSCS's results but also demonstrates the feasibility of developing a single equation for describing the Δ 14 C-DOC-S 275-295 relationship for global oligotrophic oceans.By aggregating the data from the NSCS and other ocean regions, we propose a tentative generalized oligotrophic ocean Δ 14 C-DOC-S 275-295 relationship, ln(ÀΔ 14 C-DOC) = (À0.85AE 0.06) Â ln(S 275-295 ) + (8.78 AE 0.19) (n = 45; R 2 = 0.83; p < 0.0001), which provides a practical and efficient tool to map large-scale Δ 14 C-DOC distributions from CDOM absorption spectra that were collected from various ocean basins (Nelson et al. 2010;Nelson and Siegel 2013) or will be collected in future surveys.Evidently, more research efforts, particularly paired Δ 14 C-DOC and S 275-295 measurements, are needed to expand the Δ 14 C-DOC and S 275-295 dataset, thereby improving the proposed equation and/or identifying more precise, region-specific Δ 14 C-DOC-S 275-295 relationships.
The tracer functionality of S 275-295 for Δ 14 C-DOC is not expected to hold for coastal waters impacted by terrestrial runoff.Riverine DOM is generally enriched with modern DOC (i.e., high Δ 14 C-DOC) (Raymond and Bauer 2001b;Mortazavi and Chanton 2004) but featured with lower spectral slopes compared to open-ocean surface DOM (Del Vecchio and Blough 2004;Nelson et al. 2010), leading to a S 275-295 -Δ 14 C-DOC relationship opposite to that for open oceans.

Fig. 1 .
Fig. 1.Average vertical profiles of a CDOM (330) (a), C H (b), C P (c), and S 275À295 (d) in the NSCS and at Sta. F2 in the western Pacific.Error bars signify one standard deviation.Insets in (a) and (b) are close-up views of the respective profiles at depths <200 m.

S
275-295 : Proxy of DOC radiocarbon age Using the Δ 14 C-DOC data reported by Ding et al. (2020) (their Supplementary Table

Fig. 3 .
Fig. 3. Box plots of a CDOM (330) (a) and S 275À295 (b) in different water masses.Lines in the boxes denote medians.The boxes extend from the lower to the upper quartile values of the data and the whiskers show the data range, with open gray circles indicating outliers.
(Nelson and Siegel 2013;Catal a et al. 2015)emical tracers peaks of a CDOM (330) and C H occurred within 50-100 m depths where subsurface Chl a maxima were located (Supporting Information Fig.S3), suggesting biological production of CDOM and FDOM H .The absorption coefficient at 330 nm and C H increased gradually from 200 to 1000 m, implying microbial production CDOM and FDOM H(Nelson and Siegel 2013;Catal a et al. 2015).S 275-295 decreased rapidly with depth in the upper 1000 m, with a minimum approximately mirroring the maxima of a CDOM (

Table 1 .
Mean values (AESD) and RSD of a CDOM (330), C H , C P , and S 275-295 for each water mass SW, surface water; SSW, subsurface water; IW, intermediate water; DW, deep water.