Spatial Variability of Air–Sea CO₂ Flux and Their Carbon Sources During Early Spring in the Yangtze River Estuary and Adjacent Coastal Areas

Abstract

Air–sea CO₂ flux (FCO₂) in the estuary–coastal continuum plays a vital role in global carbon sequestration; however, the mechanisms governing FCO₂ spatial heterogeneity during early spring remain poorly understood, particularly the roles of distinct dissolved inorganic carbon (DIC) sources. In March 2025, we investigated the FCO₂ spatial variability and DIC sources across the Yangtze River estuary and adjacent coastal areas using DIC concentration, pH, and δ¹³C_DIC analyses. The study area was a net CO₂ source (7.3 ± 8.7 mmol m⁻² d⁻¹), with the intensity declining progressively from the inner estuary to offshore areas. Physical mixing of three principal water masses established the following pattern: high-pCO₂ Changjiang Diluted Water and Yellow Sea Coastal Current drove CO₂ outgassing, while low-pCO₂ East China Sea Shelf Water weakened it. Quantitative apportionment revealed atmospheric CO₂ invasion as the dominant DIC source, followed by carbonate dissolution and organic matter degradation, with the latter declining from the inner estuary to offshore areas. The spatial variation in DIC source contributions further confirms that, superimposed on the physical mixing, biogeochemical processes—particularly biological activity—modulated reginal source intensities. This early-spring case captures a critical transitional window and highlights the necessity of integrating multi-factor regulation with DIC source partitioning to resolve carbon dynamics in the estuarine–coastal continuum.

1. Introduction

The estuary–coastal continuum serves as a critical transition zone characterized by intense land–sea interactions, and functions as the gateway through which terrestrial materials are transported into the ocean [1,2]. Annually, global terrestrial ecosystems uptake approximately 4.1 Pg C yr⁻¹ from the atmosphere, of which 0.95 Pg C yr⁻¹ is transported via rivers into the estuary–coastal continuum. Of this influx, approximately 0.40 Pg C yr⁻¹ is buried in sediment, while an equivalent amount is up taken by the air–sea interface, resulting in a net carbon transfer from the estuary–coastal continuum to the ocean of 0.95 Pg C yr⁻¹ [3]. Despite the small proportion of this region in the global ocean surface area, its carbon sink capacity accounts for 17% of the global ocean carbon sink (2.8 ± 0.4 Pg C yr⁻¹) [4]. The air–sea CO₂ flux (FCO₂) is primarily regulated by the CO₂ transfer velocity (k) and the surface partial pressure of CO₂ (pCO₂), which together determine the strength of its carbon source or sink capacity [2,5]. The sources of dissolved inorganic carbon (DIC) and its equilibrium processes within the estuary–coastal continuum govern pCO₂ in the water column, thereby influencing FCO₂ (Figure S1) [5,6]. The interactions between the CO₂ solubility pump, carbonate pump, and biological pump can further modulate the transformation processes of DIC, thereby driving the conversion between carbon sources and sinks in the estuary–coastal continuum [7].

Substantial uncertainties persist regarding the spatial distribution of FCO₂ in the estuary–coastal continuum, due to the spatial variability of physical, chemical, and biological factors resulting from hydrological processes and wind speed [2,5]. K, as a major factor influencing FCO₂, is primarily driven by turbulence at the aqueous boundary layer, which is generated by wind and increases with higher wind speeds [5]. pCO₂ is primarily regulated by the carbonate chemistry balance, which is influenced by multiple physical and biogeochemical processes, including the CO₂ solubility pump, carbonate pump, and biological pump [8,9]. Physical and chemical factors, such as the pH and temperature, can affect the carbonate buffer system via the CO₂ solubility pump [10,11] and influence the dissolution and precipitation of carbonate via the carbonate pump [12]. The biological pump affects the absorption and release of CO₂ through photosynthesis, respiration, and microbial degradation, which are affected by both physical and biological factors (e.g., chlorophyll a (Chla) and nutrient content) [13,14]. Riverine waters typically exhibit high pCO₂ and nutrient concentrations, and the mixing between riverine and sea waters can both directly influence the spatial distribution of pCO₂ and indirectly affect it through the biological pump [6,15]. Additionally, the spatial distributions of physical, chemical, and biological factors vary due to the vertical mixing driven by hydrological processes [15,16]. In particular, reduced river discharge, enhanced spring winds, and strong tidal forcing during spring can lead to significant vertical mixing in the estuary–coastal continuum, resulting in uncertain CO₂ source or sink dynamics [5].

Diverse DIC sources in the estuary–coastal continuum play a fundamental role in regulating the FCO₂ by modulating the carbonate chemistry balance, while their control mechanisms on FCO₂ remain poorly understood [17,18]. The principal DIC sources, comprising atmospheric CO₂ invasion, carbonate and silicate mineral weathering, and organic matter degradation (see conceptual framework in Figure S1) [17,19], elevate DIC concentrations and perturb the CO₂(g)–DIC equilibrium. This perturbation directly alters the water-column pCO₂, the key driver governing both the directionality (source vs. sink) and magnitude of FCO₂ [17,18]. Previous research has indicated that the Yangtze River estuary and adjacent coastal areas are CO₂ absorbers in spring, due to the biological uptake of CO₂ [5,20], but act as CO₂ sources in winter, due to the biological addition of DIC [15,21]. However, the relationships between specific DIC sources and the resultant FCO₂ remain unclear, impeding prediction of the impacts of climate change on FCO₂. Stable carbon isotope signatures (δ¹³C_DIC) provide a vital means of partitioning the contributions of these sources, thereby revealing how the origins of DIC driven by these pumps govern DIC dynamics and ultimately influence FCO₂ in the estuary–coastal continuum [22].

The Yangtze River estuary and adjacent coastal areas, as one of the world’s largest estuarine–shelf systems, are characterized by high biological production, complex circulation patterns, and intense land–sea interactions, exhibiting significant spatial differences in carbon sources and sinks [5]. The inner estuary typically acts as a carbon source of atmospheric CO₂, driven by elevated pCO₂ in inflowing freshwater, whereas the outer estuary and offshore areas exhibit marked seasonal uncertainty in their designation as either sinks or sources of atmospheric CO₂ [5,23]. Hydrological conditions in this region exhibit marked seasonal variability, which in turn modulates the spatial distribution of FCO₂ throughout the seasons [17,23]. Although air–sea CO₂ exchange in the Yangtze River estuary and adjacent coastal areas has been widely investigated, important uncertainties remain regarding (1) how the spatial pattern of FCO₂ is jointly shaped by water mass mixing and biogeochemical processes during early spring, and (2) how different DIC sources contribute to the observed source–sink mosaic across the estuary–coastal continuum [16,24]. Early spring represents a transitional hydrographic period characterized by strong water mass interactions; however, the corresponding carbon-source apportionment remains limited [12]. Therefore, this study is presented as an early-spring case study with strong regional significance for understanding seasonal carbon dynamics in this study region. Specifically, we investigated the DIC concentration, pH, and δ¹³C_DIC in surface and bottom waters and calculated FCO₂ across the Yangtze River estuary and adjacent coastal areas during the NORC2025-03 cruise. The specific objectives of this study were (1) to elucidate the synergistic controlling mechanisms governing the spatial variability of FCO₂ under the combined influences of meteorological, physical, chemical, and biological factors, with particular emphasis on the early-spring transitional period; and (2) to quantify the proportional contributions of major DIC sources and evaluate how these sources modulate the spatial patterns of FCO₂.

2. Materials and Methods

2.1. Characteristics of the Yangtze River Estuary and Adjacent Coastal Areas

The study region covers an area ranging from 29.5 to 32.5° N and from 121 to 124° E (Figure 1), located in the Yangtze River estuary and adjacent coastal areas. Following the delineation of Zhai et al. [25], the area between 121° E (west of Chongming Island) and 122° E, where the surface salinity is typically <10, is defined as the inner Changjiang estuary. This inner estuary is bifurcated by Chongming Island into two main tributaries: the South Branch and the North Branch. Extending from the South Branch to the East China Sea, the system can be further subdivided based on the surface salinity: the outer estuary, where the salinity ranges from 10 to 31, and the offshore area of the East China Sea, where the salinity exceeds 31 [25]. The hydrodynamic conditions in this region are complex and exhibit pronounced seasonal variations, influenced jointly by the East Asian monsoon and the western boundary currents of the Pacific Ocean. In the Yangtze River estuary and adjacent coastal areas, the hydrodynamic conditions are primarily modulated by the Changjiang Diluted Water (CDW) sourced from Yangtze River freshwater [26], the East China Sea Coastal Current (ECSCC), Taiwan Warm Current (TWC), Yellow Sea Coastal Current (YSCC), and the Yellow Sea Warm Current (YSWC) [14,26]. During early March with the cold/dry climate, the area is dominated by the CDW, ECSCC, TWC, and YSCC, driven by the northeast East Asian monsoon [27].

2.2. Sample Collection and In Situ Measurements

Sample collections and field observations were conducted from 5 to 11 March 2025 across the Yangtze River estuary and adjacent coastal areas during the open research cruise NORC2025-03, supported by the NSFC Shiptime Sharing Project (Figure 1). At each sampling station, vertical profiles of the water depth, temperature, salinity, and turbidity were obtained using a shared CTD instrument (SBE 911plus CTD, Sea-Bird Scientific, Bellevue, WA, USA). Water samples were collected from the surface (2 m depth) and bottom (2 m above the seafloor) layers using Niskin bottles mounted on the CTD frame. Samples for pH determination were collected in 100 mL narrow-neck glass bottles. Dissolved oxygen (DO) samples were collected in 250 mL brown glass-stoppered bottles following the Winkler method. Samples for DIC analysis were collected in 250 mL borosilicate glass vials and poisoned immediately with 100 μL of saturated mercuric chloride (HgCl₂) solution to halt the biological activity.

2.3. Laboratory Analysis

The concentrations of DIC were measured using a total organic carbon (TOC) analyzer (TOC-VCPH, Shimadzu Co., Kyoto, Japan). Samples were combusted at 680 °C in a combustion tube packed with a platinum catalyst. Certified reference materials (CRM) from Andrew G. Dickson’s lab at Scripps Institute of Oceanography were used for calibration and accuracy assessment, yielding a precision of ±2 µmol kg⁻¹. δ¹³C_DIC values were analyzed using a Finnigan MAT-252 mass spectrometer (Thermo Fisher Scientific, Darmstadt, Hesse, Germany), with a precision of 0.15‰. The pH was measured immediately onboard on an NBS scale using a FE28 pH meter (Mettler Toledo, Greifensee, Switzerland), with the accuracy monitored against seawater CRMs (also from A.G. Dickson’s laboratory, Scripps Institution of Oceanography). The precision of the pH measurements was ±0.01 units. For DO analysis, samples were fixed immediately after collection and titrated onboard following the classic Winkler titration procedure.

Concentrations of the major cations (Ca²⁺ and Mg²⁺) were measured via inductively coupled plasma optical emission spectrometry (ICP-OES, Optima 5300DV, PerkinElmer, Waltham, MA, USA), with an analytical precision of ±2%. For oxygen isotope (δ¹⁸O-H₂O) analysis, 2 mL of filtered seawater was analyzed using a liquid water isotope analyzer (LWIA, Model DLT-100; Los Gatos Research Inc., San Jose, CA, USA). Calibration procedures were traceable to the primary isotopic reference material, Vienna Standard Mean Ocean Water (VSMOW), yielding an average analytical precision of ±0.20‰. For nutrient analysis (NO₂⁻, NO₃⁻, NH₄⁺, PO₄³⁻ and SiO₃²⁻), water samples were filtered through 0.45 μm cellulose acetate membranes, stored frozen, and analyzed in the laboratory using a Seal QUAATRO 39 automated nutrient analyzer (SEAL Analytical, Southampton, UK). For chlorophyll a (Chla) determination, a 300 mL water sample was filtered under low vacuum pressure through a Whatman GF/F glass fiber filter. The filter was extracted in a 9:1 (v/v) acetone: water solution for >12 h, followed by centrifugation at 4500 rpm. The supernatant was measured using a Hitachi F-7100 fluorescence spectrophotometer (Hitachi, Tokyo, Japan). Calibration was performed using a pure Chla standard (99%, Sigma-Aldrich, St. Louis, MO, USA).

2.4. Calculations of Air–Sea CO₂ Flux

FCO₂ (mmol CO₂ m⁻² d⁻¹) was calculated using the following equation [5]:

$$\begin{gathered} F_{\mathrm{C}\mathrm{O}2}=k(\mathrm{c}\mathrm{m}{\mathrm{h}}^{-1})\times k_{\mathrm{H}}(\mathrm{m}\mathrm{o}\mathrm{l}{\mathrm{L}}^{-1}{\mathrm{a}\mathrm{t}\mathrm{m}}^{-1})\times {\mathsf{\Delta }p\mathrm{C}\mathrm{O}}_2(\mathsf{\mu }\mathrm{a}\mathrm{t}\mathrm{m}) \\ \times \left({\displaystyle \frac{1\mathrm{L}}{{1000\mathrm{c}\mathrm{m}}^3}}\times {\displaystyle \frac{24\mathrm{h}}{1\mathrm{d}\mathrm{a}\mathrm{y}}}\times {\displaystyle \frac{{1000\mathrm{c}\mathrm{m}}^2}{1{\mathrm{m}}^2}}\times {\displaystyle \frac{1\mathrm{m}\mathrm{m}\mathrm{o}\mathrm{l}\mathrm{e}}{1000\mathsf{\mu }\mathrm{m}\mathrm{o}\mathrm{l}\mathrm{e}}}\right) \end{gathered}$$

(1)

where ΔpCO₂ (μatm) represents the difference in pCO₂ between the surface water and the atmosphere. The atmospheric CO₂ concentration was obtained from the Tae-ahn Peninsula observation site (36.7376° N, 126.1328° E; Republic of Korea) (available at https://gml.noaa.gov/data/data.php, accessed on 20 December 2025), after correction for water vapor pressure to 100% humidity with in situ temperature and salinity data [5]. pCO₂ and TA (total alkalinity) were calculated from the DIC (μmol kg⁻¹) and pH based on the NBS scale using the Excel version of CO2SYS_v3.0_Err [29], with the carbonic acid dissociation constants K₁ and K₂ described by Lueker et al. [30], K_HSO4 described by Dickson [31], and total boron described by Lee et al. [32], according to the recommendations from previous studies [13,33]. Based on the uncertainties associated with the DIC and pH measurements, the estimated uncertainty in the calculated pCO₂ was approximately ±14 μatm. The solubility of CO₂ (k_H, mol L⁻¹ atm⁻¹) at a specific temperature (T_k, K) and salinity (S, PSU) was calculated using the following equation [34]:

$$\begin{gathered} {\mathrm{l}\mathrm{n}k}_{\mathrm{H}}\left({\mathrm{C}\mathrm{O}}_2\right)=-58.0931+90.5069\times \left({\displaystyle \frac{100}{T_{\mathrm{k}}}}\right)+22.294\times \mathrm{l}\mathrm{n}\left({\displaystyle \frac{T_{\mathrm{k}}}{100}}\right) \\ +S\times (0.027766-0.025888\times (T_{\mathrm{k}}/100) \\ +0.0050578\times {\left({\displaystyle \frac{T_{\mathrm{k}}}{100}}\right)}^2) \end{gathered}$$

(2)

The gas transfer velocity (k, cm h⁻¹) mainly depends upon the wind stress and associated near-surface turbulence, ocean waves, bubble injection, surface–surfactant conditions, and the temperature and near-surface humidity [35]. The wind variability is a key driver of changes in many of these factors (e.g., turbulence and bubble injection) and is more widely measured than other variables; thus, k is typically parameterized as a function of the near-surface wind speed [35,36]. While the wind regime is the dominant factor influencing k in the open ocean, the situation is more complex in coastal and shallower estuaries, due to the effects of other factors such as the tidal currents, bottom stress, and water depth on k [5]. Given the limited measurements of k in estuaries, we estimated k based on the wind speed dependence in this study.

Given the lack of a universally accepted relationship between k and wind speed [5,6], we compared three proposed relationships in this study: Nightingale et al. [37] (hereafter referred to as N00), Ho et al. [36] (H06), and Wanninkhof et al. [35] (W19). N00 and H06 used dual-deliberate tracer methods of ³He and SF6 to directly estimate k, with N00’s experiments conducted in coastal seas and H06’s in the Southern Ocean. In contrast, W19 proposed using nonzero intercepts to account for zero wind speed gust environments or zero wind-driven processes and utilizing global bomb ¹⁴C constraint and literature data to determine k, which are well suited for deep-water bodies. The three relationships have been used to estimate the FCO₂ in the Yangtze River estuary and adjacent coastal areas in previous studies [5,38]. Their respective equations are as follows:

$$\mathrm{N}00:k=0.333\times \mu +0.222\times {\mu }^2\times {\left(Sc/660\right)}^{-0.5}$$

(3)

$$\mathrm{H}06:k=0.266\times {\mu }^2\times {\left(Sc/660\right)}^{-0.5}$$

(4)

$$\mathrm{W}19:k=0.251\times {\mu }^2\times {\left(Sc/660\right)}^{-0.5}$$

(5)

where μ is the monthly wind speed at 10 m height, calculated using the daily wind speed (m s⁻¹) obtained from the Cross-Calibrated Multi-Platform (CCMP) analyses produced by Remote Sensing Systems, with a spatial resolution of 0.25° × 0.25° grids (available at http://www.remss.com/measurements/ccmp/, accessed on 20 December 2025).

The Schmidt number (Sc) CO₂ in water was computed from in situ temperature data. In seawater, Sc is 660 at 20 °C with a salinity of 35, while in freshwater (salinity < 30), it is 600 at the same temperature. Sc is calculated by using the following equation [39]:

$$Sc=1923.6-125.06T+{4.3773T}^2-{0.085681T}^3+{0.00070284T}^4$$

(6)

2.5. Three End-Member Mixing Models of Stable Isotope

2.5.1. The Impacts of Physical Mixing and Biogeochemical Processes on DIC and δ¹³C_DIC

The DIC concentrations and δ¹³C_DIC values in the Yangtze River estuary and adjacent coastal areas are influenced by multiple processes, including physical mixing, air–sea gas exchange, carbonate precipitation/dissolution, and biological activities. Water temperature is widely regarded as a relatively conservative parameter in marine carbonate systems, and its relationship with salinity can further indicate the mixing between freshwater and seawater end-members [17]. A temperature–salinity (T–S) diagram (Figure S2a) illustrates only CDW in the inner estuary and a three end-member mixing scheme of water masses in the outer estuary and offshore area, comprising CDW, YSCC, and the East China Sea Shelf Water (ECSSW) (Figure S2b). This distribution pattern is consistent with the previously reported observations [40]. Based on this framework, we established a three end-member mixing model with the following definitions in the outer estuary and offshore area: CDW, defined as water with salinity < 30.5; ECSSW, defined as water with salinity > 30.5 and temperature > 9 °C; YSCC, defined as water with salinity > 30.5 and temperature < 9 °C. Representative end-member stations were selected based on their distinct hydrographic characteristics. Station C, exhibiting the lowest salinity, was chosen as the CDW end-member. Station E, characterized by the highest temperature and salinity (attributable to the influence of the warm and saline Taiwan Warm Current), was selected as the ECSSW end-member. Station Y, with the lowest temperature and high salinity, was designated as the YSCC end-member. Detailed information on the end-member stations is provided in Table S1.

A three end-member mixing model, based on the mass conservation of temperature and salinity, is employed to predict the conservative concentrations of a chemical of interest in the outer estuary and offshore area [17]:

$$f_{\mathrm{C}\mathrm{D}\mathrm{W}}+f_{\mathrm{E}\mathrm{C}\mathrm{S}\mathrm{S}\mathrm{W}}+f_{\mathrm{Y}\mathrm{S}\mathrm{C}\mathrm{C}}=1$$

(7)

$$S_{\mathrm{C}\mathrm{D}\mathrm{W}}\times f_{\mathrm{C}\mathrm{D}\mathrm{W}}+S_{\mathrm{E}\mathrm{C}\mathrm{S}\mathrm{S}\mathrm{W}}\times f_{\mathrm{E}\mathrm{C}\mathrm{S}\mathrm{S}\mathrm{W}}+S_{\mathrm{Y}\mathrm{S}\mathrm{C}\mathrm{C}}\times f_{\mathrm{Y}\mathrm{S}\mathrm{C}\mathrm{C}}=S_{\mathrm{o}\mathrm{b}\mathrm{s}}$$

(8)

$$T_{\mathrm{C}\mathrm{D}\mathrm{W}}\times f_{\mathrm{C}\mathrm{D}\mathrm{W}}+T_{\mathrm{E}\mathrm{C}\mathrm{S}\mathrm{S}\mathrm{W}}\times f_{\mathrm{E}\mathrm{C}\mathrm{S}\mathrm{S}\mathrm{W}}+T_{\mathrm{Y}\mathrm{S}\mathrm{C}\mathrm{C}}\times f_{\mathrm{Y}\mathrm{S}\mathrm{C}\mathrm{C}}=T_{\mathrm{o}\mathrm{b}\mathrm{s}}$$

(9)

$${\delta }^{18}{\mathrm{O}}_{\mathrm{m}\mathrm{i}\mathrm{x}}={\delta }^{18}{\mathrm{O}}_{\mathrm{C}\mathrm{D}\mathrm{W}}\times f_{\mathrm{C}\mathrm{D}\mathrm{W}}+{\delta }^{18}{\mathrm{O}}_{\mathrm{E}\mathrm{C}\mathrm{S}\mathrm{S}\mathrm{W}}\times f_{\mathrm{E}\mathrm{C}\mathrm{S}\mathrm{S}\mathrm{W}}+{\delta }^{18}{\mathrm{O}}_{\mathrm{Y}\mathrm{S}\mathrm{C}\mathrm{C}}\times f_{\mathrm{Y}\mathrm{S}\mathrm{C}\mathrm{C}}$$

(10)

$$C_{{\mathrm{D}\mathrm{I}\mathrm{C}}_{\mathrm{m}\mathrm{i}\mathrm{x}}}=C_{{\mathrm{D}\mathrm{I}\mathrm{C}}_{\mathrm{C}\mathrm{D}\mathrm{W}}}\times f_{\mathrm{C}\mathrm{D}\mathrm{W}}+C_{{\mathrm{D}\mathrm{I}\mathrm{C}}_{\mathrm{E}\mathrm{C}\mathrm{S}\mathrm{S}\mathrm{W}}}\times f_{\mathrm{E}\mathrm{C}\mathrm{S}\mathrm{S}\mathrm{W}}+C_{{\mathrm{D}\mathrm{I}\mathrm{C}}_{\mathrm{Y}\mathrm{S}\mathrm{C}\mathrm{C}}}\times f_{\mathrm{Y}\mathrm{S}\mathrm{C}\mathrm{C}}$$

(11)

$$\begin{gathered} {\delta }^{13}{\mathrm{C}}_{{\mathrm{D}\mathrm{I}\mathrm{C}}_{\mathrm{m}\mathrm{i}\mathrm{x}}}\times C_{{\mathrm{D}\mathrm{I}\mathrm{C}}_{\mathrm{m}\mathrm{i}\mathrm{x}}} \\ ={\delta }^{13}{\mathrm{C}}_{{\mathrm{D}\mathrm{I}\mathrm{C}}_{\mathrm{C}\mathrm{D}\mathrm{W}}}\times C_{{\mathrm{D}\mathrm{I}\mathrm{C}}_{\mathrm{C}\mathrm{D}\mathrm{W}}}\times f_{\mathrm{C}\mathrm{D}\mathrm{W}} \\ +{\delta }^{13}{\mathrm{C}}_{{\mathrm{D}\mathrm{I}\mathrm{C}}_{\mathrm{E}\mathrm{C}\mathrm{S}\mathrm{S}\mathrm{W}}}\times C_{{\mathrm{D}\mathrm{I}\mathrm{C}}_{\mathrm{E}\mathrm{C}\mathrm{S}\mathrm{S}\mathrm{W}}}\times f_{\mathrm{E}\mathrm{C}\mathrm{S}\mathrm{S}\mathrm{W}} \\ +{\delta }^{13}{\mathrm{C}}_{{\mathrm{D}\mathrm{I}\mathrm{C}}_{\mathrm{Y}\mathrm{S}\mathrm{C}\mathrm{C}}}\times C_{{\mathrm{D}\mathrm{I}\mathrm{C}}_{\mathrm{Y}\mathrm{S}\mathrm{C}\mathrm{C}}}\times f_{\mathrm{Y}\mathrm{S}\mathrm{C}\mathrm{C}} \end{gathered}$$

(12)

where f denotes the mixing fraction. The subscripts obs and mix correspond to the measured value and the conservative mixing value of the sample, respectively. Equations (10)–(12) were used to predict the conservative mixing values for δ¹⁸O-H₂O, DIC concentration (C_DIC), and δ¹³C_DIC. δ¹⁸O-H₂O was used as a conservative tracer to validate the model prediction of the conservative mixing values. The strong agreement between the observed and predicted δ¹⁸O-H₂O (Figure S3; R² = 0.98, slope of the fitted line = 1.03) provided robust support for the validity of our model approach.

Deviations of the observed DIC concentrations and δ¹³C_DIC signals from their predicted conservative mixing values represented the impact of processes other than physical mixing. This is because processes influencing the C_DIC exhibit distinct δ¹³C_DIC values and isotope fractionation. Specifically, carbonate and atmosphere CO₂ dissolution results in increases in both C_DIC and δ¹³C_DIC. Primary production or CO₂ outgassing leads to decreases in C_DIC and increases in δ¹³C_DIC. CaCO₃ precipitation causes decreases in both C_DIC and δ¹³C_DIC. Conversely, organic carbon degradation results in increases in C_DIC and decreases in δ¹³C_DIC [22,41]. The deviations of the observed values from their conservative mixing values can be calculated using the following equations [17,22]:

$$∆C_{\mathrm{D}\mathrm{I}\mathrm{C}}={\displaystyle \frac{C_{{\mathrm{D}\mathrm{I}\mathrm{C}}_{\mathrm{o}\mathrm{b}\mathrm{s}}}-C_{{\mathrm{D}\mathrm{I}\mathrm{C}}_{\mathrm{m}\mathrm{i}\mathrm{x}}}}{C_{{\mathrm{D}\mathrm{I}\mathrm{C}}_{\mathrm{m}\mathrm{i}\mathrm{x}}}}}$$

(13)

$${∆\delta }^{13}{\mathrm{C}}_{\mathrm{D}\mathrm{I}\mathrm{C}}={\delta }^{13}{\mathrm{C}}_{{\mathrm{D}\mathrm{I}\mathrm{C}}_{\mathrm{o}\mathrm{b}\mathrm{s}}}-{\delta }^{13}{\mathrm{C}}_{{\mathrm{D}\mathrm{I}\mathrm{C}}_{\mathrm{m}\mathrm{i}\mathrm{x}}}$$

(14)

where C_DICmix and δ¹³C_DICmix are given by the CDW end-member (Station C) in the inner estuary and by Equations (11) and (12) in the outer estuary and offshore area, respectively. The primary biogeochemical processes influencing the distributions of δ¹³C_DIC and C_DIC can be inferred from the slopes of the relationship between Δδ¹³C_DIC and ΔC_DIC.

2.5.2. Potential Sources Analysis of DIC

The principal sources of DIC in the Yangtze River estuary and adjacent coastal areas include atmospheric CO₂ invasion, carbonate and silicate mineral weathering, and organic matter degradation [17,19]. Each of these sources imparts a distinct isotopic signature to the DIC pool, thereby influencing the observed δ¹³C_DIC values. Atmospheric CO₂ invasion occurs via air–sea gas exchange and the chemical weathering of carbonate and silicate minerals driven by atmospheric CO₂. This process typically yields δ¹³C_DIC values ranging from −6.3‰ to 0‰ [18]. Carbonate mineral weathering involves the dissolution of carbonate minerals by CO₂ derived from the atmosphere or organic matter degradation. The resulting DIC thus carries a mixed isotopic signal, comprising both the CO₂ source (atmospheric or organic) and the intrinsic δ¹³C signature of the carbonate minerals themselves. Organic matter degradation refers to the microbial breakdown of organic matter, producing CO₂ that is incorporated into the DIC pool. This source primarily reflects the δ¹³C signatures of terrestrial and marine organic matter. To quantitatively apportion these three DIC sources, we constructed an isotopic mass balance model following the approach of Wang et al. [18]:

$${\delta }^{13}{\mathrm{C}}_{\mathrm{D}\mathrm{I}\mathrm{C}\mathrm{x}}={\delta }^{13}{\mathrm{C}}_{\mathrm{a}\mathrm{t}\mathrm{m}}{\times f}_{{\mathrm{C}}_{\mathrm{a}\mathrm{t}\mathrm{m}}}+{\delta }^{13}{\mathrm{C}}_{\mathrm{c}\mathrm{a}\mathrm{r}\mathrm{b}}\times f_{{\mathrm{C}}_{\mathrm{c}\mathrm{a}\mathrm{r}\mathrm{b}}}+{\delta }^{13}{\mathrm{C}}_{\mathrm{o}\mathrm{m}}\times f_{{\mathrm{C}}_{\mathrm{o}\mathrm{m}}}$$

(15)

$$f_{{\mathrm{C}}_{\mathrm{a}\mathrm{t}\mathrm{m}}}+f_{{\mathrm{C}}_{\mathrm{c}\mathrm{a}\mathrm{r}\mathrm{b}}}+f_{{\mathrm{C}}_{\mathrm{o}\mathrm{m}}}=1$$

(16)

where δ¹³C_atm represents the δ¹³C values of DIC sources from atmospheric CO₂ invasion via air–sea CO₂ exchange and chemical weathering and is assigned as −3.2 based on atmospheric CO₂ dissolution [18]. δ¹³C_carb represents the δ¹³C values of DIC sources from the intrinsic δ¹³C signature of the carbonate minerals during chemical weathering and is assigned as −8.5 (−11.7~−5.2‰), based on the δ¹³C values of particulate inorganic carbon (PIC) observed in this region [17,18]. δ¹³C_om represents the δ¹³C values of DIC sources from organic matter degradation of river and seawater, with δ¹³C_om as −24.27 (−26.7~−22.3‰) derived from the δ¹³C values of particulate organic carbon (POC) and surface sediment total organic carbon (TOC) in this region [42,43]. f_Catm, f_Ccarb, and f_Com denote the proportional contributions of these respective DIC sources.

Given the constraint that the end-member contributions should sum to 1, the isotopic mass balance model is described by two equations (Equations (14) and (15)), which is one fewer than the number of end-members. To solve this under-determined system, the Bayesian mixing model (MixSIAR, version 3.1.11) was employed to calculate the proportional contributions of each potential DIC source [44]. MixSIAR models provide probabilistic solutions to mixing systems [45] and have been successfully applied to quantify DIC sources in rivers and estuaries [11,17]. To ensure accurate results, the Gelman–Rubin and Geweke diagnostic tests were employed to assess the convergence of Markov chain Monte Carlo (MCMC) chains [46]. The MCMC run length was set to ‘long’ to enhance the robustness of the model output [11].

2.6. Statistical Analysis

The Mantel test was applied to evaluate the correlations between the FCO₂ matrix and the factors matrix, and Pearson correlation analyses were employed to investigate the correlations among the factors. The Mantel test provided a preliminary understanding of the association strength without assuming causal directionality and was performed using the “mantel” function from the “vegan” and “LinkET” packages in RStudio 4.3.0 (RStudio, PBC, Boston, MA, USA) [47]. Multiple linear regression (MLR) was applied to evaluate the relative importance of k, water temperature, and pCO₂ in explaining the spatial variance in FCO₂. Meteorological, physical, chemical, and biological factors were selected for analysis based on their established ecological relevance in prior studies [2,5]. The meteorological factor primarily includes the wind speed, a key driver of turbulent gas transfer. Physical factors (salinity) reflect hydrodynamic conditions, water mass mixing, and water temperature, which affect the solubility of CO₂ and the biological pump. Chemical factors comprise key carbonate system variables (pCO₂, DIC, pH, and TA); nutrients (NO₂⁻, NO₃⁻, NH₄⁺, PO₄³⁻) and DO as regulators of the biological pump; major cations (Ca²⁺ and Mg²⁺) indicative of chemical weathering; and SiO₃²⁻, which relates to both biological processes and chemical weathering. Biological factors mainly consist of Chla, a proxy for phytoplankton biomass. Simple linear regression was employed to examine the relationships between the variables in R studio 4.3.0. MLR was conducted using the SPSS Statistics 22.0 software (IBM Corp., Armonk, NY, USA).

3. Results

3.1. Spatial Distribution of Air–Sea CO₂ Flux

Figure 2 presents the spatial distributions of FCO₂ in March 2025, calculated using the N00, H06, and W19 gas transfer velocity formulas.

Table 1. Air–sea CO₂ flux (FCO₂) calculated using three different gas transfer velocity (k) formulas in the Yangtze River estuary and adjacent coastal areas (March 2025).

Region	Water Types	Wind Speed (m s−1)	k (cm h−1)			Air–Sea CO2 Flux (mmol m−2 d−1)
			kN00	kH06	kW19	FcN00	FcH06	FcW19	Mean
Inner estuary		4.7 ± 0.9	5.4 ± 2	5 ± 2.1	5 ± 1.2	11.8 ± 10.3	10.8 ± 10.1	10.2 ± 9.5	11 ± 9.9
Outer estuary and offshore area	All	7.8 ± 0.6	13.1 ± 2.4	13.2 ± 2.5	10.5 ± 1.9	6.7 ± 8.7	6.7 ± 8.6	6.3 ± 8.1	6.6 ± 8.5
	CDW	7.6 ± 0.4	12 ± 1.6	12 ± 1.7	9.6 ± 1.3	14.5 ± 7.2	14.4 ± 7.1	13.6 ± 6.7	14.2 ± 7
	ECSSW	7.9 ± 0.5	13.9 ± 2.1	14 ± 2.3	11.2 ± 1.7	1.9 ± 6.8	1.8 ± 6.8	1.7 ± 6.4	1.8 ± 6.7
	YSCC	7.1 ± 0.6	10.1 ± 1.8	10 ± 1.9	8.2 ± 1.4	8.8 ± 6.3	8.7 ± 6.2	8.2 ± 5.9	8.6 ± 6.1
Carbon sink		7.8 ± 0.5	13.5 ± 2.1	13.6 ± 2.2	10.8 ± 1.7	−4.7 ± 4.7	−4.7 ± 4.8	−4.5 ± 4.5	−4.6 ± 4.7
Carbon source		6.7 ± 1.3	9.9 ± 3.2	9.8 ± 3.4	8.1 ± 2.3	10.3 ± 7.2	10.1 ± 7.1	9.5 ± 6.7	9.9 ± 7
Entire study area		7.5 ± 1.1	12.2 ± 3.3	12.3 ± 3.6	9.9 ± 2.5	7.5 ± 9	7.3 ± 8.8	6.9 ± 8.3	7.3 ± 8.7

further summarizes the FCO₂ values for each water mass, for designated CO₂ source and sink areas, and for the entire study area. The spatial patterns derived from the three formulas were broadly consistent, and the corresponding mean FCO₂ values for the entire study areas were comparable. Overall, the entire study area functioned as a CO₂ source, with an average flux of 7.3 ± 8.7 mmol m⁻² d⁻¹. A progressive decline in source strength was observed from the inner estuary to the outer estuary and offshore area. The mean fluxes were 11.0 ± 9.9 mmol m⁻² d⁻¹ in the inner estuary and 6.6 ± 8.5 mmol m⁻² d⁻¹ in the outer estuary and offshore area. Within the outer estuary and offshore area, individual water masses exhibited contrasting source magnitudes. The CDW displayed the strongest CO₂ outgassing (14.2 ± 7.0 mmol m⁻² d⁻¹), with values substantially exceeding those recorded in the inner estuary. YSCC ranked second (8.6 ± 6.1 mmol m⁻² d⁻¹), whereas the ECSSW region acted as a weak CO₂ source, with a mean flux of 1.8 ± 6.7 mmol m⁻² d⁻¹. Across the entire study region, areas characterized as CO₂ sinks exhibited a mean flux of −4.6 ± 4.7 mmol m⁻² d⁻¹, while source areas averaged 9.9 ± 7.0 mmol m⁻² d⁻¹.

3.2. Multi-Factor Drivers of Air–Sea CO₂ Flux Spatial Variability

Figure 3 displays the correlations between FCO₂ and meteorological, physical, chemical, and biological factors, examined across different regional settings and water mass regimes to assess their influence on the spatial variability of FCO₂. Based on Equation (1), FCO₂ is directly influenced by k, water temperature, and pCO₂. Mantel tests revealed that pCO₂ exhibited a significant positive correlation with FCO₂ over the entire study area and within each sub-region (Figure 3). Multiple linear regression results (Table S2) further confirmed that pCO₂ was the primary driver of FCO₂ variability across the entire study area. In the entire study area, pCO₂ showed significant positive correlations with turbidity, NO₃⁻, SiO₃²⁻, DO, and Chla and significant negative correlations with water temperature, salinity, and pH (Figure 3a). Within the outer estuary and offshore area, pCO₂ was significantly positively correlated with turbidity and significantly negatively correlated with water temperature, salinity, and pH (Figure 3c). Examination of individual water masses revealed distinct controlling factors: in the ECSSW region, pCO₂ displayed a significant negative correlation with pH (Figure 3e), whereas in the YSCC region, pCO₂ was significantly negatively correlated with SiO₃²⁻ and Chla (Figure 3f).

3.3. Potential Processes Controlling Dissolved Inorganic Carbon Spatial Variation

Three-dimensional scatter plots (Figure 4) revealed that the observed DIC concentrations deviated notably from the predicted conservative mixing values, with some samples exceeding and others falling below the theoretical mixing line (Figure 4a). Correspondingly, δ¹³C_DIC values displayed bidirectional deviations, with a portion of samples enriched relative to the theoretical mixing line and others depleted (Figure 4b). A quadrant analysis of C_DIC and δ¹³C_DIC deviations (Figure 5) provided further insight into the dominant biogeochemical processes affecting the distributions of δ¹³C_DIC and C_DIC across different regions. In both surface and bottom waters, the majority of stations fell within Quadrant II (depleted in C_DIC but enriched in δ¹³C_DIC), suggesting the effects of primary production or CO₂ outgassing. Some stations were located in Quadrant I (characterized by enrichment in both C_DIC and δ¹³C_DIC), indicating the predominant influence of carbonate mineral dissolution or atmospheric CO₂ invasion. A small subset of surface stations in the ECSSW region fell within Quadrant III (depleted in both C_DIC and δ¹³C_DIC), indicative of carbonate precipitation. Conversely, a few surface stations in the inner estuary and CDW region occupied Quadrant IV (enriched in C_DIC but depleted in δ¹³C_DIC), pointing to organic matter degradation.

Regional patterns further illuminated the spatial heterogeneity of these processes. In the inner estuary, stations were predominantly distributed between Quadrants II and IV, suggesting a mixture of primary production or CO₂ outgassing and organic matter degradation. In the CDW region, stations were mainly situated in Quadrants I and IV, reflecting the combined influences of carbonate dissolution or atmospheric CO₂ invasion and organic matter degradation. In the ECSSW region, the vast majority of stations were located in Quadrants I, II, and III, reflecting the combined influences of carbonate dissolution or atmospheric CO₂ invasion, primary production or CO₂ outgassing, and carbonate precipitation. In the YSCC region, most stations fell within Quadrant II, indicating the importance of primary production or CO₂ outgassing. Despite these qualitative insights, the relative contributions of the three key DIC-adding processes, carbonate dissolution, atmospheric CO₂ invasion, and organic matter degradation, remain unresolved and warrant further quantitative investigation.

3.4. Potential Sources of Dissolved Inorganic Carbon

The relative contributions of potential DIC sources, quantified using an isotope mass balance model, are summarized in

Table 2. Quantitative proportional contributions of DIC sources in the Yangtze River estuary and adjacent coastal areas.

Region	Water Types	Atmospheric CO2 Invasion	Carbonate Mineral Dissolution	Organic Matter Degradation
Inner estuary		50.3 ± 9.6	30.2 ± 6.5	19.5 ± 3.2
Outer estuary and offshore area	All	80.7 ± 6.6	10.0 ± 4.4	9.3 ± 2.4
	CDW	74.4 ± 7.2	14.3 ± 4.5	11.3 ± 2.7
	ECSSW	83.3 ± 4.5	8.2 ± 3.0	8.6 ± 1.8
	YSCC	82.1 ± 5.3	9.3 ± 3.3	8.6 ± 2.0
Carbon sink		82.4 ± 5.6	8.7 ± 3.8	8.9 ± 2.1
Carbon source		68.8 ± 15.8	17.8 ± 10.2	13.4 ± 5.7
Entire study area		74.0 ± 14.6	14.4 ± 9.7	11.6 ± 5.0

Note: The values in this table are in the form of mean ± SD.

. Across the entire study area, atmospheric CO₂ invasion via air–sea CO₂ exchange and chemical weathering emerged as the dominant DIC source, contributing 62.3 ± 13.5% of the total DIC pool. Carbonate mineral dissolution and organic matter degradation accounted for 19.4 ± 6.1% and 18.2 ± 7.7%, respectively. A distinct spatial gradient was observed from the inner estuary to the outer estuary and offshore area. The contribution of atmospheric CO₂ invasion increased progressively from 41.5 ± 5.4% in the inner estuary to 68.1 ± 8.2% in the outer estuary and offshore area. In contrast, the contributions of organic matter degradation and carbonate minerals dissolution declined along this gradient, from 30.1 ± 5.2% to 14.9 ± 4.2% and from 28.4 ± 1.2% to 16.9 ± 4.2%, respectively. Within the outer estuary and offshore area, further spatial heterogeneity was evident among individual water masses. The contributions of atmospheric CO₂ invasion increased from 65.5 ± 9.8% in the CDW region to 69.5 ± 6.7% in the ECSSW region and 69.3 ± 8.0% in the YSCC region. Conversely, the contributions of carbonate minerals dissolution decreased from 18.4 ± 5.1% (CDW) to 15.6 ± 2.9% (ECSSW) and 16.8 ± 4.1% (YSCC). Similarly, the contributions of organic matter degradation declined from 16.0 ± 4.8% (CDW) to 15.0 ± 3.8% (ECSSW) and 13.9 ± 3.9% (YSCC). Notably, the contribution of atmospheric CO₂ invasion increased from carbon source areas to carbon sink areas, while the contribution of carbonate minerals dissolution and organic matter degradation decreased.

4. Discussion

4.1. Controlling Mechanisms of Air–Sea CO₂ Flux Spatial Variability

4.1.1. Effects of Water Mixing Processes on Air–Sea CO₂ Flux Spatial Variability

The spatial variability of FCO₂ in March was primarily governed by the physical mixing of three water masses—the CDW, the ECSSW, and the YSCC—consistent with previous findings that winter-to-spring carbonate dynamics in this region are mixing-dominated [28]. The inner estuary acted as a CO₂ source due to elevated pCO₂ in the Yangtze River discharge (Figures S4–S6), in line with its widely recognized role as a persistent CO₂ outgassing zone [26,31]. In agreement with Li et al. [23], the outer estuary and offshore area exhibited a progressive decline in CO₂ source intensity from CDW to ECSSW and YSCC. This may be high-pCO₂ CDW- and YSCC-sustained CO₂ outgassing, whereas the warmer low-pCO₂ ECSSW functioned only as a weak source [20,21]. Lower-than-average March discharge (Figure S7) likely reduced the freshwater dilution and facilitated the intrusion of shelf waters, highlighting this early-spring case as a reference for how physical forcing preconditions subsequent biological CO₂ drawdown [20].

Water mass mixing also indirectly influenced FCO₂ by modulating the temperature, nutrients, and turbidity [23,48]. Despite reducing the CO₂ solubility, the increase in water temperature from the inner estuary to the outer estuary and offshore area likely enhanced FCO₂ through complex thermodynamic and biological feedbacks [10]. In the inner estuary and CDW region, elevated nutrients and terrigenous organic carbon input (Figure 5 and Figures S8 and S9) likely stimulated organic matter degradation, intensifying the CO₂ source [6,14]. These linkages underscore that physical mixing establishes both the chemical and biogeochemical backdrop that modulates source–sink strength, an interplay especially pronounced in early spring before the phytoplankton bloom peak.

4.1.2. Influences of Biogeochemical Processes on Air–Sea CO₂ Flux Spatial Variability

Given that pCO₂ outweighed water temperature and k as the primary driver of FCO₂, biogeochemical processes that alter DIC concentrations can profoundly modulate FCO₂ in early spring [8,9]. The coexistence of positive and negative DIC concentrations deviations from conservative mixing lines (Figure 5) indicates that DIC-adding and DIC-removing processes operated simultaneously, a dual regulation that distinguishes early spring from late spring when biological uptake dominates [20,48]. In the inner estuary, terrestrial inputs of nutrients, minerals, and organic carbon likely stimulated organic matter degradation, carbonate dissolution, and atmospheric CO₂ invasion, amplifying CO₂ outgassing [6,14]. This aligns with the view that inner estuaries function as biogeochemical reactors, where terrestrial inputs are rapidly transformed into CO₂, rather than merely acting as passive conduits [25]. In the CDW region, these DIC-adding processes persisted, in agreement with previous wintertime findings [23]. In the YSCC region, DIC concentrations dynamics were dominated by outgassing driven by high-pCO₂ YSCC waters [20], whereas in the ECSSW region, they exhibited a balance between carbonate-related DIC addition and removal via primary production and carbonate precipitation [23]. Significant correlations of pCO₂ with DO and nutrients across the entire study area (Figure 3a) further reinforce the role of biological processes in modulating FCO₂ spatial variability [6,15]. Resolving which DIC sources primarily control FCO₂ spatial variability, however, requires further quantitative source partitioning.

4.2. Effects of DIC Sources on Air–Water CO₂ Flux Spatial Variability

Quantitative DIC source apportionment confirms that the spatial variability of FCO₂ in early spring is governed by the interplay of water mass mixing and biological processes [17,49]. In the inner estuary, the CO₂ source signature is largely inherited from the high-pCO₂ river discharge, with atmospheric CO₂ invasion and carbonate dissolution via chemical weathering constituting the dominant DIC sources (Table 2), consistent with earlier findings for the lower Yangtze River [49]. The higher contribution of organic matter degradation in the inner estuary relative to offshore waters (Table 2) indicates that terrestrial organic carbon inputs actively fuel remineralization, amplifying CO₂ outgassing beyond conservative mixing expectations [50]. Primary production likely moderates this effect, explaining why the source intensity in the inner estuary remained weaker than in the CDW region [6,14]. In the CDW region, freshwater–seawater mixing reduces the nutrient availability, which may suppress primary production and thereby sustain stronger outgassing [23].

In the YSCC region, elevated pCO₂ is sustained by DIC derived from long-term carbonate dissolution and atmospheric CO₂ invasion via chemical weathering [18,49]. In the ECSSW region, the low-pCO₂ reflects the offshore origin of this water mass from the East China Sea [21]. Here, the high total alkalinity and strong buffering capacity facilitate atmospheric CO₂ uptake as the dominant DIC source [12]. Moreover, the elevated water temperatures likely enhance photosynthetic CO₂ uptake, further weakening the CO₂ source [10,11]. Although atmospheric CO₂ and carbonate dissolution dominating the DIC pool is consistent with earlier work [18,49], relying solely on ¹³C without resolving isotopic differences between terrestrial and marine organic matter may introduce uncertainties and limitations [51]. Future studies in the Yangtze River estuary and adjacent coastal areas should consider using dual-isotope mass balance equations and accounting for the differences between terrestrial and marine organic matter δ¹³C.

4.3. A Comparison of Air–Sea CO₂ Flux with Other Research

4.3.1. Uncertainty Analysis of Calculating Air–Sea CO₂ Flux

Although the inner estuary acted as a CO₂ source in our early spring survey (11.0 mmol m⁻² d⁻¹), consistent with Li et al. [23], for February and March, it was markedly lower than the April estimate (23.5 mmol m⁻² d⁻¹) of Zhai et al. [25]. This discrepancy may reflect reduced DIC inputs from declining nutrient and terrestrial organic matter loading, potentially linked to pollution control or hydrological shifts following the Three Gorges Dam construction [52]. For the outer estuary and offshore area, our March FCO₂ (6.6 mmol m⁻² d⁻¹) was slightly higher than that of Li et al. [5] (2.58 mmol m⁻² d⁻¹;

Table 3. Comparisons of air–sea CO₂ flux (FCO₂) in the Yangtze River estuary and adjacent coastal areas across different seasons.

Region	Season	Year	Month	Wind Speed (m s−1)	FCO2 (mmol m−2 d−1)	Reference
Inner estuary	Spring	2006	April	3.5 (2.7~7.2)	23.5 (14~99)	[25]
		2014	February–March		14.17 ± 9.93	[23]
		2025	March	4.7 ± 0.9	11.0 ± 9.9	This study
	Summer	2003	August	5 (0~8.0)	65.5 (0~168)	[25]
		2014	July		26.04 ± 8.19	[23]
		2020	August		17.5 ± 47.9	[38]
	Autumn	2005	October	5 (1.7~8.2)	33.7 (3.9~91)	[25]
	Winter	2005	December	6 (2.4~9.0)	37.8 (6.0~85)	[25]
Outer estuary and offshore area	Spring	2005–2008	April	5.6 ± 1.0	−8.8 ± 5.8	[48]
		2012	May	5.47	−9 (−26~11)	[20]
		2014	February–March		2.58	[23]
		2018	March	3.43 ± 1.31	−1.25 ± 1.7	[5]
		2025	March	7.8 ± 0.6	6.6 ± 8.5	This study
	Summer	2003–2007	August	4.7 ± 1.1	−4.9 ± 4.0	[48]
		2006	August		−0.12 ± 0.16	[53]
		2010	August	5.54	−16 (−31~14)	[20]
		2014	June	3.2	−4.7	[21]
		2018	July	3.64 ± 0.92	1.71 ± 5.9	[5]
	Autumn	2006–2007	September	5.1 ± 0.5	2.9 ± 2.5	[48]
		2010	November	5.02	5 (−3~18)	[20]
		2018	October	4.27 ± 1.7	3.06 ± 4.1	[5]
	Winter	2006	January	10.5 ± 1.2	−10.4 ± 2.3	[48]
		2010	February	5.22	5 (−5~20)	[20]

Note: The values in this table are in the form of mean ± SD or mean (Minimum~Maximum).

). The overall spatial variability of FCO₂ in our study closely mirrored the findings of Li et al. [23].

Accurate FCO₂ calculation depends on the surface water–atmosphere pCO₂ gradient and k [2,5]. Here, pCO₂ in water bodies was calculated via CO₂SYS from DIC and pH measurements [33], with an uncertainty of approximately ±14 μatm, small compared to the magnitude of pCO₂. Given the absence of region-specific k formulations, three wind speed-dependent formulas (N00, H06, and W19) were applied, yielding similar spatial distributions and comparable area-averaged fluxes (Table 1). At the relatively low wind speeds prevailing in March (7.5 ± 1.1 m s⁻¹), W19 produced lower k values, due to its non-zero intercept term, resulting in slightly lower FCO₂ (Table 1). While k primarily depends on wind regime in the open ocean, it is more complex in coastal and shallower estuaries, where other factors such as tidal currents, bottom stress, and water depth can also influence k [5]. The use of wind speed-driven models in the inner estuary area, neglecting the influence of hydrodynamics on k, may introduce certain uncertainties.

Monthly average wind speeds were used for the k calculation (inner estuary: 4.7 ± 0.88 m s⁻¹; outer estuary and offshore area: 7.8 ± 0.59 m s⁻¹; Table S3). The nonlinearity factor C₂, representing sub-monthly wind variability effects on k, was 1.25 in the inner estuary and 1.18 in the outer estuary and offshore area (Figure S10 and Table S3), assuming that long-term winds follow a Raleigh (Weibull) distribution [54]. These values were slightly lower than the global average of 1.27 [35]. The highest C₂ value in the three areas was 1.06- to 1.11-fold the lowest value in March, indicating that this variability did not induce significant intra-monthly changes in the calculated FCO₂. Nevertheless, the variations in estimated FCO₂ during early spring (Table 3) suggest that further work is needed to constrain the uncertainties associated with different estimation approaches and to incorporate multi-year observations [5,38]. Future research is required to validate the most appropriate gas transfer velocity model in the estuary systems with complex hydrodynamics. To improve FCO₂ estimations in this region, high-frequency and long-term observations of pCO₂ and supporting carbonate chemistry data are essential.

4.3.2. Seasonal Variations in Air–Sea CO₂ Flux

Seasonal comparisons (Table 3) show that the inner estuary remains a perennial CO₂ source sustained by high-pCO₂ freshwater and terrestrial inputs [25,52], whereas the outer estuary and offshore area shifts from a carbon sink in late spring to a source in autumn, with transitional behavior in summer and winter. This seasonal pattern is largely regulated by the mixing of CDW, YSCC, and ECSSW in winter, spring, and autumn and primarily CDW and ECSSW in summer under elevated discharge [5]. The higher carbon sink capacity in April and May compared to March suggested the increased biological uptake of CO₂, driven by rising nutrient inputs from CDW and increasing water temperature [54]. Our early-spring survey thus captures a critical transitional window, providing regionally meaningful evidence for understanding source–sink shifts along the estuary–coastal continuum. In summer, strong primary production supported by nutrient-rich CDW and warming temperatures drives net CO₂ uptake [20]. Phytoplankton absorb CO₂ from surface water through photosynthesis and transport it to deeper water via the biological pump, resulting in decreased surface pCO₂ and increased atmospheric CO₂ uptake [15,21]. In autumn, the net CO₂ release may be related to the collapse of summer stratification and the autumn mixture between the surface water and high-pCO₂ bottom water [5,20]. From autumn to winter, continued cooling and CO₂ degassing may shift the region from a carbon source [20,23] to a carbon sink [14,48].

Anticipated anthropogenic and climatic perturbations may fundamentally alter the seasonal and spatial variation in FCO₂ in the Yangtze River estuary and adjacent coastal areas [5]. Projected declines in freshwater discharge, driven by reservoir expansion and increased water consumption [5,23], could diminish summer plume stratification, enhancing the vertical advection of high-pCO₂ subsurface waters [20], and reduce nutrient delivery, inhibiting phytoplankton production [2]. Warming trends and elevated terrestrial nutrient loads due to land-use changes or wastewater discharge [23] may exacerbate summer eutrophication, potentially altering the FCO₂ [55]. Rising atmospheric CO₂ levels may enhance atmospheric CO₂ dissolution, influencing the long-term stability of estuarine carbon sinks, and leading to the acidification of this region. The resulting decrease in pH due to acidification may further increase CO₂ dissolution in water bodies by altering the carbonate chemistry balance [3,24]. To enhance our understanding of the impacts of future human activities and climate change on carbon sources and sinks in this region, high-frequency high spatial resolution and long-term observations of pCO₂ and supporting carbonate chemistry data are essential.

5. Conclusions

This study investigated the spatial variability of FCO₂ and its controlling mechanisms in the Yangtze River estuary and adjacent coastal areas during early spring (March 2025) based on the NORC2025-03 cruise. The entire study area functioned as a net CO₂ source, with an average flux of 7.3 ± 8.7 mmol m⁻² d⁻¹, with the source intensity declining progressively from the inner estuary to the outer estuary and offshore area. Within the outer estuary and offshore area, the CDW region exhibited the strongest outgassing, followed by the YSCC region, while the ECSSW region acted only as a weak CO₂ source.

The spatial pattern of FCO₂ was primarily set by the physical mixing of three water masses (CDW, ECSSW, and YSCC), with pCO₂ as the dominant direct driver. Biogeochemical processes superimposed on this physical template further modulated the source intensity regionally, while water mass mixing also exerted indirect controls through temperature, nutrient availability, and turbidity, which in turn influenced organic matter degradation and biological production.

Quantitative source apportionment revealed that atmospheric CO₂ invasion via air–sea exchange and chemical weathering dominated the DIC pool, followed by carbonate mineral dissolution and organic matter degradation. The enhanced FCO₂ from the inner estuary to CDW region can be attributed to freshwater–seawater mixing, which reduces nutrient concentrations, diminishes primary production, and decreases the contribution of organic matter degradation. In the ECSSW region, elevated water temperatures likely enhance photosynthetic CO₂ uptake, further weakening the CO₂ source. These contrasting source contributions mechanistically link the carbon source–sink mosaic to distinct DIC supply pathways.

By capturing a transitional hydrographic window which is typically underrepresented in seasonal studies, this early-spring case provides regionally meaningful evidence on how physical forcing preconditions subsequent biological CO₂ drawdown. The findings underscore the value of integrating multi-factor regulation and DIC source partitioning to resolve carbon dynamics in a large river-dominated estuary–coastal continuum. Future work should prioritize higher-frequency observations and assess the impacts of climate change and anthropogenic activities on regional carbon cycling.