Field investigations were performed in a multi-species saltmarsh, Andong Shoal, located on the protruding section of the tidal shoal on the south bank of Hangzhou Bay, China ( Figure 1 ). Andong Shoal is an alluvial coast and has a subtropical monsoon climate with a mean annual temperature of 17.1°C and mean annual rainfall of 1381 mm (Cao et al., 2020). It is formed by the accumulation of sediments from the Yangtze and Qiantang River (Wu et al., 2008). The intertidal zone of Andong Shoal is 7–8 km wide (Song et al., 2014), with a developed creek system due to macrotidal conditions (Li and Xie, 1993). An irregular semidiurnal tide exists in this region, with a mean tidal range of 5.5 m (Huang et al., 2020). The vegetation in the saltmarsh includes not only local species such as Scirpus mariqueter, Suaeda glauca, and Phragmites australis (C3 plant species), but also invasive species such as Spartina alterniflora (C4 plant species) (Wang et al., 2015).

The ²²²Rn mass balance model (Burnett and Dulaiova, 2003) has been widely used to quantify advective porewater flux in saltmarshes and mangroves (e.g., Santos et al., 2019; Correa et al., 2021; Chen et al., 2022). The model integrates all ²²²Rn sources (e.g., imports from bay water during the flood tide, diffusion from sediments, and ingrowth from dissolved ²²⁶Ra) and sinks (exports during the ebb tide, atmospheric evasion, and radioactive decay). Surface water samples were collected during the highest tide level, because the sampling boat can reach the upstream sites at this time. Assuming all the creek water is discharged during ebb tide, the missing ²²²Rn represents porewater exchange in each tidal cycle. At steady state, integrating all of fluxes over a complete day, the porewater exchange flux (F_pw , Bq m^-2 d^-1) can be estimated as follows:

where ²²² Rn_sw is the average ²²²Rn activity (Bq m^-3) in surface water during the flood tide, Δv is the difference of water volume in creeks between high tide level and low tide level (m³ d^-1), F_atm is the ²²²Rn flux to the atmosphere (Bq m^-2 d^-1), which can be calculated by concentration gradients, wind speed and current (Chen et al., 2020a), λ is the ²²²Rn decay constant (0.182 d^-1), ²²²Rn_sea is the ²²²Rn activity (Bq m^-3) of seawater end-member, F_sea is the ²²²Rn flux via sediment diffusion (Bq m^-2 d^-1), ²²⁶ Ra_sw is the ²²⁶Ra concentration in surface water, and A is the inundated area (m²).

Similar to the ²²²Rn calculation, carbon outwelling (F_outwelling , mmol m^-2 d^-1) from intertidal creeks was estimated as follows:

where C_sw is the concentration of carbon species in surface water (mmol L^-1) and C_sea is the corresponding seawater endmember of the carbon species (mmol L^-1).

Greenhouse gas emissions at the water-air interface were calculated from a bulk flux equation (Wanninkhof, 2014); therefore, greenhouse gas emissions (F_emissions , mmol m^-2 d^-1), including CO₂ and CH₄, from creek water were estimated as follows:

where C_sw is the greenhouse gas concentration in surface water (mmol L^-1), C_air is the greenhouse gas concentration in air (mmol L^-1), and k is the gas transfer velocity (m d^-1), which was the mean value derived from three gas transfer models (Borges et al., 2004; Ho et al., 2016; Rosentreter et al., 2017), α is the solubility coefficient of greenhouse gas. Uncertainties regarding the ²²²Rn mass balance model and carbon fluxes are estimated based on the basic rules of error propagation.

During the surface water observation, temperature and salinity were found with spatial gradients, whereas DO and pH were relatively stable ( Figure 2 and Table 1 ). The surface water temperature increased from downstream (20.4°C) to upstream (28.7°C) while the salinity showed a contrasting trend decreasing from 13.7 to 0.5. Surface water DO and pH were irregularly distributed in ranges of 54–107% (4.7–8.6 mg L^-1) (mean: 94 ± 10%, 7.6 ± 0.7 mg L^-1) and 7.1–8.2 (mean: 7.9 ± 0.3), respectively, and the lowest pH were measured at upstream of creeks. ²²²Rn, carbon, and greenhouse gases showed large spatial heterogeneity, indicating the necessity for spatial investigation ( Figure 3 ). Surface water parameters varied over a range of 10–521 (mean: 81 ± 103) Bq m^-3 for ²²²Rn, 1.54–7.90 (mean: 2.31 ± 1.36) mmol L^-1 for DIC, 0.23–0.79 (mean: 0.45 ± 0.14) mmol L^-1 for DOC, 29–179 (mean: 51 ± 32) μmol L^-1 for CO₂ and 94–939 (mean: 235 ± 140) nmol L^-1 for CH₄. Carbon species in surface water were dominated by DIC with various DOC proportions (12%–26%) and negligible greenhouse gases ( Figure 4 ).

All hydrological parameters in porewater samples showed spatial heterogeneity ( Figures 2 , 3 and Table 2 ). Temperature, salinity, DO and pH changed from 22.6 to 26.3°C, 4.7 to 11.7, 0.5 to 58.6% (0.03 to 4.73 mg L^-1) and 7.04 to 7.54, with mean values of 24.6 ± 1.2°C, 8.71 ± 2.46, 13.8 ± 19.3% (1.2 ± 1.5 mg L^-1) and 7.32 ± 0.16, respectively. Porewater ²²²Rn activities varied from 1.01×10³ Bq m^-3 to 4.91×10³ Bq m^-3 with the mean value of (1.95 ± 2.02)×10³ Bq m^-3, which was approximately 24-fold higher than that in surface water. Carbon and greenhouse gases displayed considerable variability in porewater. As expected, DIC (range: 5.6–15.4 mmol L^-1, mean: 10.1 ± 3.0 mmol L^-1), DOC (range: 0.28–1.78 mmol L^-1, mean: 0.85 ± 0.41 mmol L^-1), CO₂ (range: 325–1280 μmol L^-1, mean: 709 ± 277 μmol L^-1) and CH₄ (range: 551–138400 nmol L^-1, mean: 6580 ± 4510 nmol L^-1) were highly enriched in porewater, which were approximately 4.5, 1.9, 14.2 and 28.1 times their respective concentrations in surface water. Similarly, in porewater, the major carbon species was also DIC, while DOC and greenhouse gases were minor components ( Figure 4 ). This is because sulfate reduction coupled to pyrite formation effectively convert sediment organic carbon into bicarbonate (the main component of DIC) (Reithmaier et al., 2021).

According to different photosynthetic processes, plants can be divided into C3 and C4 plants, with distinct differences in the proportion of ¹³C isotopes. Generally, the δ¹³C values of C3 plants (range from -34‰ to -23‰) were more negative than those of C4 plants (range from -17‰ to -9‰) (Chmura and Aharon, 1995). In the Andong Shoal, local saltmarsh species, Scirpus mariqueter, Suaeda glauca, and Phragmites australis, were typical C3 plants with δ¹³C in range from -29‰ to -27‰, but the C4 plant Spartina alterniflora had a much higher δ¹³C of approximately -14‰ ( Table 3 and Figure 5 ). For each plant species, δ¹³C values were relatively constant for various C/N ratios in the different organs ( Figure 5 ).

Combining organic δ¹³C values with C/N ratios in sediments and porewater can trace the source of organic carbon (Meyers, 1994). In sediment samples, results of δ¹³C (range from -25.0‰ to -23.8‰) and C/N ratio (range from 9.79 to 11.42) implied that the C3 plant Scirpus mariqueter was the major source of sediment organic carbon ( Figure 5 and Tables 3 , 4 ). A low C/N ratio (<10) may indicate that organic carbon was provided by lake or marine algae (Meyers, 1994). While lake algae would not be the source of organic carbon as no direct connection with lake ecosystems, marine algae can be a potential source because of frequent algal blooms around Hangzhou Bay (Liu et al., 2013). Sediment organic carbon was not influenced by the invasion of Spartina alterniflora, because of the low density in our study region. Furthermore, in the porewater samples, δ¹³C values of DOC were relatively constant at -26‰, but C/N ratios varied from 22.28 to 79.48 ( Table 4 ). Here, we suggest that porewater-derived DOC flux was mainly provided by the biomass of Scirpus mariqueter due to approximate δ¹³C value ( Figure 5 ).

To access the ²²²Rn fluxes via porewater exchange in the saltmarsh, all ²²²Rn sources and sinks were quantified ( Figure 6 ). According to the ²²²Rn mass balance model (Equation 1), the ²²²Rn flux via porewater exchange was 196 ± 87 (Bq m^-2 d^-1), which accounted for 64% of the total ²²²Rn sources. In contrast, ²²²Rn fluxes (Bq m^-2 d^-1) via influx during flood tide, sediment diffusion, and ²²⁶Ra decay accounted for 22%, 13%, and 0.1% of ²²²Rn sources, respectively. Similar results of sediment diffusion and ²²⁶Ra decay were found in other coastal wetlands (Santos et al., 2019; Chen et al., 2021a; Chen et al., 2021b; Wu et al., 2021). In terms of the ²²²Rn sinks, outflux during the ebb tide, atmospheric evasion, and ²²²Rn decay accounted for 55%, 41%, and 4%, respectively. ²²⁶Ra decay and ²²²Rn decay were minor components in the ²²²Rn mass balance model and were negligible because of the relatively low percentage of ²²²Rn sources or sinks.

Determining the porewater endmember is the key step in estimating the porewater exchange rate, which has been considered a major source of uncertainty (Moore and Arnold, 1996). As the large spatial variation of natural tracer concentrations in aquifers (Peterson et al., 2009), collecting numbers of representative sample can help deal with uncertainty (Correa et al., 2021). In this study, to reduce the uncertainty caused by porewater endmember, we conducted a spatial investigation of porewater samples (n=9, Figure 1 ). Here, the porewater endmember was defined as the difference between the median porewater concentration and the median concentration of surface water at the creek mouth (Taillardat et al., 2018). Hence, the radon-derived porewater exchange rate was estimated to be 5.60 ± 2.78 cm d^-1 using ²²²Rn flux via porewater exchange divided by the porewater ²²²Rn endmember. This porewater exchange rate is in the range (3.4–12 cm d^-1) of other saltmarsh studies, as summarized by Liu et al. (2021).

Porewater exchange drives the transport of carbon species from the sediment to surface water. This can be proven by the significant positive correlations ( Figure 7 ) between ²²²Rn and DIC (R^{2 =} 0.49, p<0.001), DOC (R^{2 =} 0.27, p<0.001), CO₂ (R^{2 =} 0.60, p<0.001), and CH₄ (R^{2 =} 0.62, p<0.001) in surface water and higher concentrations of ²²²Rn and carbon species in porewater than those in surface water ( Figure 7 ). Therefore, we used the difference in median carbon or greenhouse gas values between porewater and surface water as the porewater endmember. By multiplying the porewater exchange rate with the corresponding carbon and greenhouse gas concentrations of porewater endmembers, the porewater-derived carbon and greenhouse gas fluxes (mmol m^-2 d^-1) were estimated to be 447 ± 227 (DIC), 26 ± 20 (DOC), 40 ± 21 (CO₂), and 0.25 ± 0.13 (CH₄) ( Table 5 ). Most of the porewater-derived carbon flux was contributed by DIC (~90%) rather than DOC or greenhouse gases. Similar results for DIC, DOC, and CO₂ were obtained from a saltmarsh in the USA (Correa et al., 2021). However, porewater-derived carbon flux is mainly sourced from DOC in some other saltmarshes (Santos et al., 2019; Chen et al., 2022). This may be due to their dominant species having higher net primary productivity, leading to the rapid accumulation of organic carbon that can be enriched in porewater (Chen et al., 2022).