An in situ OC addition experiment was conducted in August and September 2023 to test the response of pioneer marsh and intertidal flat systems to elevated inputs of OC with different compositions. We choose acetate, a chemically simple organic compound that has been detected at other salt marshes (Hyun et al., 2007; Kostka et al., 2002a) and at a site nearby (Llobet-Brossa et al., 2002), and humic acid (in the form of Pahokee Peat humic acid), a more complex OC source, as a proxy for natural organic matter. The two OC sources were chosen, knowing their thermodynamical difference from previous studies (Gunina and Kuzyakov, 2022), to present a fermentation product (acetate) and a proxy for terrestrial organic matter (in addition to the difference with respect to complexity). Studying both labile and complex OC additions is ecologically relevant as salt marshes receive OC from multiple sources (Howard et al., 2023; Temmink et al., 2022). Eutrophication of coastal waters and/or root exudates can supply readily degradable OC to salt marshes, while increased organic matter load in rivers can deliver more complex OC compounds to salt marshes. The applied OC compounds in our study, therefore, represent environmental scenarios and allows us to investigate how these OC sources influence GHG release under realistic conditions. The injection solutions were prepared in the laboratory prior to use in the field. Solutions of 1 g OC L⁻¹ were prepared with either acetate or humic acid as a carbon source. For the preparation of the solutions, sodium acetate or Pahokee Peat humic acid (obtained from the International Humic Substances Society (IHSS), Table S1) were dissolved in deionized water (Barnstead MQ system, Thermo Fisher Scientific, Germany), the pH was adjusted to 7.07–7.81 and NaCl was added (20 g L⁻¹). The solution for the control only contained NaCl. Additionally, 25 mM bromide (Br⁻) (in the form of NaBr) was added into the carbon and control solutions as an inert tracer in the field. All solutions were purged with nitrogen and stored at 4 °C in the dark until used in the field. Details of the preparation process are provided in Supplement Sect. S1.3.

The in situ experiment was performed in the pioneer marsh (54°00^′43.14^′′ N, 08°50^′12.9^′′ E) and intertidal flat (54°00^′40.45^′′ N, 08°50^′02.27^′′ E) (Fig. 1a), with the same setup in both zones. The assigned plots in the field were selected as visually similar triplicates (spatial triplicates), at a distance of ∼ 5 m between plots of the same treatment and ∼ 10 m between different treatments and the control plots. In the pioneer marsh, plots were placed outside of vegetated areas, i.e., the actual plot area of injection and sampling were free of vegetation although vegetation was present in the vicinity (Fig. S1a, b). For the intertidal flat, no vegetation was present, neither in the surroundings nor inside of the plots (Fig. S1c, d). Each plot consisted of a 16 cm diameter PVC cylinder with both ends open which was pushed 15 cm deep into the sediment (Fig. 1b). Holes (diameter 1 cm) were drilled in the cylinder wall to allow underground water movement. A porewater sampler (Rhizon sampler CSS, 0.12–0.18 µm pore size, Rhizosphere Research, Netherlands) was inserted in the middle of each plot at a depth of 5–10 cm and remained in the sediment over the duration of the experiment. The cylinder reached 5 cm out of the sediment, allowing a gas flux chamber to be mounted gastight for gas measurements. During the experiment, the cylinder stayed in the field. To decrease the influence of disturbances due to the setup of the experiment, we waited at least three days between setting up and the first measurements. Apart from the incubation time for gas sampling, the sediment within the cylinder was exposed to the atmosphere (low tide) or covered with seawater (high tide) (Fig. 1b).

The experiment was conducted similarly in both zones, the pioneer marsh and intertidal flat, using spatial triplicates and comprising four injection cycles (Fig. 1b). One injection cycle consisted of (i) the injection of the anoxic sterile solution (OC or control solution), followed by (ii) the first gas sampling 1.5 h after the injection. Gas sampling was repeated (iii) 24 h after the injection and (iv) 48 h after the injection. After the 48 h gas sampling, (v) porewater was collected. Injection of the solutions and sampling was done during low tide. The solution for the different treatments was slowly injected in step (i) with a bent needle at a depth of approximately 5–10 cm in the middle of each plot into the sediment. The injected solution was filtered (PES filter, pore size 0.22 µm, pre-rinsed with double deionized water) during injection to ensure that the solutions were sterile.

In each injection cycle, the native OC was increased by 12.0 mg C L⁻¹ in the pioneer marsh and 12.6 mg C L⁻¹ in the intertidal flat, assuming an even distribution across the experimental cylinder (calculation in Supplement Sect. S1.4). After one cycle was completed, the system had three days to recover before the next injection. At the end of the four injection cycles, sediment samples were collected (details below). The applied approach allowed us to assess short-term OC process response in minerogenic salt marshes, rather than long-term responses.

Gas sampling was conducted using an opaque, static, non-flow gas chamber made of polypropylene (chamber volume 3000 cm²). The gas chamber was placed on the cylinder and gas samples were collected at 20 min intervals over an incubation period of 1 h using a 50 mL gastight syringe with a three-way valve. For sampling, the chamber gas was gently mixed by pumping the syringe plunger three times before withdrawing 35 mL gas. The first 5 mL were used to flush the attached needle, and the rest was transferred immediately into a pre-evacuated 12 mL Exetainer^® vial (Labco, UK). The samples were measured with a Greenhouse GC equipped with two Pulsed Discharge Detector (PDD) (ThermoFisher Scientific TRACE™ 1310 GC-Analyzer, USA – custom designed by S+H-Analytik) and two column structure (first structure 30 m long, 0.53 mm ID TGBondQ column, second structure 30 m long, 0.53 mm ID Molsieve column (for CH₄) and 30 m long 0.32 mm ID TGBondQ+ column (for CO₂, N₂O)). Calculation for the gas fluxes and cumulative CO₂ emissions are given in Supplement Sect. S1.5.

Porewater samples were collected via the pre-installed porewater sampler. The pH (Mettler Toledo SevenGo, Germany) and salinity (refractometer) were measured in the field. The collected porewater was fixed for DOC (acidification with 2 M HCl), iron speciation (acidification with 1 M HCl), and total sulfide (S(II)_tot) (alkalinization with zinc acetate). Dissolved inorganic carbon (DIC) samples were transferred into vials without headspace immediately after sampling and capped. The remaining porewater was anoxically stored in nitrogen flushed bottles in the dark at 4 °C for Br⁻, SO42- and chloride (Cl⁻) measurements.

At the end of the experiment, sediment samples were collected for geochemical analysis (TOC, solid iron speciation, sulfide species) and microbial analysis. For this, push cores (inner diameter of 2.5 cm, and a length 10 cm) were taken from the middle of each plot at the same positions as the porewater samples and immediately frozen until further analysis. Sediment samples for the molecular biology analysis were stored at -80 °C upon bringing them back to the laboratory.

DOC (as non-purgeable OC) and DIC (as the difference of total carbon and OC) was determined by a TOC analyser (multi N/C 2100S, Analytik Jena GmbH, Germany). To analyse the iron speciation in the porewater, the ferrozine assay (Stookey, 1970) was used. The S(II)_tot was quantified by the Cline assay (Cline, 1969). Sulfate, Br⁻, and Cl⁻ were analysed by an ion chromatograph (Metrohm 930 Compact IC Flex, Switzerland).

For all sediment analyses, sampled cores were thawed in an anoxic glovebox to prevent oxidation (UNIlab plus Glovebox, MBRAUN, Germany), sliced in two depths (0–5 and 5–10 cm), and each depth fraction was homogenized by hand.

To target the poorly and higher crystalline iron phases in the sediment samples, parallel iron extractions were performed under anoxic conditions, adapted from Moeslundi et al. (1994) and Lueder et al. (2020). From each depth (0–5 and 5–10 cm) ∼ 0.2 g wet sediment sample (in triplicates) were added into an Eppendorf tube. To extract poorly crystalline iron minerals, we used an extraction with anoxic 0.5 M HCl (Heron et al., 1994). We expect that poorly crystalline iron (oxyhydr)oxides as well as ferrous iron (Fe(II)) phases such as carbonates and sulfides (e.g., FeCO₃ or FeS) would be extracted by this acidification. One mL of anoxic 0.5 M HCl was added to the sediment, vortexed, and incubated for 2 h in the dark at room temperature in the glovebox. For the extraction of iron minerals of higher crystallinity, targeting more crystalline Fe(II) (except pyrite) and Fe(III) phases (Cornwell and Morse, 1987; Heron et al., 1994), 1 mL of anoxic 6 M HCl was added to the sediment, and the sediment was vortexed and vertically rotated for 24 h under anoxic conditions at room temperature (30 rpm, dark). For both HCl extractions, the samples were then centrifuged (5 min, 13400 rpm), the supernatant was diluted with 1 M HCl, and iron speciation and concentration of the supernatant was determined using the ferrozine assay (Stookey, 1970). Total iron and Fe(II) concentrations of the supernatant were measured directly, and Fe(III) was determined by subtracting Fe(II) from total iron. To obtain the higher crystalline fraction separately, the poorly crystalline fraction (0.5 M HCl extract) was subtracted from the 6 M HCl fraction. We acknowledge that the weaker acid extraction extracted Fe(II) from carbonates and sulfides in addition to iron (oxyhydro)oxides. We therefore used this approach to determine the crystallinity of iron minerals and call it poorly (extracted by 0.5 M HCl) and higher (extracted by 6 M HCl) crystalline iron minerals (and not (oxyhydr)oxides).

To determine the mass of acid volatile sulfide (AVS) in the sediment similar to Burton et al. (2007), ∼ 2 g of wet sediment was weighed in centrifuge tube in a glovebox. A smaller tube filled with anoxic 1 M zinc acetate + anoxic 2 M NaOH (v/v 15:85) was placed into the sediment-filled centrifuge tube. To the sediment, 10 mL anoxic 6 M HCl + 2 mL anoxic 1 M L-ascorbic acid was added and immediately closed. The double tubes were placed in an ultrasonic bath for 30 s, followed by horizontal shaking overnight (150 rpm). The sulfide released from the sediment was captured in the zinc acetate solution and was analysed according to Cline (1969).

To quantify sediment TOC, samples collected in 2023 were dried under anoxic conditions, while those from 2022 were dried under oxic conditions at room temperature until constant weight was reached. Sediments from both years were milled and analysed by a TOC analyser (SoliTOC Cube Elementar, Germany).

Bromide was used in the in situ experiment as an inert tracer to test the washing out of the injection solution from the experimental plot over the sampling time of one injection cycle (48 h). The native Br⁻ concentration was 0.66 ± 0.02 mM in the pioneer marsh and 0.63 ± 0.01 mM in the intertidal flat. Assuming an equal distribution of the injected solution in the experimental cylinder, we expected the Br⁻ concentration to increase by 0.30 mM to a final concentration of 0.99 mM in the plots of the pioneer marsh and by 0.32 mM to a final concentration of 0.97 mM for the plots in the intertidal flat (calculations in Supplement Sect. S1.4 – expected Br⁻ concentration). Throughout a test injection cycle with the same sampling intervals as the experimental injection cycles, the Br⁻ concentration remained above the background level with a gradual decrease over time in both zones (Fig. S3a, b). Overall, after 48 h (one injection cycle), Br⁻ levels in the porewater remained elevated in each cycle for both zones. In the pioneer marsh, an average of 77.5 ± 5.9 % of the expected level of Br⁻ remained in the porewater of the experimental cylinder. The intertidal flat had a slightly lower average residual fraction of 72.1 ± 4.4 % of the expected Br⁻ level across all injection cycles (Fig. 3). Here, residual fraction is defined as the ratio between the Br⁻ concentration measured in the porewater 48 h post injection and the expected total Br⁻ concentration in an experimental cylinder (Eq. 1); details in Sect. S1.4). The expected total Br⁻ concentration includes both the native Br⁻ and the added Br⁻ during the experiment (expected Br⁻) after accounting for dilution in the sediment. Details on the calculation of the Br⁻ residual fraction are provided in Supplement Sect. S1.4. 1Brconcentration at the end of an injection cycle-Brexpected-=residual fraction The residual fraction of DOC is defined in the same way as for Br⁻, representing the proportion of measured DOC after 48 h to the expected DOC (native DOC + added acetate/humic acid) and was calculated analogously to Br⁻, with DOC concentrations used instead (Eq. 1) and Sect. S1.4. Comparing the average residual fraction of DOC and Br⁻ (Fig. 3) across all injection cycles, the DOC fraction was significantly lower in both the pioneer marsh and intertidal flat (Br⁻ vs. acetate and Br⁻ vs. humic acid) (p≤0.001, Table S5). In the pioneer marsh, 40.0 ± 1.2 % of the injected DOC in the acetate treatment remained, on average across all injection cycles, while the corresponding value for the humic acid treatment was 52.9 ± 4.5 %. In the intertidal flat, relatively less carbon remained, with a smaller difference between the carbon sources. Here, the mean residual fraction was 38.2 ± 4.8 % for the acetate treatment and 37.3 ± 3.6 % for the humic acid treatment.

Figure 4 presents the CO₂ release for each injection cycle with the individual CO₂ fluxes 1.5, 24, and 48 h post injection (Fig. 4a) and the cumulative CO₂ emissions (Fig. 4b) in the pioneer marsh. For all four injection cycles, the CO₂ fluxes of the acetate treated plots exceeded the CO₂ fluxes of the humic acid and control plots (Fig. 4a). For the first injection cycle, the acetate treatment was significantly higher compared to the humic acid treatment (p<0.05, Table S6) and slightly above the threshold for statistical significance (p=0.08, Table S6) compared to the control. In the following injection cycle, the CO₂ fluxes of the acetate treatment were also higher compared to the humic acid and control plots but not significantly (p>0.05). Hence, the acetate treated plot consistently exhibited the highest fluxes. The difference between the humic acid and control plots was statistically negligible (p>0.05). Similarly, the cumulative CO₂ emissions from the acetate treated plots were the highest while the emissions from the humic acid and control plots were in a similar range for all four injection cycles (Fig. 4b). The CO₂ emitted from the acetate treated plots was up to 83.2 ± 53.7 % higher than for the humic acid treated plots. Similarly, the emissions of the acetate plots were up to 47.4 ± 36.4 % higher relative to the control plots. No statistical differences were measured between the cumulative CO₂ emission of the acetate treated plots and the control or humic acid treatment. Overall, these differences were smaller than those seen at individual CO₂ fluxes at specific time points (Fig. 4a), likely due to high variability in fluxes that resulted in variable cumulative CO₂ emissions. In all treatments and the control, no CH₄ as a flux was detected (lower than detection limit (0.28 ppm; Table S3).

The CO₂ fluxes showed a positive correlation with air temperature and a moderate positive correlation with the tidal cycle in the pioneer marsh (Table S7). Specifically, CO₂ fluxes were found to be higher as the spring tide receded. Lower CO₂ fluxes were measured during the first and third injection cycles, which occurred close to spring tides.

Further differences between the treatments and control were observed in the DOC concentrations (Fig. S4) and the residual fraction of DOC (Fig. 3a) in the pioneer marsh. Over all four injection cycles, the average DOC concentration in the acetate treated plots did not significantly differ from the control (p>0.05), while the DOC concentrations of the humic acid treated plots were significantly higher than the concentrations of acetate treated plots and the control plots (p<0.05, Table S8). These differences are consistent with the residual fraction of DOC (Fig. 3a). Across all four injection cycles, the average residual fraction in the humic acid treatment was significantly higher compared to the acetate treatment but significantly lower than the Br⁻ residual fraction (p<0.002, p<0.001 respectively, Table S5). No difference in the TOC content was found between treatments and control in both depths. The content ranged from 0.9 ± 0.1 % to 1.2 ± 0.1 % in the upper 5 cm, with a slight decrease at lower depth (Table S9).

In addition to the gas fluxes, we also measured DIC and pH of the porewater for each treatment and control for each injection cycle (Fig. S5). As the differences in the CO₂ fluxes were more pronounced and our focus was on GHG release, we only present and later discuss these data.

In the pioneer marsh, acetate treated plots had significantly higher aqueous Fe(II) concentrations compared to humic acid and control plots when considering the average concentration across all injection cycles (acetate vs. humic acid p=0.007 and vs. control p=0.002, Table S10). Although the differences were not always statistically significant for the individual cycles, the Fe(II) concentration in the porewater of the acetate treated plots was consistently higher (Fig. 5a): 1st cycle 124.52 ± 9.44 µM, 2nd cycle 114.20 ± 19.98 µM, 3rd cycle 114.54 ± 23.80 µM, and 4th cycle 88.08 ± 24.58 µM. Humic acid treated plots and the control plots showed similar aqueous Fe(II) concentrations. For humic acid treated plots the aqueous Fe(II) concentration ranged between 53.36 ± 9.05 µM (2nd cycle) and 79.96 ± 22.16 µM (1st cycle), and for the control between 55.69 ± 8.04 (2nd cycle) and 96.01 ± 36.61 µM (1st cycle) (Fig. 5a).

A similar trend can be seen in the solid phase for the HCl extractable Fe(II), for both extractions approaches (0.5 M and 6 M HCl) (Fig. 5b). Comparing the poorly crystalline Fe(II) fraction, the acetate treatment had the highest Fe(II) content compared to the humic acid treatment and the control at both depths. Acetate treated plots showed an Fe(II) content of 57.87 ± 3.44 µmol g⁻¹ sediment in the upper 5 cm and 34.91 ± 5.45 µmol g⁻¹ sediment from 5–10 cm. Humic acid and control plots had similar levels of Fe(II): for the humic acid plots, the content in the upper 5 cm was 43.74 ± 4.18 µmol g⁻¹ sediment and from 5–10 cm, it was 29.13 ± 2.12 µmol g⁻¹ sediment. The control plots showed 46.28 ± 4.32 µmol g⁻¹ sediment in the upper 5 cm and 23.93 ± 1.75 µmol g⁻¹ sediment between 5–10 cm. We observed the same trend for the higher crystalline Fe(II) content. The acetate treatment had the highest contents with 42.04 ± 3.12 µmol g⁻¹ sediment (0–5 cm) and 36.38 ± 5.79 µmol g⁻¹ sediment compared with humic acid or control plots. For both depth and crystallinities, the acetate treatment showed significantly the highest content, in nearly all comparisons to the humic acid or control plots (Table S11). Except for poorly crystalline Fe(II) at 5–10 cm depth, humic acid and control plots did not differ significantly (Table S11).

No consistent difference was detected in S(II)_tot in the porewater of the pioneer marsh (Fig. S6a). The S(II)_tot concentrations were in the same range: acetate from 3.16 ± 0.01 µM (4th cycle) to 5.59 ± 1.21 µM (1st cycle), humic acid from 3.0 ± 0.11 µM (4th cycle) to 6.34 ± 1.56 µM (1st cycle) and control from 3.63 ± 0.34 µM (1st cycle) to 4.39 ± 1.13 µM (4th cycle). Also, statistical analysis did not reveal a difference between the different treatments and control S(II)_tot averaged over all cycles (p>0.05). Similarly, AVS measurements of the solid sulfide species showed no difference between treatments and the control (Fig. S6b). In the upper 5 cm, contents were similar (acetate: 6.13 ± 1.40 µmol g⁻¹ sediment, humic acid 3.89 ± 1.19 µmol g⁻¹, and control 7.37 ± 1.76 µmol g⁻¹ sediment). Similar contents were measured at 5–10 cm depth, with no coherent trend between the layers.

The impact of the added OC on the bacterial community was analysed by qPCR. The analyses were based on DNA (microbial abundance) and RNA (metabolically active microorganisms). The results show the gene copies of both treatments normalized to the control as log₂ fold change (log₂ FC) (Fig. 6). Statistics are based on the absolute gene copy numbers (Fig. S7). For the acetate treated plots a significantly higher bacterial 16S rRNA gene copy number (DNA- and RNA-based) was measured compared to the control plots across all analysed depths (Fig. 6a; p<0.05). In comparison to the control, DNA-based bacterial 16S rRNA gene copies increased by a factor of 0.4 ±  0.07 (log₂ FC) under acetate treatment and their potential activity, indicated by RNA, increased by 1.58 ± 1.46 log₂ FC in the upper 5 cm. This remained similar at the depth of 5–10 cm, with an increase in DNA-based 16S rRNA gene copies by 0.43 ± 0.09 log₂ FC and an increase in activity (RNA) by 5.09 ± 1.86 log₂ FC. For the comparison between plots amended with humic acid and the control, variations were observed but no significant differences were measured (p>0.05). Microbial activity (RNA) of Geobacter spp., were higher in the acetate and humic acid treatment compared to the control in the upper 5 cm, however, slightly over the significance criterion (Fig. 6b; p=0.051, p=0.057, respectively). For the acetate treated plots, the activity (RNA-based) of Geobacter spp. was higher compared to the control by a factor of 0.98 ± 0.49. The humic acid treatment also had upregulated activity by a factor of 0.72 ± 0.31. The higher microbial activity of Geobacter spp. remained significantly enhanced (p=0.008) for the acetate treatment compared to the control at the lower depth (RNA-based: log₂ FC: 0.66 ± 0.15), too. The addition of OC did not affect the microbial activity (RNA) of dsrA genes at both depths (Fig. 6c) in the pioneer marsh. However, the abundance (DNA) of dsrA genes in the upper layer was significantly higher for both treatments (acetate p=0.006, humic acid p=0.015). Absolute gene copy numbers are given in Supplement, Fig. S7 and statistical details in Table S12.

Figure 7a presents the CO₂ release from the intertidal flat over three injection cycles 1.5, 24, and 48 h post injection. Acetate treated plots released the highest CO₂ in all three injection cycles compared to the humic acid and the control plots. Similar to the pioneer marsh, no strong differences were observed between humic acid treated plots and the control plots. Consistently, the maximum cumulative CO₂ emissions were observed in the acetate treated plots (Fig. 7b). Due to nonlinearity of CO₂ release over the incubation time of gas sampling, some data points are missing; therefore, statistical comparison of CO₂ release between treatments and the control was not done. Nevertheless, plots amended with acetate consistently showed higher CO₂ releases across all injection cycles. Methane was not detected in the fluxes of any treatment in the intertidal flat plots (lower than detection limit (0.28 ppm); Table S3). Similar to the pioneer marsh, we focus here on CO₂ data although DIC and pH were measured (Fig. S8).

No difference in the DOC concentrations between the treatments and the control were measured (Fig. S9). Similarly, no difference was measured between the residual fraction (recovery of injected DOC) between acetate and humic acid treatments (Fig. 3b). The TOC content among the treatments and the control was also in the same range (0.4 %–0.5 %) (Table S9).

In the intertidal flat, aqueous Fe(II) concentrations ranged from 9.25 ± 0.21 to 24.82 ± 4.50 µM in both treatments and the control, with no significant differences (p>0.05) (Fig. S10a). Additionally, no difference between the treatments and the control was detected in the solid phase for both crystallinities (0.5 and 6 M HCl extraction) and depths. In the upper 5 cm, the content of the poorly crystalline Fe(II) ranged from 18.28 ± 1.28 to 19.60 ± 0.88 µmol g⁻¹ sediment among all treatments and control, while in the deeper layer, the range was from 18.11 ± 0.71 to 24.59 ± 1.22 µmol g⁻¹ sediment. The content of the higher crystalline Fe(II) ranged from 11.68 ± 0.99 to 20.06 ± 3.64 µmol g⁻¹ sediment in the upper 5 cm and from 12.98 ± 1.03 to 17.50 ± 2.75 µmol g⁻¹ sediment at a depth of 5–10 cm (Fig. S10b).

For porewater S(II)_tot, no large differences were measured between the treatments and control in the intertidal flat. The concentrations of both treatments and the control were in a similar range over all injection cycles, 0.62 ± 0.01 to 1.73 ± 0.67 µM S(II)_tot (Fig. S11). In contrast, the AVS in the solid phase exhibited a significant difference between the treatments and the control (Fig. 8). Higher AVS contents were measured in the acetate plots. This difference was strongly pronounced in the upper 5 cm in the acetate plots relative to the setups amended with humic acid and the control (p<0.001 and p=0.007, respectively, Table S13). AVS content in the upper 5 cm was 5.38 ± 0.57 µmol g⁻¹ sediment from the acetate plots, 2.03 ± 0.28 µmol g⁻¹ sediment from the humic acid plots, and 3.40 ± 0.35 µmol g⁻¹ sediment from the control. In the deeper layer (5–10 cm), the differences between treatments and control were statistically negligible (p>0.05).

In the intertidal flat, an increase in DNA- and RNA-based gene copy numbers of the bacterial 16S rRNA gene was detected in the acetate treatment compared to the control at both depths (0–5 and 5–10 cm) (Fig. S12a). This increase was significant for DNA and RNA in the upper layer and remained significant for RNA in the lower layer (p=0.008, p=0.01, and p=0.008, respectively). The acetate treatment showed a 3.62 ± 2.41 log₂ FC increase in the metabolic activity (RNA) of the total microbial community compared to the control in the upper layer and an increase of 2.73 ± 1.41 log₂ FC from 5–10 cm. For Geobacter spp. (Fig. S12b) the RNA-based gene copies were significantly higher compared to the control (p<0.001) by a factor of 0.6 ± 0.10 in the upper 5 cm. No significant increase in the RNA-based copy numbers of dsrA genes in both depths (Fig. S12c) were observed; however, we detected slightly higher RNA-based dsrA gene copies in the acetate treatments (0.33 ± 0.06 log₂ FC) compared to the control in the upper layer. Absolute gene copy numbers are presented in Fig. S13 and statistical details in Table S14.

Bromide, as an inert tracer, was injected along with the OC/control solution into each experimental cylinder for each injection cycle. This allowed us to follow the distribution and retention time of the injected solution over one injection cycle. The Br⁻ concentration over each injection cycle was higher than the background Br⁻ concentration (Fig. 3, S3a, b), indicating that the injected solution was partially retained within the experimental cylinder over one injection cycle. The observed decrease in Br⁻ concentration over one injection cycle, more pronounced in the intertidal flat (Fig. S3b), was likely due to flushing out by tidal water and belowground water movement. We observed a slightly lower retention of Br⁻ in the intertidal flats compared to the pioneer zone. We attribute this difference to the higher sand fraction in the intertidal flats, which likely increased permeability and led to stronger tidal flushing of the injected solution as well as subject to greater tidal inundation.

As the Br⁻ concentrations and the respective calculated residual fractions were similar for each cycle (Fig. 3), we infer that there was no residue of Br⁻ and thus no injected OC was carried over between cycles. Furthermore, the retained Br⁻ was similar between the plots of a treatment within one zone, indicating similar belowground conditions within one zone and thereby validating the experimental setup. By placing the experimental plots outside of vegetated areas in the pioneer marsh, we tried to avoid geochemical influence on the sediment by plants. However, it is possible that benthic organisms such as worms may have influenced the biogeochemistry in both the pioneer marsh and intertidal flat, causing some variability.

To test our hypothesis that that microbially mediated CO₂ release is OC limited in the pioneer marsh, we injected two different OC sources into the sediment and monitored the subsequent release of CO₂ 1.5, 24, and 48 h post injection. We observed that acetate treated plots emitted the highest CO₂ throughout the experiment (Fig. 4). No difference in the CO₂ release was measured between the humic acid treatment and the control despite the additional availability of OC. This is supported by the work of Gunina and Kuzyakov (2022) who showed that reduced and complex organics in soils are predominantly thermodynamically preserved due to insufficient energy yield upon decomposition. Humic acid, as complex OC, may be thus preserved in our study, as reflected in higher DOC concentrations. In contrast, acetate, with a simpler chemical structure, is favourable for microbial decomposition even under reducing conditions (Boye et al., 2017; LaRowe and Van Cappellen, 2011) and thus could be readily utilized and oxidized to CO₂. Notably, plots treated with humic acid received the same OC concentrations as the acetate plots, indicating that OC composition is crucial for the CO₂ release from minerogenic pioneer marshes. Based on the geochemistry at the field site, which suggested that the ecosystem is limited by the availability of OC, the results from the in situ experiment further support our hypothesis, while simultaneously highlighting the importance of OC composition. This is contrasting to other terrestrial wetlands, which are characterized by a depletion of TEAs (Schlesinger and Bernhardt, 2013b).

The increased turnover of acetate to CO₂ relative to humic acid is also evidenced in the DOC concentrations (Fig. S4) and the retention fraction of both DOC sources (Fig. 3a). Although the same mass of OC was added to both treatments, the DOC concentrations in the acetate plots were significantly lower than that in the humic acid plots, suggesting that more acetate was utilized, reflected in higher CO₂ release. However, this result does not explain the lower retention of DOC relative to the retention of Br⁻ from the humic acid plots (Fig. 3a). This suggests that in addition to flushing out, adsorption likely occurred. Previous studies have shown that minerals within the subsurface adsorb organic compounds (Kahle et al., 2003; Kleber et al., 2021). This adsorption may be preferential, favouring more aromatic and high molecular weight compounds, particularly when metal (oxyhydr)oxides and clay minerals are present (Kaiser and Guggenberger, 2000; Lv et al., 2016; Voggenreiter et al., 2024). Based on the results of our study and a previous study conducted at the same site (Kubeneck et al., 2024), both iron(oxyhdyr)oxides and clay minerals are present. As adsorption suppresses the decomposition of OC (Kleber et al., 2021), we speculate that the lower retention fraction of humic acid compared to Br⁻ was primarily due to adsorption onto the sediment rather than decomposition. This is consistent with the CO₂ fluxes: humic acid treated plots showed lower retention fractions compared to the Br⁻ retention fractions, however, the CO₂ fluxes were comparable to the control. This suggests that little to no decomposition occurred and that adsorption was the dominant process. It is worth noting that anoxic decomposition of humic acid is generally possible but the turnover time would have exceeded the duration of the experiment (Lipczynska-Kochany, 2018). These results highlight that it is not only the presence of OC that affects short-term OC release from coastal wetlands; the composition of OC is the primary determining factor.

Furthermore, it is important to note that the OC concentrations used in this experiment are higher than those expected for naturally occurring OC inputs, such as root exudates, which are typically released at lower concentrations with a continuous input. Thus, upscaling the enhanced CO₂ fluxes measured in our study might result in overestimation of CO₂ release from minerogenic salt marshes. Our findings rather reveal, on a process level, that the addition of labile OC stimulates microbially mediated CO₂ release. Enhanced CO₂ release from the acetate amended plots was measured at nearly all sampling time points (1.5, 24, and 48 h) without a clear trend, while the concentration of the inert tracer showed a slight decrease over the same period (Fig. S3) – indicating slight dilution and flushing of the injected OC. This suggest that the elevated CO₂ release was driven by enhanced availability of labile OC independently of its concentration. These findings allow us to generalize that the system is likely limited by labile OC availability, regardless of the concentration; however, further work should quantify how the magnitude of CO₂ promotion corresponds to OC concentration, particularly under low, naturally sustained OC input rates. In conclusion, we can reliably predict the direction of increased OC inputs to minerogenic salt marshes, but further studies are needed to predict the specific long-term magnitude of changes in the carbon cycle in these ecosystems.

This section combines the observed effect of OC input on the geochemistry of the porewater and sediment with the microbial growth and metabolic activity and links them to the CO₂ release from the pioneer marsh. The bacterial 16S rRNA gene copy number, an indicator of the total bacterial abundance, was significantly higher in the acetate treatment compared to the control for both DNA and RNA at both depths. This suggests greater bacterial abundance and metabolic activity, which likely led to higher CO₂ fluxes from these plots. No significant difference was noticeable in the 16S rRNA gene copy number between plots amended with humic acid and the control. Overall, the 16S rRNA gene copy numbers (DNA) are in the range of other studies from coastal wetlands and marine sediment (Petro et al., 2019; Zhou et al., 2017).

The enhanced metabolic activity in the acetate treated plots is further reflected in the elevated aqueous and solid phase Fe(II). The acetate treatment showed higher aqueous Fe(II) concentrations in all four injection cycles, while no significant difference was observed between the humic acid and control plots. In the solid phase, the Fe(II) content was the highest in the acetate treatment. Thus, the higher expression of Geobacter spp. in the acetate treated plots corresponds well to our geochemical observations. Geobacter spp. have been shown to use acetate as a carbon source to gain energy (Coates et al., 1996). The higher Fe(II) levels observed in both aqueous and solid phase of acetate treated plots along with elevated Geobacter spp. gene copies indicate that Fe(III) reduction was stimulated by increased availability of labile OC.

Ferric iron and SO42- reduction have been reported as OC decomposition processes in salt marshes (Hyun et al., 2007; Kostka et al., 2002a; Lowe et al., 2000). Sulfide concentrations in the porewater as well as in the solid phase gave evidence that SO42- reduction occurred; however, no differences between the treatments and controls were seen. The functional gene analysis provided further evidence for this, as sulfate-reducing bacteria (SRB) were present (absolute gene copy numbers in Supplement Fig. S7); however, none of the treatments led to an increase in their metabolic activity compared to the control (Fig. 6). Thus, we speculate that the higher CO₂ release from the acetate treatment was mainly driven by the enhanced Fe(III) reduction and not by SO42- reduction. Our finding that acetate is utilized follows the conventional thermodynamic sequence of iron reduction being more favorable than SO42- reduction (Schlesinger and Bernhardt, 2013b) – even if the concentrations and availability of SO42- were much higher.

Adding OC to the intertidal flat, a zone more influenced by tides than the pioneer marsh, resulted in CO₂ trends similar to the pioneer marsh. Acetate addition led to a noticeable increase in CO₂ fluxes. In contrast, the CO₂ fluxes from the humic acid and control plots were similar. This supports our hypothesis that the electron donor limits CO₂ release from the ecosystem. Moreover, it highlights that irrespective of tidal influence, the system is limited by the availability and composition of OC.

Total microbial 16S rRNA gene copies were significantly higher in the acetate treated intertidal flat plots compared to the control at both depths (RNA-based), whereas plots amended with humic acid showed no significant difference from the control plots (Fig. S12a). In contrast to the significant increase in Geobacter spp. gene copies, no difference in the aqueous Fe(II) or in the Fe(II) content in the upper sediment layer (0–5 cm) was measured for the acetate treatment, except for a higher Fe(II) content in the deeper sediment layer. Conversely, we measured higher AVS contents in the upper layer for the acetate treatment, but this is not reflected in higher dsrA copies numbers. Based on these mixed results, we consider two hypotheses: (i) acetate promoted increased Fe(III) reduction, which is supported by higher gene copies of Geobacter spp. However, this is not clearly reflected in the Fe(II) data. Furthermore, with increased Fe(III) reduction and a constant SO42- reduction rate, we would have expected a depletion of S(II)_tot in the porewater of the acetate treatment due to iron-sulfur mineral formation; however, this was not observed. Thus, we consider a second hypothesis that (ii) acetate promoted SO42- reduction, supported by increased AVS content in the acetate treatment in the upper sediment layer. In contrast, the number of dsrA gene copies was not higher in the acetate treatment. Although a small increase (RNA-based) in the gene copies of dsrA was observed in the upper layer, this was not statistically significant. Additionally, no differences were observed in the S(II)_tot concentrations, which were generally low and should therefore be interpreted with caution. Based on our data, we cannot clearly reject either hypothesis. We therefore suggest Fe(III) and SO42- reduction both lead to higher CO₂ release, stimulated by higher supply of labile OC in the intertidal flat.