Before rewetting, surface water samples for nutrients (NO3-, NO2-, NH4+, and PO43-) and chlorophyll a were collected from the inner Kubitzer Bodden (referred to in the following as “inner bay”) at station D1 (Fig. 2a) and irregularly at a second station right in front of the now-removed dike section, which was abandoned after rewetting and therefore merged with station D1. Environmental variables (water temperature, dissolved oxygen (O2), and salinity) were measured on site. Both stations were reached from the land, and sampling was conducted monthly from June to November 2019, except for the month of August.

After rewetting, surface water samples were collected with a small boat, and the sampled variables were extended for the concentrations of GHGs (CO2, CH4, and N2O) and dissolved organic carbon (DOC). The first sampling took place 1 week after the dike removal. Sampling was continued over 1 year (25 sampling dates until December 2020) at weekly (December 2019 to January 2020) or biweekly (February 2020 to September 2020, except for August) intervals. From October 2020 to December 2020, sampling was conducted monthly. In the inner bay, three stations (D1, D3, and D14) were sampled, and in the flooded peatland, six stations (D4, D5, D11, D12, D15, and BTD8) were sampled (Fig. 2a). The inner bay station D14 was sampled from March 2020 onwards. DOC sampling started in April 2020. For the air–sea gas exchange calculation, data from station D10_1, located in the channel, were also included.

Moreover, surface water samples for the analysis of GHG concentrations (CO2, CH4, and N2O) were sampled at eight stations along a transect (Fig. 2b). This sampling was carried out simultaneously with the sampling described in Sect. 2.2.2 to link the GHG air–sea exchange calculations based on surface water samples with chamber-based flux measurements.

Surface water temperature, O2, and salinity were measured directly in the field using a Hach HQ40D multimeter (Hach Lange GmbH, Germany) equipped with two outdoor electrodes (LDO10105 and CDC40105). Depending on the prevailing water depth, additional measurements were conducted in the peatland 15 cm above the soil surface (excluding the ditches) on 22 of the 25 sampling dates. The precision of the electrodes was ±0.3 ∘C, ±0.8 %, and ±0.1 for temperature, O2 saturation, and salinity, respectively.

Surface water samples were taken using a horizontal 7 L Niskin bottle to sample the upper 20 cm of the water column. These included 250 mL subsamples for CH4/N2O analysis (bottles capped with butyl rubber stoppers and crimp-sealed), analysis of the CO2 system (one bottle each for total CO2 (CT), total alkalinity (AT), and pH) and 15 mL subsamples for the analysis of nutrients and DOC. Water for chlorophyll a determination was taken using 3 L canisters.

In the laboratory, CH4/N2O and CO2 samples were poisoned with 500 and 200 µL of saturated HgCl2, respectively, and stored in the dark at 4 ∘C until analysis. Subsamples for nutrients and DOC were filtered in the field with pre-combusted (450 ∘C for 4 h) 0.7 µm glass fiber filters (GF/F, Whatman^®) and stored at -20 ∘C. Samples for chlorophyll a were filtered in the laboratory with non-combusted 0.7 µm glass fiber filters (GF/F, Whatman^®) and also stored at -20 ∘C.

Starting in June 2019, nearly 6 months before rewetting, GHG exchange was regularly measured using dynamic closed chambers (Livingston and Hutchinson, 1995) along a transect representing a soil humidity gradient (Fig. 2b). The measurements were conducted twice a month for a total of 11 sampling days at six peatland stations and two additional stations in the north–south-oriented main ditch. Each station was sampled up to eight times per sampling day, resulting in overall 418 CO2 and 184 CH4 pre-rewetting flux measurements.

For each measurement, the chambers were placed on permanently installed collars and connected through an airtight seal, with a closure period between 180 and 300 s. Between the measurements, chambers were lifted to vent them until atmospheric GHG concentrations were reached. To ensure coverage of photosynthetic and respiration activity, CO2 measurements on the terrestrial peatland were conducted using opaque and transparent chambers for NEE (net ecosystem exchange) and Reco (ecosystem respiration) determination, respectively. To cover a broad spectrum of solar radiation, two additional measurements were conducted with cloth-covered transparent chambers, resulting in a reduced photosynthetically active photon flux density (PPFD). The GHG concentrations of the ditch stations were determined in three consecutive measurements with floating opaque chambers placed on the water surface. Changes in GHG concentrations in the chamber headspace were measured using a portable laser-based analyzer (Picarro GasScouter G4301, Santa Clara, USA, LI-820, LI-COR Biosciences, Lincoln, USA, and an ultraportable greenhouse gas analyzer (UGGA), Los Gatos Research Inc., Mountain View, CA, USA).

After rewetting, the stations along the transect covered a gradient of ground elevations, including stations that fell dry at low water levels and stations that remained permanently flooded. Atmospheric GHG fluxes were measured twice a month using floating opaque chambers positioned above the same sampling locations of the flooded peatland. Since the flooding caused most plants to die, and almost all measurement locations were covered by water during the study period, we reduced the number of NEE measurements with transparent chambers to stations and days with a low water table. Approximately six measurements per station were made during 23 sampling days between December 2019 and December 2020, with a total of 698 CO2 and 482 CH4 fluxes determined during the post-rewetting year.

Nutrient analyses were carried out according to standard photometric methods (Grasshoff et al., 2009) by using a continuous segmented flow analyzer (SEAL Analytical QuAAtro, SEAL Analytical GmbH, Norderstedt, Germany). Detection limits were 0.2 µM for NO3-, 0.05 µM for NO2-, 0.5 µM for NH4+, and 0.1 µM for PO43-. Measurements of the nutrient concentrations were partly below the detection limit for the peatland, the inner bay, and the central bay (10.12754/data-2022-0003, Pönisch and Breznikar, 2022). For such measurements below the detection limit, using the actual values of these measurements is recommended (e.g., Fiedler et al., 2022) to achieve a robust statistical analysis. Since these data were not available, we decided to use randomly generated values between 0 and the respective detection limit with a uniform distribution for these measurements.

Chlorophyll a was extracted from glass fiber filters (GF/F, Whatman^®) by incubation with 96 % ethanol for 3 h and analyzed afterwards by using a fluorometer (TURNER 10-AU-005, Turner Designs, Inc., San José, USA) at 670 nm, according to Wasmund et al. (2006). DOC was analyzed after high-temperature combustion, using a multi 2100S instrument (Analytik Jena GmbH, Jena, Germany) and detected by non-dispersive infrared spectrometry according to ISO 20236, ISO 8245 I, and EN 1484.

Coastal nutrient data (NO3-, NO2-, NH4+, and PO43- concentrations) from a monitoring station in the Kubitzer Bodden (central bay; Fig. 1b), ∼ 15 km away from the study area, were obtained as reference. Monitoring data from 2016 to 2020 were included. These data were used (1) to compare them with nutrient concentrations from the inner bay before and after rewetting to detect potentially higher concentrations resulting from nutrient leaching within the peatland and a subsequent export into the inner bay and (2) to calculate the total possible export out of the peatland (Sect. 2.4.3) by using the monitoring station as a second, unaffected, endmember besides the inner bay, which is by contrast potentially affected by the rewetting. Due to transformations and potential losses along the way to the monitoring station, especially of the nitrogen species, the calculated total possible exports are meant to be maximum estimates.

To calculate the bulk exchanges of dissolved inorganic nitrogen (DIN-N) and PO4-P between the flooded peatland and the inner bay/central bay, the water level was transformed to water volume by creating a hypsographic curve with increments of 0.1 m and a resolution of 1 × 1 m (Fig. A3). Water level data from a nearby monitoring station (Barhöft; 54.43∘ N, 13.03∘ E) and topography data with a resolution of 1 × 1 m were obtained from the Wasserstraßen- und Schifffahrtsamt Ostsee (WSA) and the Landesamt für innere Verwaltung MV, respectively. To ensure that the water level data of the monitoring station were valid for the peatland, the water level data of the latter, measured between August and December 2020, were compared with the data from the monitoring station, which showed a strong correlation (rs=0.95; p<0.001; 15 min intervals; data not shown).

A water level of -1.6 m a.s.l., as the lowest recorded water level within the last 25 years, was used as the starting point to derive the cumulative water volumes of the peatland. The water volumes were then assigned to the corresponding water levels to finally calculate the water volume changes (Q; in m3 s-1), according to Eq. (1): 1Qt=dVdt, where V is the water volume and t the time. Positive volume changes (Q>0) indicate an inflow of water into the peatland, and vice versa. For each season, the mean inflow (Qin) and outflow (Qout) volumes were calculated, according to Eqs. (2) and (3): 2Qin=1ΔT∫tt+ΔTQpositivedtforQ>03Qout=1ΔT∫tt+ΔTQnegativedtforQ<0, where ΔT denotes the season length. Note that Qout is negative. Seasonal mean values of nutrient concentrations (DIN and PO43-) were calculated and converted from micromoles per liter (µM) to kilograms per cubic meter (kg m-3) by using the molecular masses of the basic elements N and P to derive DIN-N and PO4-P. After the conversion, nutrient masses of the peatland (cpeatland) and the inner bay (cIB) vs. peatland and central bay (cCB), respectively, were multiplied by Qout and Qin and integrated to calculate the net nutrient transport (NNT, in tonnes, equivalent to megagrams), according to Eqs. (4) and (5): 4NNT=∫tt+ΔTQincIBdt+∫tt+ΔTQoutcpeatlanddt5NNT=∫tt+ΔTQincCBdt+∫tt+ΔTQoutcpeatlanddt.

Negative values indicate a net nutrient export from the peatland into the inner bay/central bay, and positive values display a net nutrient import into the peatland. Uncertainty ranges for the seasonal NNT (uNNT, as the 95 % confidence level) were calculated by using an error propagation, according to Eq. (6): 6uNNT=cbaydtuQin2+cpeatdtuQout2+Qoutdtucpeat2+Qindtucbay2‾, where terms with u denote the respective 95 % confidence level. To derive the annual uncertainty range of the NNT, all seasonal errors were added up.

The inorganic carbon system was determined by analyzing the total CO2 (CT), total alkalinity (AT), and pH of the water samples. CT was measured with an automated infrared inorganic carbon analyzer (AIRICA, serial no. 027; MARIANDA, Kiel, Germany). The system acidifies a discrete sample volume (phosphoric acid, 10 %), whereby the inorganic carbon species of CT are shifted to CO2(g). A carrier gas stream (99.999 % N2) transfers the gaseous components to a Peltier device and a Nafion^® drying tube (Perma Pure Nafion^®, ANSYCO GmbH, Karlsruhe, Germany) to remove water residues. The produced CO2(g) is detected by an infrared detector (LI-7000; LI-COR Environmental GmbH, Bad Homburg, Germany). Certified reference materials (CRMs; Scripps Institution of Oceanography, University of California, San Diego, USA; Dickson et al., 2003) were used for calibration. Triplicate measurements were conducted for each sample, and a precision of ±5 µmol kg-1 was achieved.

AT was measured by potentiometric titration (glass electrode type LL, Electrode Plus, 6.0262.100; Metrohm AG, Filderstadt, Germany) in the open-cell configuration, after Dickson et al. (2007). The system was calibrated with the same CRMs as used for CT and resulted in the same precision.

The pH was analyzed spectrophotometrically using the pH-sensitive indicator dye metacresol purple (mCP, 2 mM; CONTROS Systems & Solutions GmbH, Kiel, Germany). The measurement principle and instrumental setup are described elsewhere (Dickson et al., 2007; Carter et al., 2013). In brief, absorption was measured using the Agilent 8453 UV-visible spectroscopy system (Agilent Technologies, Waldbronn, Germany); pH parameterization for brackish water was calculated following Müller and Rehder (2018). Quality control was performed by measuring buffer solutions (salinity of 20) prepared according to Müller et al. (2018). An external buffer solution with a salinity of 35 (Scripps Institution of Oceanography, University of California, San Diego, USA) was additionally used. All pH values are reported given on the total scale (pHT).

The CO2 partial pressure in the water phase (pCO2), the value of which was required for the CO2 air–water flux calculations (Sect. 2.5.3), was calculated from CT and pH using the R packages seacarb (Gattuso et al., 2019), with K1 and K2 from Millero (2010), Ks from Dickson (1990), and Kf from Dickson and Riley (1979). CT and pH were preferred because non-oceanic components, in particularly organic acid-base systems, contribute significantly to AT (Kuliński et al., 2014). AT was also calculated from CT and pH and the values compared with measured values, thus revealing the magnitude of the contributions of those components to AT.

Dissolved CH4 and N2O concentrations were determined by gas chromatography on an Agilent 7890B instrument (Agilent Technologies, Santa Clara, USA) coupled to a flame ionization detector (FID) and an electron capture detector (ECD). A purge-and-trap technique, explained in detail in Sabbaghzadeh et al. (2021), was used. In brief, a helium gas stream was used to purge 10 mL of seawater to extract volatile compounds. The gas stream passed through a purifier (VICI – Valco Instruments Co. Inc., Houston, USA) and was dried using a Nafion^® tube (Perma Pure Nafion^®, ANSYCO GmbH, Karlsruhe, Germany) and a SICAPENT^® tube (Merck KGaA, Darmstadt, Germany). The relevant compounds were enriched by cryofocusing on a trap filled with HayeSep D^® (CS – Chromatographie Service GmbH, Langerwehe, Germany) maintained at -120 ∘C using an ethanol/nitrogen cooling bath. After 10 min of heating in a 95 ∘C water bath, the compounds were desorbed and separated by two capillary columns linked to the detectors by a Deans Switch (Pönisch, 2018).

For quality control, a calibration standard (gas composition of 9.9379 ± 0.0159 ppm CH4 and 1982.07 ± 3.77 ppb N2O, where ppm is parts per million, and ppb is parts per billion) was measured daily before and after the sample measurements; the standard deviation was <1 %. The calibration range was adjusted using multi-loop injection of the calibration gas to ensure that the samples were within the limits of the calibration. The standard was recalibrated according to high-precision standards (ICOS-CAL, Max Planck Institute, Jena, Germany).

CO2 and CH4 fluxes were calculated using the ideal gas law (Livingston and Hutchinson, 1995), as formulated in Eq. (7): 7F=MpVRTA⋅dcdt, where F is the GHG flux (g m-2 h-1), M is the molar mass of the gas (g mol-1), p is the standard air pressure (101 300 Pa), V is the chamber volume (m3), R is the gas constant (m3 Pa K-1 mol-1), T is the temperature in the chamber (K), A is the surface area of the measurement collar (m2), and dc/dt is the change in concentration over time. The latter was derived from the slope of a linear regression based on the medians of the gas concentrations. The atmospheric sign convention was applied; thus, positive fluxes indicated a release of GHG by the soil and negative fluxes GHG uptake by the soil. The fluxes were estimated using the function fluxx() from the R package flux (Jurasinski et al., 2022) and the SLP (slope based, i.e., median-based regression; Komsta, 2019) method. Outlier identification (using a histogram) resulted in the exclusion of CO2 fluxes which were smaller than -2.5 and larger than 4 g m-2 h-1. Similarly, CH4 fluxes larger than 200 mg m-2 h-1 were discarded due to a high risk of capturing ebullition-based CH4 emissions instead of diffusive fluxes.

The air–sea gas exchange (F; g m-2 h-1) is a function of the gas transfer velocity (k) and the concentration difference between the bulk liquid (Cw) and the top of the liquid boundary layer adjacent to the atmosphere (Ca). It was calculated as reported in Wanninkhof (2014) and as shown in Eq. (8): 8F=k(Cw-Ca), where k was derived from an empirical relationship between a coefficient of gas transfer (0.251) and the wind speed 〈U2〉 (Wanninkhof, 2014) and Schmidt number (Sc), as expressed by Eq. (9): 9k=0.251〈U2〉(Sc/660)(-0.5).

Wind speeds originated from the nearby (∼ 15 km away) monitoring station of Putbus and were measured at 10 m height (DWD; 54.3643∘ N, 13.4771∘ E; WMO-ID 10093). The average wind speed was defined in this study as ±3 h from midday because the wind speed over 24 h was lowest at night and highest at midday and because sampling was usually conducted within the selected time interval. The Schmidt number was approximated by a linear interpolation between the freshwater and seawater values. Atmospheric equilibrium conditions (Ca) were calculated using the atmospheric data for CO2 and CH4 obtained from the ICOS station of Utö (Finnish Meteorological Institute, Helsinki). Due to the seasonal changes in the atmospheric dry molar fraction of CO2 and CH4, mean values for each season were computed. For N2O, the atmospheric dry mole fraction from the station of Mace Head was selected (National University of Ireland, Galway; data from the NOAA Global Monitoring Laboratory (GML) carbon cycle cooperative global air sampling network; Dlugokencky et al., 2019a, b). A mean value of the atmospheric N2O concentration during the investigation period was calculated due to its minor seasonality. Equilibrium concentrations were then calculated using the solubility coefficient (K0) from Weiss and Price (1980). We acknowledge that the air–sea exchange model we used (Wanninkhof, 2014) was developed for open-ocean waters and is a questionable approach for deriving fluxes in small enclosed areas such as our study area. However, the lack of an appropriate parameterization and the convincing result of the comparison of our two approaches (see below and Appendix C) justify our approach.

We evaluated the comparability of the two previously described methods by comparing the results of a representative station (BTD7) for each post-rewetting season. The comparison showed no significant differences between the fluxes of CO2 and CH4 derived with the different methods, and therefore, it seems appropriate to combine the fluxes for each GHG into one pooled post-rewetting data set. The pooled post-rewetting flux values were compared with the pre-rewetting values to investigate the effect of rewetting on CH4 and CO2 fluxes (Table 3). For more details concerning the comparability assessment, see Appendix C. Due to the large variability and the pooling of chamber-based measurements with k model data, the GHG fluxes after rewetting are hardly suitable for upscaling, and thus, the single values in the published data should be used.

In the inner bay, all N nutrient concentrations were substantially higher at the first sampling after rewetting than prior to rewetting, while PO43- concentrations were only slightly higher post-rewetting (Fig. 4). This increase in N nutrients led to a drastic increase in the N : P ratio from ∼ 73 in autumn 2019 before rewetting to ∼ 1600 shortly after rewetting in winter 2019. A comparison of the same pre- and post-rewetting seasons (summer and autumn 2019/2020) showed generally higher N nutrient concentrations in the inner bay after rewetting, which could not be confirmed statistically (Mann–Whitney U test; Table 3).

During winter, all N nutrients were high in the peatland and inner bay. After a rapid decrease in spring, N nutrient concentrations reached their lowest values during summer, with NH4+ and NO2- increasing in autumn again. PO43- concentrations followed a different pattern, with the highest concentrations determined in summer and fewer fluctuations over the year.

The spatial differences in nutrient concentrations between the inner bay and the peatland after rewetting varied greatly between the nutrient species. From the N nutrients, only NO2- concentrations were significantly higher once in winter, shortly after rewetting, whereas NH4+ and NO3- concentrations showed no significant differences in any season (Table 2). Significantly higher PO43- concentrations in the peatland occurred during spring and summer (p<0.05). Some significant correlations between nutrient species were found (Fig. D1), especially between NO2-/NH4+ and NO3-/NO2-, both in the peatland and the inner bay.

Nutrient concentrations of the monitoring station (central bay) showed a low interannual variability during the years 2016–2020 and often lower concentrations than the inner bay (Fig. 5). A detailed comparison of nutrient data from the monitoring station with those from the inner bay showed that, before rewetting, only the NH4+ concentrations were significantly higher in the inner bay. After rewetting, NO3- and NO2- concentrations in the inner bay increased and were significantly higher than in the central bay (p<0.001 and p<0.05, respectively). In spring, N nutrient concentrations were similar at the two locations, whereas in summer, all N nutrients were significantly higher in the inner bay (p<0.01). In autumn, NO2- and NH4+ concentrations increased again and, thus, showed significantly higher concentrations in the inner bay. PO43- again followed a pattern different to that of the N nutrients. Shortly before rewetting, its concentrations in the inner bay were significantly lower than those in the central bay (p<0.05). After rewetting, PO43- concentrations showed no significant differences in any season.

The rewetted peatland was a net source of DIN-N and PO4-P for the inner bay (Table B1 in Appendix B). During the first year after rewetting, 10.8 ± 17.4 t yr-1 DIN-N and 0.24 ± 0.29 t yr-1 PO4-P were exported into the inner bay (given as mean ± 95 % confidence level, equivalent to Mg yr-1). DIN-N export was highest during the winter directly after rewetting (8.6 ± 9.9 t) and lowest during summer (0.3 ± 0.5 t). DIN-N and PO4-P were only exported from the peatland into the inner bay in all seasons.

N nutrient concentrations showed a gradient from the peatland through the inner bay to the central bay. Therefore, nutrient data from the central bay were also taken into account to estimate the total possible export from the peatland to the sea. This resulted in an estimated total net export of 33.8 ± 9.6 t yr-1 DIN-N. In contrast to the comparison of the peatland and the inner bay, PO4-P was once imported from the central bay into the peatland in autumn (0.03 ± 0.10 t). Additionally, it was noticeable that the PO4-P concentrations in the central bay were permanently higher than in the inner bay, leading to a lower annual export of 0.09 ± 0.32 t yr-1 PO4-P.

During the winter after rewetting, the differences in the CO2 system (CT, AT, pH, and pCO2) between the inner bay and the peatland were not significant (Figs. 6, 7a). All variables increased slightly until spring, coinciding with a slight increase in salinity over the same period. From spring onwards, however, the components of the CO2 system followed contrasting patterns, with CT and AT remaining relatively constant in the inner bay but reaching significantly higher values in the peatland (p<0.05), including maximum values in summer (Table 2). The pH also showed significant seasonal differences, with lower values and a minimum in summer in the peatland (p<0.05). CT and AT values in the inner bay and in the peatland aligned in autumn, whereas the pH remained significantly different (p<0.05). The mean pCO2 (calculated from CT and pH) of the surface water in winter was 1050.0 ± 55.7 µatm in the inner bay and 1403.9 ± 674.8 µatm in the peatland (Fig. 7a). The pCO2 values were highest during the first few weeks after inundation and then steadily decreased, with the lowest mean values occurring in spring (peatland) and summer (inner bay). The summer was characterized by high pCO2 values in general, including earlier and stronger increases in the peatland than in the inner bay that resulted in significant differences in spring and summer (p<0.05 for both seasons). pCO2 values were highest in summer, with 4016.7 ± 2120.0 µatm (peatland) and 1161.7 ± 1275.5 µatm (inner bay; Table 2). In October, all of the examined CO2 quantities had a short-term inversion of the prevailing pattern.

During the first few months after flooding (in winter), the CH4 concentrations in both the inner bay and the peatland were low and did not differ significantly (Fig. 7b; Table 2), i.e., 48.0 ± 49.5 nmol L-1 (peatland) and 81.4 ± 107.0 nmol L-1 (inner bay). From mid-spring onwards, CH4 concentrations in the inner bay and the peatland increased such that, during summer and autumn 2020, the differences at the two areas were significant (p<0.05). Mean CH4 values were highest in summer and amounted to 1502.4 ± 693.4 nmol L-1 in the peatland and 502.5 ± 479.3 nmol L-1 in the inner bay. Furthermore, the peatland was characterized by a considerable short-term variability in spring and summer, which is expressed in four peaks representing elevated concentrations. A positive significant correlation (rs=0.73; n=72; p<0.001) was found in the peatland between the surface water CH4 concentrations and a water temperature >10 ∘C but not <10 ∘C.

The highest N2O concentration of 486.3 nmol L-1 was measured in the peatland 1 week after rewetting (Fig. 7c), followed by 4–5 weeks of still-elevated N2O concentrations between 19.9 and 91.8 nmol L-1. During winter, significant positive correlations were determined in the peatland between N2O and NH4+ (rs=0.61; n=45; p<0.001) and between N2O and NO2- (rs=0.46; n=45; p<0.01). From spring onwards, N2O decreased rapidly, both in the peatland and the inner bay, with the lowest values of 4.7 to 7.9 nmol L-1 reached in summer. Other positive correlations of N2O with N nutrients in the peatland included NO3- (rs=0.74; n=35; p<0.001) and NO2- (rs=0.70; n=35; p<0.001) in spring and all N species in autumn (NO3- has rs=0.85, n=30, and p<0.001; NO2- has rs=0.70, n=30, and p<0.001; NH4+ has rs=0.80, n=30, and p<0.001).

Spatial differences in N2O concentrations between the inner bay and the peatland were low and not significant in winter, spring, or autumn, whereas significantly lower concentrations were measured in the peatland during summer (Table 2).

The carbon system in our study area is governed by a variety of processes (e.g., Wolf-Gladrow et al., 2007; Kuliński et al., 2017; Schneider and Müller, 2018). CT and AT were transported with the brackish water from the central bay and ultimately from the Arkona Basin. Additional alkalinity can be added either by a supply of freshwater, which in the southwestern Baltic Sea is characterized by higher alkalinities than the brackish or even saltwater endmember (Beldowski et al., 2010; Müller et al., 2016), or can be introduced by mineralization processes from the seafloor in the inner bay and the flooded peatland. Primary production (i.e., carbon fixation) will decrease CT, lower the pCO2, and increase pH during the formation of organic matter. The mineralization of OM from various sources (new primary production, mineralization of the inundated former vegetation, and from the underlying peat) will enhance CT and AT concentrations, increase pCO2, and decrease pH. Air–sea exchange during our study is fostered by a pCO2 that is above atmospheric levels throughout the year, except for a short period in spring in the inner bay and the peatland, where outgassing of CO2 occurred, resulting in lower pCO2 and a decrease in CT.

We observed the following three main developments in the surface water CO2 system and air–sea flux pattern: (i) in winter 2019/2020, the CO2 system hardly differed between the peatland and the inner bay; (ii) from spring to autumn, there were significant differences in the CO2 system between the peatland and the inner bay, with higher pCO2, CT, and AT values and lower pH in the peatland coinciding with an enrichment in chlorophyll a; and (iii) overall, the first post-rewetting year showed sustained high but less variable CO2 fluxes compared to pre-rewetting conditions. In the following, we will discuss these three observations and put them into context.

The first weeks after the rewetting were characterized by high nutrient concentrations, a continuous increase in AT, CT, and pH and a decrease in pCO2 (Figs. 4, 6, 7). The increase in CT and AT coincided with a steady increase in salinity (Fig. 3), which is in line with a general increase in AT with increasing salinity that is known for the western Baltic Sea (e.g., Kuliński et al., 2022).

Still, the AT values at the given salinity were higher in the inner bay and the peatland than would be expected from a linear AT–S relationship found for surface waters in the open Baltic Sea from the central Gotland sea to the Kattegat (Beldowski et al., 2010; Müller et al., 2016). Thus, the high AT in the inner bay and peatland were likely associated with local carbonate (CaCO3) weathering from terrestrial sources and/or a transport by groundwater (Schneider and Müller, 2018). CT and AT values during this period were consistently higher by ∼ 70–80 µmol kg-1 in the peatland than in the inner bay, consistent with enhanced leaching from the recently inundated peat. Besides, local CaCO3 weathering and local anoxic processes, such as SO42- reduction, may have increased the AT in the submerged soil and finally contributed to higher AT values compared to the inner bay.

The oversaturation in pCO2 and potentially the excess leaching of alkalinity from the soil might have contributed to the decrease in pCO2 and increase in pH in the peatland in winter 2019/2020. This was apparently reinforced by a short episode of primary production between the middle and end of January, indicated by a steeper decline in the pCO2 and a steeper pH increase. This coincided with a short increase in chlorophyll a (∼ 30 µg L-1) and a slight intermittent increase in the surface water temperatures (Fig. 3). This short, unusually early productive period might have resulted from the high nutrient availability induced by the rewetting of the peatland (Sect. 3.2.1), in particular the high NH4+ levels, which simultaneously showed a sharp intermittent minimum.

In late winter and the first half of spring, pCO2 continuously decreased in the peatland and in the inner bay. The lowest pCO2 was measured between March and May and coincided with enhanced chlorophyll a concentrations and a high availability of nutrients in the peatland and in the inner bay, which decreased until mid-spring. This resulted in a slight CO2 uptake in the peatland of -0.005 g m-2 h-1 for a short period of time, so that spring was the only season in which pCO2 was, on average, below atmospheric concentrations in the inner bay (Fig. 8). This finding can be attributed to the onset of the productive period, at still-moderate surface water temperatures below 10 ∘C until mid-April. During this period, productivity clearly exceeded mineralization, as suggested by the decreasing pCO2 and increasing pH, despite rising temperatures, and increasing O2 oversaturation in the surface waters. These trends were slightly more pronounced in the peatland than in the inner bay, in accordance with higher nutrient concentrations available for production.

From mid-spring until late summer, the peatland was characterized by increased pCO2 and a variable CO2 system together with high mean chlorophyll a concentrations of up to 106.0 µg L-1. N nutrients were very low, and the system was clearly nitrogen limited, with only slightly elevated NH4+ concentrations in late summer (Figs. 3, 4). Furthermore, the O2 saturation shifted from over- to undersaturated conditions. These observations suggest that the peatland and the inner bay were characterized by simultaneous production and mineralization processes from mid-spring until autumn that kept the N nutrients (except for PO43-) low. Mineralization of OM in the water column, sediment, and soil dominated over production, leading to the observed high pCO2, lowered pH, and enhanced AT and CT concentrations. Mineralization during this period was more pronounced in the peatland than in the inner bay, leading to the higher pCO2, AT, and CT values in the peatland, and a stronger and more pronounced reduction in the pH. This stronger mineralization, in particular in the warm summer months, also led to higher DOC concentrations in summer, with a maximum in June/July coinciding with maximum surface water temperatures. The enhanced mineralization in the peatland was likely fueled by higher OM availability from high decomposition rates of fresh plant substrate from inundated plant residuals (Glatzel et al., 2008; Hahn-Schöfl et al., 2011). In addition, aerobic and anaerobic oxidation of CH4, which was produced in anoxic zones, might have led to increased CO2 production, especially during increased water temperatures (e.g., Treude et al., 2005; Dean et al., 2018), due to the availability of SO42- and O2.

The calculated AT (from CT and pH) in the peatland was consistently lower than the measured AT, with a difference in the range of 55–122 µmol kg-1 and thus of 2.7 %–4.7 % (data not shown). This difference was higher than in the Baltic Sea, where the contribution of organic AT is estimated to be 1.5 %–3.5 % (Kuliński et al., 2014). Due to closer vicinity to the coast and the high amount of degradable OM, this higher contribution of organic AT was to be expected. The highest discrepancy between measured AT values and those calculated from pH and CT occur in early summer, simultaneous with the highest values in DOC, in particular in the peatland (Fig. 3). This suggests that the organic AT related to the occurrence of DOC (and thus dissolved organic matter, DOM), contributed to the excess of AT. The higher DOM formation in summer in the peatland might partly explain the difference in AT between the inner bay and the peatland.

The amplitude of the CO2 fluxes from formerly drained parts of the study area decreased after rewetting with brackish water, while the amplitude of CO2 fluxes from the ditch (inundated after flooding but with deeper, probably incompletely exchanged water) did not differ strongly before and after rewetting (Fig. 8). An increased water table is the main driver for the reduction in CO2 emissions on formerly drained locations. A similar scenario has been reported for terrestrial sites (Bubier et al., 2003; Strack, 2008). In a nearby coastal peatland, both photosynthesis and ecosystem respiration were strongly reduced after rewetting (Koebsch et al., 2013). The rewetting of our study area probably caused a die-back of the highly productive grassland vegetation, which most likely led to a reduction in the CO2 flux amplitude.

Average summer/autumn CO2 fluxes after rewetting had a mean of 0.26 ± 0.29 g m-2 h-1 and thus remained relatively high compared to those fluxes from 2019. They were also higher than the fluxes determined in studies of shallow coastal or near-shore waters in the northwestern Bornholm sea, of up to 0.01 g m-2 h-1 (Thomas and Schneider, 1999), or the Bothnian Bay, of around ∼ 0.0007 g m-2 h-1 (Löffler et al., 2012). In a nearby coastal fen recently influenced by brackish water inflow, ecosystem respiration was 2 orders of magnitude lower (Koebsch et al., 2020) compared to our study site, where the ongoing decomposition of submerged substrate from plant residuals and the fresh soil may have fueled the continuously high CO2 fluxes in the first year after rewetting (Hahn-Schöfl et al., 2011). The mineralization of OM from primary production driven by the high initial nutrient availability, and aerobic and anaerobic oxidation (Dean et al., 2018) of easily degradable substrates or CH4, might have additionally contributed to these CO2 fluxes. We expect that CO2 emissions will further decrease, likely because substrates become exhausted, and a novel ecosystem will be established (Kreyling et al., 2021), with developing algae fostering CO2 fixation.

We observed the following three main developments in surface water methane concentrations and flux patterns: (i) a short-term, very moderate increase in CH4 concentrations directly after rewetting in winter 2019/2020, (ii) an increase in the CH4 concentrations, mainly from spring to autumn, that was significantly higher and more variable in the peatland than in the inner bay and correlated with water temperature, and (iii) in the first year after rewetting, much lower CH4 fluxes than reported for nearby peatlands rewetted by freshwater. These three observations are discussed and put into context in the following.

The measurements in winter, immediately after rewetting, showed a short-term but moderate increase in the CH4 concentrations (Fig. 7). The rewetting resulted in the inundation of the degraded peat and the remaining vegetation. Therefore, it is assumed that methanogenesis was not limited by the availability of high-quality OM, which is often a major controlling factor (Heyer and Berger, 2000; Parish, 2008) and likely originated from the decomposition of the residual plant material. However, since CH4 concentration remained low, a temperature control is assumed, which has been frequently described in the literature. A major control of temperature has been reported, for example, for a nearby shallow coastal area of the Baltic Sea, between the islands of Rügen and Hiddensee, where low CH4 emission rates and variability were found together with low temperatures (Heyer and Berger, 2000).

The rewetting transported water with a salinity of 6–7.4 into the peatland, such that there were no significant differences in salinity compared to the inner bay in winter (the same as for temperature; Table 2). Thus, sulfate reached the peatland immediately after rewetting. As a terminal electron acceptor (TEA), SO42- promotes the activity of sulfate-reducing bacteria (SRB), which outcompete methane-producing microorganisms (methanogens) for substrates (Segers and Kengen, 1998; Jørgensen, 2006; Segarra et al., 2013). This process was shown to play an important role in flat brackish water systems (e.g., Heyer and Berger, 2000). The availability of other TEAs, such as NO3- that had high concentrations of ∼ 100 ± 58 µM in our study, could have further suppressed methanogenesis (Table 2; Jørgensen, 2006). Beside competitive mineralization, aerobic and anaerobic CH4 oxidation may have reduced the CH4 concentrations (Heyer and Berger, 2000; Reeburgh, 2007; Knittel and Boetius, 2009; Steinle et al., 2017), supported by the effective exchange of water masses. Overall, the rewetting with brackish water during the cold winter season apparently inhibited methanogenesis and/or facilitated effective CH4 oxidation, resulting in low CH4 concentrations and a small CH4 flux into the atmosphere.

The temperature increase from spring to autumn was accompanied by elevated, albeit variable, CH4 concentrations. Temperature is of crucial importance for controlling the CH4 cycle in shallow coastal brackish water (Bange et al., 1998; Heyer and Berger, 2000) and in the North Sea (e.g., Borges et al., 2018). Similar relationships have been described for wetlands, e.g., for permanently inundated wetlands (e.g., Koebsch et al., 2015) and in a peatland close to our study site during the first year after rewetting (Hahn et al., 2015). Accordingly, CH4 concentrations in the peatland (rs=0.75; n=74; p<0.05) and the inner bay (rs=0.55; n=29; p<0.05) also correlated significantly and positively with temperature. In the study of Heyer and Berger (2000), the temperature range influenced the temporal variability in CH4 emissions, which were highest in late spring. Since the temperature range in the peatland of our study was variable (e.g., maximum difference of ∼ 6 ∘C between samplings), with the highest values between spring and autumn (7.4–23.1 ∘C), this variability may have strongly contributed to the observed CH4 dynamics.

The peatland and the inner bay were clearly influenced by the same hydrographic conditions, evidenced by their very similar salinities and temperatures. However, the peatland showed higher CH4 concentrations from spring to late autumn, likely due to the high availability of OM, as described by Heyer and Berger (2000) and Bange et al. (1998). Incubation experiments of a degraded fen grassland demonstrated the accumulation of fresh plant litter in a new sediment layer after flooding that resulted in high rates of CH4 and CO2 production (Hahn-Schöfl et al., 2011). A further potential driver of OM availability is the sedimentation of freshly produced OM originating from primary production, as described for shallow areas in the Baltic Sea (Bange et al., 1998) and for a shallow bight in the North Sea, which in the latter led to a yearly peak in the seasonal CH4 cycle (Borges et al., 2018). Although our observations were not made in OM-poor sediments, an impact of freshly produced OM on enhanced CH4 concentrations in the OM-rich Drammendorf peatland is possible, given the significant positive correlation of the surface CH4 concentrations and the chlorophyll a concentration (rs=0.41; n=56; p<0.05).

Despite the high surface water CH4 concentrations in the peatland and their inter-seasonal and spatial variability, rewetting with brackish water resulted in CH4 emissions being considerably lower than those from temperate fens rewetted with freshwater, where CH4 emissions strongly increased (Augustin and Chojnicki, 2008; Couwenberg et al., 2011; Hahn et al., 2015; Franz et al., 2016; Jurasinski et al., 2016).

At our study site, although average CH4 fluxes on formerly terrestrial locations increased significantly by 1 order of magnitude after rewetting, the overall increase from 0.13 ± 1.01 to 1.74 ± 7.59 mg m-2 h-1 (Fig. 8) was lower than that reported for freshwater rewetted fens under similar climatological boundary conditions (e.g., Hahn et al., 2015; Franz et al., 2016). Even several years after rewetting, the annual CH4 budgets of a shallow lake on a formerly drained fen varied between 13.2 and 52.6 g m-2 yr-1 (Franz et al., 2016), which corresponds to approximately 1.5 to 6.0 mg m-2 h-1. Our CH4 fluxes were also lower than the emissions reported from coastal near-shallow waters of the Baltic Sea, where fluxes of 39.9–104.2 mg m-2 h-1 were measured in June/July (Heyer and Berger, 2000). For the same months, mean CH4 fluxes at the formerly dry stations in our study site were 0.5–4.9 mg m-2 h-1. However, compared to CH4 fluxes from continental shelves (0.015–0.024 mg m-2 h-1; adapted from Bange et al., 1994), the fluxes of our study site were 2 orders of magnitude higher. Despite the low average fluxes, emission peaks could be distinguished with the highest flux from the now inundated ditch of 149.2 mg m-2 h-1 in September 2020 and 108.3 mg m-2 h-1 in October 2020. While these values were still lower than the maximum value of 243.0 mg m-2 h-1 reported by Heyer and Berger (2000), it is important to stress that our study site was already a source of CH4 in its drained state, especially within the drainage ditch, where CH4 fluxes were comparable to the ∼ 0.2 mg m-2 h-1 reported from undrained fens (Danevèiè et al., 2010).

The lower CH4 emissions of the brackish rewetted Drammendorf peatland might be attributed to the availability of TEAs, especially SO42-, which (1) may have contributed to a suppression in methanogenesis by competitive inhibition (Segers and Kengen, 1998; Jørgensen, 2006; Segarra et al., 2013) or (2) fostered the anaerobic oxidation of methane (AOM) as an effective pathway to reduce CH4 emissions and by (3) fast aerobic CH4 oxidation mediated by oxygen-rich water. The high variability in CH4 concentrations may also be related to changing rates of AOM, as the process is sensitive to the introduction of O2 mediated by sporadic wind-driven resuspension (Treude et al., 2005). Since our study area was shallow and likely experiences regular wind-driven resuspension, spatially and temporally dynamic AOM can be assumed. The low CH4 fluxes suggested that effective aerobic and anaerobic oxidation of CH4 likely occurred. Moreover, higher CH4 concentrations in the peatland compared to the inner bay in combination with the high lateral water exchange due to frequent changes in the water level (Fig. A3) might have driven a net advective export of CH4-enriched water to the inner bay. This would have further contributed to the low peatland CH4 emissions and the observed high variability.

While CH4 production and emission were likely prevented by rewetting with oxygen-rich, sulfate-containing brackish water, the possibility remains that the total CH4 release was underestimated by insufficient accounting for ebullition. In the marine environment, bubble-mediated transport is attributed to gassy sediments and an effective mechanism of vertical CH4 migration (e.g., Borges et al., 2016). Although neither of the methods used to determine CH4 fluxes specifically account for ebullition, we estimated that 6.9 % of all analyzed chamber-based fluxes were partly bubble influenced. We estimated this percentage by counting the measurements which showed irregular data points in graphical depiction but did not influence the linear slope. We observed further that, in another 9.6 % of the chamber-based flux measurements, the CH4 concentration patterns indicated ebullition (flux measurements with exponential slope, which was clearly steeper than the linear regression of the majority of data points), but these were not accounted for in the final calculations of diffusive fluxes. Thus, given that only 16.5 % of the chamber-based flux measurements indicated bubble-mediated CH4 transport, and in almost half of those cases, the resulting perturbation was small and was included in the flux amplitude, the magnitude of the ebullition-driven underestimation of our flux estimates is considered to be small.

In summary, the increase in CH4 concentrations after rewetting in winter was small, short-lived, and associated with the die-back of plants. CH4 fluxes in the first year after rewetting remained relatively low and were lower than typical of post-rewetting conditions. They also followed a seasonal pattern common for shallow organic-rich systems, with a strong correlation with temperature in spring and summer. We anticipate that continuing reduction in OM availability after the initial die-back of vegetation will likely lead to a further decrease in CH4 emissions in subsequent years.

The rewetted peatland was a source of N2O in the first year after rewetting, although the mean annual N2O flux of 0.02 ± 0.07 mg m-2 h-1 was very low (Fig. 8). This was expected since a permanent inundation leads to anoxic conditions in the peat, preventing the production of N2O by nitrification, in addition to denitrification, due to the lack of NO3- (e.g., Succow and Joosten, 2001; Strack, 2008). However, the range of post-rewetting N2O fluxes in the first 3 months (winter) was clearly much larger than during the rest of the year, which indicated that N2O was strongly and immediately affected by the rewetting, as shown elsewhere (Goldberg et al., 2010; Jørgensen and Elberling, 2012). The highest N2O flux (0.4 mg m-2 h-1) and the highest NH4+ concentration (78.0 µM) was measured 1 week after rewetting and a significant positive correlation between these two variables was found in winter (rs=0.61, n=45, p<0.001). Additionally, N2O had a significantly positive correlation with NO2- in winter (rs=0.46, n=45, p<0.01), whereas no correlation with NO3- was found. The accumulation of N2O, and also of NO2- and NO3-, can generally be interpreted as a result of shifting O2 conditions in the freshly inundated ecosystem, such that incomplete process chains of, for example, nitrification and denitrification were favored (Rassamee et al., 2011). However, it seems likely that the high N2O concentrations in winter originated from nitrification due to the correlations of N2O with its substrate (NH4+) and its main accumulating intermediate product (NO2-), in addition to a trend of increasing NO3- concentrations throughout the winter.

During late spring and early summer, an undersaturation of the surface water with N2O, compared to the atmosphere, pointed to consumption within suboxic/anoxic zones of the peat. Consumption in the surface water was unlikely because anoxic conditions were never found near the peat surface. The undersaturation of N2O a few months after rewetting evidenced the change in O2 conditions in the peat, from oxic to hypoxic/anoxic, turning the rewetted peatland into an N2O sink, at least temporarily. This change was likely driven by the higher availability of fresh OM (measured as chlorophyll a) in the peatland compared to the inner bay, finally leading to significantly lower N2O concentrations in the peatland in summer (p<0.001; Table 2).

Previously reported N2O fluxes in drained peatlands range from 0.002 to 0.45 mg m-2 h-1, with a clear trend towards higher fluxes in fertilized or naturally N-rich areas (Flessa et al., 1998; Glatzel and Stahr, 2001; Augustin, 2003; Strack, 2008; Minkkinen et al., 2020). Augustin et al. (1998) examined multiple degraded fens in Mecklenburg-Vorpommern and Brandenburg (Germany) and calculated N2O fluxes of 0.04 to 0.10 mg m-2 h-1 in extensively and intensively used fen grasslands, respectively (Augustin et al., 1998). N2O fluxes in drained peatlands result from a low water level which allows the permanent penetration of atmospheric oxygen into the peat to fuel N2O-producing processes that are dependent on oxygen (Martikainen et al., 1993; Regina et al., 1999). As the water level in our study site was permanently below the soil surface before rewetting, it is likely that the drained peat was a source of N2O. The mean post-rewetting N2O flux determined in our study area (0.02 ± 0.07 mg m-2 h-1) is in the lower range of reported fluxes from drained peatlands. Therefore, as shown in other studies (Succow and Joosten, 2001; Minkkinen et al., 2020), the rewetting probably led to a reduction in N2O fluxes, since they were likely high before the rewetting.

In general, the N2O fluxes in rewetted peatlands are in the same range as fluxes from pristine ones (Minkkinen et al., 2020), indicating that rewetting is a very effective measure to reduce N2O emissions to natural levels. The literature values range from up to 0.01–0.02 mg m-2 h-1, for rewetted and undrained boreal peatlands (Minkkinen et al., 2020), respectively, to 0.08 mg m-2 h-1, for a rewetted riparian wetland near a freshwater meadow (Kandel et al., 2019). Although it is difficult to compare the N2O fluxes determined in this study with those from other sites with different salinity, hydrology, and history of use, our mean annual post-rewetting value is in the lower range of N2O fluxes previously reported for rewetted and pristine peatlands.