RV Alkor Cruise AL543 took place in August 2020. The geographical positions and water column properties of all sampling stations are provided in Table 1 . Water column samples and temperature and salinity data were obtained by deploying a HydroBIOS MWS 12 rosette water sampler equipped with a CTD. A novel trace metal-clean bottom water sampler (Benthic Trace Profiler) (Plass et al., 2022) was also deployed at Station 8 to collect waters close to the seafloor. Short sediment cores were retrieved together with the overlying bottom water using a mini multiple corer (MUC). The sediment cores were transferred to the home laboratory for subsampling and pore water recovery (see Scholz et al. (2011) for details). The bottom water was siphoned off with a tube and subsequently treated like pore water samples. Sediments were sub-sampled in an argon-flushed glove bag by cutting the core into 1 - 4 cm thick slices with the highest resolution at the surface. The pore water was separated from the solid phase by centrifuging for 20 minutes at 4000 rpm. The supernatant water was filtered through 0.45 μm cellulose acetate syringe filters and split into several aliquots for the determination of Fe²⁺, H₂S, anion and metal concentrations. Aliquots for the metal analyses were stored in acid-cleaned LDPE vials and acidified with concentrated HNO₃ (sub-boiling distillation). A wet sediment aliquot was stored in air tight pre-weighed plastic cups for the determination of water content and porosity followed by total digestion for element and isotopic analyses. As the main source of sediments to Kiel Bight is coastal erosion (see above), we also analyzed solid phase samples taken at the most prominent coastal cliffs in Stohl and Schönhagen ( Figure 1 ) to determine the regional terrigenous background ratio of Ba to aluminum (Al) and δ¹³⁸Ba.

Dissolved O₂ concentrations in the water column were determined by Winkler titration. The seawater samples for dissolved nutrient analysis (NO_3⁻ , PO₄ ³⁻ and SiO₄ ⁴⁻) were filtered through 0.2 µm PES-filters and stored frozen in polypropylene bottles until analysis in the shore-based laboratories using a Quaatro Seal Autoanalyser. Concentrations of NH_4⁺ , PO₄ ³⁻, SiO₄ ⁴⁻ and Fe²⁺ in bottom water and pore water samples were analyzed by standard spectrophotometric techniques (Stookey, 1970; Grasshoff et al., 1999). Hydrogen sulfide (ΣH₂S = H₂S + HS^- + S^2-) concentrations in the water column and pore waters were also determined photometrically by applying the methylene blue method (Grasshoff et al., 1999). Bottom water and pore water SO₄ ²⁻ concentrations were analyzed by ion chromatography. The concentrations of Ba and Mn in the water column and pore waters were determined by inductively coupled plasma optical emission spectroscopy (ICP-OES, VARIAN 720-ES) and inductively coupled plasma mass spectrometry (ICP-MS, Agilent Technologies 7500 Series).

The total organic carbon (TOC) content of freeze-dried and ground sediment samples was determined with an element analyzer (Euro EA, HEKAtech). For the analysis of total element concentrations, ~100 mg of freeze dried and ground sediment samples and coastal cliff samples were digested in HF (40%, supra pure), HNO₃ (65%, supra pure) and HClO₄ (60%, p.a.) on a hotplate and the resulting solutions were analyzed by ICP-OES. The accuracy of the digestion procedure was monitored by inclusion of method blanks and the reference standards PACS-3 (marine sediment, Canadian Research Council), MESS-3 (Marine Sediment Reference Material, Canadian Research Council) as well as the in-house standard OMZ-2. Average values of replicate digestions were generally well within the recommended ranges with relative standard deviations (RSD) being <1% for Al and <2% for Ba (n=5).

Ba isotope compositions were measured using a double spike (¹³⁰Ba-¹³⁵Ba) technique (Yu et al., 2020), which was modified from the one of Horner et al. (2015). Aliquots containing ~50 ng Ba were taken from seawater, pore waters and total sediment digestions for the determination of Ba isotope compositions. Note that due to limited volumes of seawater samples from the CTD casts, samples from the shallow water column with low Ba concentrations had to be combined to obtain enough Ba for isotope analyses. In brief, appropriate quantities of the ¹³⁰Ba-¹³⁵Ba double spike solution were equilibrated with sample aliquots on a hotplate at 80°C for at least 24 hours. A pre-treated Ba-free Na₂CO₃ solution was added stepwise to seawater samples to co-precipitate Ba with CaCO₃. The resulting precipitate was centrifuged and dissolved in 1 mL 1 M HCl. The water and total digestion samples were evaporated and then re-dissolved in 1 mL 1 M HCl. Chemical separation was performed twice using cation exchange resin (AG50W-X8, 200-400 mesh, 1.4 mL volume). The matrix elements were first eluted with 8 mL of 1 M HCl and 8 mL of 3 M HCl, followed by collection of Ba with 10 mL of 2 M HNO₃. Finally, the Ba cut was evaporated and dissolved in 2% (v/v) HNO₃ for Ba isotope analyses. Ba isotope analyses were conducted at GEOMAR Helmholtz Centre for Ocean Research Kiel using a Neptune Plus (Thermo Finnigan) MC-ICP-MS. The purified sample solutions were introduced as a dry aerosol using an Aridus II desolvator system and an Aspire PFA micro-concentric nebulizer (uptake rate of ~50 μL min^-1). Tuning for an appropriate Normalized Ar Index (NAI; an index of plasma temperature) prior to measurement attenuated possible matrix effects (Yu et al., 2020). The geometrical procedure described in Siebert et al. (2001) was used for data reduction. Each sample solution was analyzed three to five times in two different measurement sessions, with external sample reproducibilities between ±0.01 and ±0.07 ‰ (2SD). Long term repeated measurement of reference materials at GEOMAR over a course of three years yield δ¹³⁸Ba_NIST values of +0.62 ± 0.03 ‰ (2 SD, n = 6) and 0.29 ± 0.03 ‰ (2 SD, n = 19) for seawater reference samples SAFe surface and SAFe 1000 m, respectively, and 0.29 ± 0.03‰ (2 SD, n = 27) for carbonate reference material JCP, which are all within error identical to those reported by three other laboratories (Hsieh and Henderson, 2017; Geyman et al., 2019; Cao et al., 2021). Long term measurements of in-house seawater reference samples BATS 15 m and 2000 m also show similar reproducibility: 0.55 ± 0.05 ‰ (2 SD, n = 26) and 0.47 ± 0.04 ‰ (2 SD, n = 18) respectively. The Ba isotope compositions are reported as δ¹³⁸Ba values relative to the Ba SRM NIST3104a:

Benthic diffusive fluxes (F) across the benthic boundary were calculated applying Fick’s first law of diffusion (Schulz, 2006):

In Eq. 2, ϕ is porosity, D_sed is the diffusion coefficient of ions in sediment and d[C]/dx denotes the ion concentration gradients between the uppermost pore water sample (0 – 1 cm) and the overlying bottom water. Negative values indicate a flux from the pore water into the bottom water and positive values represent a flux from the bottom water into the pore water. The diffusion coefficients of ions in seawater (D_sw) under standard conditions (Li and Gregory, 1974) were adjusted to in situ temperature, pressure and salinity applying the Stokes-Einstein equation. Diffusion coefficients of ions in sediments (D_sed) were calculated from the diffusion coefficients in seawater (D_sw) taking into account tortuosity (θ₂):

Tortuosity was calculated from porosity using the following relationship from Boudreau (1996):

The saturation index (SI) for barite in the water column and sediment pore waters was calculated applying the geochemical model PHREEQC by Parkhurst and Appelo (2013) using the PHREEQC data base. Saturation indexes greater than 0 indicate that aqueous solutions were oversaturated with respect to barite whereas SIs below 0 indicate undersaturation with respect to barite and potential barite dissolution. A saturation index close to 0 indicates that dissolution and precipitation of barite reached equilibrium. Input data of the chemical equilibrium calculations include temperature, alkalinity and concentrations of all major ions in seawater (including SO₄ ²⁻) and Ba²⁺.

During Cruise AL543 in August 2020, temperature and salinity in the upper water column (above 10 - 15 m) ranged from 18.1 to 19.4°C and from 14.7 to 16.8, respectively ( Figures 2A, B ). Below this depth range, temperature dropped and salinity increased sharply with depth, ranging from 13.5 to 18.3°C and from 16.4 to 21.6, respectively. The temperature and salinity profiles reflect pronounced pycnoclines and are consistent with stable stratification of the water column during late summer (Lennartz et al., 2014). The pycnocline was shallower at Station 8 (~8 m depth; roughly corresponding to the depth of the surrounding moraine shoals) and deeper at Stations 10 and 13 (~15 m depth). Dissolved O₂ concentrations were high in surface waters (~250 µM) and dropped sharply below 7 m depth. The deepest samples at Stations 10 and 13 were still oxic (O₂ concentrations of 31 µM and 80 µM, respectively). The deep-water at Station 8 was anoxic below ~15 m depth ( Figure 2C ). The concentrations of SiO₄ ⁴⁻ and PO₄ ³⁻ were low in the upper water column above the pycnocline and increased sharply below ( Figures 2D, E ). In contrast, NO_3⁻ concentrations were low throughout the water column, except for somewhat elevated concentrations in the bottom water at Stations 8 and 10 ( Figure 2F ). The deep-water was thus characterized by a nitrate deficit relative to phosphate when compared to the average nitrogen to phosphorus ratio in phytoplankton (Redfield ratio), as indicated by a negative N* (N* = NO₃ ⁻ - 16 · PO₄ ³⁻) (Gruber and Sarmiento, 1997) ( Figure 2G ).

At Station 8, dissolved Mn concentrations increased below 10 m water depth and reached maximum concentrations of ~12 μM in the deep-water. Hydrogen sulfide concentrations up to ~45 µM were detected in the deep-water at Station 8 ( Figure 2H ). In contrast, at Stations 10 and 13 dissolved Mn concentrations varied little and remained below 1.2 µM throughout the water column ( Figure 2I ). Above the pycnocline, dissolved Ba concentrations were relatively constant between 41 and 116 nM at all stations. Below the pycnocline, dissolved Ba concentrations were higher and ranged from 147 to 938 nM. The highest dissolved Ba concentrations were observed in the deep-water of Station 8 ( Figure 2J ). Seawater above the pycnocline was undersaturated with respect to barite (SI_Barite< 0), whereas the deep-water was oversaturated (SI_Barite > 0) ( Figure 2K ).

Pore water dissolved Mn profiles were characterized by a peak within the upper 5 cm of the sediment (12.8 - 70.6 µM) and decreased (Stations 8 and 13) or increased again (Station 10; up to 126.5 µM) towards the lower end of the core ( Figure 3A ). Pore water dissolved Fe concentrations were high (up to ~48 μM) in the uppermost 3 cm at Station 10 but low throughout the profiles at the other two stations ( Figure 3B ). Pore water SO₄ ²⁻ concentrations decreased ( Figure 3C ) and those of H₂S, NH_4⁺ , PO₄ ³⁻ and SiO₄ ⁴⁻ increased with depth at all three stations ( Figures 3D–G ). The pore waters at Station 13 were more depleted in SO₄ ²⁻ (approaching zero at the lower end of the core) and more enriched in H₂S and NH_4⁺ than at Stations 8 and 10.

Pore water dissolved Ba concentrations reached a maximum in the uppermost 1 or 2 cm of the sediment column and decreased sharply below ( Figure 3H ). At Stations 8 and 10 pore water Ba concentrations remained essentially constant below 5 cm sediment depth. In contrast, at Station 13, dissolved Ba concentrations increased again below 4 cm depth and reached ~1000 nM at the lower end of the core ( Figure 3H ). Bottom waters and pore waters above 4 cm were oversaturated with respect to barite at all three stations, especially at the depths corresponding to the dissolved Ba maximum (SI_barite up to 0.80). Below 4 cm, most pore waters were close to equilibrium with respect to barite (0.05 ≤ SI_Barite ≤ 0.30) ( Figure 3I ). At Station 13, the pore water were undersaturated with respect to barite below 18 cm depth (SI_Barite< 0) ( Figure 3I ), coincident with decreasing sulfate but increasing dissolved Ba and H₂S concentrations ( Figures 3C, D, H ).

The concentrations of PO₄ ³⁻, SiO₄ ⁴⁻, H₂S, NH_4⁺ , Fe, Mn and Ba in the uppermost pore water samples (0.5 cm) were higher than their respective concentrations in the overlying bottom water (0.0 cm) (except H₂S at Station 10 and Fe²⁺ at Station 8). These concentration gradients across the sediment-water interface are indicative of a diffusive flux of these solutes from the sediments into the bottom water ( Table 2 ).

Solid phase Ba, Al and TOC concentrations were corrected for the amount of salt and pore water Ba that was transferred to the solid phase upon freeze drying (based on water content, pore water salinity and Ba concentrations). At all three stations, the concentrations of Ba and Al were lowest in the uppermost sediment layer (0 - 1 cm). Below the surface sediment, corrected Ba concentrations in the sediment ranged from 315 μg g^-1 to 416 μg g^-1 and bulk Al concentrations varied between 30.1 mg g^-1 and 57.0 mg g^-1, both remaining essentially constant with depth ( Figure 4 ). The coastal cliff material had an average Ba concentration of 332 ± 38 μg g^-1 and a Ba/Al of 0.010 ± 0.002 (SD, n = 36). Pore water corrected Ba/Al ratios of the uppermost sediment samples at Stations 8 and 13 exceeded this regional background ratio, while the ratios of other sediment core samples were lower than the regional terrigenous background value ( Figure 4 ). Organic carbon concentrations were highest in the uppermost sediment intervals (up to 8.7 wt.%) and decreased with depth ( Figure 4 ).

Above the pycnocline, where dissolved Ba concentrations were low, the seawater δ¹³⁸Ba values varied little with depth and ranged from +0.36 ‰ to +0.42 ‰ ( Figure 5 ). Below the pycnocline at Stations 8 and 10, where dissolved Ba concentrations increased, the δ¹³⁸Ba was generally lower than in the surface waters (average of +0.28 ± 0.04 ‰, n = 9). Below the pycnocline at Station 13, seawater δ¹³⁸Ba also decreased at first but then increased again (average of +0.35 ± 0.04 ‰, n = 4). Dissolved Ba and δ¹³⁸Ba distributions in the water column display a mirror image with high dissolved Ba concentrations corresponding to low δ¹³⁸Ba values ( Figure 5 ).

The dissolved Ba concentration maximum at the sediment surface was characterized by a δ¹³⁸Ba similar to that in the overlying bottom and deep-water ( Figure 5 ). Below the concentration maximum at the sediment surface, pore water δ¹³⁸Ba first decreased within the upper few centimeters and then either remained constant (Stations 8 and 10) or increased again to reach +0.34 ‰ at Station 13, coincident with downcore increasing Ba concentrations ( Figure 5 ). Unlike in the water column, dissolved Ba and δ¹³⁸Ba distributions in the pore water did not display a mirror image. Instead, the highest Ba isotope values coincided with high dissolved Ba concentrations at Station 13 ( Figure 5 ). The δ¹³⁸Ba values of solid phase samples are uniform within analytical error (-0.01 ± 0.03‰, SD, n = 18) ( Figure 4 ) and identical to the δ¹³⁸Ba of the terrigenous sediment source from the coastal cliffs (+0.02 ± 0.03, SD, n = 6).

Baltic Sea water is generated by mixing of freshwater from the Baltic Sea drainage area and North Atlantic seawater entering via the Danish straits. Freshwater is generally characterized by higher Ba concentrations and a lower δ¹³⁸Ba than seawater (e.g., Cao et al., 2016; Cao et al., 2020; Bridgestock et al., 2021a; Bridgestock et al., 2021b; Rahman et al., 2022). Therefore, it is likely that the Ba concentration and isotopic distribution in Kiel Bight is to some extent a function of mixing of freshwater and seawater.

The dissolved Ba concentration and isotope composition resulting from mixing of freshwater and seawater are given by the following equations:

In Eq. 5 and 6, [Ba] denotes Ba concentrations and f_fresh and f_sw are the fractions of freshwater and seawater, respectively (f_fresh + f_sw = 1). Salinity along the estuarine mixing gradient is calculated analogous to Eq. 5. Because no Ba isotope data for rivers entering the Baltic Sea are available, the Ba concentration (116 nM) and δ¹³⁸Ba (+0.23 ‰) of Eurasian riverine freshwater (Bridgestock et al., 2021b) were used in Eq. 5 and 6. The catchment areas of both the Baltic Sea and Eurasian rivers are located at mid to high northern latitudes thus justifying this choice of δ¹³⁸Ba for the freshwater end member. The Ba concentration (43 nM) and δ¹³⁸Ba (+0.53 ‰) of North Atlantic surface seawater (Bridgestock et al., 2021b) was chosen for the seawater end member.

Dissolved Ba concentrations in laterally well-mixed surface waters (S = 15) plot on a mixing line between freshwater and North Atlantic seawater ( Figure 6A ). In contrast, denser (S = 19.2 - 21.6) and more restricted deep-waters are strongly enriched in dissolved Ba compared to the freshwater-seawater mixing line. The δ¹³⁸Ba of water column samples plot both above and below the mixing relationship between freshwater and North Atlantic seawater ( Figure 6B ) suggesting that the mean δ¹³⁸Ba of the water column Ba inventory can be explained by freshwater-seawater mixing. Importantly, however, the deviation towards higher δ¹³⁸Ba in surface waters and lower δ¹³⁸Ba in Ba-rich deep-waters cannot be explained by mixing of freshwater and seawater. Instead, Ba removal from well-mixed surface water coupled to Ba recycling in the restricted deep-water and/or mixing of surface water Ba with isotopically light Ba originating from the seafloor must be responsible for the observed trend towards higher Ba concentrations and lower δ¹³⁸Ba in the deep-water of Kiel Bight ( Figure 6C ).

With the exception of the uppermost samples at Stations 8 and 13, sediments in Kiel Bight are characterized by Ba/Al below the local terrigenous background (glacial till at coastal cliffs) ( Figures 4B, E, H ) indicating that Ba is lost from the terrigenous material upon cliff erosion, sediment transport or diagenesis. The Ba concentrations in the glacial till are higher than those reported for carbonates but generally consistent with those of aluminosilicate minerals (Dehairs et al., 1980; Nan et al., 2018). It is thus possible that Ba is lost from the solid phase due to ion exchange or alteration of aluminosilicate minerals upon contact with seawater. Since the δ¹³⁸Ba signature of the sediments is, within error, identical to the lithogenic background ( Figures 4C, F, I ), Ba release from aluminosilicates may contribute to the generally low δ¹³⁸Ba of brackish water in Kiel Bight compared to open-marine seawater. However, the isotope composition of the terrigenous Ba (~±0.0 ‰) is too low to exclusively explain the δ¹³⁸Ba of deep-waters and pore waters of surface sediments (~+0.25 ‰). Furthermore, as demonstrated in Section 5.2.2, simple mixing of sediment-derived Ba (e.g., from aluminosilicate minerals) and seawater Ba cannot explain the Ba versus δ¹³⁸Ba relationship in the water column ( Figure 7 ).

The highest dissolved Ba concentrations are observed in the pore water of the uppermost 1 or 2 cm of the sediment ( Figure 5 ) and the concentration gradient between surface sediments and bottom waters drives a diffusive Ba flux across the sediment-water interface. As demonstrated above, this efflux contributed to the relatively light isotopic signature of the bottom water ( Figures 5 , 7A ). High dissolved Ba concentrations within the pore water of surface sediments coincided with low solid phase total Ba concentrations but elevated Ba/Al (above the local terrigenous background at Stations 8 and 13) and TOC ( Figures 4B, E, H ). This observation indicates that both non-lithogenic Ba and fresh organic material are present within the uppermost sediment layer. Furthermore, intense remineralization of fresh organic material also occurs within the uppermost 1 - 2 cm of the sediment (Dale et al., 2013; Perner et al., 2022), which can explain Ba release into the pore water ( Figure 4 ).

We do not observe a correlation between pore water Ba and the major nutrients PO₄ ³⁻, SiO₄ ⁴⁻ and NH_4⁺ . However, such a correlation is not expected given that below the surface sediments (>1 - 2 cm), nutrient and Ba profiles are governed by different processes: PO₄ ³⁻, NH_4⁺ and SiO₄ ⁴⁻ concentrations continue to increase ( Figures 3E-G ) due to remineralization of refractory organic matter (Dale et al., 2013; Perner et al., 2022) and opal dissolution, respectively. In contrast, dissolved Ba concentrations decrease and approach equilibrium with respect to barite ( Figures 3H, I ). This observation suggests that Ba is precipitated as authigenic barite. In subsurface sediments, pore water and solid phase are in continuous contact, which is why dissolved Ba is more likely to reach thermodynamic equilibrium with respect to barite compared to the shallow water column of Kiel Bight. Below 18 cm sediment depth at Station 13, the pore water became undersaturated with respect to barite and dissolved Ba concentrations increased again ( Figures 3H, I ). This shift in saturation state was related to intense organic carbon degradation by sulfate reduction causing near-complete depletion of sulfate within the recovered sediment interval at this site ( Figures 3C-E ). Increasing concentrations of Ba in the pore water are thus attributed to well-known barite dissolution in the sulfate-methane transition zone (Torres et al., 1996; Riedinger et al., 2006).

In contrast to the water column in Kiel Bight and the open ocean (Horner and Crockford, 2021), pore water Ba concentrations and δ¹³⁸Ba in subsurface sediments do not show a mirror image but are correlated, i.e., low Ba concentrations (coinciding with SI_barite = ~0) are accompanied by low δ¹³⁸Ba whereas high Ba concentrations (coinciding with SI_barite < 0) are accompanied by high δ¹³⁸Ba ( Figure 5 ). This observation implies that authigenic barite precipitation at all our sites preferentially removes the heavy Ba isotopes and barite dissolution returns this heavy isotope signature back to solution (Station 13). Importantly, this finding is in apparent conflict with the results of barite precipitation experiments and water column studies showing that the light Ba isotopes are preferentially incorporated into the solid (von Allmen et al., 2010; Horner et al., 2017; Crockford et al., 2019). We note, however, that Ba isotope fraction during barite precipitation from pore water has not been studied to date. Moreover, the thermodynamic and/or kinetic circumstances of Ba isotope fractionation in sediments, suspended particles and laboratory experiments are likely to differ substantially, which is why differing extents and directions of isotope fractionation are conceivable. Different Δ¹³⁸Ba_{seawater-barite} obtained in experiments (~+0.3 ‰) and inferred from water column data (~+0.5 ‰) (Horner and Crockford, 2021) may also reflect differing environmental conditions during barium isotope fractionation. Support for our finding of a relatively heavy δ¹³⁸Ba in authigenic barite comes from a recent study about Black Shale weathering (Charbonnier et al., 2022). These authors demonstrated that weathering of organic carbon-rich sedimentary rocks mobilized Ba with a relatively heavy δ¹³⁸Ba compared to siliciclastic material.