Ten sediment cores (50 cm multi-corer) and 120 l of bottom water were recovered from 27 m water depth (below the euphotic zone) at Boknis Eck, Eckernförde Bay, with FK Littorina on January 13th 2022. These cores contained 20–24 cm of sediment and between 1.5 l and 2 l of bottom water. After recovery, the cores containing mainly fine-grained sediments of the clay to fine sand fraction, were sealed top and bottom with rubber plugs, held in upright position, and taken to a cool laboratory (12.5°C). One core was sliced for pore water and sediment analyses and one core was subsampled for bacterial 16S rRNA amplicon sequencing immediately after recovery. They served to provide data characteristic of the initial sediment conditions. Eight cores were kept for the incubation experiments. Dunite and calcite were used as alkaline materials to investigate their efficiency with respect to OAE. Dunite (Olivine AFS 80) was obtained from Sibelco^TM that ultimately derives from several quarries in Årheim (Norway). Via sieving, the 20–30μm fraction was extracted. For the mineralogical composition see Supplementary Table 1S. Since the dunite used in this study contains 88.8% olivine (Supplementary Table 1S) and this mineral dissolves much faster than the other components (Pokrovsky and Schott, 2000), it was assumed that olivine is the major dissolving silicate phase. For correct nomenclature, though, dunite will be used to describe the added material, whereas olivine will be used to describe the dissolution kinetics and stoichiometry of this mineral. Ground limestone (CaCO₃ > 99%, Jurassic, grain size < 0.1 mm) obtained from the German limestone association (Supplementary Table 2S) was first rinsed with deionized water (18.2 MΩ cm⁻¹-Milli-Q system). In order to avoid an unknown fine fraction remaining in suspension during the experiment, the material was first mixed with deionized water in a settling column (height 20 cm), and particles that remained in suspension after 20 s were decanted. This procedure was repeated until the supernatant water was clear after 20 s to ensure that after adding the material to the sediment cores, all particles sink to the sediment surface after several minutes. Due to the high calcite content, this material will henceforth be referred to as calcite.

For dunite grains, the reactive surface was calculated as the surface of spherical grains as the majority of studies report dissolution kinetics based on this assumption (Rimstidt et al., 2012).

For calcite grains, the reactive surface was determined via gas adsorption using the Brunauer-Emmett-Teller (BET) method following Brunauer et al. (1938) on a Stöhlein^TM AREA-Meter II. For this method two equally sized vessels (see Figure 1), one containing the sample and the other empty, are filled with nitrogen at atmospheric pressure at room temperature. Subsequently, both adsorption flasks are immersed in liquid nitrogen, cooling them to the boiling point of liquid nitrogen. At this temperature, the sample adsorbs nitrogen, resulting in a pressure difference between the adsorption flask and the reference flask, which is measured using a differential manometer. The specific surface area is calculated from this pressure difference, the initial pressure, and the sample weight. Results of the measurements and calculations are summarized in Table 1.

The eight sediment cores were placed in a rack in an upright position. The bottom water was carefully removed via suction and replaced with a known volume (1.5 l−2.0 l) of filtered (0.2 μm) Baltic Sea bottom water in order to remove pelagic auto- and heterotrophs and suspended particles. The volume of water added depended on the height of sediment in each core which varied slightly due to the recovery method. After this procedure, a gaseous headspace of ca. 10 cm was left in each core. Furthermore, the cores were equipped with adjustable stirring heads that contained ports for inserting optodes to continuously record pH and O₂ concentrations in the overlying water (for details see Section 2.4). In order to prevent anoxic conditions developing, ambient air was bubbled into the water column. The water column in each core was slowly and continuously flushed with a constant throughflow of 40 μl min⁻¹ from a single reservoir of bottom water. The residence time of the water inside the cores was thus about 4–5 weeks.

The cores were kept untreated for 3 weeks prior to mineral addition in order to equilibrate with the laboratory environment and to obtain a quasi-steady state natural background for each core. After this period, calcite (Cal1–Cal3) and dunite (Dun1–Dun3) were carefully spread on the sediment surface with three replicates for each treatment. Two control cores were left untreated (C1, C2). A precise description of each core is given in Table 1. The incubation was carried out in darkness for approximately 3 months at 12.5°C.

Bottom water samples were taken from the outflow of each core over a time period of several hours. Thus, samples represent the average outflow over the respective time period. Sampling intervals increased from daily during the first 2 weeks to every 3–4 days and weekly toward the end of the experiment. All samples were filtered through a 0.2 μm cellulose membrane filter and refrigerated in 25 ml Zinsser^TM scintillation vials. A 5 ml aliquot was frozen directly after the sampling procedure for later nutrient analysis. Nutrient measurements were performed either via manual photometric measurement (NH₄) or using a Seal—Analytical^TM QuAAtro autoanalyzer (PO43-, NO2-, NO3-). Samples for TA were analyzed directly after sampling by titration of 1 ml of bottom/pore water with 0.02N HCl following Ivanenkov and Lyakhin (1978). Titration was ended when a stable purple color appeared. During titration, the sample was degassed by continuous bubbling with nitrogen to remove any generated CO₂ and H₂S. The acid was standardized using an IAPSO seawater standard. Anion element concentrations (SO42-, Cl⁻, Br⁻) were determined using ion chromatography (IC, METROHM 761 Compact, conductivity mode). Acidified sub-samples (30 μl suprapure HNO3- + 3 ml sample) were prepared for analyses of major and trace elements (Si, Na, K, Li, B, Mg, Ca, Sr, Mn, Ni and Fe) by inductively coupled plasma optical emission spectroscopy (ICP-OES, Varian 720-ES).

At the end of the experiments, the bottom water was removed via suction and the cores were sliced for pore water, solid phase and bacterial 16S rRNA analysis. The pore waters were recovered by centrifuging each respective sediment layer in 50 ml falcon tubes at 3,000 rpm for 10 min. Afterwards, the supernatant water was transferred to polyethylene (PE) vials in an Ar-filled glove bag to minimize contact with oxygen. In addition to the parameters listed above, pore waters were analyzed for H₂S and Fe²⁺. For the analysis of dissolved Fe²⁺ concentrations, sub-samples of 1 ml were taken within the glove bag, immediately stabilized with ascorbic acid and analyzed within 30 min after complexation with 20 μl of Ferrozin. For H₂S, an aliquot of pore water was diluted with appropriate amounts of oxygen-free artificial seawater and the H₂S was fixed by immediate addition of zinc acetate gelatin solution. Further details are described by Dale et al. (2014, 2016).

Subsamples for bacterial community analyses were collected from surface sediments (0–2 cm) of all treatments and controls at the beginning of the experiment, after 42 days and at the end of the incubation (72 days, t6). At the end of the experiment (91 days, t7), not only the surface sediment but the entire core of one control (C1) was subsampled, with a resolution of 1 cm in the first 4 cm, followed by 2 cm sampling below the 4 cm horizon. The same resolution was used for the above-mentioned environmental start core that was also subsampled in a whole. All microbiological samples were immediately frozen at −80°C until further processing.

Bottom water pH and O₂ were continuously measured using logging devices (Pyroscience^TM PHROBSC-PK8 and OXROB3, respectively) calibrated following the instructions as provided by the manufacturer. Three TRIS-buffers were used to calibrate the pH probe for brackish bottom water conditions (Pratt, 2014; Müller et al., 2018). pH is reported on the total scale.

Total carbon (TC) and total organic carbon (TOC) as well as total nitrogen (TN) and total sulfur (TS) were measured in freeze-dried and ground sediment samples by flash combustion using the Carlo Erba Element Analyzer (NA-1500). Total inorganic carbon (TIC) was calculated by subtracting TOC from TC. The accuracy of the method was tested by method blanks and two reference standards [2.5-Bis(5-tert-butyl-2-benzo-oxazol-2-yl)thiophene (BBOT, HEKAteck^TM) and one internal sediment standard]. For more details see Bohlen et al. (2011) and Dale et al. (2011).

Sediment micro profiling was performed using a motorized Unisense^TM micromanipulator (MMS). Oxygen, pH and H₂S profiles were measured with Unisense^TM sensors Ox-50, pH-100 and SULPH-100, respectively. Vertical step sizes were adjusted to the respective sensor (minimum size was twice the needle diameter) and desired measuring resolution to keep the time for profiling as short as possible (100 μm for all O₂ profiles, 250μm for first pH and H₂S profiles, later 300 μm). All sensors were calibrated in accordance with manufacturer recommendations for saline solutions (Unisens^TM calibration kit for SULPH-100 with a calibration concentration of 109 μmol l⁻¹). Additionally to WTW^TM standard pH buffers (4.01, 7.00 and 10.01), the pH sensitivity of the pH-100 in brackish waters and the off-set was determined using three TRIS-buffers (Pratt, 2014; Müller et al., 2018). After obtaining H₂S and pH profiles, the total dissolved sulfur concentration was calculated according to the Unisense^TM guidelines based on Millero et al. (1988) and Jeroschewski et al. (1996). Unless otherwise stated, H₂S henceforth refers to total dissolved hydrogen sulfide (H₂S + HS⁻).

Solute fluxes were calculated based on concentrations in the inflow and the outflow following:

where F_e is the flux of any component (e) over the entire sediment surface (mol cm⁻² d⁻¹), dCedt is the measured concentration change of the respective element over a specific time (mol l⁻¹ d⁻¹) assuming that the concentration of this element is not affected by any process in the water column. k_w is the water exchange rate in d⁻¹, C_out is the outflow concentration (mol l⁻¹), C_in is the inflow concentration (mol l⁻¹), V_MUC is the bottom water volume (l) in each core and A_sed (cm²) is the sediment surface area of each core.

These calculations were performed for all measured solutes except oxygen and H₂S. Additionally, diffusive fluxes were calculated based on pore water profiles following:

where, F_sed is the flux of a respective solute (μmol cm⁻² d⁻¹), ϕ is the porosity, D_s is the molecular diffusion coefficient in the sediment and dC/dz is the concentration gradient where the flux is determined. D_s was calculated as:

where, D_M is the molecular diffusion coefficient in seawater and m is an empirical factor (here m = 3) (Boudreau, 1997). The D_M values were calculated following Boudreau (1997).

For a clearer visualization, the flux data were smoothened using a moving average. For further calculations, raw flux data were used (Supplementary material S1.1 and Supplementary Figure 1S).

In order to compare results from the different cores, the experiment was segmented into different time intervals. These time intervals were defined individually for each core in a way that each interval started intermediate between the times that microsensor profiles were obtained. The time intervals for each core are listed in Table 2 and reported as starting time to end time in days after addition of substrates.

The pore water O₂, H₂S and pH were assumed to be constant in each time interval. Subsequently the fluxes for O₂ and H₂S calculated from the microsensor data were averaged over the respective time interval to provide values in μmol cm⁻² d⁻¹ (Table 2). The oxygen penetration depth was defined as the depth where the oxygen concentration reached detection limit (~0.2 μM). pH values are reported as an average over the entire depth where each profile was obtained. The depth where H₂S first appeared below the sediment surface was defined as the depth where measured H₂S exceeded the analytical precision (~2 μmol l⁻¹). H₂S inventory for each time period were calculated by integrating bulk sediment concentrations over the upper 6 cm height and are reported as μmol cm⁻².

For studying the composition of the microbial community, a 16S amplicon library compatible with the Illumina MiSeq System was prepared as described previously (Böhnke et al., 2019; Perner et al., 2022). Briefly, RNA was extracted from around 300 mg sediment sample using the NucleoBond RNA Soil Mini kit (Macherey-Nagel, Düren, Germany) followed by subsequent DNaseI digestion using the RapidOut DNA Removal kit (Thermo Fisher Scientific GmbH, Waltham, Massachusetts, USA). Total environmental RNA was transcribed into cDNA with Invitrogen's SuperScript^® VILO^TM cDNA Synthesis Kit (Life Technologies^TM, Darmstadt, Germany). Purified cDNA (Zymo Clean & Concentrator-5, Irvine, CA, USA) was then used to generate 16S ribosomal amplicons for the bacterial hypervariable V3 and V4 regions using the primer pairs S-D-Bac-0341-b-S-17 and S-D-Bac-0785-a-A-21 as described by Klindworth et al. (2013). Sequencing was done in collaboration with the Competence Center for Genomic Analysis (CCGA) of the Institute of Clinical Molecular Biology (IKMB, CAU Kiel, Germany) in a 2x300 bp paired-end sequencing run on the Illumina MiSeq platform (Illumina, St. Diego, USA).

For sequence processing the Qiime2 workflow was applied (Bolyen et al., 2019). Here, filtering and merging of demultiplexed raw reads were performed using the dada2-plugin with default settings and removal of the primer sequences (Callahan et al., 2016). For taxonomic assignments, the SILVA database release 138 (99% OTUs full-length sequences) (Quast et al., 2013) was pretrained with the respective primer pairs for bacteria (Pedregosa et al., 2011; Bokulich et al., 2018). The assignments were computed using the feature-classifier plugin (classify sklearn) with default settings and the pre-trained SILVA classifier (Bokulich et al., 2018). The phylogeny was calculated using the “align-to-tree-maffft-fasttree” pipeline (Price et al., 2010). Statistical analyses and plots were performed using R software (R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/) using the “microeco” package (Liu et al., 2021).

Pore water analysis of one core was performed at the beginning of the experiment, and in all cores at the end of the experiment (Figure 1A). In all cores, SO42- concentrations decreased with depth. In the initial core, SO42- decreased to < 1 mM. Concentrations and penetration depths increased significantly in all cores over the course of the experiment, whereas the bottom water concentration increased only slightly from ~17 mmol l⁻¹ to ~18 mmol l⁻¹ (Figure 1A). In deeper sediment layers, the concentration increase was much larger, varying from 3 mmol l⁻¹ in Dun3 to 6 mmol l⁻¹ in Dun2 at 20 cm depth.

This downcore decrease in SO42- was mirrored by a downcore increase in H₂S (Figure 1B). At the end of the incubation, H₂S concentrations in all cores were lower compared to initial values. The H₂S loss relative to the initial values was slightly smaller than the SO42- increase with large variations between the different cores (~800 μmol l⁻¹ in Dun2 to ~2,000 μmol l⁻¹ in C2 at the end of the experiment compared to ~4,000 μmol l⁻¹ at the beginning in 20 cm depth). In the initially sampled core, the H₂S front (i.e. the depth where H₂S increases sharply) was about 3 cm. By the end of the experiment, this depth horizon had shifted downwards to ~5 cm and to ~7 cm in the cores Dun2 and C1, respectively. Exceptions to this trend are Cal2 and Dun3, where no change in the H₂S depth horizon from the initial value was observed.

In all cores, TA increased with depth (Figure 1C), but decreased in each depth layer over the course of the experiment by 10 mmol l⁻¹ (C2) to 15 mmol l⁻¹ (Cal2) compared to the initial values.

A different behavior was observed for NH4+. Compared to the initially sampled core, lower concentrations were measured in the upper sediment sections (0–7 cm) toward the end of the experiment in all cores. However, the concentrations in deeper layers did not decrease and fluctuated around initial values (Figure 1D).

Initial Si concentrations (Figure 1E) increased sharply in the upper 5–7 centimeters. This concentration gradient weakened with depth and reached asymptotic values of about 750 μmol l⁻¹ below 15 cm depth. In the cores sampled after the experiment, concentrations in all cores increased to ~850 to ~950 μmol l⁻¹ at ~10 cm depth. In the upper 5 cm, concentrations were mostly lower compared to initial values.

Initial Ca concentrations increased linearly from ~6.4 to ~7.0 mmol l⁻¹ at 7 cm depth, and then fluctuated around these values in deeper layers. Final concentrations in the upper sediment were markedly higher than initial ones, and up to ~6.8 mmol l⁻¹ (C1) and ~8.2 mmol l⁻¹ (Dun3). In most incubated cores, Ca concentrations showed peaks between 2.5 cm and 5 cm. An exception to this was core C2 with highest values (~7.8 mmol l⁻¹) observed at 10 cm depth. Below the peak, Ca concentrations decreased in all cores to values between ~6.4 mmol l⁻¹ (Dun1) and ~6.8 mmol l⁻¹ in C1.

After the addition of calcite and dunite, the substrates initially covered the sediment surface as an approximately 0.5 mm thick layer. Within the first 2 weeks the grains had been entrained into the sediment and were continuously transported further downwards until they reached a depth of 1–2 cm after about 6 weeks. By the experiment, the substrates were visible as a brighter layer in the sediment through the walls of the cores clearly distinguishable from the matrix (Supplementary Figure 4S).

In the initially sampled core, TIC content varied between 0 and ~ 4.5 wt.% with a slight increase downcore (Figure 2). In calcite-treated cores, TIC content at the surface at the end of the experiment was up to 16 % in the uppermost sample. C1 also showed values of ~6% in the upper 3 cm. After the experiment, TIC contents in all cores below 3 cm were in the same range, except in Dun3 where TIC contents up to 10% were measured at 15 cm depth. This core showed overall higher concentrations compared to the other treatments.

The pH distribution in pore water profiles was highly dynamic (Figure 3) with two major regimes that occurred at different times in nearly all cores (exceptions are Dun3 and C2). Initially, pH decreased by around 0.5 units in the oxic surface layer (~1– 2 mm, Table 2) and then increased in deeper sediments by up to 1.4 units (peaking around 4 cm) before decreasing again below 5 cm. Lowest surface pH (~7.0) was observed in Dun1 and Dun2 (Figure 3), along with highest values in deeper sediments (~8.4). Over the course of the experiment, this distribution gradually shifted to the second major regime that was initially observed in Dun3 and C2, that is, a clear pH peak with values up to 8.5 (Dun3) just below the surface. With increasing depth, pH decreased down to values of ~5.6 (C2).

TA concentrations in bottom waters increased over time in all cores over the first month of the experiments (Figure 4A). These changes are mirrored by corresponding TA fluxes (Figure 4B). Addition of alkaline minerals led to an almost immediate increase in TA fluxes for all calcite and dunite-treated cores. The reference core (C2) and dunite-treated core (Dun3) initially showed a very rapid TA increase to peak values of 4 mmol l⁻¹ (Dun3) to 5.5 mmol l⁻¹ (C2). Toward the end of the experiment, TA fluxes generally declined to about 1 - 2 μmol cm⁻² d⁻¹ for all cores treated with alkaline minerals. The two reference cores (C1, C2) covered a surprisingly large range in TA concentrations and fluxes despite no minerals being added.

Dissolved silicon concentrations (Figure 5A) increased in all cores during the early stage of the experiment to peak values between ~190 μM (Dun3) and ~375 μM (Cal1). The corresponding fluxes (Figure 5B) averaged between 0.28 μmol cm⁻² d⁻¹ and 0.65 μmol cm⁻² d⁻¹ in the respective cores during this period. During the 8 days prior to addition of substrates, the concentrations in all cores slightly decreased, as reflected in decreasing fluxes. An increase in Si concentrations between day −1 and day 0 was associated with elevated fluxes just prior to mineral addition. A strong decrease in average fluxes around day 0 led to a slight decrease in Si concentrations. Afterwards the increasing trend in concentrations was a common feature for all cores with no significant trend ascribable to a specific treatment. Over the course of the experiment, the fluxes smoothened out and varied between 0.16 μmol cm⁻² d⁻¹ (Cal1) and 0.3 μmol cm⁻² d⁻¹ (C1).

Dissolved calcium (Ca) concentrations showed more muted trends than TA fluxes, yet clearer trends than Si concentrations (Figure 6A). Ca concentrations decreased in all cores shortly before mineral addition and then showed an overall increase in all cores over the course of the experiment. The highest Ca concentrations were observed in those cores that also reached anomalously high TA values (C2, Dun3, Figures 4A, 6A). The addition of calcite led to an almost immediate increase in dissolved Ca fluxes (Figure 6B, Cal1–Cal3). Ca fluxes decreased after a few days and reached values in the range of 0.3–0.8 μmol cm⁻² d⁻¹ toward the end of the experiments for all calcite-treated cores. Interestingly a similar development of fluxes was observed in dunite treated cores.

Throughout the experiment, nickel concentrations stayed below detection limit (~0.2 μmol l⁻¹) both in pore waters and in bottom waters (data not shown).

Oxygen concentrations in bottom waters (Supplementary Figure 2S) varied widely during the early stage of the experiment when oxygen was only supplied by diffusion across the air/water interface. The lowest concentrations were measured between day 2 and 3 in Cal2 and C1 (~90 μmol l⁻¹). After bubbling the water column with ambient air directly in the liners, oxygen concentrations increased and stabilized around full saturation (~292 μmol l⁻¹) in all cores. Higher values probably result from small gas bubbles stuck to the tip of the optode. Temporarily lower values occurred in Dun3 (~120 μmol l⁻¹, day 69), Cal2 (day 1) and C2 (~90 μmol l⁻¹, day 2).

Bottom water pH increased in all cores over the course of the experiment (Supplementary Figure 3S). The earliest and strongest increase occurred in C2 (pH 8.2 on day 15) and Dun3, in accordance with the high TA fluxes observed in these cores. Although pH remained at a high level in C2 (~8.15), it decreased in Dun3 after day 15 to values around 7.7. In the other cores, pH values fluctuated around 7.2–7.4 until the addition of substrates and the beginning of the bubbling of bottom waters. Subsequently, a strong increase to values around 7.9 was observed until day 15 that was most pronounced in the calcite treated cores and C1. The weakest increase was observed in Dun2 where pH reached values of ~7.75. After day 25, all cores (except C2 and Dun3) followed a rather similar trend of a slow pH increase. Final values ranged from 7.9 (Dun1) to 8.2 (C1).

The assessment of the time intervals in each core reveals common trends in all cores except Dun3. Highest calcite dissolution potentials (Diss_pot, defined in Section 4.4) were observed toward the end of the experiment. The overall range of these values was between 0.0017 cm (Cal2, t3) and 3.8898 cm (C2, t8). pH values decreased in all cores over the course of the experiment, yet again with the exception of Dun3. Another common feature is the shallowing of the H₂S front between 53.5 mm (Cal1, t1) and 11.0 mm (Dun3, t7). This decrease is not clear in all cores, however. The H₂S fluxes varied over the course of the experiment, with a slight tendency toward lower values at the end of the experiment. TA and Ca fluxes varied with highest values around t2 in most cores. Si fluxes did not show any obvious time dependency.

Based on RNA-profiling of 16S rRNA genes, the active parts of the microbial communities of the different incubations were dominated by (i) phylogenetically distinct chemolithoautotrophic, sulfide oxidizing Gammaproteobacteria, primarily related to Beggiatoaceae (mostly Candidatus Parabeggiatoa and Candidatus Isobeggiatoa), Thiotrichaceae (uncultured), Ectothiorhodospiraceae (mostly Thiogranum) and Chromatiaceae, (ii) sulfate-reducing Desulfobacterota of the Desulfosarcinaceae and Desulfobulbaceae, (iii) members of the Planctomycetes, mostly heterotrophic Pirellulaceae (Rhodopirellula, Blastopirellula) and the Pla4 lineage as well as Cyanobacteria affiliated with Nostocaceae (Aphanizomenon) and Cyanobiaceae (Figure 7). Sulfurimonas sp. (Campylobacterota), hallmarked by versatile metabolisms (hydrogen, sulfur and nitrogen cycling) (Han and Perner, 2015), and the uncultured Sva0081 (Desulfobacteraceae), which has been suggested as important H₂-scavenger in marine sediments (Dyksma et al., 2018), were abundant throughout the treatments (Figure 7). Cable bacteria associated with Candidatus Electrothrix were found in all incubations, albeit in low numbers (Figure 7). Maximum values at the beginning of the incubations were found in Cal1 (1.8%), Cal2 (3.6%), Dun2 (2.3%) and Dun3 (1.4%), and after 72 days of incubation in C2 (1.8%), Cal3 (2.2%), Dun1 (3.0%), Dun2 (1.8%) and Dun3 (1.6%) (Figure 7). Abundances of Candidatus Electrothrix in other cores were below 1%.

When interpreting results from laboratory experiments, it is instructive to scrutinize the set-up of the experiment for deviations from the natural system. Key considerations of potential importance for interpreting the results of this study include the following:

A careful discussion of pore water alteration is essential when interpreting the results of laboratory experiments. In this study, several key considerations highlight the importance of assessing the potential deviations from the natural system. Despite these complexities, it is important to acknowledge that these challenges are inherent to studies involving such kind of laboratory experiments, and the data obtained from such experiments thus remain valuable and informative for understanding benthic weathering processes under experimental conditions.

The most striking change in pore water solutes was the loss of H₂S and a concomitant increase of SO42- concentrations between the initially sampled core and the final sampling (Figures 1A, B). H₂S production is mainly driven by the anaerobic oxidation of methane (AOM) and particulate organic carbon oxidation via SO42- reduction (POCSR) (Barnes and Goldberg, 1976; Froelich et al., 1979; Mark Jensen et al., 2003; Torres et al., 2020):

AOM:

POCSR:

Thus, both processes convert SO42- to H₂S with the formation of 2 moles (AOM) and 2.28 moles (POCSR) of TA per mole of SO42-.

Since the concentration gradients of SO42- and H₂S at the bottom of the initial core were close to zero (Figures 1A, B), a reduced H₂S flux from deeper sediments as a consequence of plugging the bottom of the cores can be ruled out as an explanation for the lower final H₂S concentrations compared to the initially sampled core. Toward the end of the experiment, the constant SO42- concentrations down to ~3 cm below the sediment surface point toward reduced POCSR, which is also observed in the lower NH4+ concentrations (Figure 1D). This decrease is likely a consequence of a lack of supply of fresh POM during the incubation. The deeper penetration of SO42- and the lower H₂S and TA concentrations below 3–5 cm could point toward lower POCSR, but the NH4+ rather indicate little change from the initial POCSR. Thus, it appears that the production of H₂S independently from SO42- was limited during the experiment. The most likely explanation is that the plug introduced at the bottom of the cores shut down the upward transport of methane (CH₄) from deeper layers as described for Boknis Eck sediment e.g. by Maltby et al. (2018). This cutoff from CH₄ supply would result in a decline in AOM rates below 20 cm where the AOM zone was likely situated. Considering the stoichiometry of AOM (Equation 7), this change allowed for a decrease in SO42- consumption and deeper SO42- penetration and a decline in dissolved H₂S and TA production rates and concentrations at depth (Equation 7).

The apparently stronger decrease of H₂S over time compared to the increase in SO42- may be a consequence of shifted Fe-H₂S balance and a possibly stronger formation of FeS. Also, the elevated laboratory temperatures compared to initial in situ conditions can foster the formation of pyrite (Thiel et al., 2019). The decrease in H₂S concentrations and flux to the upper layers of the sediment may have affected the microbial community and thus the experimental results.

In addition to reduced AOM, the lower TA values in deeper sediment can point toward authigenic formation of CaCO₃. This is in line with lower Ca²⁺ concentrations in deeper sediments toward the end of the experiment (Figure 1F). This formation does not appear to be influenced by the treatment of the different cores, but is most likely a consequence of the increase in temperature during the incubation since CaCO₃ solubility decreases with increasing temperature (Zeebe and Wolf-Gladrow, 2001).

Overall, the pH profiles of most sediment cores looked very similar at the start of the experiment and then evolved over time (Figure 3). Pore water pH is driven by a variety of oxidation and reduction processes which either release or consume protons. These processes are mainly driven by microbes that utilize the respective Gibbs free energy that the different chemical reactions yield (Fenchel and Jørgensen, 1977; Froelich et al., 1979).

Previous work has demonstrated that the pH profiles in the organic rich sediments of Boknis Eck are dominated by Fe³⁺ reduction coupled to H₂S oxidation and subsequent Fe²⁺ oxidation via manganese oxide (MnO₂) (Preisler et al., 2007). A suboxic zone between the H₂S horizon and the maximum oxygen penetration depth (Table 2) is commonly observed in sediments inhabited by the filamentous H₂S -oxidizing bacteria of the family Beggiatoacea that use nitrate to oxidize H₂S via Dissimilatory Nitrate Reduction to Ammonium (DNRA) (Schulz and Jørgensen, 2001; Preisler et al., 2007):

These bacteria can store nitrate within their intracellular vacuoles and transport it down into the sediment, leading to the formation of a suboxic zone (Jørgensen and Nelson, 2004). H₂S oxidation by Beggiatoa spp. is typically characterized by low pH values in the oxic zone and a clear peak around the H₂S horizon (Jørgensen and Revsbech, 1983; Sayama et al., 2005). In most cores (except Dun3 and C1), this situation is reflected in the initial pH conditions at the beginning of the experiment (Figure 3) and the presence of Beggiatoa spp. is supported by bacterial 16S rRNA amplicon sequencing data (Figure 7 and Discussion Supplementary material S3.1).

Over time, though, in all cores except Dun3 and C2, the pH profile shifted toward a shape that was observed initially in Dun3 and C2, indicating a change in the predominant H₂S oxidation pathway. Instead of decreasing, pH values increased in the upper millimeter of the sediment, peaking at around the maximum oxygen penetration depth (Figure 3 and Table 2). These maxima were situated above a very strong decrease in pH toward the lowest measured values of ~5.7 in core C2. The pH minima were observed slightly above the H₂S horizon. The latter type of profile shape has previously been associated with the activity of cable bacteria (Nielsen et al., 2010; Malkin et al., 2014; Meysman et al., 2015). Like Beggiatoa spp., these bacteria oxidize H₂S to SO42-, but in two spatially separated reactions, termed long-distance electron transfer (Nielsen et al., 2010; Risgaard-Petersen et al., 2012; Meysman et al., 2015):

Step one takes place around the H₂S horizon. The electrons released are then transported upwards within the bacterial filaments to the oxic layer where step two takes place. Thus, per mole of H₂S oxidized, 10 moles of protons are released in deeper sediments and 8 moles of protons are consumed in the oxic surface layer. The combination of these two processes would explain the pH peak at the surface and low pH values around the H₂S horizon (Figure 3 and Table 2) (Malkin et al., 2014; Meysman et al., 2015, 2019). Thus, the pH profiles in cores Dun3 and C2 resemble profiles that would be generated by cable bacteria activity, but over time this activity appeared to cease, which is shown by the less pronounced surface peak and increasing pH values in pore waters. This hypothesis stands in contrast to the RNA-profiling data where, according to 16S rRNA gene proportions, cable bacteria, namely Candidatus Electrothrix increased in C2 and remained unchanged in Dun3 over time (Figure 7). Although the pH profiles of most incubations support increasing activity of cable bacteria over time of the experiments, we could not verify this by identifying increasing proportions of 16S rRNA genes from known cable bacteria. A detailed discussion on the microbial development is provided in the Supplementary material S3.1. With regards to the aims of this study, it is important to highlight that it cannot be resolved conclusively whether the changes in pore water pH are a naturally occurring feature or were induced by the laboratory incubation conditions. This will be considered for the evaluation of enhanced weathering in the further discussion.

The easiest approach to evaluate calcite dissolution is the assessment of bottom water concentrations of the weathering products (TA and Ca) and the respective fluxes via the sediment surface. Figures 4A, B indicate a slightly increased TA accumulation and elevated fluxes in cores Cal1 – Cal3 directly after the addition of calcite. These fluxes combined with elevated Ca concentrations and fluxes (Figures 6A, B) point toward dissolution in bottom waters during addition. Still, with regards to the overall experiment, there is no clear trend in Ca fluxes, or in TA fluxes or corresponding bottom water concentrations that could be attributed to a certain treatment. The reason for this observation is obviously the large inhomogeneity of the different cores with regards to pH values which are a major driver for calcite dissolution (Walter and Morse, 1985). In order to reliably quantify weathering processes in the different cores it is necessary to evaluate them under the geochemical conditions that foster or inhibit calcite dissolution during different time intervals (Table 2).

The rate of calcite dissolution is driven by saturation state with respect to calcite (Ω_cal; Walter and Morse, 1985 and references therein). In this study, CO32- concentrations were calculated using measured pH and TA pore water values, leading to estimations of Ω_cal:

where [Ca] and [CO32-] are the concentrations of calcium and carbonate, respectively, and Ks_cal is the solubility product of calcite at the respective salinity, temperature and pressure (Zeebe and Wolf-Gladrow, 2001). To provide temporal estimates of Ω_cal, the initial and final values of TA and Ca were interpolated linearly for each depth layer. We recognize the obvious uncertainty associated with this approach and the following results should be interpreted with this in mind.

The total amount of calcite that can be dissolved is driven by the degree of undersaturation over the whole core. Since calcite only dissolves when ambient pore waters are undersaturated with respect to this mineral (Ω_cal < 1), a dissolution potential [Diss_pot (cm)] was defined by integrating Ω_cal over the region where undersaturation was observed:

where Ω_cal(z) is the saturation state at depth z (cm) in the sediment. z(Ω < 1) and z(Ω>1) refer to depths where Ω_cal values fall below 1 and above 1, respectively. Thus, Diss_pot combines the degree of undersaturation and the depth interval where undersaturation occurs (Figure 8).

Diss_pot values depend on the dynamic interplay between pH and calcite dissolution. When CaCO₃ dissolves in pore waters, the saturation state with respect to CaCO₃ increases due the accumulation of weathering products. This has the effect of lowering Diss_pot. At the same time, a higher sedimentary flux of the weathering products to the bottom water should be observed. Consequently, Ca and TA fluxes together with Diss_pot values at a given time indicate how much calcite is being dissolved compared to how much calcite could be dissolved. The higher the calcite content of a sediment that bears a source of protons, the higher the ratio of Ca and TA fluxes to Diss_pot, the more effective is the dissolution process. The maximum dissolvable calcite content would then be indicated by high fluxes combined with Diss_pot values close to 0.

Highest Diss_pot values are associated with elevated TA and Ca fluxes during the discrete time intervals in cores C2 and Dun3 (Figures 9A, B). This clearly indicates that the natural inventory of calcium carbonate in the form of either calcite or aragonite (Figure 2) is being dissolved in corrosive pore waters. To disentangle these natural fluxes from enhanced fluxes induced by the addition of calcite, the pH profiles were compared to profiles obtained at Boknis Eck in situ under natural conditions (Preisler et al., 2007). At the time of sampling, these sediments were dominated by the activity of Beggiatoa spp. and were more similar to the initial profiles in e.g. Dun1 and Cal1. Therefore, based on these in situ profiles and Dun1 and Cal1 profiles, a realistic maximum value for Diss_pot was defined (dashed line in Figures 9A, B). This approach was chosen for because it only considers naturally observed and thus unambiguously realistic weathering regimes. In addition, not all cores showed extreme Diss_pot values. By considering time intervals with similar Diss_pot values, comparisons between different treatments are more justified.

For the evaluation of enhanced carbonate weathering, the fluxes of the different cores during time periods where Diss_pot was below this value were then averaged and compared (Table 3). This procedure excludes values obtained in C2, as in this core pH was very low over the entire course of the experiment. The results of the comparison show that TA fluxes were 0.961 ± 0.096 μmol cm⁻² d⁻¹ higher in calcite treated cores than in the control experiment. The ratio of F_TA/F_Ca of 1.45 (Table 3) in the calcite treated cores is lower than the theoretical value of 2 (Equation 2), which points toward a loss of alkalinity which is not associated with Ca. We suggest that TA derived from CaCO₃ dissolution was diminished by the oxidation of reduced species in the oxic surface sediment (Wallmann et al., 2008; Dale et al., 2011, 2021; Rassmann et al., 2020; Perner et al., 2022). The ΔF_TA/ΔF_Ca values which represent the ratio of enhanced fluxes compared to the control experiment, though, are close to 2 (1.99), which would be the ratio expected for calcite dissolution. Hence, the additional fluxes were most likely induced by the addition of calcite to the sediment. Overall, it is important to acknowledge that the accumulation of weathering products in bottom waters and the low pCO₂ have led to higher Ω_cal values than expected in the natural system (Melzner et al., 2013). In more corrosive bottom waters the dissolution of calcite might be higher than observed in this study.

To assess the suitability of weathering products as a proxy for mineral dissolution, TA and Ca concentrations in bottom waters were directly compared over the entire experiment (Figure 9C). The data clearly show that the ratio of TA/Ca is not always 2 (indicated by the thin red line, Figure 9C), as expected from equation 2. Instead, TA appears to increase before Ca concentrations. This is an especially well pronounced feature in C2 and Dun3, in which highest TA concentrations were observed during the early stage of the experiment. In the other cores this enhanced TA/Ca ratio was observed toward the end of the experiment (color coding in Figure 9C).

When comparing the different pH profiles in C2 and Dun3 (Figure 3), it is clear that very low values and subsequently high Diss_pot values dominated during the early stage of the experiment. During this time, the TA:Ca ratio in bottom waters is clearly higher than 2:1 (blue to green symbols in Figure 9C). In the other cores, similar conditions appeared later in the experiment (yellow to red symbols in Figure 9C). Since the low pH values presumably result from cable bacteria, their metabolism (Equations 9, 10) might lead to an altered TA/Ca ratio. The different diffusive coefficients of Ca²⁺ and HCO3-, which is the main contributor to TA, might alternatively be responsible for this effect, as D_Ca < D_HCO3 (Boudreau, 1997). This might also explain why the ratio of TA/Ca decreased in C2 and Dun3 toward the end of the experiment. Along with the higher consistency of measured TA values and calculated fluxes, this suggests that a combination of TA and Ca may be the best way to assess weathering efforts, even though TA is the main driver for CO₂ uptake in the surface ocean (Wolf-Gladrow et al., 2007) and thus of more relevance with respect to CDR studies.

Dunite is an igneous rock with an olivine content of over 90% (Hochstetter, 1859) and displays rapid dissolution rates in seawater (Wolff-Boenisch et al., 2006; Rimstidt et al., 2012; Rigopoulos et al., 2018; Fuhr et al., 2022). Assuming that dunite dissolution represents olivine dissolution, the rate can be described as a function of temperature and pH (Rimstidt et al., 2012):

where r_ol is the dissolution rate given in mol m⁻² s⁻¹, T is temperature in K. In the temperature and pH range of this study, a pH increase from 7 to 8 would lead to an identical 55% decrease in rate as a theoretical temperature change from 12.5 to 6.59°C. Thus, the pH changes observed in our experiment (Figure 3) must play an important role for olivine dissolution.

The chemical process of olivine dissolution ought to lead to an accumulation of magnesium (Mg²⁺), Si and TA (Equation 1). The high Mg background of ~32 mmol l⁻¹ in seawater at Boknis Eck prevents detection of the additional Mg due to analytical precision (±2%). Hence, the candidate proxies for detecting olivine dissolution are Si and TA concentrations. However, Si concentrations and fluxes over entire course of the experiment showed no trends between the differently treated cores (Figure 5). TA concentrations and fluxes (Figures 4A, B) slightly increased directly after the addition of dunite in Dun1 and Dun2, pointing toward olivine dissolution (Fuhr et al., 2022). Since Dun3 was dominated by CaCO₃ dissolution in the early stage of the experiment, the TA fluxes from olivine dissolution in this core were strongly overprinted.

Pore water pH values were compared to Si fluxes averaged for each time step (Figure 10). A slightly lower gradient of the linear regression between pH and Si fluxes in cores treated with dunite might be interpreted as increased contribution from olivine dissolution for lower pH values. Still, the correlation for all cores is weak (R² values between 0.241 and 0.072). Overall, the increase in Si flux with higher pH values in all cores points toward dissolution of either siliceous detrital material or biogenic debris such as diatomaceous frustules that are highly abundant in Baltic Sea sediments (Lewin, 1961; Gasiunaite et al., 2005). These processes result in a high background signal that overprinted Si contribution by olivine weathering at lower pH values.

When comparing the different treatments during comparable time intervals (Table 3), cores treated with dunite showed slightly enhanced TA fluxes (0.634 μmol cm^−1. d⁻¹) compared to the control experiment (0.412 μmol cm^−1. d⁻¹). The very large standard deviations of 40 % and 73 %, though, diminish the significance of these data. Overall, the data does not allow for olivine dissolution to be evaluated in the experiment neither qualitatively nor quantitatively using these proxies.

Dissolved nickel, which was proposed as a dissolution proxy by Montserrat et al. (2017), was below detection limit (~0.2 μmol l⁻¹) in all samples. Consequently, although it cannot serve as a dunite dissolution tracer, it can likely be excluded as a possible toxic threat to fauna (Flipkens et al., 2021) in the current benthic environment. Likewise, the formation of secondary mineral formation associated with olivine weathering (Fuhr et al., 2022) cannot be resolved based on the present data set.

Marine alkalinity enhancement via enhanced weathering at the seabed requires monitoring to quantify rates of mineral dissolution and alkalinity production and ultimately the uptake flux of atmospheric CO₂ in the ocean induced by deliberate mineral addition. To this end, a methodology needs to be established and suitable tracers have to be defined that allow for a reliable quantification of mineral dissolution rates.

Dissolved Si would be a suitable tracer to quantify dissolution rates of olivine and other reactive silicate minerals added to the seabed. However, our experimental data clearly show that benthic Si fluxes are dominated by the dissolution of biogenic opal and other sedimentary Si phases such that olivine dissolution rates could not be quantified using this tracer. Since biogenic opal produced by diatoms and other siliceous plankton is deposited and dissolved in almost all coastal and shelf sediments (Tréguer et al., 2021), the attribution problems that were encountered during this study will very likely also emerge during field experiments and large-scale application of benthic marine alkalinity enhancement. It may be possible to solve this attribution problem using stable Si isotopes (δ³⁰Si) since biogenic opal and detrital silicate minerals have distinct isotopic compositions such that benthic fluxes induced by the dissolution of these separate phases can be differentiated (Geilert et al., 2020). Future work should, hence, test whether δ³⁰Si may be a suitable tracer of the dissolution of deliberately added silicate minerals.

Dissolved Ca could be used to quantify dissolution rates of carbonates (calcite, aragonite) added to the seabed. Data obtained in this study indicate that this may be possible but that the interpretation of benthic Ca flux data is complicated by the dissolution of calcite delivered by coastal and seabed erosion in the study area (Wallmann et al., 2022). Similar problems may arise in other coastal and shelf settings where carbonate minerals produced by erosion and calcareous plankton are deposited and dissolved at the seabed. Since the stable isotopic composition of Ca and Sr bound in natural and deliberately added carbonate minerals may differ (Teng et al., 2021), employing these isotope tracers (e.g. δ^44/40Ca, δ^88/86Sr) might improve attribution.

According to our experimental results, TA may be the most suitable tracer to quantify the dissolution of alkaline minerals (olivine, calcite) added to the seabed. However, large TA fluxes were not only induced by the dissolution of added minerals but also by the dissolution of naturally-occurring carbonate minerals. Moreover, observed TA was likely consumed via H₂S and iron oxidation. The resulting benthic TA fluxes were, hence, strongly modulated by microbially catalyzed redox reactions and very sensitive to shifts in the microbial community (primarily Beggiatoa spp. vs. other microbes including cable bacteria). Since these TA-producing and consuming microbial metabolisms occur in all coastal and shelf sediments (Boudreau, 1997), attribution problems will also arise in field experiments and applications. Attribution of TA fluxes to the dissolution of added alkaline mineral will, therefore, only be possible after accounting for highly dynamic natural TA background fluxes that can change on seasonal and shorter time scales in most benthic environments (Dale et al., 2013). Overall, the difficulties encountered with attribution will subsequently complicate monitoring, reporting and verification (MRV) in a possible field application scenario.