The studied lakes span across a 400 km latitudinal gradient in Finland, all within the boreal zone (Fig. 1). The lakes and sampling sites were chosen based on their water quality, using aerial and satellite images showing the evidence of extensive areas of macrophyte vegetation on Finnish lakes, and accessibility of the sampling sites (Figs. S1 to S9b).

The clearwater and mesotrophic Lake Vesijärvi is the smallest and southernmost lake sampled in the study, while the humic and oligotrophic Lake Kallavesi is located in central Finland, and the humic and dysoligotrophic Lake Oulujärvi with large (2.7 m) water level regulation is the largest and northernmost lake (Tables 1 and S1). All three lakes are located north of the youngest of the Salpausselkä ice marginal sand ridges that are among the most conspicuous glacial formations in Finland (Rainio et al., 1995). The Salpausselkä I and II formations were laid down during the Younger Dryas cold climatic period, and constitute a zone approximately 20–50 km wide and 600 km long across southern Finland, which is composed mainly of glaciofluvial gravel and sand (Tikkanen, 2006; Rainio et al., 1995). In the regions north of the ice marginal formations, large deposits of gravel and sand are also found in eskers, representing the locations of subglacial streams and oriented approximately in the direction of glacial movement (Tikkanen, 2006; Hyvärinen, 1973). Other widespread landforms dating from the deglaciation and subsequent land uplift are large clay plain and mires (Tikkanen, 2002). This mosaic of surficial materials contributes to the soils and thus also the freshwater sediments in the regions of the three study lakes.

Lake Vesijärvi is located between Salpausselkä I and II formations and, thus, the region is mainly covered by surficial deposits of glaciofluvial sands (GTK, 2024) (Fig. S10). Lake Kallavesi is located in a region with a notable presence of eskers (Glückert, 1974) and its surroundings are mainly covered by bedrock outcrops, sandy eskers and mixed soil (GTK, 2024) (Fig. S10). Because of its northern location, lake Oulujärvi's surroundings also present some sandy eskers in the surficial deposits, but are mainly characterized by mixed soils and a thick peat layer (GTK, 2024) (Fig. S10).

The sampling campaign was conducted in August 2023 in the vegetated shoreline areas of Vesijärvi, Kallavesi and Oulujärvi. From each lake, three sites, each including a depth transect of 3 zones, were sampled yielding a total of nine sediment cores per lake (Fig. 2). At each site, sampling was performed along a depth transect across the vegetation patch (one core from each of the following zones: landside, transitional, waterside; Table S2; Figs. 2 and S1 to S9d). Each site was selected based on existing evidence of macrophyte patches from satellite (Koponen et al., 2022) and aerial (Finnish National Geoportal (Orthophotos), 2023) images (Figs. S1 to S9b), and aligned with sites where previous macrophyte campaigns had also been carried out. Regular grids (10×10 m or 5×5 m, depending on the size of the vegetation patch) were created for each macrophyte patch. Based on satellite images (Koponen et al., 2022), the closest grid points to the lake shoreline were selected as potential starting points (i.e., the “landside” sampling point) of each transversal transect. For each sampling site (vegetation patch), the starting point was randomly selected using the “Random selection” tool of QGis (QGIS Development Team, 2024) (Figs. S1 to S9c). The starting points were always within the vegetation patch but could be either submerged or waterlogged (standing water depths ranging between 0.05 and 0.60 m). “Transitional” and “waterside” sampling points followed a transect perpendicular to the shoreline across the vegetation patch (Figs. S1 to S9d). “Waterside” sampling points were the furthest grid points from the starting points (“landside”) within the vegetation patch based on satellite images (standing water depths ranging between 0.37 and 1.23 m). “Transitional” sampling points were selected as the middle grid points between “landside” and “waterside” sampling points (standing water depths ranging between 0.05 and 1.13 m).

Sediment sampling was performed using a combination of two hand-held coring systems: a short piston corer (0.5 m length, 4 cm internal diameter) and a long box corer (100 cm length, 8×8 cm cross-section area). The corers were inserted into the sediments and pushed down until 1 m was reached, or until no further penetration was possible. Final depths obtained varied by site and zone (20–100 cm), and were mostly determined by the transition from organic-rich surface material to more compacted organic-poor material deeper in the core. Each of the 27 cores was sliced at 2 cm resolution in the first upper 20 cm, and 10 cm resolution until the bottom of the core. Sediment cores that had a clear compositional transition (e.g. from fine grained organic rich sediments to organic poor clay) below the top 20 cm were sliced at 5 cm resolution until reaching that transition, and then at 10 cm resolution downwards.

Samples (n = 359) were bagged and kept refrigerated until weighed in the laboratory. They were then frozen, freeze-dried and weighed again to determine dry mass [g] and water content [%]. Because some of the samples contained large rhizomes and pebbles, each sample was broken up with mortar and pestle and sieved through a 2 mm mesh for further analysis. The fraction retained in the sieve (> 2 mm) was weighed, and visually categorized as mainly “biogenic”, “lithogenic” or a “mixture” of both (Figs. S11 to S13). For the purposes of visualization, we classified the sieved samples (< 2 mm) into three categories based on grain size (d50, Sect. 2.3) and total organic carbon (TOC) content: “sands” (d50 ≥ 62.5 µm), “OM-poor (grey) silts” (d50 < 62.5 µm and TOC ≤ 5 % dry weight (dw)) and “OM-rich silts” (d50 < 62.5 µm and TOC > 5 % dw).

In each zone, in addition to the sediment core, vegetation stems for all macrophyte species were counted and identified in a quadrat of 1 m².

Organic matter (OM) content was measured by dry combustion at 550 °C (loss on ignition, LOI) using a LECO TGA701 analyser (LECO Corp., St Joseph, MI, USA; precision of 0.02 % RSD – relative standard deviation), while TOC was determined by dry combustion using a LECO CN828 analyser (LECO Corp., St Joseph, MI, USA). TOC limit of detection was 0.000017 %. OM was measured in all samples, whereas TOC was determined in 173 out of 359 samples taken alternately in each of the sediment cores, i.e. at every other sample (Fig. 2). These data were used to estimate TOC content in the remaining 186 samples by linear regression (a conversion factor of 0.44; r2=0.98) (Fig. S14).

Grain size was measured on the sieved and inorganic fraction of the sediments using a Microtrac FlowSync particle size analyser and is reported as % of fine-grained sediments. The precision of measurement based on replicates from an in-house sediment reference sample was 3 %.

The longest (0.9 m) sediment core (lake Kallavesi, Site E, zone transitional) was selected to be used as a reference core for sediment dating. Additionally, three samples within single sediment cores from the remaining eight sites were also selected (Fig. 2). All selected samples were analysed for ¹³⁷Cs activity using BrightSpec gamma spectrometer (3600 s counting time, Geological Survey of Finland). We prospected for a single major peak in the profile assumed to correspond to the 1986 Chernobyl nuclear accident (Ojala et al., 2017), as evidence for continuous sediment accumulation in the reference core.

Dry bulk density (DBD) was determined from the sediment density corrected by organic matter content (Eq. 1), the water content and the sample volume. Wet and dry sample weights were used to calculate the sediment content [% relative to sample weight] (Eqs. S1 and S2), which was in turn converted into sediment volume [% relative to sample volume] and porosity Φ [fraction of 1] (Eqs. S3 to S5). Finally, porosity [fraction of 1], sediment density [g cm⁻³] and sediment content [%] were used to calculate wet bulk density (WBD) (Eq. 2) and dry bulk density (DBD) (Eq. 3). 1Sedimentdensitygcm-3=1.4×OM100+2.65×1-OM100, where 1.4 g cm⁻³ is the assumed density for sediments rich in OM and 2.65 g cm⁻³ is the grain density of quartz (Pajunen, 2000; Burdige, 2006). 2WBDgcm-3=Φ×1+(1-Φ×Sedimentdensitygcm-3),3DBDgcm-3=Sedimentcontent%100×WBDgcm-3, where Φ= volumetric porosity. TOC inventory for a given sample [g] was calculated multiplying the TOC content [%] by the DBD [g cm⁻³] and the sample volume [cm³]. For the C stock calculations, only samples with > 1 % OM were considered (Kortelainen et al., 2004). The TOC inventories were summed for all sampled depths to obtain a cumulative inventory [g] per core. Lastly, the cumulative TOC inventories [g] were divided first by 1000 (to obtain data in kg) and then by the coring device cross-section areas [m²] to obtain sedimentary organic carbon (SOC) stocks [kg m⁻²] (Eq. 4a). 4aSOCstockkgm-2=∑corebottomcoretopTOC×DBD×volume/1000area,4bSOCstock(0-20)kgm-2=∑20cmdepthcoretopTOC×DBD×volume/1000area, where SOC stock is sedimentary organic carbon stock [kg m⁻²], TOC is total organic carbon concentration [%] in the samples (where OM > 1 %), DBD is the dry bulk density of each sample [g cm⁻³], depth is the thickness of sampled material [cm] and area is the cross-section areas of the coring devices [m²].

To remove the effect of different core lengths in the SOC stock calculation, SOC stocks were also calculated for the top 20 cm of each core (Eq. 4b). This depth (20 cm) was selected as it is the length of the shortest core (landside zone of site D in Kallavesi). Additionally, SOC accumulation values were calculated normalizing SOC stocks by the depth of the organic-rich layer in each sediment core (Eq. 5). 5SOCaccumulationkgm-2cm=SOCstockkgm-2depthoforganic-richlayercm, where SOC stocks [kg m⁻²] were calculated in Eq. (4a) and the depth of the organic-rich layer [cm] is defined by samples containing more than 1 % OM (Kortelainen et al., 2004).

Maps were produced using QGIS version 3.28.4 (QGIS Development Team, 2024), plots and statistical analyses were performed using R version 4.3.3 (R Core Team, 2024), and package ggplot2 was used for graphics (Wickham, 2016). TOC content and SOC stock estimates were log-transformed prior to statistical analysis to meet assumptions of normality and homogeneity of variance. Analyses of variances (ANOVA) were computed to compare grain size distribution (% of fine-grained sediments), TOC contents and SOC stock estimates across ecosystems, considering categorical parameters. For the grain size distribution, lakes and zones were used as categorical parameters. For the TOC content analysis, lakes, zones and grain-size classes were used as categorical parameters. For the SOC stock analysis, the used categorical parameters were lakes, zones, grain-size classes and macrophyte types. A post hoc Tukey Pairwise Comparison was performed to identify groups that differ significantly from each other.

Additionally, SOC stock estimates were also compared to environmental numerical parameters, i.e., density of macrophyte stems [stems m⁻²], water depth [m], slope of each site (calculated as the distance between the landside and waterside sampling locations divided by the difference in their depth), and the average fetch length. For each sampling site (A–I), the distance to the nearest shoreline (either lake shoreline or islands) was calculated considering 48 angles (from 0 to 345°, by 15° increments) (Murtojärvi et al., 2007; Hayes, 1996). The average of these 48 distances was considered as the “average fetch length”. R packages terra (Hijmans et al., 2024) and waver (Marchand and Gill, 2023) were used to import shapefiles into R and calculate the fetch lengths, respectively. Lastly, a generalized linear model (GLM; using multiple regression and family gamma) was employed to examine if there were significant relationships between our SOC stock estimates and the environmental numerical parameters (density of macrophyte stems, water depth, slope of each site, and the average fetch length using R package glm2 (Marschner and Donoghoe, 2022). The data was tested for multicollinearity using the variance inflation factor (VIF) analysis and they did not present multicollinearity.

The signal of the 1986 Chernobyl nuclear accident was not clearly visible in the ¹³⁷Cs profiles of our samples (Fig. S15). The maximum activity observed in the longest core was around 160 Bq kg⁻¹, while peak values can reach up to 4 800 Bq kg⁻¹ dry wt in deeper areas of the studied lakes (Junna et al., 2025). The low ¹³⁷Cs signal values observed in the teal carbon areas imply mixing and removal of material, either by in-situ bioturbation (e.g. from roots or sediment dwelling invertebrates) or from physical action. The lack of clear ¹³⁷Cs profile with a peak value suggests that there is no net sediment accumulation at our study sites, and we do not report data of long-term sequestration. Therefore, the shoreline areas do not act as permanent deposition locations for carbon, but rather zones of temporary carbon storage. This interpretation is supported by clear unconformities in most cores between a distinct surface layer rich in organic material and basal organic-poor material (Figs. S11 to S13, and S16). As such, the calculation of the sediment and carbon mass accumulation rates were not possible and instead we based our interpretation and discussion on stock estimates (mass per unit area) of carbon.

A similar behaviour might be seen in deltaic environments and alluvial floodplains, where sediments are subject to a repeated series of depositional, burial, and erosional events (Stanley and Hait, 2000; Sutfin et al., 2016). In the latter study, the authors suggest that the potential for long-term storage of SOC depends on the source and form of the OC and the residence time of sediment. However, Gurwick et al. (2008) studied the effect of microbial activity on OC burial in riparian soils and they emphasized that the abundance and quality of organic matter at the time of burial may be more important to SOC than the duration since burial.

The sediments from all cores can be classified as organic-poor sands, organic-rich silts, or as organic-poor silts. The grain size (d50) varied between 7 and 819 µm while the percentage of fine-grained sediments (silt + clay) varied between 0 % and 95 % (Fig. S17; Table 2). Lakes Kallavesi and Oulujärvi had a larger fraction of fine-grained sediments (means and median between 50 % and 65 %) than lake Vesijärvi (30 % and 24 %, respectively) (Fig. 3a). There was a clear difference also between shoreline zones (Fig. 3b). Larger fractions of fine-grained sediments were observed in the landside and transitional zones (means and median between 47 % and 62 %), while samples from the waterside presented a coarser grain size distribution, with fine-grained sediments presenting a mean of 37 % and a median of 28 %.

Most of the sediment cores presented a top layer typically rich in organic matter content [% dw] that was overlying sandier sediments and/or organic-poor silts (Figs. S11 to S13, and S16), both poor in organic matter (Fig. 4). The top 30 cm encompasses the entire OM-rich layer in many cases, and only at sites A, B (lake Oulujärvi) and E (lake Kallavesi) was the whole sediment core filled with organic-rich material (Figs. S11 and S12). The percentage of TOC across all samples varied between 0 and around 40 % dw in each of the three lakes, and the medians were 1.0 % dw, 0.4 % dw, and 9.8 % dw, respectively (Figs. 3c and 3d, Table 2). Significant differences were observed among the three lakes (ANOVA F-value = 18.62; p-value < 0.05), shoreline zones (ANOVA F-value = 6.441; p-value < 0.05), and grain-size classes (ANOVA F-value = 83.73; p-value < 0.05), with the samples containing silts presenting overall larger TOC content (ANOVA results in Table S3 and Tukey test results in Table S6). Overall, TOC content was higher than 10 % only when the amount of fine-grained sediments was higher than 50 % (Fig. 3e and f).

The whole core SOC stocks varied between 0 kg m⁻² (lake Vesijärvi, waterside zone, fine and very fine sands, dominated by Phragmites) and 40.8 kg m⁻² (lake Oulujärvi, transitional zone, silts, dominated by non-Phragmites), with a large variability being observed between lakes and sites (Fig. 5a). Larger stocks were generally observed in Oulujärvi, and smaller stocks in Vesijärvi (Fig. 5b). For the shoreline zones, smaller SOC stocks were observed on the waterside (Fig. 5c). Significant differences were observed among different types of vegetation (ANOVA F-value = 5.102; p-value < 0.05; Table S3), with the sites dominated by “non-Phragmites” vegetation presenting generally larger SOC stocks (Fig. 5e). As the ANOVA results for TOC content [% dw], the most significant differences between SOC stocks were observed among cores of different grain size classes (ANOVA F-value = 14.650; p-value < 0.05), with larger amounts of SOC stock observed only in the presence of silts (Fig. 5d) (multiple factor ANOVA results in Table S4 and Tukey test results in Table S7).

Similar patterns were observed when only the top 20 cm were considered for the SOC stock calculation (Fig. S18) and for SOC accumulation (Fig. S19), with larger amounts of SOC stock₍₀₋₂₀₎ and SOC accumulation estimates being observed only in the presence of silts and most commonly in sites dominated by “non-Phragmites”. Their geospatial patterns were also very similar, with slightly larger mean and median values being observed in Oulujärvi and lower values in the waterside.

To assess the relationship between environmental predictors and SOC stocks, we compared our cumulative SOC stocks with the density of macrophyte/helophyte vegetation, water depth, slope of each site, and average fetch length (Fig. 6). When all sites were considered together, the stem density does not seem to be a key parameter explaining carbon accumulation on lakes shorelines (Fig. 6a). However, when lakes and zones are considered separately (Fig. S20), the vegetation density becomes somewhat important on the waterside zone, and in lakes Kallavesi and Vesijärvi (r2>0.5), with higher stem densities corresponding to higher SOC stocks.

For water depth and slope, large SOC stocks (> 10 kg m⁻²) were observed across the entire sampled range, from shallow (< 0.4 m) to deep (< 1.0 m) sites (Fig. 6b), and from gradual (slope index > 250) to steep (slope index < 50) areas (Fig. 6c). The average fetch length, on the other hand, presented the clearest pattern with SOC stocks, with estimates larger than 10 kg m⁻² being observed only when fetch lengths were smaller than 500 m (Fig. 6d). It was also the most significant (t-value = 4.788; p-value < 0.05) component in our generalized linear model (Table 3).

When considering only the top 20 cm for the SOC stock calculations or SOC accumulation, the same relationships were observed, with larger values being observed only when the average fetch length was smaller than 500 m (Figs. S21 and S22). The similarity between the results reinforces the robustness of using the whole core for the SOC stock calculation.

All three SOC calculations (SOC stocks [kg m⁻²], SOC stocks₍₀₋₂₀₎ [kg m⁻²] and SOC accumulation [kg m⁻² cm⁻¹]) presented the same spatial distribution (Table 2), with the highest values being observed in sites A, B (Oulujärvi) and E (Kallavesi), and the lowest in site I (Vesijärvi).

When compared with the sedimentological parameters, both TOC content [% dw] and SOC stocks [kg m⁻²] showed a strong dependency on the grain size (Figs. 3e and f, and 5d, respectively). Overall, higher TOC contents (> 10 % dw) were observed only when the proportion of fine-grained sediments exceeded 50 %. This association is well known, due to the higher specific surface area of lower-grain size materials allowing more OM to be sorbed on the particle surfaces (Keil et al., 1994; Keil and Cowie, 1999). Additionally, OM is more likely to be transported and sedimented with fine inorganic materials due to the hydrodynamics of the environment, i.e., a relatively quiescent setting will have more fine-grained sediments and more OM sedimentation (Milliman, 1994). Lastly, fine-grained sediments tend to have smaller pore spaces, which might hinder microbial degradation (Bianchi, 2007).

Interestingly, the relationship between TOC content and grain size varied between the studied lakes (Fig. 3e and f). Because Vesijärvi is located between the two ice-marginal ridges Salpausselkä I and II, most of its surficial deposits are sands and thus, there is less fine-grained material that can be discharges to the lake. Kallavesi and Oulujärvi, on the other hand, are located further north and more influenced by the input of reworked glaciolacustrine clays (Lunkka et al., 2024). This likely causes the two latter lakes to receive more fine-grained sediments. Additionally, the water level of Oulujärvi has been altered in the last decades, which might have impacted the sediment input to the lake. Specially, between 1940 and 1960, several dams and power plants were constructed around Oulujärvi (Järvi-meriwiki: Oulujärvi, 2025). Dams built on rivers flowing towards the lake tend to retain larger sediment sizes, allowing a higher proportion of fine-grained sediments to arrive to the lake.

As pointed out by Tangen and Bansal (2020) in inland freshwater wetlands, SOC stocks are highly variable in space because of the suite of natural and anthropogenic controlling factors. Our SOC stock estimates presented a large spatial variability, ranging from 0 to 40.8 kg m⁻² (Fig. 5a), and were mainly controlled by the sediment grain size (Fig. 5d). Several environmental parameters could influence the entrapment of organic-rich fine-grained sediments, such as the presence of vegetation acting as barriers for the water flow and mixing, promoting deposition of particulate matter (Leonard and Luther, 1995), and the exposure of the shoreline to waves and currents, which can erode the deposited material (van Rijn, 1993).

Macrophyte presence can promote sediment deposition either through sediment adhesion to vegetation or by attenuating vegetation-induced hydrodynamics (Li and Yang, 2009). However, the role of specific species promoting sediment deposition is not clear (Nikora et al., 2008; Leonard et al., 2002). In our study, we did not observe any clear pattern between SOC stocks and type of vegetation. Even though sites without any Phragmites and dominated by Equisetum spp. and Carex spp. presented overall higher SOC stocks, high SOC stocks were observed also in sites covered only by Phragmites (Fig. 5e and Table S7). Therefore, the type of vegetation is not the main factor affecting SOC stocks, but it might play a secondary role.

Fine-grained sediments usually are dominant in low-energy conditions with gentle slopes, such as tidal flats and protected and sheltered basins (van Rijn, 1993). Among the tested properties (Fig. 6), the exposure degree of the shoreline presented the clearest relationship with SOC stocks. Significant SOC accumulation (mean of 21.9 ± 9.4 kg m⁻² for the whole core, and 7.1 ± 1.9 kg m⁻² for the top 20 cm) was observed only when the average fetch length was smaller than approximately 500 m (Fig. 6d). In these sites (Figs. S1, S2 and S5), the sheltered shorelines prevent the sediments (OM and fine-grained mineral material) from being washed away, promoting their deposition and allowing a deeper storage of OC. The remaining sites, with smaller SOC stock estimates (mean of 2.9 ± 2.3 kg m⁻² for the whole core, and 1.8 ± 1.3 kg m⁻² for the top 20 cm), were all located in more exposed shorelines (Figs. 5d, S3, S4, S6 to S9), where the fine-grained sediments and OM are more easily carried away, for example by waves. Our results suggest that in these exposed sites, only a small amount of SOC is stored in the more surficial portion of the sediments, especially in the root zone (Figs. S11 to S13). The presence of vegetation might buffer external disturbances and, therefore, promote the accumulation of fine-grained sediments and organic matter. Thus, vegetation density might play an additional role in controlling SOC accumulation in more exposed areas (Fig. S24).

In shallow estuaries dominated by wetland vegetation, the wave activity contributes to processes such as sediment re-suspension and transport (Karimpour et al., 2017; Green and Coco, 2014). Similarly in freshwater teal carbon environments, waves may play an important role in sediment re-suspension and erosion from shallow vegetated shorelines. The wave height, and hence the energy of the waves, increases with the wind speed and the fetch (the unobstructed distance over which the wind can blow) (Fagherazzi and Wiberg, 2009; Lamont-Smith and Waseda, 2007). Assuming our studied lakes are subjected to rather similar meteorological conditions, e.g. the number of days with constant wind are similar in the whole lake (Fig. S23), the length of the fetch in certain direction will be the main factor controlling wave heights, with longer fetch generating waves with larger heights (Lamont-Smith and Waseda, 2007).

When taller waves (i.e., of longer fetch) reach the shoreline, their orbital motion extends to the bottom and their shear stress can exceed the critical shear stress against which the sediment resists motion. On the other hand, taller waves are more prone to breaking as they shoal, and the dissipation of wave energy in the breaker zone can erode the bottom and generate a net offshore transport (van Rijn, 1993). During this interaction, wind-induced waves tend to remove sediments from the shoreline, and fine-grained sediments tend to be washed out faster than sands. On the contrary, sheltered areas (smaller fetch) are exposed mainly to waves with smaller heights, that tend not to affect deep into the water column or break when reaching flat areas, even promoting a net onshore transport of sediments (van Rijn, 1993). The lower energy environment promotes the deposition and preservation of fine-grained sediments. Because of their high surface/volume rate and chemical charges (Keil et al., 1994; Keil and Cowie, 1999), fine-grained sediments tend to adsorb more OM and, therefore, to accumulate more organic carbon.

In addition to the shoreline exposure, we tested the effect of bathymetry on SOC accumulation. Most of the sites with SOC > 10 kg SOC m⁻² were found around 0.5 m of water column, but one sampling site located deeper than 1.0 m water column also presented SOC stocks larger than 10 kg m⁻² (Fig. 6b). We also studied the relationship of the shorelines' slope on average SOC stock estimates. While sites A (average of 15.6 kg m⁻²) and B (average of 24.7 kg m⁻²) are located in a region with a gentle and subtle slope (slope indices of 294 and 195, respectively), which promotes fine-grained sediments and OM accumulation, site E (average of 25.5 kg m⁻²) was located in the steepest site (slope index of 22) (Fig. 6c). Therefore, neither bathymetry nor the shoreline's slope seem to have a consistent role on SOC stocks.

The spatial variability of the macrophyte belts zonation in freshwater wetlands that generally changes from more terrestrial and drier zones towards more aquatic and flooded zones has previously been described to control OC accumulation in shallow freshwater teal carbon ecosystems. Bernal and Mitsch (2012), Tangen and Bansal (2020) and Taran et al. (2023) found higher concentrations and stocks of OC in the inner, more aquatic portion, of the wetlands. They attribute this to the combined effect of the permanent anaerobic conditions and the high macrophyte productivity of the inner zone. Further, they observed a change in the species composition from their more landward/upper zone towards the water, which is different to what we observed in our boreal lake sites, where most macrophyte belts were monospecific. However, in a study focusing in monospecific reed belts, Buczko et al. (2022) observed larger SOC stocks in the terrestrial zone when compared to the littoral, more aquatic zone. They attributed this pattern to generally larger biomass being produced by Phragmites at the terrestrial edges of wetlands.

In our study, however, we did not observe any clear and general spatial pattern between SOC stocks and zones or macrophyte biomass (Figs. 4c and S20). In Kallavesi, for example, the largest SOC stocks were indeed observed in the landside zone in all three sites, but the landside zone was not always the most densely vegetated. In site E, for example, the largest SOC stock was observed in the landside zone, despite it having the lowest vegetation density (less than 100 stems per m⁻²) (Table S2, Figs. 4a and S20). In Vesijärvi, the largest SOC stocks were observed in the transitional zone, with monospecific (only Phragmites) or mixed macrophyte belts (Figs. 4a and S20). In this lake, a denser vegetation was associated with more OC accumulation, but even in the most densely vegetated study area (with almost 400 stems per m⁻²), the SOC stock did not exceed 10 kg TOC m⁻² (Tables 2 and 3, Fig. S20). Lake Oulujärvi had the largest SOC stocks but the weakest relationship between the stocks, zones and macrophytes density (Table S2, Figs. 4a and S20). In Oulujärvi, the highest stock was found in a different zone in each site. Also, the largest stock estimate we observed was found in Oulujärvi, in an area with only 62 stems per m⁻². This suggests that the zonation and the vegetation structure (species composition and stem density) have a secondary role in the amount of SOC stock in these teal carbon areas of boreal lakes. As mentioned before, when only the more exposed areas are considered, the vegetation density had a larger role in controlling role in SOC accumulation (Fig. S24), possibly due to the effect of vegetation in buffering external disturbances and, subsequently, promoting the accumulation of fine-grained sediments and organic matter. Thus, although macrophyte belts, especially Phragmites, can thrive in a wide range of environmental conditions (Murphy, 2002), only sheltered areas of teal carbon environments seem to accumulate larger amounts of OC into deeper sedimentary layers.

To put the teal carbon stocks from boreal zones in a global perspective, we compare them with the reported SOC stocks estimated for brackish and freshwater vegetated wetlands. Overall, our SOC stock estimates are well within the existing published values (Fig. 7). The SOC stock estimates from Oulujärvi and Kallavesi are comparable to those observed in a monospecific Phragmites reed belt from coastal lagoons in northern Germany (Buczko et al., 2022) and in several freshwater marshes from southeastern Australia (Carnell et al., 2018).

In the German coastal lagoons studied by Buczko et al. (2022), only one site (Michaelsdorf) presented organic-rich material throughout the whole sediment core (1 m length), reaching SOC stock estimates of 75 kg m⁻². For the remaining sites the SOC stock estimates ranged between 1.76 and 35 kg m⁻², and presented a clear decline in TOC content [% dw] with core depth. The Michaelsdorf sampling site is located in a relatively narrow channel connecting Saaler and Bodstedter water bodies, and is surrounded mainly by pastures. Even though the sampling site itself is not as sheltered as the sampling sites in Kallavesi (site E) and Oulujärvi (sites A and B), the geomorphology of the channel might prevent strong water currents and waves, allowing the settlement of fine-grained sediments, which in turn promotes the accumulation of TOC. Therefore, while Michaelsdorf sampling site probably stores more OC via the same mechanism as sites A, B and E in our study, the other sites in the studied German coastal lagoons probably resemble our more exposed sampling sites.

In the study by Carnell et al. (2018), data of 57 sediment cores from shallow freshwater marshes and 100 cores from deep freshwater marshes were combined. Therefore, the information about the geomorphology of each site is not available. Carnell et al. (2018) mention, however, the influence of the proximity to major rivers. Wetlands located closer to fluvial inputs were associated with higher sediment accretion and potentially carbon sequestration. The authors point out that this might be caused by anthropogenically clearing of terrestrial vegetation in the catchment area and changes in river hydrology, leading to larger inputs of autochthonous organic matter to the adjacent wetlands and macrophyte belts.

Compared to coastal marine blue carbon environments (mangrove forests, salt marshes and seagrass meadows), the mean of SOC stock estimates from Oulujärvi and Kallavesi are around one third of the SOC stocks observed in mangrove forests (Alongi, 2022). When compared to salt marshes (Maxwell et al., 2023) and seagrass meadows (Kennedy et al., 2022), however, the SOC stocks from boreal teal carbon environments are notably larger.

The overall magnitude of OC stocks observed in our study highlights the potential importance of shallow vegetated areas to OC accumulation in inland waters. As pointed out by Grasset et al. (2025), vegetated shoreline areas can act globally as net atmospheric C sinks, as opposed to pelagic zones that generally act as net C sources and thus decrease whole-lake atmospheric C emissions when included in lake budgets. Boreal latitudes contain the largest abundance of lakes in the world, especially considering lakes smaller than 0.1 km² (Verpoorter et al., 2014). Due to the high perimeter/surface area ratio of inland waters in the boreal zone, shoreline areas and associated teal carbon environments in boreal lakes may be important components of global C cycling and therefore should be considered when budgeting SOC of lake sediments. In order to upscale our SOC estimates to obtain a boreal SOC budget, subsequent studies might consider calculating the fetch distances of lacustrine shorelines to classify them as exposed or protected. Once the shorelines are classified, our SOC stock estimates can be multiplied by the area covered by macrophytes, which will produce SOC pool data (in mass of SOC).