Evaluation of seagrass as a nature-based solution for coastal protection in the German Wadden Sea under end of the century sea level rise projections

Coastal zones, typically characterized by the combination of low land elevation and high population density, as well as the presence of highly valuable ecosystems (Costanza et al. 1997), are among the most vulnerable regions to climate change (Neumann et al. 2015). Rising sea levels, altered storm regimes, ocean warming, and acidification threaten both ecosystems and human communities (Pachauri et al. 2014).

1.1 Study area

This study focuses on the tidal inlet systems of the North and East Frisian Wadden Sea (NFWS, EFWS), which are part of the UNESCO-listed Wadden Sea located in the southeastern North Sea (German Bight; Fig. 1). The entire intertidal wetland area is highly vulnerable to sea level rise, which threatens to permanently flood at least part of the area and disrupt the endemic ecosystem. At the same time, the low lying hinterland makes coastal communities highly exposed to SLR, making dyke protection essential.

1.2 Climate Change implications

The Intergovernmental Panel on Climate Change (IPCC) has progressively updated its projections of global mean sea-level rise (GMSL) upward. The Fourth Assessment Report (AR4) projected a rise of 0.18-0.59 m from 1990 to 2100, though some studies suggested 1-2 m (Houston 2012). AR5 estimated a ‘likely’ (66% probability) rise of 28-61 cm under drastic emission reductions (RCP2.6) and 52-98 cm under unmitigated emissions (RCP8.5), relative to 1986-2005 (Ipcc 2019; Horton et al. 2014). AR6, using updated models and scenarios, further raised these estimates, projecting 0.15-0.23 m by 2050 and 0.28-0.55 m by 2100 under very low emissions (SSP1-1.9), and 0.20-0.29 m by 2050 and 0.63-1.01 m by 2100 under high emissions (SSP5-8.5) (Lee et al., 2023, medium confidence). Although uncertainties remain, rapid and sustained emission reductions could substantially slow the rate of sea level rise and reduce long-term commitments.

This long-term trend is accompanied by an increase in the frequency and intensity of extreme sea-level (Menéndez and Woodworth 2010; Tebaldi et al. 2021) and wave (Izaguirre et al. 2011) events, though with strong regional variability. In the Southern North Sea, for instance, Grabemann and Weisse (2008) projected increases in mean and extreme wave heights in the eastern region by the end of the century. Similarly, Bonaduce et al. (2019), using regional projections under RCP8.5 for 2075-2100, found significant spatial variations, with extreme wind speed, surface Stokes drift, and significant wave height increasing in the Northeast Atlantic but decreasing in parts of the North Sea.

The consequences of sea-level rise are substantial, leading to permanent inundation, increased flooding, shoreline erosion, and salinity intrusion (Nicholls and Cazenave 2010; Houston 2012; Vousdoukas et al. 2017; Dronkers and Stojanovic 2016). Hence, there is a need for effective and resilient adaptation measures.

1.3 Nature based solutions

Nature-based solutions (NbS), as defined by the International Union for Conservation of Nature (IUCN), "leverage nature and the power of healthy ecosystems to protect people, optimise infrastructure, and safeguard a stable and biodiverse future." In coastal contexts, this often involves biogenic habitats such as mussel beds, salt marshes, seagrasses, and kelp forests, which can attenuate wave energy, reduce currents, and stabilize sediments (Narayan et al. 2016; Ferrario et al. 2014; Temmerman et al. 2013).

NbS are increasingly recognized as cost-efficient and resilient complements to traditional grey infrastructure (Temmerman et al. 2013; Van Wesenbeeck et al. 2016; Cohn et al. 2021). In contrast, engineering solutions such as seawalls and breakwaters are becoming economically and ecologically unsustainable to protect coastal communities (Morris et al. 2018). Consequently, restoration or creation of natural habitats has been widely recommended to enhance coastal defense (Boudouresque et al. 2021; Govers et al. 2022).

In the Wadden Sea, seagrass meadows are mainly composed of Zostera marina Linnaeus, 1753 (Z. marina) and Zostera noltei Hornemann, 1832(Z. noltei), the latter being physiologically smaller but more abundant in the region (Dolch et al. 2013).

1.3.1 Nature Based Solutions under Future Climate

Coastal ecosystems can provide effective coastal protection and at present may show partially recovery trends (Dolch et al., 2013, i.e. in our study area) but are in general increasingly vulnerable to climate change and direct anthropogenic pressures (ankering, dredging, eutrophication, ...), with rising temperatures being a significant stressor (Valle et al. 2014; Ondiviela et al. 2014). In a worst-case scenario for the Mediterranean Sea, Chefaoui et al. (2018) predict that the endemic seagrass Posidonia oceanica (Innaeus) Deille, 1813 could face extinction by 2100. Along the French Atlantic coast, Zostera noltei, which primarily inhabits intertidal zones, is also at risk. Valle et al. (2014) assessed the species’ response to climate change in the Oka Estuary (Basque Country, Bay of Biscay), considering both temperature increases and SLR. Their findings suggest that by the end of the 21st century, warming will shift Z. noltei’s suitable habitat 888 km northward, while SLR will drive landward migration, expanding intertidal habitat by 14-18\(\%\), though this will be constrained by anthropogenic barriers.

Within in our region, habitat suitability modeling for 2050 climate scenarios in a representative intertidal basin of the East Frisian Wadden Sea (Singer et al. 2017) projects a significant expansion of Z. noltei seagrass beds on lower mixed-sediment intertidal flats. In their review, Ondiviela et al. (2014) conclude that seagrasses cannot protect all shorelines under every scenario and that optimal protection occurs in shallow, low-wave energy environments with high water-seagrass interaction. They also note that combined climate change and anthropogenic threats may prevent seagrasses from adapting quickly enough to continue offering coastal defense. Therefore, they suggest various adaptation measures, both artificial and natural, to mitigate the decline of seagrasses.

1.4 Scope of this study

While previous studies have demonstrated that seagrass meadows can effectively reduce wave energy (Paul et al. 2012; Horstman et al. 2014; Pillai et al. 2022), currents (Fonseca et al. 1982; Pillai et al. 2022), sediment resuspension (Jacob et al. 2023), and erosion (Chen et al. 2024, 2022) under present day conditions, it remains unclear how these provided ecosystem services (ESS) persist under rising sea levels, though the NbS effectiveness is expected to decline (Ondiviela et al. 2014). Building on Jacob et al. (2023), this study investigates whether seagrass can continue to mitigate coastal hazard risks in the German Bight intertidal Wadden Sea as water depths increase (R1) and whether these impacts differ between the East and North Frisian Wadden Sea compartments (R2). Within our study, "coastal hazards" are interpreted as high-energy conditions that could negatively affect the coastal system, rather than single catastrophic events. Using one-year time slices with hourly resolution, we implicitly evaluate both short-term high-energy to extreme conditions, represented by the 95th percentile of wave heights and currents, which predominantly capture strong wind events and peak tidal flows. Longer-term, potentially accumulating effects are reflected by temporal mean conditions that indicate sediment transport and stability benefits, even in the absence of fully dynamic morphodynamic feedbacks.

As a novel approach this study combines regionalized climate downscaling (Pein et al. 2023) with hydrodynamic modeling that incorporates seagrass in a first application for the Wadden Sea region.

2.1 Modeling approach

The modeling framework and grid configuration follow Jacob et al. (2023), with the main difference being the climate forcing. present day and future projections use the downscaling approach of Pein et al. (2023), based on coupled MPIOM-REMO outputs and intermediate 2D and 3D SCHISM setups (see Section 2.1.6). Model component descriptions are therefore presented only in abbreviated form here; for details, we refer to Jacob et al. (2023).

Simulations were performed with the SCHISM framework (Zhang et al. 2016), coupling the hydrodynamic core together with the source-code-level coupled third-generation wave model WWMIII (Roland et al. 2012) and the sediment model SED3D (Pinto et al. 2012), all operating on a shared grid.

2.1.1 Grid configuration

For this study, we applied the SCHISM German Bight configuration (Stanev et al., 2019; Jacob et al., 2023, the domain extent is given in Fig. 1a), discretized on an unstructured triangular grid with 476,000 nodes and 932,000 elements. Horizontal resolution refines from 1.5 km at the open boundary to 50-100 m in estuaries, while the vertical structure is resolved with 21 terrain-following sigma layers. Within the East and North Frisian Wadden Sea subdomains, resolution ranges from 100-300 m, with the highest refinement in tidal channels. Wetting and drying are treated with a 5 cm threshold, and simulations use an 80 s time step. Vegetation effects are represented in the hydrodynamic (Zhang et al. 2020) and wave model components.

2.1.2 Hydrodynamic model

The Semi-implicit Cross-scale Hydroscience Integrated System Model (SCHISM) is derived from SELFE Zhang and Baptista, 2008, v3.1dc), extended for large-scale eddying regimes and seamless cross-scale applications (Zhang et al. 2016), and distributed under the open-source Apache v2 license. It solves the Reynolds-averaged Navier-Stokes equations under hydrostatic and Boussinesq approximations on unstructured grids. The numerical discretization uses a mixed finite element–volume formulation. The hydrodynamic core uses semi-implicit time stepping and a higher-order Eulerian-Lagrangian scheme for momentum advection, with tracers solved using a 2nd-order TVD scheme. Turbulence closure follows the generic length scale (GLS) formulation with k-\(\varepsilon \) parameterization (Umlauf and Burchard 2003). Wetting and drying are treated with a minimum water depth of 5 cm, to resolve drying and flooding of tidal flats.

The frictional effect of seagrass vegetation is implemented as cylindrical form drag in the momentum equations, added as an additional term zhangspsselfesps2008Zhang et al. 2020; Jacob et al. 2023):

with the implicit terms in

In the drag term of (1)

\(\alpha \)([m\(^-1\)]) denotes the vegetation-induced friction calculated as the product of shoot diameter (D\(_v\)), vegetation density (N\(_v\), stems per m2), and the bulk form drag coefficient CD\(_v\). It was chosen as 1 in line with Jacob et al. (2023). The vegetation term is active within the 3D water column where submerged (or emerged) vegetation is present, implement as:

where \(\mathcal H\) is the Heaviside step function

and \(z_v\) the z-coordinate of the canopy.

Furthermore, turbulence generation by vegetation is included as an additional term in the turbulent kinetic energy and mixing length equations of the GLS model, proportional to the cubed velocity magnitude (see Zhang et al., 2020).

2.1.3 Wave model

The wind wave model III (WWMIII Roland et al., 2012) is a third-generation spectral wave model that solves the wave action balance equation on an unstructured triangular mesh using ST4 physics. Shallow-water wave breaking is parameterized following, while bottom friction dissipation is based on JONSWAP (Hasselmann et al. 1973). The model runs with a 240 s time step and exchanges water levels and velocities from SCHISM to WWM, and radiation stress from WWM to SCHISM with the hydrodynamic model (80 s time step) every third iteration. The directional-spectral domain in our application is resolved with 24 directional bins and 36 frequency bins, spanning the range from 0.04 to 0.6345 Hz. Wave dissipation by seagrass vegetation is implemented following Zhang et al. (2020) and Suzuki et al. (2012), incorporating a vegetation-induced dissipation term that scales proportionally with vegetation density and the applied drag coefficient (for further details, see Jacob et al., 2023).

2.1.4 Sediment model

The sediment model SED3D (Pinto et al. 2012), adapted from the community Sediment Transport Model (Warner et al. 2008), was implemented for unstructured grids within SCHISM. The model simulates erosion, deposition, bedload, and suspended load transport of non-cohesive sediments. The model computes sediment transport as advection–diffusion of multiple sediment classes, including vertical settling and mixing, and accounts for the influence of near-bed hydrodynamics on erosion and deposition It was configured with eight sediment classes of median grain sizes (d50) ranging from 0.06 mm to 2.0 mm (0.06, 0.07, 0.1, 0.125, 0.25, 0.5, 1.0, and 1.99 mm) and a density of 2650 kg/m\(^3\). Sediments are erodible from an infinite bottom pool, with bed fractions derived from maps by Milbradt et al. (2015), based on surveys by the Bundesanstalt für Wasserbau (Valerius et al. 2015). Bed shear stress is computed dynamically from the hydrodynamic model, including options for linear, quadratic, or logarithmic bottom friction, and the impact of sediment concentration on water density is accounted for. At the open boundaries, a zero-gradient condition is applied for sediment. The model was operated morphostatically in this study.

2.1.5 Limitations of seagrass representation

As in Jacob et al. (2023), experiments were conducted with the SCHISM modeling system using the vegetation module described by Zhang et al. (2020), which accounts for vegetation effects within the hydrodynamic core through vegetation-induced turbulence and friction, as well as wave attenuation in the WWM wave module. It was shown by Zhang et al. (2020) that it is capable of simulating the effects of first-order vegetation on hydrodynamics, but has the following limitations reported in Jacob et al. (2023): 1) seagrass is represented using a standard rigid-cylinder canopy formulation, which is commonly applied in coastal hydrodynamic–vegetation models. Dynamic blade bending, reconfiguration and posture changes are not represented. This assumption allows a robust and numerically stable parameterisation of vegetation drag and turbulence generation, but may overestimate flow resistance under strong currents and waves and thus the attenuation there of. The approach is generally capable of representing the effects of high-density vegetation (Vargas-Luna et al. 2016). 2) Vegetation-flow interactions are unidirectional, the model considers only the influence of seagrass on flow turbulence and friction as a function of the 3D resolved vegetation drag, but not the influence of the flow on seagrass. Biological dynamics of seagrass, including seasonal cycles, are not represented. Therefore, we only consider the maximum extent of seagrass in summer, without taking into account the reduction in population in autumn and winter or the damage that seagrass suffers during storm events, which can be severe.

2.1.6 Forcing for present and future scenarios

Time-slice simulations were carried out for the years 1997 and 2090, selected because they are separated by an integer multiple (5\(\times \)) of the 18.6-year lunar nodal tidal cycle. This choice minimizes tidal variability between the two periods. Each simulation was run for a full year with hourly model output.

Climate scenario - RCP8.5

The chosen climate scenario is based on the IPCC AR5 RCP8.5 ensemble median. Although more recent assessments suggest that RCP8.5 may not represent the most likely emissions pathway in the near future, it remains valid for this study because of its scope and focus. Given the long-term nature of sea-level rise projections and the complexity of hydrodynamic modeling, it is important to consider a broad range of possible futures, particularly worst-case impacts. RCP8.5 therefore continues to provide a useful upper-bound scenario for coastal risk evaluation under unmitigated emissions. Due to computational constraints and the high cost of running fully coupled annual simulations, this study focuses on a single scenario, instead of exploring multiple emission pathways.

Model forcing

The model forcing for the SCHISM German Bight implementation, includes hourly values of elevation, 3D velocity, temperature, and salinity at the open-ocean boundary (red line in Fig. 1a)). Temperature and salinity are further nudged in a 20 km relaxation zone adjacent to the boundary, with an exponential decay of the nudging weight applied at daily frequency. Atmospheric forcing consists of atmospheric pressure, 10 m wind, 2 m air temperature, specific humidity, and solar radiation. River discharge is prescribed using climatological data for the Ems, Weser, Elbe, and Eider rivers.

For both present-day and future climate simulations, forcing is based on the multi-stage downscaling framework of Pein et al. (2023). In the workflow schematized in Supplementary Fig. S1, climate variability and atmospheric forcing are taken from the regionally coupled MPIOM–REMO (Elizalde et al. 2014) climate model system, which provides hourly outputs for sea level and meteorological forcing projections (Mayer et al. 2022). These outputs represent the large-scale ocean atmosphere dynamics over the North Sea and European domain, with a horizontal resolution of approximately 5 km for the ocean component and 24 km for the atmospheric component.

The global mean sea-level rise is prescribed separately using the ensemble median of probabilistic estimates under the RCP8.5 scenario from the UKCP18 project (Palmer et al. 2020, 2018).

The consistently combined climate forcing, consisting of regionalized mean sea level together with hourly sea-level elevation and wind fields, is first applied at the open boundaries of a 2D barotropic southern North Sea (SNS) model. This model uses a downscaling resolution from approximately 5 km on the shelf to about 1 km toward the German Bight and ensures dynamically consistent tidal propagation under sea-level rise.

The 2D simulations are then combined with monthly REMO baroclinic currents, salinity, and temperature to force a 3D baroclinic model of the SNS. Output from this 3D model is subsequently used to initialize and drive our nested high resolution German Bight model (horizontal interpolation by inverse distance weighting, vertical interpolation linear), which is run in a fully baroclinic 3D configuration The horizontal resolution increases from the kilometer scale at the open boundary to approximately 200 m in the coastal Wadden Sea.

This hierarchical nesting strategy propagates climate-scale signals across scales from the shelf to the coastal zone while preserving dynamic consistency between sea level, tides, circulation, and density-driven flows.

Beyond Pein et al. (2023), this study also accounts for wave forcing. For the time-slice simulations, wave conditions were provided by a WaveWatch III (WW3) configuration of the Atlantic and Northwest Shelf (details in the Supplement). WW3 was forced with hourly REMO winds, and its bathymetry was adjusted according to projected MSL (+\(\sim \)0.7 m), ensuring a simplified representation of altered water depths. WW3 then delivered spectral boundary conditions for the WWM component of the German Bight SCHISM configuration.

2.1.7 Calibration and validation

A validation of the German Bight SCHISM configuration re-applied in this study, albeit under different forcing, is based on a modeling system that has been calibrated and validated in previous work using realistic hindcast forcing. A comprehensive validation of the fully coupled hydrodynamic–wave–sediment–vegetation configuration for the German Bight was provided by Jacob et al. (2023), including comparisons against tide gauge records and wave buoy observations. In Jacob et al. (2023), modeled water levels along the German Bight showed correlations between 0.95 and 0.99 with observations and root mean squared errors typically between 0.2 and 0.3 m. Storm surge peaks were reproduced with correct timing, though with some underestimation of extreme levels in the East Frisian Wadden Sea. Significant wave heights at the FINO3 and Westerland stations were simulated with correlations of about 0.96 and only moderate underestimation of extremes. These results demonstrate the overall realism and internal consistency of the coupled modeling system.

In the present study, the model is forced with scenario-based climate forcing derived from MPIOM–REMO and regionalized mean sea-level rise projections under the RCP8.5 scenario following Pein et al. (2023) and Palmer et al. (2018). The 1997 simulation therefore represents a statistically consistent realization (ensemble member) of the RCP8.5 scenario for that period rather than the actual historical climate state. Consequently, a direct validation against observations is not meaningful.

Instead, this study builds on the prior calibration and validation of the same model configuration under realistic forcing conditions and uses the scenario simulations to investigate the dynamical response of the system under physically and statistically consistent climate and sea-level forcing. The boundary forcing has been applied inheriting the bias correction conducted by Pein et al. (2023) to ensure consistency with observed mean sea level at Cuxhaven (on the southeastern coast of the model domain) in the reference period.

2.2 Bathymetry

For the projections, dikes are assumed to remain in their current positions (coinciding with the model land boundaries), with crest heights adjusted to prevent hinterland flooding. While the Wadden Sea is expected to adapt to sea-level rise through natural sediment accumulation, its long-term morphodynamic evolution remains uncertain, particularly regarding sediment availability. To reduce this uncertainty, the bathymetry in 2090 is assumed to be identical to that of 1997, representing a scenario in which the Wadden Sea does not accrete.

2.3 Seagrass scenarios

To simulate cases with and without NbS, we redeploy two of the five vegetation scenarios from Jacob et al. (2023): E2 (Blank) and E3 (Veg-Max). E2 represents the complete absence of seagrass, while E3 assumes a hypothetical maximum expansion (shown as the green area in Fig. 1b,c), representing the upper bound of potential coverage. In E3, all areas within the depth range suitable for seagrass are vegetated, with this depth range being extrapolated from the observed vertical distribution reported by Jacob et al. (2023). The E3 vegetation scenario is applied commonly in both the 1997 and 2090 time-slice simulations in order to isolate the effect of changing climatic and sea-level conditions as the sole experimental driver, rather than combining these with changes in seagrass extent.

The focus on two scenarios reflects the high computational cost of annual coupled simulations (achieving a \(\sim \)40x speedup compared to a real-time year with 10 nodes a 128 cores requiring \(\sim \) 2190 node-hours on Levante DKRZ).

We therefore adopted an end-member approach, in which the maximum expansion scenario (E3) was included to evaluate the potential impact of seagrass under SLR conditions, while the vegetation-free scenario (E2) provides a clear reference for quantifying vegetation effects.

The seagrass physiological characteristics used to calculate \(\alpha \) (Eq. 1) for E3 follow Jacob et al. (2023), where a hypothetical "hybrid" plant was defined by averaging the properties of the endemic species Z. marina and Z. noltei, resulting in a shoot diameter of \(D_v = 1.44\) mm and canopy height \(H_z = 14.4\)cm. While the set of vegetation parameters may vary between grid nodes, the model does not allow to represent multiple species within a single node. This hybrid approach therefore implicitly represents the presence of both species within a grid cell. The selected shoot density of \(N_v = 7360\) shoots m\(^{-2}\) corresponds to the highest observed density reported in Jacob et al. (2023).

We emphasize that the E3 scenario is not intended to represent a realistic future state of seagrass distribution in the Wadden Sea. Instead, it represents an on purpose idealised configuration with high density and maximum spatial extent serving as upper bound scenario. This synthetic configuration should rather be interpreted as a system-wide sensitivity experiment designed to provide an upper-bound estimate of the potential hydrodynamic impact of seagrass. In reality, seagrass meadows are spatially heterogeneous, and such a configuration would not occur everywhere simultaneously.

Future climate projections were constructed relative to climate forcing, assuming seagrass distributions remain constant despite changes in intertidal extent. This assumption allows us to isolate the influence of rising sea levels on seagrass attenuation capacity, rather than explore realistic recolonization scenarios. As in Jacob et al. (2023), the focus lies on the one-way impact of seagrass on hydrodynamics under SLR.

Both the vegetation-free (E2) and maximum expansion (E3) scenarios were each simulated for 1997 (present day conditions) and for 2090 (future sea levels under RCP8.5).

The seagrass expansion scenario (E3) and the vegetation-free control scenario (E2) were simulated for both the year 1997 (representing present day conditions) and for the projected sea level of 2090 under RCP8.5 (future projections).

2.4 Analysis

Following the methodology of Jacob et al. (2023), simulation outputs were subjected to statistical analysis for each year and scenario. The focus of this analysis is on key hydrodynamic state variables, specifically examining changes in the annual mean and the 95th percentile. These statistics were derived from daily, hourly-resolved SCHISM model outputs stored in NetCDF format, processed using Python’s xarray library.

To ensure accuracy, only time periods when the model grid cells were submerged were included in the statistical analysis. Dry periods were masked to avoid skewing results, resulting in a spatially variable number of valid time steps, particularly in regions affected by tidal wetting and drying.

The effects of seagrass were quantified as both absolute and relative differences of annual mean and 95th percentile between scenarios with and without vegetation. These differences represent the vegetation-induced attenuation of hydrodynamic forces. Comparative analyses were carried out for both the EFWS and NFWS, contrasting the present day (1997) and future (2090) sea-level conditions. Plots were created using matplotlib and cartopy, with color scales restricted to the 95th percentile range to better highlight spatial variability in attenuation.

To analyze the relationship between attenuation effects and water depth, values for both years (1997 and 2090) were grouped into depth bins of 0.4 meters (with respect to the present day bathymetry datum). Area-weighted averages for each bin were then visualized as bar plots to illustrate depth-dependent trends in seagrass-induced changes and complement the presented maps.

3.1 Sea surface height

3.1.1 Baseline in present and future scenarios

Figure 2 depicts the simulated annual mean and 95th percentile sea surface height (SSH) in the German Bight for 1997 and 2090, along with their respective differences (right column). The mean sea level in the German Bight increases by approximately 82 cm between 1997 and 2090 (Fig. 2a). Within the Wadden Sea, the average rise is smaller (30-50 cm) due to elevated (supra) littoral areas that remain dry at low tide (in 1997), which is expected since the mean and 95th percentile are calculated only over wet periods.

The 95th percentile SSH exhibits a similar trend, with a spatially averaged increase of 0.83 m ( numbers avg=x in panel text overlays represent the spatially weighted domain average of the displayed variable), slightly higher than the mean due to larger changes in the Wadden Sea. Overall, the response is relatively homogeneous, with slightly higher increases toward the southeast, particularly in regions of amplified tidal amplitudes, such as tidal channels and estuarine thalwegs. Additional increases occur along the northwestern boundary, likely influenced by the limited propagation of the amphidromic point located just outside the domain in the southern North Sea.

As a consequence of sea level rise part of the 1997’s tidal flats will become permanently submerged causing a seaward loss of tidal wetlands, while on the landward boundary of the Wadden Sea part of the supra tidal will become extend tidal area.

3.1.2 Seagrass-induced attenuation in present and future scenarios

As the temporal mean and the 95th percentile generally exhibit very similar spatial patterns of seagrass-induced attenuation, we focus in the following on the 95th percentile fields, which are more relevant for high-energy and hazard-related conditions. To avoid redundancy and overcrowding the manuscript with figures, the corresponding depictions of temporal mean changes are provided in the Supplementary Material (Figs. S2–S11).

The influence of seagrass on sea surface height (SSH) in the EFWS and NFWS is depicted in Figs. 3 and 4, respectively. Panel a) show the 95th percentile SSH for 1997 and 2090, highlighting the strong bathymetric and topographic control from tidal channels to intertidal and supratidal areas. Panels b) and c), respectively show the absolute and relative changes in the 95th percentile SSH induced by seagrass. The corresponding temporal mean SSH fields are provided in the Supplementary Material Fig. S2. Furthermore, the corresponding depth-binned distributions of vegetation-induced changes are included as the first row in the all variable encompassing summary bar plots Fig. 13 (temporal mean) and Fig. 14 (95th percentile), allowing a direct quantitative comparison of the magnitude and depth-dependent response between mean and extreme conditions.

1997

In 1997, seagrass induces substantial attenuation of sea surface height (SSH) in the vegetated intertidal areas of both Wadden Sea regions. For the temporal mean, depth-binned statistics (Fig. 13) indicate SSH reductions of about 10–28 cm in the vegetated intertidal areas of the EFWS (approximately 20–30%), with localized maxima up to 70 cm. Within tidal channels towards the intertidal transition zone, SSH increases by approximately 4 to 10 cm (Fig. 13, Fig. S2) due to the local water-level pile-up caused by drag that delays floodwater spreading across the shoals (see also Jacob et al., 2023; Pillai et al., 2022).

Reductions in NFWS appear at similar patterns but with peak reductions 5-10 cm lower than for the EFWS in the intertidal zone (-1 to +1 m depth, Fig. 13).

For the 95th percentile, spatially averaged reductions in the Wadden Seas amount \(\sim \)11 cm in EFWS and \(\sim \)9 cm in NFWS. Peak attenuation occurs in the -1 to +1 m depth range (Fig. 14), with reductions up to 25 cm in EFWS and NFWS (Fig. 14). In channels and deeper zones (>2.5 m), effects are weaker (\(\sim \)4-6 cm; <5% relative change). In NFWS, the impact extends over a broader depth range with further landward reach (up to -5 m depth), with stronger reductions in intertidal flats. Overall, absolute and relative reductions remain small (10-12% of baseline), falling below 5% in channels.

Area-averaged mean SSH reductions over the displayed sub-domains amount to \(\sim \)8 cm in the EFWS and \(\sim \)4 cm in the NFWS. When restricting the analysis to areas shallower than 15 m -effectively covering only the intertidal flats and tidal channels- the reductions increase to about 10 cm in the NFWS and 5 cm in the EFWS (Fig. 13)).

2090

Under SLR, the effectiveness of seagrass-induced SSH attenuation is substantially reduced. For the temporal mean, depth-binned statistics (Fig. 13) show that vegetation-induced SSH pile-up at the transition between channels and intertidal areas in the EFWS increases by about 4–5 cm (around 2.5 m depth). In the core intertidal zone (approximately \(-2\) to \(+2\) m), the mean attenuation efficiency decreases to about 30–50% of its present-day magnitude, corresponding to a reduction of roughly 10–30% relative to 1997. The spatial patterns of these mean changes are shown in Supplementary Fig. S2. While localized peak absolute attenuation still reaches up to about 0.5 m (e.g., in Jade Bay Fig. S2), the relative reductions are generally smaller than under present-day conditions.

The zone of effective SSH reduction shifts landward (Fig. 13 black vs green bars), and attenuation along channel margins becomes more spatially extensive. For the NFWS a similar pattern shows, with mean reductions diminished to 30-50% of 1997 values.

For the 95th percentile, prolonged submergence under SLR reduces frictional control, leading to a more homogeneous SSH response. In EFWS, absolute reductions drop to 10-15 cm in embayment areas (Ems Dollard, Jade Bay, Fig. 3b), with relative reductions of \(\sim \)5% (Fig. 3c). Tidal channels show a loss of 5 cm, with local reversals in Jade Channel. NFWS shows comparable absolute (Fig. 4b) and relative reductions (Fig. 4c). Overall, seagrass continues to marginally attenuate SSH, but efficiency is reduced under SLR, especially in intertidal flats and channels.

3.2 Velocity magnitude

3.2.1 Baseline in present and future scenarios

1997

Depth-averaged velocity magnitudes peak between 0.8 and 1.5 m s\(^{-1}\) within the tidal channels in terms of the annual mean (Supplementary Figs. S4 and S5)). For the 95th percentile, peak velocities within channels range between 1.4 and 1.8 m s\(^{-1}\) (Figs. 5a and 6a).

2090

Under SLR, baseline annual mean velocities in the EFWS increase by approximately 10% (10-15 cm/s) in mid-channel areas and by around 5 cm/s in adjacent areas. For the 95th percentile, the increase is roughly twice as high, with local increases of up to 0.3 m/s on intertidal flats. These values gradually decrease below 0.1 m/s toward the coast.

3.2.2 Seagrass-induced attenuation in present and future scenarios

1997

In the 1997 time-slice simulation, seagrass substantially reduces depth-averaged currents in both subdomains. For the temporal mean, depth-binned statistics (Fig. 13) show reductions of up to about 20 cm s\(^{-1}\) on tidal flats in both the EFWS and NFWS, corresponding to relative reductions exceeding 60%. In shallow coastal areas shallower than \(-2.5\) m, mean reductions of about 5 cm s\(^{-1}\) persist, corresponding to relative reductions of roughly 30–40% in the EFWS and 40–60% in the NFWS due to the lower baseline velocities there. The spatial patterns of these mean changes are shown in Supplementary Figs. S4 and S5.

Peak reductions in 95th percentile reach about twice as high as the mean reductions. These peak reductions are a few cm/s higher in the EFWS, while in the NFWS the plateau of elevated reductions extends across a wider range of depth bins (Fig. 14, current row).

In the EFWS, the relative reduction reaches up to 60%, without surpassing 70%, which is also the case for mean reductions. Spatially averaged absolute reductions are approximately 5 cm/s higher in the EFWS (-19 cm/s) than in the NFWS (-14 cm/s), so absolute reductions are greater in the EFWS but not in relative terms.

In contrast, spatially averaged relative reductions are about 7% lower in the EFWS, although a substantial 40-60% reduction extends into deeper bins in the NFWS (Fig. 14, right panel). In the NFWS, bin-averaged reductions exceed 10 cm/s across much of the profile, whereas absolute reductions due to seagrass in the EFWS drop by half toward the landward end.

2090

Under the 2090 SLR scenario, seagrass continues to attenuate currents in the EFWS. For the 95th percentile velocities, localized enhancement of velocity damping is observed in very shallow seagrass-covered flats and in parts of tidal inlets (blue regions in the difference plots, right panels of Fig. 5b ; very similar patterns are found for the temporal mean (see Supplementary Fig. S4). This local increase in attenuation is caused by more frequent and prolonged flooding, which allows seagrass to interact with stronger currents in near-coastal areas that were only occasionally submerged under present-day conditions.

However, while the overall spatial pattern of velocity reduction remains similar, the magnitude of attenuation decreases in most tidal flat areas (Fig. 5b left vs center figure).

However, while the overall spatial pattern of velocity reduction remains similar in both years, the magnitude of attenuation decreases across most tidal-flat and channel-adjacent areas under SLR (Fig. 5b, left vs. center panels). In large parts of the EFWS , particularly in channels and adjacent flats, seagrass-induced velocity reductions at the 95th percentile still locally reach about 0.3 m s\(^{-1}\), but the spatial extent of these strong reductions (dark blue areas) is substantially reduced. The difference maps indicate a loss of attenuation of typically 15–20 cm s\(^{-1}\), with local maxima exceeding 30 cm s\(^{-1}\) (dark red patches in Fig. 5b, right panel). This corresponds to a reduction in attenuation efficiency of approximately 10–20% (Fig. 5c, right panel), particularly in coastal embayments such as Jade Bay.

A consistent response is also found for the temporal mean velocities (Supplementary Fig. S4). Here, relative efficiency losses occur in a similar range as for the 95th percentile, but the absolute reductions are smaller–typically of about 5–10 cm s\(^{-1}\), since the underlying mean velocities (and thus the baseline seagrass effect) are lower than for the 95th percentile. Locally, reductions still reach up to about 15–20 cm s\(^{-1}\) on tidal flats. These losses also extend over a slightly larger spatial area in regions of strongest attenuation decline, such as Jade Bay and the Ems Dollard.

Overall, although seagrass continues to reduce current velocities under SLR, its effectiveness is substantially diminished. In some very shallow and inlet regions, attenuation can locally increase due to enhanced current exposure, but in general the efficiency of current attenuation decreases by up to about 25%, especially in coastal embayments and channel-adjacent areas. On intertidal flats, relative efficiency losses typically range between 5 and 10%.

Baseline responses to SLR in the NFWS are more spatially heterogeneous. Although spatially averaged 95th percentile ( 0.6 m/s in both 1997 and 2090) and mean and velocities ( 0.37 m/s in both 1997 and 2090) remain relatively stable (number avg in Fig. 6a), Fig. S6a) text overlays), distinct regional patterns emerge for SLR induce baseline velocity change : velocities decrease seaward of the Wadden Sea (to the west), whereas they increase within the tidal inlets and main channels to accommodate the larger tidal prism. Notably, some reductions are observed in back-barrier areas, particularly north of Sylt. In this basin, deeper channels occupy a larger portion of the intertidal area, potentially altering the hydrodynamic regime and the spatial extent of flow attenuation.

As in the EFWS, seagrass-induced reductions in the NFWS remain largely consistent for both mean and 95th percentile velocities, with a general declining trend under SLR. Absolute reductions of the attenuation provided by seagrass reach up to 20 cm s\(^{-1}\) for the 95th percentile (Fig. 6b) and about 10 cm s\(^{-1}\) for the mean (Fig. S5b), with relative efficiency losses between 5% and 10% and localized peaks of up to 25% similarly for the 95th percentile (Fig. 6c) and the mean (Fig. S6c).

Seagrass continues to play a vital role in attenuating tidal currents, particularly on shallow tidal flats, where reductions of up to 50% are observed. However, its overall effectiveness is somewhat reduced under future sea-level rise conditions, with efficiency losses ranging from 5% to 25%, depending on location and flow conditions. Reduction in absolute terms most strongly decrease around a present day depth bin of 1 m.

3.3 Wave dynamics

3.3.1 Baseline in present and future scenarios

1997

Here again we focus here on the 95th percentile of significant wave height (\(H_s\)), which better represents high-energy conditions relevant for coastal impact and attenuation processes. Under present-day conditions, the 95th percentile of \(H_s\) seawards of the barrier islands ranges between approximately 1.5 and 2.0 m in both the EFWS and NFWS (Figs. 7b, 8b). Entering the back-barrier Wadden Sea, wave heights rapidly decrease due to shallow bathymetry, enhanced bottom friction, and wave breaking, dropping below 0.5 m and approaching zero in the shallowest intertidal areas.

The corresponding temporal mean wave heights exhibit similar spatial patterns but with substantially smaller magnitudes (roughly 0.6–0.8 m offshore), and are shown in the Supplementary Material (Figs. S6 and S7 for EFWS and NFWS, respectively).

2090

In the future sea-level scenario, the offshore wave climate exhibits opposite trends for mean and extreme conditions. For the 95th percentile significant wave height, wave forcing in the open North Sea slightly increases compared to 1997 (EFWS: Fig. 7a; NFWS: Fig. 8a). In contrast, the temporal mean offshore wave height decreases on average.

Within the Wadden Sea, however, wave heights increase for both metrics despite the reduced mean offshore forcing. This is caused by the increased water depth, which delays and reduces wave breaking and bottom friction, allowing waves to propagate further landward before dissipating. As a result, the Wadden Sea is exposed almost constantly to higher wave energy even when the offshore mean wave climate is weaker (Supplementary Figs. S6 and S7).

Overall, the baseline change from 1997 to 2090 therefore shows a relatively homogeneous increase of about 20–30 cm in the 95th percentile significant wave height, extending from the seaward side of the barrier island chain across large parts of the Wadden Sea (Figs. 7 and 8a, rightmost panels). In contrast, the temporal mean exhibits a dipole pattern, with decreases offshore (by about 10–20 cm; Supplementary Figs. S6 and S7) and increases in the back-barrier regions due to reduced dissipation in deeper water. Consequently, in contrast to the changes in the other ocean variables, wave heights in these particular climate scenraios display a stronger statistically metric-dependent response to sea-level rise. Nevertheless, since the Wadden Sea experiences increasing wave energy in both cases, the qualitative spatial pattern of the seagrass attenuation response (discussed in the next subsection) remains broadly similar, except in the very offshore regions near the barrier islands.

3.3.2 Seagrass-induced attenuation in present and future scenarios

At first glance, the wave reduction induced by seagrass appears largely unchanged between present-day and future sea-level scenarios when considering spatially averaged values. Domain-averaged absolute reductions in the 95th percentile significant wave height are nearly identical in both time slices, amounting to about 12-13 cm in the EFWS (Fig. 7b) and about 9-10 cm in the NFWS (Fig. 8b). Analogously, the spatially averaged reductions of the temporal mean significant wave height remain nearly unchanged, with values of about 5 cm in the EFWS (Supplementary Fig. S6) and about 4 cm in the NFWS (Supplementary Fig. S7) in both time slices. Thus, while the sub domain-averaged magnitude of wave damping by seagrass remains essentially constant, the patterns of attenuation reduction (Figs. 7b 8b, right most panels) is spatially heterogeneous.

In both regions, the seaward-facing vegetated zones (positioned in front of the barrier islands) show decreased absolute attenuation at their offshore edges with respect to the annual mean for 2090 compared to 1997 (red areas in the difference plots, Supplementaty Figs. S6 and S7), with reductions of up to 10-15 cm. This pattern corresponds to a slight decrease in landward-directed wave forcing on average. Conversely, landward of these zones, attenuation generally increases by 5-10 cm (blue areas), reflecting the ability of seagrass to interact with and dissipate the elevated wave energy reaching further across the tidal flats. On the tidal flats themselves, attenuation tends to decrease slightly (red).

Accordingly, for the temporal mean wave climate, the seaward-facing vegetated zones in front of the barrier islands exhibit decreased absolute attenuation at their offshore edges in 2090 compared to 1997 (red areas in the difference plots; Supplementary Figs. S6 and S7), with reductions of up to 10–15 cm. This reflects the reduced mean offshore wave forcing in the future scenario. Further landward, attenuation locally increases by about 5–10 cm (blue areas), because increased water depth delays wave breaking and allows more wave energy to penetrate into the back-barrier basin. On the tidal flats themselves, mean attenuation tends to decrease slightly.

In the EFWS, the large-scale wave damping effect of seagrass remains similar between present and future scenarios (Fig. 7c). Seagrass reduces the 95th percentile significant wave height by approximately 5–10 cm near channels (Fig. 7d) and achieves relative reductions of about 20–50% across intertidal flats (Fig. 7f). The modest increase in future tidal-flat wave heights is accompanied by slightly higher absolute attenuation (typically about +5 cm), while relative efficiency remains similar overall, with only minor reductions of about 5–10%.

In the NFWS, a more heterogeneous pattern emerges. Absolute reductions in significant wave height decrease at the ocean-facing boundaries of the inlets, especially in the southern part of the domain (red in Fig. 8), while attenuation increases in the northern part (blue). Eastward toward the coast, attenuation generally weakens.

In the EFWS, the wave damping effect of seagrass remains similarly strong at the large scale between present and future scenarios (Fig. 7c). Seagrass reduces 95th percentile Hs by \(\sim 5-10\)cm near channels (Fig. 7d) and achieves 20-50% reduction across intertidal flats (Fig. 7f). The modest increase in future tidal flat wave heights is accompanied by slightly higher absolute attenuation (typically +5 cm). Relative efficiency remains similar overall, with minor reductions of about 5-10%.

In the NFWS, a mixed pattern emerges. Absolute reductions in Hs, decrease (red in Fig. 8) at the ocean-facing boundaries of the inlets, especially in the southern part. In the northern part, attenuation increases (blue), while eastward toward the coast, attenuation generally weakens.

Regarding the 95th percentile significant wave height, future reductions by seagrass generally decline. While absolute reductions still reach up to 15 cm in many areas (blue in Fig. 8c) left and center panel), this increase in absolute damping does not keep pace with the larger wave heights penetrating into the Wadden Sea. As a result, relative efficiency declines significantly - often by more than 25% (red in Fig. 8f) right most panel) - despite seagrass mitigating higher wave quantiles. Similarly, relative reductions in mean wave heights decrease by over 20% (red in Fig. 8e) , comparable to the decline observed for the 95th quantiles.

In the EFWS, the large-scale wave damping effect of seagrass remains similar between present and future scenarios (Fig. 7c). Seagrass reduces the 95th percentile significant wave height by approximately 5–10 cm near channels (Fig. 7d) and achieves relative reductions of about 20–50% across intertidal flats (Fig. 7f). The modest increase in future tidal-flat wave heights is accompanied by slightly higher absolute attenuation (typically about +5 cm), while relative efficiency remains similar overall, with only minor reductions of about 5–10%.

In the NFWS, a more heterogeneous pattern emerges. Absolute reductions in significant wave height decrease at the ocean-facing boundaries of the inlets, especially in the southern part of the domain (red in Fig. 8), while attenuation increases in the northern part (blue). Eastward toward the coast, attenuation generally weakens.

For the 95th percentile significant wave height, future reductions by seagrass generally decline in relative terms. Although absolute reductions still locally reach up to about 15 cm in many areas (Fig. 8c, left and center panels), this increase in absolute damping does not keep pace with the larger wave heights penetrating into the Wadden Sea. As a result, relative attenuation efficiency declines substantially, often by more than 25% (Fig. 8f, rightmost panel). A similar decline is also found for the temporal mean wave height, with relative reductions decreasing by more than 20% in large parts of the NFWS (Supplementary Fig. S7), comparable to the response of the 95th percentile.

In contrast, for high-energy wave conditions represented by the 95th percentile significant wave height, the future scenario is characterized by more energetic offshore forcing together with deeper water levels. As a result, extreme waves propagate further into the Wadden Sea before breaking, allowing seagrass to dissipate part of this additional energy. This leads locally to increased absolute attenuation in landward regions. However, this increase in absolute damping does not keep pace with the overall increase in wave energy reaching the basin, so that the relative attenuation efficiency generally declines under SLR.

3.4 Bottom stress magnitude

3.4.1 Baseline in present and future scenarios

1997

Consistent with the current velocity distribution ( Figs. 5, 6), the highest bottom stress values occur within the tidal inlets and main tideways, where flow velocities are strongest. For the 95th percentile, bottom stress in the EFWS reaches values of around 2 Pa within the primary channels (Fig. 9a), while in the NFWS slightly higher values are found on average (Fig. 10a). Laterally towards the outer channel branches and adjacent areas, the 95th percentile bottom stress decreases to about 1.2 Pa. On the tidal flats, values further decline to roughly 0.2–0.6 Pa and gradually diminish towards the coast.

For the temporal mean (Supplementary Figs. S9 and S10), bottom stress in the EFWS reaches approximately 1 Pa within the main channels, decreases to about 0.5 Pa in the outer channel branches, and further drops to roughly 0.1–0.3 Pa on the tidal flats. Similar patterns are found in the NFWS, albeit with slightly higher values on average.

2090

Under SLR conditions projected for 2090, bottom stress generally increases in most regions, both in terms of the 95th percentile (Figs. 9a, 10a) and the temporal mean (Supplementary Figs. S9, S10). These changes largely mirror the spatial distribution of velocity enhancements. Within the main tidal channels, the 95th percentile bottom stress increases by up to about 0.5 Pa (0.2 Pa for the annual mean), with similar magnitudes occurring in the deeper sections between the islands. A comparable increase is also found in the southern parts of the Ems and Jade channels, including the Ems Dollart estuary (Fig. 9a). In channel-adjacent areas and in Jade Bay, the increase is more moderate, typically between 0 and 0.2 Pa (0.1 Pa for the mean).

However, localized reductions in bottom stress are also evident, such as in the southeastern extension of the Jade Bay channel within the bay. Additionally, particular areas within the back-barrier Wadden Sea, such as near tidal sub-basins separating watersheds, exhibit decreasing bottom stress under SLR, especially between the main channels. In these regions, reductions locally reach up to about 0.3 Pa for the 95th percentile (Figs. 9b, right-most panel) and about 0.1 Pa, occasionally up to 0.2 Pa, for the temporal mean (Figs. 9a, right-most panel). In the central Ems channel, the 95th percentile bottom stress even shifts to slightly negative anomalies, although the magnitude remains small and generally closer to 0 than to \(-0.1\) Pa.

As observed for velocity, the spatially averaged 95th percentile and mean of bottom stress remain relatively similar in structure comparing 1997 and 2090 in both regions (Fig. 9a, Fig. 10a, Supplementary Figs. S9a, S10a). An increase in response magnitude to SLR is evident across most tidal inlets and extended channels, reflecting the increased flow required to accommodate the larger tidal prism under sea-level rise. At the same time, reductions in bottom stress are apparent in about 1/3 in the area of the back-barrier Wadden Sea in NFWS, particularly north of the island of Sylt. These reductionsf are more pronounced and spatially extensive than those observed in the EFWS.

3.4.2 Seagrass-induced attenuation in present and future scenarios

In relative terms, this corresponds to reductions of over 80% across much of the vegetated intertidal area and up to 20% even in parts of the adjacent, unvegetated channels (Fig. 9e,f).

Analogous to the baseline change, the reduction in bottom stress due to vegetation (Fig. 9b; Fig. 10b) follows a pattern similar to that of velocity. Notable absolute reductions of more than 0.6 Pa occur in the 95th percentile on intertidal flats adjacent to channels, particularly coastward and toward the watersheds, while corresponding mean reductions are smaller, typically 0.2–0.25 Pa (Supplementary Figs. S9, S10). Some parts of the channels, such as the northeastern area of Borkum, also experience reductions of about 0.3–0.4 Pa (95th percentile) and 0.1–0.15 Pa (mean) (Fig. 9c). In relative terms, this corresponds to reductions exceeding 80% across much of the vegetated intertidal area and up to 20% even in parts of adjacent, unvegetated channels (Fig. 9c).

When evaluated through the spatially averaged absolute reductions, values for 1997 and 2090 remain largely unchanged in the EFWS (95th percentile: \(-30\) Pa in 1997 and \(-29\) Pa in 2090; mean: \(-12\) Pa for both years; Fig. 9b) and similarly stable in the NFWS (95th percentile: \(-22\) Pa for both years; mean: \(-9\) Pa; Fig. 10b). These minor changes in the integrated values reflect predominantly a redistribution of the attenuation effect under sea-level rise. Vegetated intertidal areas now experience stronger currents and stresses, allowing seagrass to attenuate more energy (blue shading, Fig. 9c right panel), while attenuation decreases in parts of the deeper channels (as indicated by partially inverted signs between Fig. 9a and Fig. 10a, right panels). Similarly, the reduction by seagrass tends to increase on tidal flats where the plants are exposed to higher currents for longer durations. In contrast, attenuation weakens where this interaction diminishes–for example, in the tidal basin area north of Sylt, where the overall effect on the tidal channels is reduced.

Overall, in absolute terms, seagrass remains similarly effective in reducing bottom stress under sea-level rise, with reductions up to 0.5 Pa and more than 80% in relative terms. The effect becomes less pronounced in transition zones between vegetated flats and larger channels (especially the Jade and Ems channels), where relative reductions drop below 40%–about 20% less than under present sea level. Nevertheless, reductions still reach over 30% in the NFWS and up to 40% in the EFWS for the 95th percentile.

3.5 Sediment dynamics

As in Jacob et al. (2023) the class integrated bottom layer suspended sediment is used here as proxy for erosion based of the morphostaic simulations.

3.5.1 Baseline in present and future scenarios

1997

The highest bottom-layer SPM concentrations in the EFWS are observed in the 95th percentile, reaching up to about 1.2 g/L along the seaward-facing beaches of the barrier islands, around ebb-tidal deltas, and along main channel thalwegs, including deep paleo-river valleys such as the Ems, Jade, and Weser channels (Figs. 11a, 12a). The corresponding annual mean concentrations are lower, generally up to 0.4 g/L, with isolated peaks exceeding 1 g/L (Supplementary Figs.S10a, S11a).

2090

In the future climate scenario, the 95th percentile SPM concentrations generally increase across most of the EFWS by up to 0.1 g/L, with local maxima exceeding 0.5 g/L in regions such as Jade Bay (Fig. 11a, right panel). Only the westernmost coastal areas in front of the barrier islands show slight decreases. Concentrations increase within main tidal channels and decrease near shallow supralittoral zones, especially along embayment edges like the Ems-Dollart and Jade Bay (Figs. 11a, right panel; Supplementary Figs. S6–S7 for reference to mean patterns).

The annual mean SPM concentrations (Supplementary Figs. S10a, S11a), which are typically about three times lower than the corresponding 95th percentile values, exhibit a more heterogeneous response. Reductions in temporal mean SPM concentrations occur north and west of the barrier islands (Supplementary Figs. S10a). This behavior is directly linked to the weakening of the mean wave climate in the coastal North Sea under the 2090 scenario relative to 1997 (see Section 3.3). Because waves play a key role in mobilizing sediments along the seaward sides of the barrier islands and feeding the Frisian littoral drift system from west to east (e.g., Staneva et al., 2009; Stanev et al., 2006), the reduced wave energy leads to a pronounced weakening of the sediment transport band in front of the islands with a 1997 to 2090 annual mean SPM reduction exceeding 0.15 g/L in these regions (Supplementary Fig. S10a).

In the NFWS, the baseline SPM response to sea-level rise is more complex and spatially heterogeneous (Fig. 12). For the 95th percentile concentrations (Fig. 12a), pronounced local increases occur near the western shores of the Danish islands and near the German islands south of Sylt, with peak anomalies reaching up to 0.2 g/L. However, these regional increases are outweighed by more widespread decreases across the basin, resulting in a slightly negative spatially averaged change of about \(-0.01\) g/L.

In contrast, the annual mean bottom-layer SPM concentrations (Fig. 12a; Supplementary Figs. S10b, S11b) exhibit a more uniform large-scale decrease across roughly the northern two-thirds of the Wadden Sea. Increases are largely confined to selected tidal channels, particularly in the northern and southern sectors. As in the EFWS, the strongest reductions occur along the exposed island beaches and around ebb-tidal deltas.

3.5.2 Seagrass-induced attenuation in present and future scenarios

As with previously analyzed variables, the spatial pattern of seagrass-induced SPM attenuation for the 95th percentile remains largely consistent between the 1997 and 2090 time slice comparisons for both the EFWS and NFWS (Figs. 11b, 12b left and central panels). At first glance, the attenuation effect appears unchanged, with spatially averaged absolute reductions (as displayed by the avg numbers in the panel titles) remaining at approximately -18 to -19 g/L in the EFWS and \(-13\) to \(-12\) g/L in the NFWS (Figs. 11b, 12b, left and central panels). Analogously, for the temporal mean, the spatially averaged reductions remain at about \(-0.06\)/\(-0.04\) g/L in the EFWS/NFWS (Supplementary Figs. S10b and S11b; Figs. 11c, 12c, left and central panels). Concerning the temporal mean, the highest reductions by seagrass (exceeding 0.25 g/L, dark blue) occur around the islands and in areas between tidal channels. Within the main channels themselves, the maximum SPM reduction reaches about 0.06 g/L, while the effect diminishes toward the supralittoral zone and approaches zero in the deeper channel centers.

Since baseline 95th percentile SPM concentrations are approximately three times higher than the annual mean, the absolute reduction is also about three times larger, with peak reductions exceeding 1 g/L, still indicating an effective attenuation:

In relative terms, the attenuation by seagrass for the 95th percentile remains broadly comparable to the mean values but tends to be slightly lower overall, reflected by a roughly 2% decrease in the spatial average. Efficiency losses in SPM attenuation closely follow areas where baseline SPM concentrations and mean SPM reduction also decrease, especially along the edges of the tidal channels. In these regions, relative efficiency drops by more than 25% (Fig. 11c, 11c, right panel). By contrast, on the tidal flats, changes in efficiency remain small, with reductions generally below 5%.

Regarding the temporal average Most notably, a reduction in attenuation magnitude ( exceeding 0.15 g/L) is observed along the northern (seaward) sides of the barrier islands, which strongly correlates with discuessed reductions in baseline SPM concentrations. South of the barrier islands, however, attenuation increases in many areas (indicated in blue in the right panel of Fig. 11b), reflecting stronger interactions between seagrass and elevated SPM under future sea-level conditions.

Despite these losses, SPM removal by seagrass remains highly effective under future scenarios and is broadly comparable to present day (1997) levels. The distribution of attenuation in the Wadden Sea across depth intervals from -5 m to 15 m (Fig. 13), indicates reductions near upper spring tidal range (-2.5 m [negative depth]) to remain substantial, removing a few tenths of a gram per liter on average.

Efficiency is slightly enhanced toward shallower areas (-2.5 m depth), although relative efficiency at these shallow depths decreases by up to 25-40% under SLR. In deeper intertidal areas, the absolute concentration reductions remain substantial, achieving spatially averaged bottom-layer SPM reductions of around 60% in the EFWS and 40% in the NFWS. On average across all intertidal flats, the SPM concentration decrease is about 4% in the EFWS and 1% in the NFWS. Hence, even where absolute reductions are maintained, they do not scale proportionally with the increased SPM concentrations under future conditions.

This pattern also holds for extreme events. For the 95th percentile SPM concentrations, seagrass continues to dampen suspended sediment effectively under projected sea-level rise, with only a modest decrease in mean efficiency (\(\sim \)5% decrease on average across the Wadden Sea). Specifically, efficiency decreases from 61.3% to 55.9% in the EFWS, and the NFWS exhibits slightly less effective attenuation than the EFWS (Fig. 14), with relative efficiency declining by about 3% spatially-averaged (from approximately 30% to 36% under future projections).

This study follows up on Jacob et al. (2023) to evaluate whether seagrass meadows in the German Bight can continue to mitigate hydrodynamic energy and sediment resuspension under projected future sea level rise (R1), and to assess differences between the East Frisian Wadden Sea (EFWS) and the North Frisian Wadden Sea (NFWS) (R2). The present results widely suggest that seagrass will retain a substantial protective function under 2090 hydrodynamic conditions, although its relative attenuation capacity decreases with increasing water depth across most hydrodynamic variables.

4.1 Ranking outcomes across variables

Among all variables, SPM and bottom stress showed the largest percentage reductions due to seagrass, both under present and future scenarios. This underscores the role of seagrass as an efficient attenuator of currents and turbulence that limits erosion. Even under projected sea-level rise (SLR), absolute bottom-stress reductions often exceeded 60% in shallow areas, and relative reductions of 40-60% were observed. This pattern reflects the tight coupling between bottom stress and depth-averaged velocity, suggesting that currents – rather than waves – remained the dominant contributor to bed shear stress.

Significant wave height responses were more complex. In the shallowest Wadden Sea areas, absolute reductions by seagrass increased under SLR because greater offshore wave energy propagated landward across the now-deeper tidal flats. Although the offshore average wave climate is projected to decrease slightly, deeper water allows higher energy waves to reach the vegetated flats. This results in larger absolute wave height attenuation at the 95th percentile under future conditions, confirming that seagrass still intercepts a substantial portion of wave energy in these areas. However, relative efficiency (i.e. proportionate reduction) decreases with water depth under SLR. This implies that while seagrass can dissipate higher baseline wave heights and therefore achieve larger absolute reductions, the fraction of energy removed declines. This is a logical consequence of orbital velocities weakening at canopy level as water depths increase. Following Fonseca and Cahalan (1992), seagrass most effectively dissipates wave energy when the canopy height is comparable to water depth. Our results confirm that as mean water depth increases by \(\sim \)0.7 m, relative wave attenuation drops up to 25% along tidal channels and barrier-island edges, where seagrass is submerged most of the time. However, these losses in relative efficiency are less pronounced on shallow flats and in areas of increasing inundation frequency. Thus, seagrass remains effective for mitigating waves in absolute terms under future conditions, especially during extreme events, but its relative attenuation capacity (in proportion to the increased baseline) is lower.

SSH exhibited the weakest relative and absolute attenuation by seagrass. These findings align with earlier studies including ours, suggesting that seagrass has minimal influence on mean water levels. Reductions in mean SSH remained very small in absolute terms ( 0.01-0.02 m), though the relative reductions appeared greater due to small baseline variations. At higher percentiles, seagrass-induced reductions also decreased by approximately 50% under SLR. Hence, any already minor effects on extreme SSH further weakened under future water-level rise.

Overall, absolute reduction efficiencies across most variables were similar to present day, with a tendency for slightly reduced attenuation on the seaward (offshore) side and a slight increase on the landward side - especially on the most exposed flats. However, the absolute energy present in the Wadden Sea is higher under SLR, implying that even small absolute reductions continue to represent a significant ecosystem service. Nevertheless, the relative efficiency of seagrass generally declined by 10-20%, a signal of diminishing interaction as water depth increases. By that means SSH was the most sensitive to SLR variable.

4.2 Contrasting the EFWS and NFWS responses

Both the EFWS and NFWS share similar qualitative response patterns under present and projected future conditions. In both areas, seagrass most effectively attenuated hydrodynamic energy and sediment resuspension across shallow flats, while relative efficiency generally decreased towards deeper, more exposed channels. However, the two systems also exhibit subtle differences driven by baseline bathymetry and hydrodynamics.

First, the EFWS is characterized by more enclosed basins, shallower average depths, and slightly larger tidal amplitudes compared to the NFWS. These factors promote lower baseline wave energy and stronger currents; thus resulting in higher peak wave attenuation in the NFWS and higher peak current attenuation in the EFWS, as well as a more uniform spatial pattern of SPM and bottom-stress attenuation. Even under future sea-level rise, efficiency losses in the EFWS remained relatively uniform across most variables.

The higher wave permeability of the barrier island chain (wider inlets) in NWFS results in stronger wave energy propagating further landward, especially under SLR, which in turn yields larger absolute attenuation on the shallow tidal flats. However, the relative efficiency losses at the landward fringe were more pronounced in the NFWS. Interestingly, when spatially averaged across the entire region, the EFWS showed a greater overall relative reduction in attenuation efficiency – despite starting from a higher baseline under present day conditions.

As water depths increase under SLR, the proportional influence of seagrass on waves and currents declines across both systems, and the similarities between the two regions are lastly more pronounced than the differences.

4.3 Limitations of the modeling approach

In our preceding study (Jacob et al. 2023) we argued: "As seagrass impacts on sea level are minor, seagrass contributions to flood protections are only indirect. The potential of seagrass to support the vertical height growth of the Wadden Sea to maintain bathymetric control under future increases in sea level is seen as the major long-term contribution of seagrass to coastal protection."

4.3.1 Stressors

However, this study focused solely on hydrodynamic effects and did not account for other critical stressors that may reduce seagrass persistence.

Such stressors include both direct and indirect impacts from anthropogenic activities, such as maintenance dredging in navigation channels, which can lead to the physical removal or burial of vegetation, as well as increased turbidity affecting light availability (Erftemeijer and Lewis 2006) as well as eutrification (Brockmann et al. 2018).

Climate change-related stressors, including rising water temperatures and shifting light regimes, may also impair seagrass health and productivity, potentially limiting its long-term stabilizing effects. Assessing the biological response of seagrass to future warming thus represents an important next step. Nevertheless, Singer et al. (2017) projected a potential expansion of the dominant local seagrass species under future climate scenarios.

4.3.2 Time-slice modeling

Our modeling employed a time-slice approach with two discrete snapshots for present day and 2090 conditions driven by RCP8.5 scenarios. While this strategy isolates the hydrodynamic response to projected sea-level rise and wave climate changes, it omits continuous morphodynamic evolution. Sediment accumulation and bed-level adjustments over time, which could enhance bed elevation and further support wave and current attenuation, were not represented. Although fully coupled morphodynamic-ecological simulations could, in principle, capture these long-term feedbacks, such simulations are computationally expensive, and their credibility decreases over longer time spans as uncertainties compound. Factors like sediment availability, spatial sediment redistribution, and ecological responses remain difficult to constrain under future scenarios. Future research incorporating morphodynamic-ecological feedbacks would provide a more comprehensive view of the long-term protective role of vegetated tidal flats under changing climate.

4.3.3 Fixed bathymetry and eco-morphostatic simulations

However, the magnitude and spatial patterns of such (eco-)morphodynamic adjustments over multi-decadal time scales remain highly uncertain and depend on sediment supply, organic matter accumulation, vegetation dynamics, and storm history (Wang et al. 2012; Jordan et al. 2021; Pineda Leiva et al. 2025; Hoagland et al. 2023). In order to avoid introducing unconstrained assumptions that could dominate the results, we adopted a fixed-bathymetry ("eco-morphostatic") approach that avoids the large uncertainties that accumulate in long-term morphodynamic simulations and isolates the hydrodynamic and sediment-transport response to sea-level rise, wave climate change, and vegetation. However, this simplification also implies that natural vertical accretion or erosion processes, which may occur alongside sea-level rise, are not accounted for:

Recent studies suggest that (eco-)morphodynamic processes can, depending on local conditions, lead to substantial vertical accretion in intertidal seagrass meadows, potentially amounting to several decimeters over multi-decadal time scales, which is non-negligible relative to projected sea-level rise. For example, Zou et al. (2021) report accretion rates on the order of 4.2 mm yr\(^{-1}\) in southern Taiwan. In the Dutch Wadden Sea, field experiments by Bos et al. (2007) indicate that intertidal eelgrass patches can accrete 5–7 mm during the growing season, although this sediment may be partially released again during winter. This suggests that the role of seagrass can range from facilitating long-term bed-level building to acting as a temporary sediment buffer. The dynamics and eventual sedimentary consolidation processes have a strong dependence on canopy density (Chen et al. 2024) and seagrass seasonal (for seasonal intertidal species ) variability (Mohr et al. 2025).

In our experiments, both bathymetry and vegetation are prescribed and fixed. Vegetation effects are highly nonlinear: shoot density, canopy height, and diameter strongly control attenuation, with threshold behavior and diminishing returns at high densities shown in XBeach sensitivity experiments using the same vegetation formulation (Chen et al. 2024).

We therefore conducted a limited sensitivity analysis with the present model under 1997 forcing (two-week simulations), varying shoot density and canopy height. For the Borkum seagrass meadow extent of the Jacob et al. (2023) scenario, the meadow-averaged 95th-percentile bottom stress (0.341 Pa without vegetation) was reduced by 40–70%, increasing systematically with both shoot density (450, 1530, 7360 shoots m\(^{-2}\)) and canopy height (0.099, 0.144, 0.19 m). For a fixed canopy height, increasing shoot density from 450 to 7360 shoots m\(^{-2}\) yielded an additional 20% reduction in the high-end stress quantile, while for fixed density, increasing canopy height from 0.099 to 0.19 m contributed an additional 10–12% reduction. Wave attenuation showed a similar but weaker response: the meadow-averaged 95th-percentile significant wave height (0.301 m without vegetation) was reduced by 15–40%, with an additional 20% reduction for increasing shoot density at fixed canopy height. Comparable sensitivities for mean values and other variables are provided in the Supplementary Material.

The intentionally very high shoot density adopted here should therefore be interpreted as an upper-bound estimate of the impact of seagrass.

4.3.4 Scientific contributions despite limitations

Despite these simplifications, this study provides a robust, process-based assessment of seagrass effects on hydrodynamics, sediment mobilization, and wave energy under changing sea levels. To our knowledge, this is one of the few high-resolution 3D hydrodynamic and wave-modeling efforts to explicitly isolate vegetation-induced attenuation under present and projected climate conditions in the German Bight. Our results demonstrate that, although relative efficiency decreases under sea-level rise, seagrass meadows continue to produce substantial absolute reductions in waves heights and currents – preserving an important protective function even as water depths and offshore energy increase.

As the water column raise on the wadden sea permitted significantly increased wave height, repeatedly points towards the implied merit of potentially furthered sediment accumulation via seagrass, as most valuable contribution, compounding over climate time scales as bathymetry exerts a stronger control on hydrodynamics than vegetation does.

Overall, this study highlights the continued relevance of seagrass meadows for coastal resilience under future climate conditions and supports their role as a nature-based solution within integrated adaptation strategies. At the same time, it identifies a clear research priority: the need to better integrate hydrodynamics, sediment dynamics, and vegetation-driven morphodynamic feedbacks in order to fully quantify the long-term coastal protection potential of vegetated tidal flats.

A multi-physics coupled model for the German Bight, incorporating vegetation dynamics, was applied to simulate scenarios with and without seagrass under both present day conditions and RCP8.5 future projections. It was used to address the research question of how sea-level rise affects vegetation-induced attenuation of hydrodynamics, wave energy, and sediment content. For each time slice, vegetation and vegetation-free control scenarios were statistically analyzed using monthly mean values and 95th percentiles to capture both average and extreme conditions, respectively reflecting potential hazard mitigation during high-energy conditions and long-term ecosystem contributions under average conditions.

The results of our study suggest that from a purely hydro-kinematic perspective, seagrass as a NBS, can continue to exert a substantial attenuating influence on currents, waves, and bottom stress under higher water levels. Overall both regions, the EFWS and NFWS respond similarly to sea-level rise, with some reduction in relative attenuation efficiency at increased depths. Even under sea-level rise, bottom stress reductions frequently remain on the order of several tens of percent in vegetated areas. Whereby these estimates represent an upper bound scenario under the assumption of a dense, static seagrass distribution.

Although limited to hydrodynamics and a time-slice approach, the study points to potential shifts in seagrass attenuation efficiency under SLR, underscoring the role of sediment accumulation balancing net water column increase for a maintained bathymetric control, advocating for seagrass-based nature-based solutions in coastal adaptation.