Using salt marshes for coastal protection: Effective but hard to get where needed most

Salt marshes provide valuable ecosystem services such as carbon sequestration, water purification and coastal protection (Barbier et al., 2011; Duarte 2013; Shepard 2011). The combination of these natural ecosystems with hard engineering like seawalls or dikes for the purpose is known ecosystem-based defence (Figure 1a). Considering projected rise in sea level increase storm intensity (IPCC, 2014; Morris 2020), may offer a more sustainable cost-effective solution to ensuring safety, compared alone (Schoonees 2019; Temmerman Vuik 2019). However, implementation solutions requires an in-depth understanding (i) where salt can be established, based on environmental characteristics; (ii) how effective are reducing uprush onto dike due waves, referred wave run-up. applicability strongly depends their effectivity attenuating waves. Dike overtopping, run-up reaches passes over crest, forms major flood-safety risk, it induce sea-wall breaching by eroding back (Schüttrumpf & Oumeraci, 2005; Zhu 2020). Hence, seawall height typically designed resist high levels during surge corresponding overtopping volume (EurOtop, 2018; Lashley, Jonkman, 2021). Elevated foreshores fronting will reduce depth so that large incident waves limited forced break (Bouma 2014). shoaling breaking induced shallow leads several nearshore processes, including mean level, setup growth long-period infragravity also ‘surfbeat’ (Keimer 2021; van der Meer, al. (2021) examined effects detail concluded have net positive effect safety extent which overtop structure behind them. attenuation bare tidal flats result having higher stable bathymetric elevation, combined imposing vegetation friction orbital motion (e.g. Bouma Leonardi Möller 2016). magnitude vary depending species, seasonality grazing, characteristics plant height, biomass, structural complexity (leaves branches complexity), stem density stiffness 2005, 2010; Maza 2022; Möller, 2006; Ysebaert Locations expected attenuate less storms, ultimately resulting Finally, locations directly exposed wind direction likely experience (Vuik would logical expect at wind-exposed than sheltered locations. develop. Marsh presence soil elevation relative (Balke 2016), sediment supply (Fagherazzi 2020; Ladd Liu 2021) local exposure (Wang 2017). Bed-level change inundation period drive marsh edge location: lower elevations longer periods, bed needed preserve seedlings 2016; Willemsen As wind-driven cause dynamics, they impede seedling establishment, causing restricted sometimes even cliff formation retreat 2014, Wang Despite growing we lack quantitative landscape-scale analyses relating long-term occurrence regional scale morphological development foreshores. Currently, measurements varies scale, variation foreshore properties bathymetry, characteristics, range exposure. To our knowledge, most work modelled tested flumes using artificial 2021), applied beaches Didier Polidoro, Field regarding remain scarce not making possible perform regional-scale (Spencer 2015; Moreover, been related ability establish across deepen insight defence, investigated Dutch Wadden Sea relates tidal-flat bathymetry presence/absence marshes, properties, intertidal area barrier islands (Reise 2010). Our study focuses stretch mainland coast between Harlingen (53°10′46.2″N, 5°24′34.3″E) Eemshaven (53°27′53.1″N 6°45′33.7″E), Netherlands 2a). This micro meso-tidal (RWS, 2013). protected against flooding 8–10 m slope average gradient ~1:4 seaward side (van Loon-Steensma After construction ‘Afsluitdijk’ (a closing off Zuider Zee from Sea) 1930s ‘Lauwersmeerdijk’ Lauwerszee) 1969, changes currents led current distribution (De Jonge 1993). Marshes currently present some areas along accretion works aimed land reclamation (Text S1). was contour lines 1 NAP (i.e. ordinance similar level) extracted two sets maps area: 2004 (5 resolution) 2012–2017 (2 (Source: Rijkswaterstaat, Ministry Infrastructure Water Management) QGIS. Contour corresponded pioneer (excluding pre-pioneer vegetation), contrasted polygon 2002 2014 obtained respectively. lines, rather maps, were used analysis because objective representation addition being area. 0.5 (hereafter called ‘upper flat’) 0 visualize location elevated relation presence. Changes width (2004 2012/2017) upper flat front. toe intersection calculated time steps drawing transects perpendicular spacing 500 Digital Shore-line Analysis System ArcGIS (Himmelstoss 2018). Width front same procedure but both steps. subtracting package ‘raster calculator’ accretion/erosion values first 100 each transect value change. Five locations, different grazers present, selected monitored 3 years (2018–2020), interface marsh-mudflat Uithuizen (53°27′26.8″N 6°40′14.0″E), Den Andel (53°25′36.1″N 6°30′53.7″E), Holwerd (53°23′18.9″N 5°56′05.9″E) Zwarte Haan (53°18′44.8″N 5°37′34.0″E) 1b). Eemshaven, Andel, easternmost ranges western Haan) 2018) All except Haan, framed brushwood groynes ~0.5 high. carried out under permission Province Friesland (01614250), It Fryske Gea (BIG/2018/6245), Staatsbosbeheer Natuurmonumenten. Transects loggers installed all measure heights calculate total transect, one logger (OSSI-10-003C) deployed 10 cm above ground next equivalent distance mudflat (~300 m) 1c,d). December 2018 November 2019 (Table 1; Table Additional significant data offshore Rijkswaterstaat six stations In this paper, focused subset 9 tides inundated sensors, them occurring winter (December–February), spring (April) autumn (September) 1). Mean sensor described script Marin-Diaz, Fivash, (2021). Values hour averaged reference analysis. Wave percentage reduction dike. wave-attenuating capacity derived measuring maximum deposit debris/beach wrack 2020) Beach measured rtk-DGPS (Leica GS12) every ~30 1–3 km reached zones only 8 January 11 February 2020. actual beach field. Vegetation cover, structures branches) dominant 1c) September February, March, August 2020 2021. For zone, covered visually assessed overall ruler 6 replicates. March 2020, counted three 20 × plots collected zone. 10–20 stems per plot, diameter species laboratory. From 2–5 (2005). Samples then dried 60 degrees until constant weight obtain dry biomass. winter, represent standing tussocks; therefore, general cover should taken into account. Soil profiles 2 resolution newest date (2017–2019) (source: Rijkswaterstaat) profiles, 400 ‘offshore elevation’ 40 (to avoid ditches) ‘elevation dike’. These statistical relate correspondent addition, seasonal field ~10 determine stability mudflats. Variability standard deviation transect. Wind speed storms Royal Meteorological Institute (KNMI), recorded weather station Lauwersoog study, fetch considered specific system direction. Fetch point waver r (Marchand Gill, inputs peak shapefile European Environment Agency. Relative (RE) multiplied speed. One-way ANOVA followed Tukey HSD post-hoc test differences variability (measured field). Spearman correlations analyse relationships heights. Pearson relationship determined multiple linear regression independent variable. Differences levels, run-up, among mixed models (LMM) random factor tests. could dates without sensors calculation, contrast calculation LMM investigate width, RE response variables represented models. First, simple model explanatory variable run isolated variables. Secondly, collinearity variance inflation factors (VIF) (Zuur We dropped RE, correlated (VIF > 3). Third, parsimonious explaining stepwise lowest Akaike's information criterion. Interactions included importance (chi-square values) small main S2). explored affected underlying again date. zero excluded Significance fixed Type II Wald chi-square Marginal conditional R2 sjstats r. Normality homogeneity checked residuals plots, log transformed when necessary comply assumptions. Statistical performed R 3.5.0 (R Development Core Team, (>0.5 NAP), hereafter flats’, human interventions (groynes/sedimentation fields) 2a; Figure extension has remained past decade Most landward erosion occurred (up −100 13 years) includes ‘Zwarte Haan’ 2b,c; S2), confirmed observations S3). Overall, 2b,c). Vice versa, expansion marsh. Expansion up 300–400 (in 200 Groningen (east). 3-year monitoring mostly (the 3.1 while 1.5 cm) (ANOVA: F8,90 = 7.5, p < 0.001) S4). dominated Spartina, Salicornia and/or Puccinellia. plateau limit dense (Spartina Puccinellia) 1c). tallest Phragmites Aster) found throughout seasons, grazed livestock, non-domesticated deer occur occasionally (Figures 1c 3a). cows, horses sheep respectively had shorter year Grazer footprints seen abundant close narrower other transects, composed biomass declined Shoot Atriplex lost flowering basal significantly varied Phragmites) S5). (MLR: 0.99, 0.001). Therefore, levels. values, eastern ~20 independently attributed east west (LMM: X2 192.03, 5a). noted influence non-linear interactions form dynamic sea, there influence, prominent east. Given always fronted mudflats, (LMM wrack: 397.12, 0.001, run-up: 155.6, 4c,d). increased Higher observed increasing depths elevations) 4b,e 5b). Waves slightly further Oude Westereems Closer Uithuizerwad, already comparable argue considered, played larger role conditions. explained astronomical S6). best dike, latter accounting S3; Figures 4 6; model: mR2 0.78, cR2 0.9 0.66, 0.85 respectively). wider therefore X2(1) 154, 0.001 131, respectively), explains wracks 5). Furthermore, near <2.5 (around soil), stronger limitation Surprisingly, neither (m), nor explain Although incoming directions, (which harbour) S7), determining overruled effects. decreased (Tables 2; S7). 4f) mudflats attenuated lesser if short grazing 4e; Within greater 4a; Holwerd, completely attenuated, smaller 4b,f; Uithuizen, reaching 4b). summary, ~300 effectively mitigated medium-intensity storms—with little no cases S8), approximately storm-surge level. At >2 Holwerd), narrow ~100 On hand, (<300 low (<1.5 NAP) marsh) effective. shows protect state vegetation, management simplify considerations include adjacent nature-based schemes. accreting flats, highlighting protection. Including benefit protection, fisheries support Temmink 2022). (above transported deposited tides, (Schuerch results show vertically expansion. While edge, promoting expansion; lead (Callaghan Hu establishment opportunities dynamics case (west area, largest 4f), low-lying flats. predominant intensities last decades 2d), promoted edges (cf provides evidence (by ~3 study) mainly forming soils 1c; S4) reduced 2). moderate consequence depth-limited (Zhu Sea, East experienced (~100 width), marshes. 300 ~1.5 effective, guarantee severe Nevertheless, highest often occurs part (Möller Spencer, 2002; 2011), still protective (1/1000 worse) encountered period. flume experiments 2014) report dampening, although sites, contribution parameters space made impossible create single fit statistics. example, ungrazed taller stiffer somewhat ratio drag forces Conversely, livestock kept shorter, non-grazed vegetation. Similarly, play summer. low-intensity recommended high-intensity over-compaction, excessive decrease biodiversity, resistance preventing dominance root (Davidson 2017; Kosmalla Govers, Pagés did harbour winds, location. 7a). vulnerable deeper waves) naturally 7b). sense, (‘where need most’) sufficient thus loadings providing adequate windows 2015, Large-scale exposure, currents, surges, supply, dredging activities well biotic Mariotti Fagherazzi, Schuerch 2014), study. Management conditions hydrodynamics) key (Hu Enough especially important keep sea-level situations able migrate inland constructions (Doody, Foreshore require actions watershed opening upstream dams river (Mariotti If possible, nourishments option (Baptist Eemshaven), creating opportunity hydrodynamics wave-breaking woody debris) Dijkema Falkenrich A ecologically alternative restore seagrass shellfish reefs (Chowdhury Walles 2015). success restoration depend hydrodynamic forcing enough develop (Marin-Diaz, Schoonees insufficient accretion, intervention, engineered floods heightening) necessarily developing ‘where most’. problem means measures locations: either stimulate those places cannot help strengthen too costly, achievable, going expense ecological loss migratory birds (Boere Piersma, 2012). Beatriz Laura L. Daphne Wal, Christopher H. Tjeerd J. Han Olff conceived ideas methodology; Kornelis de Jong Jan-Willem Nieuwenhuis contributed importantly experimental design selecting contrasting relevant compare perspective authorities. Leon Kaptein Pol Martinez-Garcia data; Marin-Diaz analysed Olff, analysis; writing manuscript. authors critically drafts gave final approval publication. Perspectief research programme All-Risk project number P15-21 B1 (partly) financed NWO Domain Applied Engineering Sciences, collaboration following private public partners: (RWS), Deltares, STOWA, authority Noorderzijlvest, Vechtstromen, Gea, HKV consultants, Natuurmonumenten, waterboard HHNK. L.L.G. funded grant 016.Veni.181.087. thank Lennart IJzerloo NIOZ technical assistance; Annette Wielemaker helping obtaining GIS maps; Nelly Eck, Jacob Hogendorf, Panagiota Stergiou Santiago Amaya laboratory; Fabris Zee, Panagiotia Stergiou, Isabelle Buyens, Sarah Paulson, Lissie Groot, Michelle Jongenelen, Thije Zuidewind, Nadia Hijner Lucia Irazabal declare conflict interests. Data available via 4TU.Research Repository https://doi.org/10.4121/872ad544-c7c7-41c2-8307-3e353c9befd2 (Marin-Diaz 2023). Text S1: Summary history creation Sea. Recording settings utilized loggers. S2: Output flotsam interactions. S3: Set random. Significant publicly series Rijkswaterstaat. Intertidal (0 2012/2017. Erosion Netherlands. S4: S5: season (a) diameters, (b) stem, (c) attached stem. S6: Relationships S7: Effect types attenuation. S8: Correlations dates. Please note: publisher responsible content functionality any supporting supplied authors. Any queries (other missing content) directed author article.