Abstract
Resilience of coastal ecosystems to climate change is largely determined by the interaction between plants and the surrounding tidal environment. Research has tended to focus on processes operating at the local scale to explain resilience mechanisms, overlooking potentially important landscape-scale processes and patterns. We show from aerial images spanning 67 years across 3 estuaries that saltmarsh loss was compensated by expansion elsewhere in the estuary when tidal channels shifted position. Compensatory expansion rates were as high as 6 m/yr. This phenomenon of “geomorphic compensation” represents a hitherto overlooked large-scale self-organizing pattern that facilitates the long-term persistence of marshes in estuaries. The geomorphic compensation pattern likely also occurs in other hydrological systems including mangrove forests, and seagrass meadows, and river islands. Compensatory erosion-expansion patterns occurred at the same time as net marsh extent increased by between 120 and 235% across all three estuaries. Marsh expansion mostly occurred in the lower parts of each estuary, where channel migration and compensatory expansion was less evident. Patterns of geomorphic compensation therefore appear to operate at discrete spatio-temporal scales, nested within a hierarchy of coastal morphodynamic processes that govern longer-term patterns of either net marsh gain or loss. Coastal ecosystem resilience can therefore only be fully appreciated when examining erosion and expansion patterns at both local and landscape scales. The intrinsic dynamics of marshes described here have important implications for the long-term delivery of ecosystem services.
Introduction
Estuaries are amongst the most productive environments on Earth, supporting salt marsh, mangrove, and sea grass ecosystems that are the focus of recent global conservation efforts (Van Coppenolle and Temmerman, 2020). Salt marshes globally are considered at-risk from climate change and human pressure (Nicholls et al., 2007; ; ), placing under threat important ecosystem services such as carbon sequestration and coastal flood protection (Möller et al., 2014; Rogers et al., 2019). Whilst large-scale changes in areal marsh extent have been anthropogenically induced (; Silliman et al., 2012), marshes expand and erode in response to natural variation in the environment (Reed et al., 2018; ; ). Understanding the mechanisms of marsh change is a first step toward elucidating the effects of human induced changes to marsh extent, including those associated with climate change. To date, studies have focused on local-scale interactions between plant growth, sediment mobility and wave/current forcing to explain why marshes contract and expand tens of meters at the seaward marsh edge (van Wesenbeeck et al., 2008; Wang et al., 2017). Fewer studies have considered how changes in forcing at the landscape scale shape marsh extent (Leonardi et al., 2016; Ladd et al., 2019a). Migrating tidal channels have been identified as a likely driver of lateral marsh dynamics across entire estuaries (Pringle, 1995; ; Traini et al., 2015), however, the principle has remained untested until now.
Estuaries are highly dynamic and subject to erosive river and wave action that influence the expanse of ecosystems. Erosional forces are concentrated in channels that drain the estuary of tidal and riverine water, and the meanderings of channels have profound influence on the shape of the estuarine landscape and the distribution and cover of its ecosystems (). Channels are known to periodically change course and the small-scale effects on ecosystems from such shifts are well described, including local habitat erosion by scouring and/or the expansion of ecosystems through the redistribution of sediment (). However, the extent to which channel shifts explain the periodic re-distributions of ecosystems around the margins of estuaries is less explored. Little is known about how channel shifts affect the sum area of ecosystems at the estuarine scale, and how resilient ecosystems are to periodic shifts in channels. Here, we use salt marshes as a model system to explore how shifting tidal channels affect landscape-scale ecosystem resilience.
Tidal channels migrate across the estuary either gradually in response to displaced bidirectional flood and ebb tidal currents (Li et al., 2008), or rapidly after river floods, storm surges or dredging activities that can move large volumes of sediment along channel margins (; ). The movement of channels scour the estuary bed where flow is greatest (Sylvester et al., 2019). A channel shift near a marsh edge will grow the elevation difference between the marsh and the downshore tidal flat. The step-change in elevation concentrates wave and current forcing which drives marsh erosion (; ; Willemsen et al., 2018). Once initiated, rates of lateral erosion can be as high as 8 m yr–1 (). Where channel flow is reduced after migration, new sediment deposition raises the tidal flat to elevations suitable for the establishment of plant seedlings and the seaward expansion of the marsh (Wang and Temmerman, 2013; ).
Whilst channel shifts undoubtedly account for some marsh erosion and expansion, the extent to which they explain longer term and estuarine scale marsh distribution remains unclear. Moreover, studies of marsh change have also tended to focus on individual marshes, even though the effects of channel migration are likely to extend well beyond single sites to influence landscape-scale marsh morphology and dynamics across the estuary (Lanzoni and Seminara, 2002; ). For example, Pringle (1995) reported the near total loss of ca. 200 ha of salt marsh over a period of 20 years following the migration of a tidal channel. On the opposite side of the estuary, tidal flats rapidly became vegetated to form a new, extensive marsh (). Such a phenomenon, where channel shifts cause erosion in one part of the estuary and, at the same time, marsh expansion in another, may be commonplace. Indeed, erosion-expansion compensation patterns have been reported at a localized scale along interior marsh creek banks (). Estuarine-scale channel-marsh dynamics would represent a hitherto unreported landscape-scale resilience mechanism, whereby net marsh area cover stays the same despite local scale erosion (Figure 1).
FIGURE 1
Here, we provide the first estuarine-scale, multi-marsh study of erosion and expansion in response to tidal channel migration. We use a timeseries of aerial photographs from 1946 to 2013 for twelve salt marshes in three estuaries in Wales, to examine coupled changes in salt marsh lateral extent and tidal channel position. We hypothesize that: (1) temporal variation in the position of the seaward edges of saltmarshes correlate with shifts in the position of tidal channels, and that; (2) channel migration couples marsh erosion on one estuarine bank with marsh expansion on the opposite bank, in a compensatory fashion that results in no net change in marsh cover at the estuarine scale (Figure 1). We also investigate whether patterns of channel-marsh change have been influenced by changes in wind-wave or river flow, which are the main drivers of channel migration.
Materials and Methods
Study Estuaries
Salt marsh extent and tidal channel migration patterns were studied in three estuaries: the Glaslyn-Dwyryd, the Mawddach and the Dyfi, located in Cardigan Bay, Wales, United Kingdom (Figure 2). All three estuaries were drowned river and glacial valleys, flooded during the later stages of Holocene transgression (Larcombe and Jago, 1996). During the 1800s, estuary sizes and tidal prisms were reduced by railway embankment and seawall construction on river flood plains (Rhind and Jones, 1994). Import of non-cohesive, medium grained terrigenous sediments via flood-dominant tides has subsequently reduced the estuary accommodation space (), lowering the average estuary depth to between 4 and 10 m (Manning and Whitehouse, 2012). Each estuary is macrotidal with extreme neap and spring tidal ranges of 0.8 and 5.0 m, respectively. Strong tidal flows along the length of the estuaries produce current velocities of 1 m s–1 during flood tides that produce tidal flows of around 280–1,400 m3 s–1 during neap and spring tides, respectively (Larcombe and Jago, 1996). Tidal flows are considerably stronger than mean freshwater flows, which are typically around 20 m3 s–1 (Shi and Lamb, 1991). The estuaries are sheltered from north-westerly wind-generated waves by sand-shingle spits at the estuary mouths, protecting the well-developed intertidal flats, salt marshes and tidal channels (Pye and Blott, 2014).
FIGURE 2
Processing Aerial Photographs
To measure change in salt marsh extent and tidal channel position over time, we used a collection of aerial photographs taken between 1946 and 2013. Only image sets that covered the entirety of each estuary were used with the exception of the image set from 1971–72 for the Dyfi estuary. This image set had two missing images, that represented 4% of the estuary coverage. Missing areas were excluded from further trend analyses (Supplementary Table 1). We scanned aerial photographs at a resolution of ca. 400 dpi and georeferenced them in a Geographic Information System (GIS) to British National Grid, using a Thin Plate Spline algorithm with Lancsoz resampling suited to coarse resolution historical image rectification. A combination of Ordnance Survey 1:25,000 raster maps (EDINA) and Environment Agency 1:7,500 orthorectified images from 2006 to 2009 (Phelan et al., 2011) provided basemaps to match the unrectified photographs to well-distributed control points including crossroads, building corners and salt marsh creeks. Resulting georeferenced images had pixel sizes corresponding to ca. 0.25 × 0.25 m in the field.
We manually delineated marshes and channels for the entirety of each estuary at a scale of 1:7,500. To aid in distinguishing between features, aerial image contrast was set to 50% to exaggerate “dark” vegetated surfaces and wet channels from “light” bare tidal flats. Polygon vertices were placed every 10 meters along the margins of salt marshes and tidal channels. Marsh patches less than 5 m in diameter, creeks narrower than 10 m and unvegetated features in the interior of salt marshes, such as salt pans, were ignored. In most cases, salt marshes were bounded by embankments that allowed for simple identification of the landward marsh edge. An error term was calculated for each measure of marsh areal extent using the method outlined in Wernette et al. (2017). The channel edge was manually identified as the transition between light (dry) and dark (wet) pixels in each image. Condition of the tide when images were captured was not known, therefore a standard channel edge between years was difficult to justify. However, potential error of several meters would be offset by variation by tens of meters in channel position over time, so this method was considered acceptable.
Quantifying Geomorphic Change of Marshes and Channels
We used the GIS layers of marsh extent to calculate marsh area change over time. Next, we created four GIS layers per estuary to interpret temporal marsh and channel change between 1946 and 2013. Marsh and channel extents in each observation year (see Supplementary Table 1) were stacked in a GIS to classify extents in four ways: (1) a “saltmarsh geomorphic state” layer depicted where marshes had consistently eroded or expanded, fluctuated between erosion and expansion phases, or remained unchanged. Areas where the marsh had consistently occurred were classified as “stable.” Areas where salt marsh gain occurred were classified as “expanded.” Areas which had been vegetated and became bare were classified as “eroded.” Areas that had fluctuated between marsh being present, absent, and present again (or absent, present, absent) were classified as “dynamic.” (2) A “saltmarsh age” layer represented the number of years from a 2020 baseline back to when the marsh first appeared. Areas where marshes consistently occurred were classified as being older than the earliest image year. Since the exact year when a marsh appears will lie sometime between two image sets, an age range was given (between the older and newer of the compared aerial photographs). (3) A “saltmarsh-mudflat transitions” layer showed the number of times an area had fluctuated from saltmarsh to mudflat or mudflat to saltmarsh. (4) A “channel locational probability” layer represented temporal change in tidal channel movement across each estuary. We used the locational probability technique outlined by to do this. This technique shows the probability that a channel will be found at a given point along the estuary, based on where the channel was previously observed. For example, if a channel was found in the same area in every photograph year, that area received a locational probability score of 100%. An additional set of maps were produced showing changes in marsh and channel position between successive image years.
To assess the relationship between marsh and channel movements, in a GIS we placed transects (perpendicular to estuary centerlines) every 200 m along the length of each estuary (Supplementary Figure 1) and measured the distances from the marsh land-boundary (back of the marsh, henceforth referred to as “land”) to (i) the seaward marsh edge and (ii) the nearest tidal channel for each timestamp. We also recorded the estuary width across each transect, distance from the top of the estuary where each transect was located, and which year an observation of marsh and channel location was made, to determine whether position in the estuary and observation year had a confounding effect on channel-marsh dynamics. Only transects where marshes occurred at least once during the survey period were used in this analysis. All distance measurements were expressed as percentage values of either estuary width (from one estuary bank to the other along a transect) or length (from the top to the bottom of an estuary), allowing for comparisons between the three estuaries.
River and Wind Influence on Channel Shifts
We collated daily average river flow, annual maximum river flow and annual flood event frequency data taken from a river gauge (float with counterweight) near the Dyfi estuary (ID: 64001; coordinates: 52°36′3″′N 3°51′17″W). Data was provided by the National River Flow Archive1 and was quality controlled before being made available for download. No post-processing of the data was therefore required (Robson and Reed, 1999). Daily wind speed and direction at Aberporth weather station (ID: 1198; coordinates: 1198, 52°8′21″N, 4°34′12″W) were taken from the Met Office Integrated Data Archive System (MIDAS) distributed by the British Atmospheric Data Centre2. We selected stations based on proximity to estuaries and availability of long temporal dataset ranges. Prior to analysis, wind speed data was screened for quality and completeness using methods outlined in Watson et al. (2015). MIDAS assign a “flag” to measurements that have an unreliable observation, and these were excluded from further analysis, as were duplicates. Any given year was only used in later calculations when the dataset held observations from ≥ 75% of days per year, and when each month had observations for ≥ 50% of days.
Statistical Analyses
We used a mixed effects model to determine the effect on the response variable “distance from the land to the seaward marsh edge” from four explanatory variables (1) “distance from the land to the nearest tidal channel,” (2) “estuary width,” (3) “distance from the estuary top,” and (4) “year.” Multicollinearity between predictor variables was established from Variance Inflation Factor (VIF) tables, whereby variables with VIF scores of > 3 were excluded (Zuur et al., 2010). Estuary width and distance from the estuary top were highly correlated (r = 0.75) so estuary width was subsequently dropped from further analysis. Since measures of distance from land to both the marsh edge and the nearest channel were made for the same transect over time, and since the use of all three estuaries in the analysis could introduce spatial autocorrelation, “year” and “estuary” were included as random factors in the maximal mixed effects model. Restricted maximum likelihood (REML) estimation and AIC scores determined that the inclusion of “estuary” significantly improved the model fit (Zuur et al., 2009), however, the inclusion of “year” did not and was subsequently dropped in favor of the most parsimonious model. The minimal adequate model was selected using likelihood ratio tests and AIC scores. All statistical analyses were carried out in the R software package. All variables were verified for homoscedasticity and normality and log transformations applied where necessary. The final model was checked for homoscedasticity using the “cftest” function of the “multcomp” package ().
We also used a time series analysis to determine whether the response variables “daily average river flow,” “annual maximum river flow,” and “annual flood event frequency” (between 1963 and 2013), as well as “daily average wind speed,” and “daily average wind direction” (between 1957 and 2017), had changed significantly over time (). Change over time was first detrended from seasonal variability using the “stl” function; a locally weighted regression procedure (). Mixed Effects models were then used to determine whether applying both “seasonal” and “long term” trend fits (model 1) better explained change in each of the response variables over time than when a “seasonal” fit alone was considered (model 2). If model 1 was significantly different (assessed by Chi2) and had a lower AIC value than model 2, this would indicate a long-term trend existed in the data. Model fixed effects were the “seasonal” and “long term” trend fits. The REML approach was used on each Mixed Effects model for a fair selection of the parsimonious model (Zuur et al., 2009). “Year” was designated as a random effect. We fitted linear regression models (continuous data) and generalized linear models with Poisson error distribution (count data) to determine whether response variables changed over time. See Supplementary Material 1 for further details on the full statistical analysis used.
Results
Channel-Marsh Evolution in Estuaries
Tidal channels and salt marshes changed considerably over the 1946–2013 period (Figure 3). Tidal channels showed progressive migration and course switching in all parts of each estuary (Figure 3; greyscale shading), except at the tops of the Mawddach (Figures 3A–F), and Dyfi (Figures 3G–I) estuaries where channel position was largely fixed. Channel shifts were frequently followed by both marsh erosion and expansion. Two marshes in the Glaslyn-Dwyryd and Mawddach (GC and GI) emerged once tidal channels moved away from the shore, allowing 66 ha of marsh to develop between 1946 and 2013 (Figure 4 and Supplementary Figures 2, 3). In contrast, channel migration toward the south-western GT marsh in the Glaslyn-Dwyryd estuary eroded the marsh by as much as 330 m between 1946 and 2013 (Supplementary Figure 2). The relative proportions represented by each category of salt marsh geomorphic behavior (stable, expanding, eroding or dynamic), age and number of state changes between marsh and mudflat were similar between estuaries. Approximately a third of marsh area per estuary remained stable over the 67 observation years (Figure 3; brown area in left panels, dark red in middle panels, light blue in right panels) and were concentrated in the inner estuary or along the rear of wide marshes (i.e., at higher elevations) in the outer estuary. The rest either eroded, expanded or were dynamic (both erosion and expansion).
FIGURE 3
FIGURE 4
Overall, there was a greater tendency for marsh expansion (35–45% in Figure 3; left panel, yellow areas) than erosion (1–10% in Figure 3; left panel, red). Expansion was mostly in the outer estuary or along the margins of stable marshes of the inner estuary (Figure 3; left panel, yellow). The number of transitions between marsh and mudflat states increased the further marshes developed out from the coast and into the estuarine basin (Figures 3C,F,I; right panel). These high-transition areas, also represented as “dynamic” areas by the blue color in left panels of Figure 3, were especially pronounced in the Dyfi estuary, representing 26% of all geomorphic states. Overall, marsh area per estuary expanded by 120, 235, and 162% for the Glaslyn-Dwyryd, Mawddach, and Dyfi estuaries, respectively (Figures 4A,C,E). Dyfi estuary experienced the fastest rise in marsh extent to a peak of 565.7 ha in 1982, followed by a moderate decline (Figure 4E). Marsh expansion was concentrated in the outer estuary and along all mid estuary marshes (except “GT”: Figure 4B) whilst inner estuary marshes remained relatively stable (Figures 4B,D,F).
All three estuaries had local marsh area loss compensated for by marsh expansion elsewhere in the estuary. We recognized three geomorphic types of compensatory coupling of expansion with erosion (Figure 5). Figure 5 and Supplementary Figures 2–4):
FIGURE 5
- 1.
“Opposite bank compensation”: marsh erosion on one estuary bank (red) coincided with marsh expansion on the opposite bank (yellow). This type was often cyclical (Figure 5A; color pattern switching banks). From 1946 to 1971 in the Glaslyn-Dwyryd estuary, the northern bank PBr marsh expanded by 4.5 m yr–1 whilst the southern bank GT marsh eroded by 1.5 m yr–1; this trend reversed in 1971–2011, with PBr eroding by −3 m yr–1 and GT expanding by 3 m yr–1 (Figure 5A and Supplementary Figures 2A–C). Similarly, in the Mawddach, the northern DW marsh eroded by 1.5 m yr–1 between 1969 and 2013, whilst the southern GI marsh expanded 6 m yr–1 to compensate (Supplementary Figures 3A–C). Between 1948 and 2013 in the Mawddach, the northern PBo marsh expanded by 1 m yr–1, whilst the southern PP marsh eroded by 1 m yr–1 (Supplementary Figures 3A–C).
- 2.
“Same bank compensation”: erosion and expansion coinciding in different marshes on the same bank (Figure 5A). In the Glaslyn-Dwyryd, erosion by −3 m yr–1 of the GT marsh mirrored a 2.5 m yr–1 expansion of the TB marsh between 1946 and 2011 (Figure 5B and Supplementary Figures 2A–C). In the Dyfi estuary, TM eroded as YG expanded between 1961 and 1971 (Supplementary Figure 4B). Between 1971 and 1982, YH eroded as YG and TM expanded (Supplementary Figure 4C). Between 2000 and 2009, marshes between TM and YG emerged as marshes along the rest of estuary eroded (Supplementary Figure 4F).
- 3.
“Same marsh compensation”: erosion and expansion coinciding in a single marsh. Between 1948 and 2013 in the Mawddach, the southern PP marsh eroded by 1 m yr–1 whilst the eastern portion expanded at a similar rate (Figure 5C and Supplementary Figures 3A–C). In the Dyfi, erosion of western YG was mirrored by expansion on the eastern side of the marsh between 1982 and 1990 (Supplementary Figure 4D). The trend continued between 1990 and 2000, as the western TM erosion was also compensated by expansion along the eastern flank during this time (Supplementary Figure 4E).
Figure 6 shows the percentage of the estuary occupied by salt marsh and tidal channel over the study period, from the top to the bottom of each estuary (left to right in x-axis). The portion of estuary width occupied by marsh decreased from 75 to 95% at the top of estuaries to < 20% at the bottom of estuaries (Figure 6; green line). Marshes tended to be more stable in the top than the bottom of estuaries (Figure 6; smaller error bars on left than right on green line). Channels occupied 15–30% of the estuary width, except near the bottom of Mawddach and Dyfi estuaries where the proportions reached 80 and 65%, respectively (Figure 6; blue line). Temporal variation in channel position (Figure 6; error bars on blue line) was high along the length of each estuary and up to 75% of estuary width, except near the estuary tops, where variation dropped as the distance between channel and marsh reduced.
FIGURE 6
Table 1 and Figure 7 summarize the results of the Mixed Effects model. The further away the channel, the wider was a marsh (Table 1 and Figure 7A). However, the closer marshes got to the outer estuary, the smaller the portion of estuarine width they occupied (Figure 7B) and therefore the association between marsh width and channel proximity weakened toward the outer estuary (hence the interaction term in Table 1). The proportion of estuarine width occupied by marshes showed a slight, but significant increase from 1946 to 2013 (Figure 7C).
TABLE 1
| Predictor variable | Estimate | SE | z-value | P-value |
| Best model fit: AIC = 1354.8 | ||||
| Channel × Estuary | –0.317 | 0.025 | –12.756 | <0.001 |
| Channel | 0.442 | 0.024 | 18.666 | <0.001 |
| Estuary | –0.447 | 0.023 | –19.390 | <0.001 |
| Year | 0.084 | 0.023 | 3.6310 | <0.001 |
Predictor variables of salt marsh width selected from a minimal adequate model (Mixed Effects model with “estuary” as a random intercept).
“Channel” is the distance of channel from the land, and “Estuary” is the distance along the estuary. Both values have been arcsine transformed. “×” indicates an interaction term.
FIGURE 7
Drivers of Marsh and Channel Dynamics
Figure 8 shows how wind parameters at a nearby weather station and river flow at the top of the Dyfi estuary varied over the study period. The wind and river parameters had high seasonal variability (Figures 8A–C; gray lines), however indications of any pattern of change (increase/decrease) over the 77-year observation period was either weak or absent: Change in annual average wind direction was weakly significant (Figure 8B), shifting toward a more westerly direction by 0.14 degrees a year between 1957 and 2017 [linear model: F(1, 59) = 5.413, p = 0.023, Adjusted R2 = 0.069]. The number of annual flood events (Figure 8D) showed a significant, but small, increase of 0.01 per year between 1962 and 2013 (Poisson model: AIC = 223.63, p = 0.043). No trend in annual average wind speed or daily average river flow could be detected (Figures 8A,C).
FIGURE 8
Discussion
We found estuarine saltmarshes exhibited large-scale resilience to erosion, despite individual marshes within estuarine complexes periodically eroding. Migrating channels drove phases of marsh erosion and expansion which were often synchronised at the estuary scale: as part of a marsh eroded, another part expanded elsewhere in the estuary over the same period. Such a “geomorphic compensation” phenomenon was evident between marshes on (i) opposite estuary banks, (ii) between singular marshes on the same bank in the mid and inner estuary, and (iii) along the margins of the same marsh. This meant change in the total area of all marshes in the mid and inner estuaries was low, despite up to 260 m of lateral change during erosion and expansion phases in some cases. The phenomenon of geomorphic compensation is likely commonplace in other biogeomorphic systems. Compensation after erosion has been described along mangrove forests in deltas (), coastal seagrass meadows (León-Pérez et al., 2020), and vegetated islands in braided rivers (). The driver of geomorphic compensation need not necessarily be restricted to channel dynamics. Patterns of compensatory marsh establishment have been identified in response to sea level rise (), storm surges (), and dredging activity (Spearman et al., 2014).
Capacity for erosion to be compensated by expansion elsewhere in the estuary has important implications for the large-scale resilience of tidal landscapes. The present paradigm of ecosystem resilience often emphasizes interactions at the scale of plants and the immediate surrounding environment: resilience is gained over the short-term through a positive feedback between plant growth and sediment trapping, that entails a loss of resilience over the longer-term as elevation gain exposes the vegetated edge to erosion in a negative feedback (; Stallins and Corenblit, 2018; ). Unless sediment released during erosion encourages plant establishment in front of eroding edges in a cyclical manner (van der Wal et al., 2008), the removal of sediment (as is the case when a tidal channel shifts) is perceived as a loss of resilience; the high-elevation and “vegetated” state has undergone a regime shift to an alternative low-elevation and “unvegetated” stable state (Marani et al., 2010; Wang and Temmerman, 2013). However, when a larger-scale perspective is taken, we show that sediment released from one regime shift (“vegetated” to “unvegetated”) can drive a regime shift in the other direction (“unvegetated” to “vegetated”) elsewhere in the estuary, resulting a persistence of net habitat extent.
In the same way that marsh edges can cycle between erosion and expansion phases (van der Wal et al., 2008), this study confirms the existence of a similar self-organizing pattern between multiple marshes operating at the estuarine scale. The importance of long-distance interactions, in which ecological processes between ecosystems kilometers apart bolster resilience, is beginning to be recognized (van de Koppel et al., 2015). Long-distance interactions between landscape units, such as the geomorphic compensation phenomenon described here, may be equally important in governing the resilience of biogeomorphic systems. Detecting large-scale resilience requires clearly defined boundaries in which compensatory erosion-expansion patterns occur and can be assessed. Such a requirement is relatively simple in estuaries, where marsh extent is confined between the fully marine and fully fluvial ends of an estuary. For other settings, such as open-coast environments, defining appropriate boundaries may be challenging and the mechanistic drivers for compensation more complex to identify. Since tidal landscape habitat types are ultimately driven by sediment flux (), identifying littoral “sediment cells” (i.e., the transport pathways and boundaries of sediment flux) could provide a way to define the template in which geomorphic compensation can be detected ().
Patterns of erosion-expansion compensation described here occurred at the same time as a net increase in marsh extent at estuarine scales. The expansion mainly occurred in the outer estuaries, reflecting a tendency for marsh expansion across western Britain caused by positive sediment supply (Ladd et al., 2019a). The present phase of sediment import may be part of a longer-term cycle of sediment import and export that occur in estuaries over centennial to millennial timescales (; Townend et al., 2007). Both geomorphic compensation and net marsh gain patterns therefore appear to be operating under a set of discrete spatio-temporal scales and drivers, as a feature of coastal morphodynamics (). At the finer-scale, channel migration drives marsh erosion and expansion patterns. At the macro-scale, marsh change is dictated by sediment import/export phases. The hierarchical nesting of discrete erosion-expansion patterns play an important role in how tidal landscapes evolve (). Identifying the mechanisms that drive resilience at each scale can help better predict response of biogeomorphic landscapes to forcing (Stallins and Corenblit, 2018).
Except for proportionally smaller marshes (where marshes had yet to become wide enough to be exposed to channel migration), and where bedrock outcrops and headlands prevented channel migration (), channel migration drove compensation patterns throughout each estuary. We therefore demonstrate the often-overlooked importance of migrating tidal channels as long-term regulators of lateral marsh dynamics. Channel shifts are a common morphodynamic feature of estuaries, lagoons, and deltas worldwide (). The role of channel shifts as a driver of coastal ecosystem dynamics is likely to be of greater importance than previously thought. High interannual variability of river and wind wave forcing likely drove the shifting of tidal channels in this study. Sea level rise, changes to sediment supply and changes in tidal prism by estuarine construction work all risk changing intertidal profiles, and hence the configuration and migration patterns of tidal channels (Olabarrieta et al., 2018; ; Ladd et al., 2019a). Interrupting channel-habitat interaction could negatively impact any compensatory expansion patterns and undermine landscape scale ecosystem resilience.
Geomorphic compensation has important practical implications for how ecosystem services at landscape scales are valued and managed. Certain ecosystem services are likely to be unaffected by erosion-expansion dynamics if net marsh extent remains consistent. For example, marshes provide a habitat provision service for nesting birds and insects. Vegetation patches of varying height and density are required for wader birds’ selection preference of nest sites to be met (Sharps, 2015). Bees rely on saltmarsh flower meadows for foraging (). If a marsh undergoes erosion at one location whilst a marsh elsewhere in the estuary expands, the habitat provision service would remain resilient, since structurally diverse vegetation and geomorphic profiles are maintained at the estuarine scale. Estuarine marshes also effectively attenuate storm surges and the upstream attenuation effect is determined by marsh extent, not specific location in the estuary (). The storm surge capacity of marshes is therefore unlikely to be affected by compensatory erosion-expansion cycles. Other ecosystem services will likely be undermined by geomorphic compensation. Marsh loss would negatively impact on localized wave attenuation, as wave dissipation is strongly correlated with marsh width (). Organic carbon stored in marsh deposits would be released during erosion, and it remains unclear whether the carbon could be reburied during marsh expansion or whether a portion of the carbon would be reintroduced into the carbon cycle (Mueller et al., 2019). Understanding how changes in the biological components of marshes affects ecosystem service delivery remains an important research gap (). Additional work should also consider the implications of large-scale resilience mechanisms in the long-term provision of ecosystem services.
Publisher’s Note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
Statements
Data availability statement
The dataset on saltmarsh extent change can be accessed at Ladd et al. (2019b), and the remaining datasets supporting our conclusions are given in Supplementary Data 1.
Author contributions
CL and MD-E digitized the aerial images. CL processed the data. CL and JP completed the statistical analyses. CL prepared the initial manuscript. All authors contributed to the editing of the final manuscript.
Funding
This work was supported by the Coleg Cymraeg Cenedlaethol (Ysgoloriaethau Ymchwil), the Welsh Government and Higher Education Funding Council for Wales (S r Cymru National Research Network for Low Carbon, Energy and Environment), the Natural Environment Research Council (C-SIDE Project) (Grant No. NE/R010846/1), and the United Kingdom Research and Innovation Global Challenges Research Fund (Living Deltas Research Hub) (Grant No. NE/S008926/1).
Acknowledgments
We acknowledge the Royal Commission on the Ancient and Historical Monuments of Wales and the Central Register of Air Photography for Wales as the source of aerial images. We also thank Prof. Tom Spencer from Cambridge University for valuable feedback on the initial manuscript.
Conflict of interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Supplementary material
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmars.2021.757715/full#supplementary-material
References
1
BalkeT.HermanP. M.BoumaT. J. (2014). Critical transitions in disturbance-driven ecosystems: identifying windows of opportunity for recovery.J. Ecol.102700–708. 10.1111/1365-2745.12241
2
BoumaT. J.van BelzenJ.BalkeT.van DalenJ.KlaassenP.HartogA. M.et al (2016). Short-term mudflat dynamics drive long-term cyclic salt marsh dynamics.Limnol. Oceanogr.612261–2275. 10.1002/lno.10374
3
BraswellA. E.HeffernanJ. B. (2019). Coastal wetland distributions: delineating domains of macroscale drivers and local feedbacks.Ecosystems221256–1270. 10.1007/s10021-018-0332-3
4
BraudrickC. A.DietrichW. E.LeverichG. T.SklarL. S. (2009). Experimental evidence for the conditions necessary to sustain meandering in coarse-bedded rivers.Proc. Natl. Acad. Sci. U.S.A.10616936–16941. 10.1073/pnas.0909417106
5
BrooksH.MöllerI.CarrS.ChirolC.ChristieE.EvansB.et al (2020). Resistance of salt marsh substrates to near-instantaneous hydrodynamic forcing.Earth Surf. Proc. Land4667–88. 10.1002/esp.4912
6
BrownJ.DaviesA. (2010). Flood/ebb tidal asymmetry in a shallow sandy estuary and the impact on net sand transport.Geomorphology114431–439. 10.1016/j.geomorph.2009.08.006
7
BrücknerM. Z. M.SchwarzC.van DijkW. M.van OorschotM.DoumaH.KleinhansM. G. (2019). Salt marsh establishment and eco-engineering effects in dynamic estuaries determined by sepceis growth and mortality.J. Geo. Res. Earth Surf.1242962–2986. 10.1029/2019JF005092
8
BurninghamH. (2008). Contrasting geomorphic response to structural control: the loughros estuaries, northwest Ireland.Geomorphology97300–320. 10.1016/j.geomorph.2007.08.009
9
BurnsC. J.AlexanderC. R.AlberM. (2021). Assessing long-term trends in lateral salt-marsh shoreline change along a U.S. east coast latitudinal gradient.J. Coast. Res.37291–301. 10.2112/JCOASTRES-D-19-00043.1
10
CampbellA.WangY. (2019). High spatial resolution remote sensing for salt marsh mapping and change analysis at fire island nationla seashore.Remote Sens.11:1107. 10.3390/rs11091107
11
CarrascoR. A.KombiadouK.AmadoM.MatiasA. (2021). Past and future marsh adaptation: lessons learned from the Ria Formosa lagoon.Sci. Tot. Env.790:148082. 10.1016/j.scitotenv.2021.148082
12
ChauhanP. P. S. (2009). Autocyclic erosion in tidal marshes.Geomorphology11045–57. 10.1016/j.geomorph.2009.03.016
13
ClevelandR. B.ClevelandW. S.TerpenningI. (1990). STL: a seasonal-trend decomposition procedure based on loess.J. Off. Stat.63–73.
14
CocoG.ZhouZ.van MaanenB.OlabarrietaM.TinocoR.TownendI. (2013). Morphodynamics of tidal networks: advances and challenges.Mar. Geol.3461–16. 10.1016/j.margeo.2013.08.005
15
CoxR.WadsworthR.ThomsonA. (2003). Long-term changes in salt marsh extent affected by channel deepening in a modified estuary.Cont. Shelf Res.231833–1846. 10.1016/j.csr.2003.08.002
16
CrawleyM. J. (2012). The R Book.Chichester: Wiley.
17
DavidsonK. E.FowlerM. S.SkovM. W.FormanD.AlisonJ.BothamM.et al (2020). Grazing reduces bee abundance and diversity in saltmarshes by suppressing flowering of key plant species.Agric. Ecosyst. Environ.291:106760. 10.1016/j.agee.2019.106760
18
DeeganL. A.JohnsonD. S.WarrenR. S.PetersonB. J.FleegerJ. W.FagherazziS.et al (2012). Coastal eutrophication as a driver of salt marsh loss.Nature490388–392. 10.1038/nature11533
19
DronkersJ. J. (2005). Dynamics of Coastal Systems.Singapore: World Scientific. 10.1142/9818
20
ElmiladyH.van der WegenM.RoelvinkD.JaffeB. (2019). Intertidal area disappears under sea level rise: 250 years of morphodynamic modeling in San Pablo Bay, California.J. Geophys. Res. Earth Surf.12438–59. 10.1029/2018JF004857
21
FagherazziS.KirwanM. L.MuddS. M.GuntenspergenG. R.TemmermanS.D’AlpaosA.et al (2012). Numerical models of salt marsh evolution: ecological, geomorphic, and climatic factors.Rev. Geophys.501–28. 10.1029/2011RG000359
22
FairchildT. P.BennettW. G.SmithG. S.DayB.SkovM. W.MöllerI.et al (2021). Coastal wetlands mitigate storm flooding and associated costs in estuaries.Environ. Res. Lett.16:074034. 10.1088/1748-9326/ac0c45
23
FrancisR. A.CorenblitD.EdwardsP. J. (2009). Perspectives on biogeomorphology, ecosystem engineering and self-organisation in island-braided fluvial ecosystems.Aquat. Sci.71290–304. 10.1007/s00027-009-9182-6
24
FrenchJ.PayoA.MurrayB.OrfordJ.EliotM.CowellP. (2016). Appropriate complexity for the prediction of coastal and estuarine geomorphic behaviour at decadal to centennial scales.Geomorphology2563–16.
25
GablerC. A.OslandM. J.GraceJ. B.StaggC. L.DayR. H.HartleyS. B.et al (2017). Macroclimatic change expected to transform coastal wetland ecosystems this century.Nat. Clim. Chang.7142–147. 10.1038/NCLIMATE3203
26
GanjuN. K.DefneZ.KirwanM. L.FagherazziS.D’AlpaosA.CarnielloL. (2017). Spatially integrative metrics reveal hidden vulnerability of microtidal salt marshes.Nat. Commun.8:14156. 10.1038/ncomms14156
27
GhoshM. K.KumarL.LangatP. K. (2018). Mapping tidal channel dynamics in the Sundarbans, Bangladesh, between 1974 and 2017, and implications for the sustainability of the Sundarbans mangrove forest.Environ. Monit. Assess.190:582. 10.1007/s10661-018-6944-4
28
GrafW. L. (2000). Locational probability for a dammed, urbanizing stream: salt river Arizona, USA.Environ. Manage.25321–335. 10.1007/s002679910025
29
GrayA. (1972). The ecology of Morecambe Bay. V. the salt marshes of Morecambe Bay.J. Appl. Ecol.9207–220. 10.2307/2402057
30
HothornT.BretzF.WestfallP.HeibergerR. M.SchuetzenmeisterA.ScheibeS. (2016). Simultaneous inference in general parametric models.Biom. J.50346–363. 10.1002/bimj.200810425
31
HughesZ. J. (2012). “Tidal channels on tidal flats and marshes,” in Principles of Tidal Sedimentology, edsDavisR. A.DalrympleR. W. (New York, NY: Springer), 269–300. 10.1007/978-94-007-0123-6_11
32
KirwanM. L.GedanK. B. (2019). Sea-level driven land conversion and the formation of ghost forests.Nat. Clim. Chang.9450–457. 10.1038/s41558-019-0488-7
33
KirwanM. L.MurrayA. B.DonnellyJ. P.CorbettD. R. (2011). Rapid wetland expansion during European settlement and its implication for marsh survival under modern sediment delivery rates.Geology39507–510. 10.1130/G31789.1
34
KochE. W.BarbierE. B.SillimanB. R.ReedD. J.PerilloG. M. E.HackerS. D.et al (2009). Non-linearity in ecosystem services: temporal and spatial variability in coastal protection.Front. Ecol. Environ.7:29–37. 10.1890/080126
35
LaddC. J. T. (2021). Review of processes and management of saltmarshes across Great Britain.Proc. Geo. Assoc.132269–283. 10.1016/j.pgeola.2021.02.005
36
LaddC. J. T.Duggan-EdwardsM. F.BoumaT. J.PagèsJ. F.SkovM. W. (2019a). Sediment supply explains long-term and large-scale patterns in salt marsh lateral expansion and erosion.Geophys. Res. Lett.4611178–11187. 10.1029/2019GL083315
37
LaddC. J. TDuggan-EdwardsM. F.BoumaT. J.PagèsJ. F.SkovM. W. (2019b). Data From: Change in Saltmarsh Extent for Six Regions Across Great Britain (1846–2016). NERC Environmental Information Data Centre.10.5285/03b62fd0-41e2-4355-9a06-1697117f0717
38
LanzoniS.SeminaraG. (2002). Long-term evolution and morphodynamic equilibrium of tidal channels.J. Geophys. Res. Oceans1071–13. 10.1029/2000JC000468
39
LarcombeP.JagoC. F. (1996). The morphological dynamics of intertidal megaripples in the Mawddach Estuary, north Wales, and the implications for palaeoflow reconstructions.Sedimentology43541–559. 10.1046/j.1365-3091.1996.d01-16.x
40
LeonardiN.DefneZ.GanjuN. K.FagherazziS. (2016). Salt marsh erosion rates and boundary features in a shallow bay.J. Geophys. Res. Earth Surf.1211861–1875. 10.1002/2016JF003975
41
León-PérezM. C.ArmstrongR. A.HernándezW. J.Aguilar-PereraA.Thompson-GrimJ. (2020). Seagrass cover expansion off Gaja de Muertos Island, Puerto Rico, as determined by long-term analysis of historical aerial images (1950-2014).Ecol. Indic.117:106561. 10.1016/j.ecolind.2020.106561
42
LiC.ChenC.GuadagnoliD.GeorgiouI. Y. (2008). Geometry-induced residual eddies in estuaries with curved channels: observations and modeling studies.J. Geophys. Res. Oceans113:C01005. 10.1029/2006JC004031
43
ManningA.WhitehouseR. (2012). Enhanced UK Estuaries Database: Explanatory Notes and Metadata. Development of Estuary Morphological Models. Technical Report. Wallingford: HR Wallingford.
44
MaraniM.D’AlpaosA.LanzoniS.CarnielloL.RinaldoA. (2010). The importance of being coupled: stable states and catastrophic shifts in tidal biomorphodynamics.J. Geophys. Res. Earth Surf.1151–15. 10.1029/2009JF001600
45
MöllerI.KudellaM.RupprechtF.SpencerT.PaulM.van WesenbeeckB. K.et al (2014). Wave attenuation over coastal salt marshes under storm surge conditions.Nat. Geosci.7727–731. 10.1038/NGEO2251
46
MuellerP.LadigesN.JackA.SchmiedlG.KutzbachL.JensenK.et al (2019). Assessing the long-term carbon-sequestration potential of the semi-natural salt marshes in the European Wadden Sea.Ecosphere10:e02556. 10.1002/ecs2.2556
47
NichollsR. J.WongP. P.BurkettV.CodignottoJ.HayJ.McLeanR.et al (2007). “Coastal systems and low-lying areas,” in Impacts, Adaptation, and Vulnerability, edsParryM. L.CanzianiO. F.PalutikofJ. P.van der LindenP. J.HansonC. E. (Cambridge: Cambridge Univ. Press), 315–357.
48
OlabarrietaM.GeyerW. R.CocoG.FriedrichsC. T.CaoZ. (2018). Effects of density-driven flows on the long-term morphodynamic evolution of funnel-shaped estuaries.J. Geophys. Res. Earth Surf.1232901–2924. 10.1029/2017JF004527
49
PhelanN.ShawA.BaylisA. (2011). The Extent of Saltmarsh in England and Wales: 2006-2009.Bristol: Environment Agency.
50
PringleA. (1995). Erosion of a cyclic salt marsh in Morecambe Bay, north-west England.Earth Surf. Process Landf.20387–405. 10.1002/esp.3290200502
51
PyeK.BlottS. J. (2014). The geomorphology of UK estuaries: the role of geological controls, antecedent conditions and human activities.Estuar. Coast. Shelf Sci.150196–214. 10.1016/j.ecss.2014.05.014
52
ReedD.van WesenbeeckB.HermanP. M.MeselheE. (2018). Tidal flat-wetland systems as flood defenses: understanding biogeomorphic controls.Estuar. Coast. Shelf Sci.213269–282. 10.1016/j.ecss.2018.08.017
53
RhindP.JonesA. (1994). Brackish salt marsh communities in the glaslyn marsh trust reserve.Field Studies8373–384.
54
RobsonA.ReedD. (1999). Flood Estimation Handbook.Wallingford: Institute of Hydrology.
55
RogersK.KellewayJ. J.SaintilanN.MegonigalJ. P.AdamsJ. B.HolmquistJ. R.et al (2019). Wetland carbon storage controlled by millennial-scale variation in relative sea-level rise.Nature56791–95. 10.1038/s41586-019-0951-7
56
SharpsE. A. (2015). The Effects of Salt Marsh Conservation Grazing on Breeding Common Redshank Tringa Totanus. Ph.D. Dissertation. Bangor: Bangor University.
57
ShiZ.LambH. (1991). Post-glacial sedimentary evolution of a microtidal estuary, Dyfi Estuary, West Wales, UK.Sediment Geol.73227–246. 10.1016/0037-0738(91)90086-S
58
SillimanB. R.Van de KoppelJ.McCoyM. W.DillerJ.KasoziG. N.EarlK.et al (2012). Degradation and resilience in louisiana salt marshes after the BP–deepwater horizon oil spill.Proc. Natl. Acad. Sci. U.S.A.10911234–11239. 10.1073/pnas.1204922109
59
SpearmanJ.BaughJ.FeatesN.DearnaleyM.EcclesD. (2014). Small estuary, big port – progress in the management of the stour-orwell estuary system.Estuar. Coast. Shelf Sci.150299–311. 10.1016/j.ecss.2014.07.003
60
StallinsJ. A.CorenblitD. (2018). Interdependence of geomorphic and ecologic resilience properties in a geographic context.Geomorphology30576–93. 10.1016/j.geomorph.2017.09.012
61
SylvesterZ.DurkinP.CovaultJ. A. (2019). High curvatures drive river meandering.Geology47263–266. 10.1130/G45608.1
62
TownendI. H.WangZ. B.ReesJ. (2007). Millennial to annual volume changes in the Humber Estuary.Proc. R. Soc. A463837–854. 10.1098/rspa.2006.1798
63
TrainiC.ProustJ.-N.MenierD.MathewM. (2015). Distinguishing natural evolution and human impact on estuarine morpho-sedimentary development: a case study from the Vilaine Estuary, France.Estuar Coast. Shelf Sci.163143–155. 10.1016/j.ecss.2015.06.025
64
Van CoppenolleR.TemmermanS. (2020). Identifying global hotspots where coastal wetland conservation can contribute to nature-based mitigation of coastal flood risks.Glob. Planet Change187:103125. 10.1016/j.gloplacha.2020.103125
65
van de KoppelJ.van der HeideT.AltieriA. H.ErikssonB. K.BoumaT. J.OlffH.et al (2015). Long-distance interactions regulate the structure and resilience of coasal ecosystems.Annu. Rev. Mar. Sci.7139–158. 10.1146/annurev-marine-010814-015805
66
van der WalD.Wielemaker-van den DoolA.HermanP. M. (2008). Spatial patterns, rates and mechanisms of salt marsh cycles (Westerschelde, The Netherlands).Estuar. Coast. Shelf Sci.76357–368. 10.1016/j.ecss.2007.07.017
67
van WesenbeeckB. K.Van de KoppelJ.HermanP. M. J.BoumaT. J. (2008). Does scale-dependent feedback explain spatial complexity in salt-marsh ecosystems?Oikos117152–159. 10.1111/j.2007.0030-1299.16245.x
68
WangC.TemmermanS. (2013). Does biogeomorphic feedback lead to abrupt shifts between alternative landscape states?: an empirical study on intertidal flats and marshes.J. Geophys. Res. Earth Surf.118229–240. 10.1029/2012JF002474
69
WangH.van der WalD.LiX.van BelzenJ.HermanP. M. J.HuZ.et al (2017). Zooming in and out: scale dependence of extrinsic and intrinsic factors affecting salt marsh erosion.J. Geophys. Res. Earth Surf.1221455–1470. 10.1002/2016JF004193
70
WatsonS. J.KritharasP.HodgsonG. (2015). Wind speed variability across the UK between 1957 and 2011.Wind Energy1821–42. 10.1002/we.1679
71
WernetteP.ShortridgeA.LuschD. P.ArbogastA. F. (2017). Accounting for positional uncertainty in historical shoreline change analysis without ground reference information.Int. J. Remote Sens.383906–3922. 10.1080/01431161.2017.1303218
72
WillemsenP. W. J. M.BorsjeB. W.HulscherS. J. M. H.van der WalD.ZhuZ.OtemanB.et al (2018). Quantifying bed level change at the transition of tidal flat and salt marsh: can we understand the lateral location of the marsh edge?J. Geophys. Res. Earth Surf.1232509–2524. 10.1029/2018JF004742
73
ZuurA.IenoE. N.WalkerN.SavelievA. A.SmithG. M. (2009). Mixed Effects Models and Extensions in Ecology With R.New York, NY: Springer.
74
ZuurA. F.IenoE. N.ElphickC. S. (2010). A protocol for data exploration to avoid common statistical problems.Methods Ecol. Evol.13–14. 10.1111/j.2041-210X.2009.00001.x
Summary
Keywords
coastal biogeomorphology, ecosystem resilience, scale-dependence, saltmarsh edge, tidal channel migration, sheltered macrotidal estuaries, ecosystem services
Citation
Ladd CJT, Duggan-Edwards MF, Pagès JF and Skov MW (2021) Saltmarsh Resilience to Periodic Shifts in Tidal Channels. Front. Mar. Sci. 8:757715. doi: 10.3389/fmars.2021.757715
Received
12 August 2021
Accepted
27 September 2021
Published
20 October 2021
Volume
8 - 2021
Edited by
Nicoletta Leonardi, University of Liverpool, United Kingdom
Reviewed by
Pat Dale, Griffith University, Australia; Peter M. J. Herman, Delft University of Technology, Netherlands
Updates
Copyright
© 2021 Ladd, Duggan-Edwards, Pagès and Skov.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: Cai John Tomos Ladd, cai.ladd@glasgow.ac.uk
This article was submitted to Coastal Ocean Processes, a section of the journal Frontiers in Marine Science
Disclaimer
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.