Changes in Vessel Traffic Disrupt Tidal Flats and Saltmarshes in the Tagus Estuary, Portugal

The Tagus Estuary is one of the largest in Europe with 320 km2, and it has been, for centuries, a gateway to Lisbon. This study focuses on the Moita-Montijo Bay and on the recent dynamics of its tidal flats and saltmarshes. Aerial photographs, orthomosaics, and very high-resolution satellite imagery were used to analyze landcover and shoreline changes. Tidal flats have shown expansion from 1977 to 1995, but since then, contraction dominated, with a change of position of the tidal flat edge of −2.8 m/year in the north bank and −4.2 m/year in the south bank of the Montijo channel. Most contraction occurred along the route of the fast transport catamarans that started operating in 1995, while in the sector without catamaran navigation, expansion was observed. Saltmarshes have been suffering contraction since 1958, with increased rates after 1995 (−0.38 to −0.44 m/year), especially along the catamaran route (−0.57 to −1.27 m/year). The analysis of the wake generated by different vessel types shows a wake increase with the catamarans, in agreement with the increase in contraction along the Montijo channel. Inside abandoned salt pans, saltmarshes expanded. Since 1995, major changes are also observed along the tidal flat margin, with the formation of coarse lag deposits of coarse sands and shells. Given the contraction increase associated with catamaran traffic and the resulting degradation of the tidal flat and the saltmarshes, it is important to introduce measures for containing contraction.


Introduction
Estuaries are coastal wetlands of transition between marine and fluvial environments. These are dynamic ecosystems, and their morphology arises as a result of the characteristics of the paleovalley that originated them. As a result of marine transgressions and fluviomarine dynamics, sediments are deposited on valley floors (Masselink et al. 2014). The attractiveness of estuaries is due to the high availability and productivity of natural resources, coupled with an ease of navigation, which have underpinned the socio-economic development and fostered settlements along the shores (Agardy et al. 2005). The increased intensity of use of these areas in Portugal occurred from the late nineteenth century onwards, due to demographic expansion and the proliferation of economic activities, such as industry (Marques 2011), which have led to the degradation and destruction of estuarine ecosystems, as is the case of saltmarshes (Freire 2003).
Estuarine intertidal areas in Portugal are naturally constituted by tidal flats and saltmarshes. The former are low intertidal areas composed of silt, clay, and sand, submerged Communicated by Brian B. Barnes 1 3 during most of the tidal cycle (Allen 2000;Townend et al. 2011), with a gentle slope (between 0.01-0.30%) without vegetation or partially covered by algae (e.g., Ulva lactuca), seagrasses (e.g., Zoostera noltii), or bivalve beds. Saltmarshes are formed in the upper intertidal areas not affected by intense wave action, being transitional environments between terrestrial and marine ecosystems (Roman and Nordstorm 1996;Townend et al. 2011). Saltmarshes result mainly from the vegetation growth or elevation gain of tidal flats and sand banks, with halophilic and halotolerant vegetation that promotes sediment retention (Boorman 2003;Freire 2003). Saltmarsh species favor soil development, generating the progressive elevation of the saltmarsh surface in relation to the sea, and reducing the flooding time, resulting in a greater biological diversity (Boorman 2003). Sea level rise and increase in tidal amplitude (Bird 2011), more frequent and more intense storm events (Boorman 2003), increased vessel traffic (Bilkovic et al. 2017), and less sediment availability can halt saltmarsh expansion and lead to saltmarsh loss.
Intertidal areas and ecosystems provide important services (Agardy et al. 2005;Boorman 2003). Saltmarshes and tidal flats have the capacity to sequester "blue carbon" (Macreadie et al. 2017), and to retain pollutants and metals present in the water, forming important contaminant sinks (Caçador and Vale 2001;Caçador and Duarte 2012). Intertidal areas and ecosystems constitute a natural defense of the coastline, which can dissipate wave energy. Following severe, but low return period events, intertidal ecosystems can regenerate themselves, especially in areas under accretion (Boorman 2003;Caçador and Duarte 2012). In the Tagus Estuary, Central Portugal, most of the intertidal areas have suffered from land occupation and reclamation, having been seized by maritime-port activities and others related to industry, saliniculture, aquaculture, agriculture, and pastoralism. These have been contaminated areas for decades, influencing shoreline dynamics, and biodiversity survival (Vale et al. 2008;Cesário et al. 2021).
For a multitude of factors, saltmarsh and tidal flat ecosystems are vulnerable to climate change and anthropogenic activity (Agardy et al. 2005;Boorman 2003;Dyer et al. 2000;Guerreiro et al. 2012;Wang et al. 2020). Sea level rise may increase saltmarsh submergence during high tides and storms and exacerbate "coastal squeeze" effect in artificialized areas (Boorman 2003;Doody 2004;Silva et al. 2009;Bilkovic et al. 2017). According to Antunes (2019), the sea level at the Cascais gauge has risen at a mean rate of 1.94 mm/year from 1920 until 2000. Since 1980, sea level rise showed a constant increase, and in the period of 1992 to 2016, the mean rate was 3.10 mm/year. In addition, wave action, infrastructure buildup, and low sediment availability and sedimentation rates, which increase tidal range and current velocities, further cause erosion (Van der Wal and Pye 2004;Bilkovic et al. 2017;Wolters et al. 2005;Safty and Marsooli 2020). Vessel traffic-induced wakes are another factor that impacts shorelines, especially in confined water bodies (Parnell and Kafoed-Hansen 2001). The characteristics and kinematics of the wakes depend on environmental factors (e.g., water depth, riverbed characteristics, tidal currents), vessel characteristics (e.g., design, displacement, loading, type of propulsion), and operation (e.g., speed, maneuvering). Thus, the wakes can be described through the dimensionless parameters of depth-based and length-based Froude numbers, which account for water depth, vessel length, and transit speed (Parnell and Kafoed-Hansen 2001). Some studies have shown that vessel wakes intensify the erosion of saltmarshes (e.g., Castillo et al. 2000;Fontolan et al. 2012;Zaggia et al. 2017;Bilkovic et al. 2017;Safty and Marsooli 2020), and of tidal flats, damaging its ecosystems, namely oyster reefs, in areas parallel or close to vessel transit routes (e.g., Grizzle et al. 2002;Wall et al. 2005). Moreover, this phenomenon tends to remobilize and release contaminants from the shore, due to erosion and resuspension of contaminated sediments that, ultimately, lead to the decrease in the quality of the estuarine system.
Despite the nefarious consequences of saltmarsh and tidal flat erosion, vessel traffic's physical impacts in estuarine ecosystems too often are not accounted for. Studies of such phenomena are still lacking and are insufficient, when compared to studies of other vessel-related impacts (Jägerbrand et al. 2019). It is therefore necessary to evaluate the resilience of these environments to the action of climatic and anthropogenic forcing agents, in the current scenario of increasing population and urban land use in coastal areas (Agardy et al. 2005;Fontolan et al. 2012;Neumann et al. 2015;Valiela et al. 2018;Seto et al. 2011). Furthermore, it is crucial to create strategies to balance coastal protection with land use (Jia et al. 2021), as well as to monitor and identify changes in wetlands (Mendes et al. 2012;Jia et al. 2021).
The present study focuses on the Moita-Montijo Bay, part of the Tagus Estuary, located in the metropolitan area of Lisbon. The main objective is to understand the influence of vessel traffic on the observed changes. Hence, we aim to (1) identify the changes in saltmarsh and tidal flat areas from 1958 to 2018, (2) quantify the rates of expansion and contraction of saltmarshes and tidal flats, and (3) investigate the causal factors for the changes that have occurred, especially on those related to vessel traffic and land use. The quantification of the changes in the waterward margin of the tidal flats and saltmarshes, as well as in land use, is based on the analysis of remote sensing imagery (aerial photos and satellite imagery), supported by field observations. The impacts of vessel traffic are assessed by accounting for changes in vessel characteristics and their influence on wake characteristics.

Study Area
Located on the Portuguese West coast, the Moita-Montijo Bay is part of the Tagus Estuary (Fig. 1), and of the Tagus and Sado Cenozoic Basin. It has a temperate climate with a rainy winter and a hot dry summer -Csa -according to the Köppen classification (IPMA 2021). The tidal regime is high mesotidal, according to the classification of Hayes (1979), with semidiurnal tides (Freire 2003;Silva et al. 2009;Guerreiro et al. 2012;Rilo et al. 2014) and mean amplitudes between 3.2 m at spring tide and 1.5 m at neap tide in Lisbon (Terreiro do Paço), and 3.6 m at spring tide and 1.6 m at neap tide in Alcochete (Costa 2021). The predominant wind directions are N, NW, NE, and SW , and the maximum fetch is around 7 km in W-E orientation.
The Moita-Montijo Bay is located in the middle sector of the Tagus Estuary, according to the morphological classification of Bettencourt et al. (2003), on its Southern shore, belonging to the municipalities of Barreiro, Moita, and Montijo. The study area has 24.2 km 2 , of which 12.1% was occupied by saltmarshes, and 64.0% by tidal flats, in 2018. The shore shows an irregular jagged outline (Mendes et al. 2012;Freire et al. 2013), and recesses coinciding with the outflows of the Moita, Sarilhos Pequenos, and Alhos Vedros creeks. Saltmarshes are mostly located next to sandbanks and the mouth of streams, in areas bordering dikes and drainage ditches, and inside inactive salt pans, tidal mills, and aquaculture farms. The shoreline also includes sandy beaches and artificialized sectors associated with infrastructures to support fishing or vessel transport activities. However, they are not analyzed in this work. Accounting for fetch and dike protection, we have classified the saltmarshes in the Moita-Montijo Bay into three exposure classes ( Fig. 1): (1) sheltered saltmarshes, which are protected by dikes in abandoned salt pans, aquaculture farms, and tidal mill areas, (2) low exposure saltmarshes, which are located outside dikes and show a fetch of less than 1 km and, (3) high exposure saltmarshes, which are located outside dikes and show a fetch of more than 1 km. Sheltered saltmarshes are not affected by severe waves, either created by wind or vessels. Low-exposure saltmarshes show a minor influence of wind waves and tidal currents. High-exposure saltmarshes, although also formed in a low energy environment, are located in exposed locations, more affected by tidal currents, wind waves, and, in some cases, by higher wakes. In the high-exposure saltmarsh margins, cliffing is frequent, due to increased wave action and/or tidal currents. There is the formation of micro-scarps and undercutting occurs, enhancing toppling, which causes the contraction of the saltmarsh margin. Cliffing and undercutting of scarps occur in the areas exposed to larger fetch and wave action. These processes are also typical consequences of wake action (Fontolan et al. 2012). Low exposure saltmarsh margins are generally more stable. Cliffing is rare and the margin shows a gentle slope, with colonization by pioneer species such as Spartina maritima, contributing to sediment retention and the progressive expansion of the saltmarsh margin.
Marks of human occupation stand out in the bay, mainly as a result of maritime-port activities, industry, abandoned salt production, aquaculture, and milling activities, which turned the area into a pole of economic development since the beginning of the twentieth century (Marques 2011). After the 1960s, the area has suffered significant urban sprawling and became a dormitory area of Lisbon, with the municipalities bordering the bay having 59,327 inhabitants in 1960, and 121,937 in 2021 (Instituto Nacional de Estatística 2022; Pordata 2021). Infrastructure buildup has occurred in a significant portion of the Moita-Montijo Bay from 1958 to 2018, which can be perceived by the progradation of the intertidal limit in that period ( Fig. 1), amounting to a loss of 1 km 2 of the intertidal area in the Bay.
In 1932, the Moita-Montijo Bay had 108 active salt pans, but this important local industry began to decline from 1970 onwards, resulting in a progressive abandonment of the salt pans, which led to the current situation, where commercial salt harvesting is absent (Marques 2011). Historically, vessel transportation was the main transport means for goods and people in the Tagus Estuary. Ports on the left bank of the Estuary were essential, as fluvial transport was the only way to travel to Lisbon (Gaspar 1970). Around 130 vessels from the ports of Moita-Montijo Bay (numbers including Samouco and Póvoa) served Lisbon in 1552 (Gaspar 1970). However, with the development of road transportation after the Second World War, fluvial transport was unable to compete and was largely reduced (Gaspar 1970). Nonetheless, its importance remains to this day, as it allows a public transportation alternative for the inhabitants to reach Lisbon, apart from the two bridges over the Tagus Estuary (Fig. 1).
In the beginning of the twentieth century, steel ferries, called cacilheiros, were introduced in the Estuary. More recently, starting in 1994-1995, high-speed aluminum catamarans started to replace the cacilheiros, as the latter traveled slower and were ageing. This followed the contemporary trend in other parts of the world, due to the need for increasing transportation speed and capacity (Varyani 2006). Nowadays, only catamarans travel in the Moita-Montijo Bay, while cacilheiros are still in use in other parts of the Estuary.
In December 2002, the larger and renovated Seixalinho harbor ( Fig. 1) opened to the public (Baptista 2012). This harbor is located further downstream in the tidal channel with vessels traveling around 2.5 km less than when the old Vapores harbor, near the center of Montijo, was in use (Fig. 1). This also meant that vessels stopped going into shallower waters that needed more frequent dredging (Rosa and Teles 2004).

Methodology
The methodological framework consisted in (1) the analysis of aerial photographs and orthophotos for a 60-year period, to map the position of the waterward boundaries of tidal flats and exposed saltmarshes, (2) the quantification and analysis of the changes on the tidal flat and saltmarsh margins, (3) the quantification of vessel traffic and type of waves generated by it, in order to determine its impacts on saltmarsh and tidal flat changes, and (4) the mapping of the saltmarsh areas, using a WorldView-3 image of 2018, to assess land cover changes.

Aerial Photographs and Orthophotomap Interpretation
Aerial photographs from 1958 and 1977 (Direção Geral do Território, DGT) were orthorectified using ENVI 5.6. Spatial resolutions from 0.78 to 2.3 m and typical mean root mean squared errors (RMSE) of 1.8 to 2 m were obtained (Table 1). In the study area, the high tide coincides with a sharp vegetation line in the saltmarsh, while the low tide coincides with the limits of the tidal flat channels. By accounting for these features, the imagery was selected for low and high tide conditions, while scenes with intermediate tidal conditions were excluded. The data set was complemented with two orthophotomaps from 1995 and 2018 at low tide (Direção Geral do Território, DGT) and an 8-band multispectral WorldView-3 image of 23rd of August 2018. For aerial photographs and orthophotomaps, the exact date is not available.
Saltmarsh and tidal flat waterward margins were delineated by visual analysis using ArcMap 10.6. Saltmarsh margins were delineated for 1958, 1995, and 2018, following the vegetation line. Tidal flat margins are defined by the channel to tidal flat sharp limit, only visible in low tide scenes, limiting their use to 1977 onwards.

Tidal Flat and Saltmarsh Change Analysis
Margin changes were calculated for both the tidal flats along the Montijo channel and the exposed saltmarshes. The DSAS v5.0 extension in ArcMap 10.6 was used to calculate the absolute change in position and change rates (Himmelstoss et al. 2018), with perpendicular cast transects 5 m apart from each other. This analysis was not conducted for the saltmarshes inside the dikes, as the shoreline shape does not allow for the correct application of DSAS. The software was developed for the analysis of continuous coastal shorelines, and not for fragmented sectors in restricted and enclosed estuarine areas. Inside the dikes, we have solely analyzed land cover changes. Saltmarsh and tidal flat boundary changes were quantified using the end point rate (EPR), which calculates rates of change, in meters per year (m/year), between the oldest and the youngest shoreline (Himmelstoss et al. 2018).
The uncertainty of the shoreline position (U SP ) and the dilution of accuracy (DOA) were calculated. U SP indicates the uncertainty related to the digitizing, which is mainly influenced by the image resolution and the orthorectification errors. The digitizing error by the operator was not accounted for. Equation (1) assesses the U SP (m) (Irrgang et al. 2018): where R is the resolution of the aerial photographs or ortophotomaps (pixel size, m) and E G is the horizontal accuracy error (RMSE, m). For the ortophotomaps, a default RMSE error of 2 m was used, as the producer does not provide it and 2 m is the maximum error acceptable by the DGT.
DOA serves as an indicator of boundary change rate uncertainty (Irrgang et al. 2018). A smaller period between the boundaries will increase the DOA, meaning less accuracy. Equation (2) assesses the DOA (m/year): where U SP1 is the position uncertainty of the youngest boundary, U SP2 is the position uncertainty of the oldest boundary, T 1 is the year of the youngest boundary and T 2 is the year of the oldest boundary. The imagery at low tide was carefully selected in order to guarantee the complete exposure of the tidal flat margin. This strict approach resulted in not making use of several other dates available in the repository and that is why our analysis is limited to 1977 to 2018 for the tidal flat margin delineation. Hence, we consider negligible the tidal effect on uncertainty. The saltmarsh waterward margin delineation is not affected by tide, since it is the vegetation boundary.

Vessel Data
Data about the types of vessels used for public vessel transport in the study area and their characteristics are shown in Table 2. The vessel types include the three catamaran types that currently navigate the bay (A1, A2, and A3), and the two types of cacilheiros that still transit in the Estuary, although not within the Moita-Montijo Bay (B1 and B2). It was not possible to gather information about the cacilheiros used in the past, but B1 and B2 are assumed to serve as a comparison.
To characterize the wake waves created by the different vessels, the dimensionless parameters of length-based and depth-based Froude numbers (F nl and F nh , respectively) were calculated. Both parameters serve as indicators of wake characteristics, since these closely depend on them (Parnell and Kofoed-Hansen 2001).
A sailing vessel moves at subcritical speed when F nh is less than 1, at critical speed when F nh is equal to 1, and at supercritical speed when F nh is higher than 1. In the subcritical range (typically F nh < 0.6-0.7), a Kelvin wake pattern forms. In the near-critical range (when F nh ~ 0.8-1.1), the vessel speed is equal or close to the speed of the wakes, resulting in the merging of the divergent and the transverse waves at the front. In this range, wake height is the largest. In the supercritical range (F nh > 1), the sailing speed surpasses wave speed. The transverse waves disappear, and the divergent waves become non-dispersive and show a concave shape close to the vessel (Kirkegaard et al. 1998;Parnell and Kafoed-Hansen 2001;Varyani 2006). F nl represents the relation between the vessel length at water level (L WL ) and the types of wakes produced. For a determined L WL , F nl increases as the speed of the vessel increases, which in turn generates larger wavelength wakes. The waves generated at the bow and at the stern interact either constructively or destructively, depending on the F nl , increasing or decreasing wake height, respectively. At the "hump speed," when F nl is 0.5, the highest wave resistance occurs (Kirkegaard et al. 1998;Houser 2010). If F nl increases over 0.5, the vessel can start planing, and wave resistance is decreased, reducing wave height. The maximum wake height is achieved at the near-critical range, when F nh = 1, and at the same time when F nl = 0.5 (Kirkegaard et al. 1998;Parnell and Kafoed-Hansen 2001;Varyani 2006;Houser 2010).
The length-based Froude number is obtained with where V S is the service speed (m/s), g is the gravity constant (≈ 9.81 m/s 2 ), L WL is the length at water level (m), and h is the water depth (m). The parameters used for the calculations, together with vessel size and capacity, are shown in Table 2. According to the Bathymetric model of the Tagus Estuary by Instituto Hidrográfico (2015), the water column height in the Montijo channel varies between 5 and 10 m below the hydrographic zero (HZ), which corresponds to heights at low tide. These limits are used as h to assess F nh .

Field Surveys
Field surveys took place in April, May, and July 2021, and in February 2022, focusing on Montijo and Baixa da Banheira ( Fig. 1), which were selected due to their good representativeness for the vegetation communities of the area. A total of 148 control points for land cover were collected, using a Garmin GPSMAP 64st GPS, for the constitution of training and validation areas for the WorldView-3 image classification. The classification was validated through the Kappa Index.

Supervised Classification
The land cover map was produced using a pixel-based supervised classification over a WorldView-3 image from the 23rd of August 2018, using ArcMap 10.6. The aim was to quantify the saltmarsh area that arose from inactive salt pans, which were still active in 1958, and to perceive how much it contributes to the total saltmarsh area of 2018. The first step was the visual interpretation of aerial photographs and delineation of the active saltpans in 1958. For the supervised classification, the following steps were conducted: (1) separation and masking of water bodies using the normalized difference water index (NDWI) (Lopes et al. 2020) based on a threshold of 0.08; (2) generation of training areas, using ground truthing, visual interpretation of the orthophotomap from 2018, and of the WorldView-3 multispectral image in false color (NIR-G-B). Four classes were defined: sandy beaches and salt surfaces (merged due to their high spectral similarity), dry scrub, tidal flats, and saltmarshes; (3) image classification, using the random trees (or random forest) classifier; (4) evaluation of the classification error, through validation areas from the visual interpretation of the orthophotomap from 2018, and control points collected in the field; (5) extraction of the saltmarsh class and selection of the area within the inactive salt pans that were still active in 1958.

Uncertainty of the Delineations and Dilution of Accuracy of the Change Rates
The uncertainties of the delineations of the margins and of the change rates are shown in Table 3. The U SP is lower in the 1977 shoreline (1.96 m) and is higher in the oldest shoreline (2.95 m), due to the lower resolution of the imagery, when compared to the others. Most change rates are well above these thresholds. The DOA is lowest from 1958 to 1995 (0.10 m/ year) and is higher from 1977 to 1995 (0.17 m/year), supporting the good quality of the results, since the uncertainty of the change rates is low.

Change Rates in Exposed Saltmarshes and Tidal Flats
Saltmarshes Saltmarsh changes were analyzed for the exposed saltmarshes for the periods of 1958 to 1995, and 1995 to 2018. In the Moita-Montijo Bay, both expansion and contraction (e.g., erosion) occurred, but taking place in areas with different characteristics and forcings. In general, contraction prevailed and has increased since 1995. Ten sectors (A to J) have been selected for the analysis (Fig. 2), based on their location, exposure level, size, and the rates observed. Six sectors are in high exposure saltmarsh (Table 4). Sector A is the closest to the sailing route before and after the transition to the Seixalinho harbor in 2002, while B, C, and F are closest to the sailing route before 2002. The other highexposure sectors, E and G, are further away from the sailing route. Two sectors are in low-exposure saltmarsh (Table 4). Sector H is a large sector in a very limited fetch area, and D is an expanding marsh margin on the opposing side of an eroding marsh front (sector C). I and J are small sectors where the rates observed were influenced by infrastructure buildup within or close to them.
The change rate results reveal prevailing contraction (erosion) in the saltmarshes, with over 70% of the transects showing negative EPR for both 1958 to 1995 and 1995 to 2018. In the 2nd period, there were −5.4% of transects with −0.5 to 0 m/year, +1.5% with −1 to −0.5 m/year, + 2.9% with −1.5 to −1 m/year, and + 1.9% with ≤ − 1.5 m/year. As such, in general, contraction increased, showing an increase of 1.5-2.9% for EPR values below −0.5 m/year in 1995 to 2018, when compared to 1958 to 1995 (Fig. 3a). The overall mean (µ) value was negative in both periods, with −0.38 m/ year from 1958 to 1995, and −0.44 m/year from 1995 to 2018 (Fig. 3b, Table 4).
From 1995 to 2018, most of the areas had contraction rates increasing even more, as is the case of sectors A, C, E, and G, with changes from −0.57 to −1.27, −1.04 to −1.14, −0.55 to −0.63, and −1.05 to −1.23 m/year, respectively. In sector F, contraction continued, but the rates diminished from −1.49 to −1.10 m/year (Fig. 3b, Table 4).
Exposed saltmarsh expansion occurred in fewer and, generally, smaller areas with less exposure, with an associated fetch of less than 1 km, such as near creeks. From 1958 to 1995, sectors B, D, and H presented positive mean EPR values of 0.01, 0.58, and 0.12 m/year, respectively (Fig. 3b, Table 4). From 1995 to 2018, EPR decreased in these sectors, and sector B experienced expansion rates change to contraction, with a decrease from 0.01 to − 0.04 m/year. Some small areas that showed contraction from 1958 to 1995 have shown expansion subsequently. These were the cases of I and J (Fig. 2b), where new artificial areas, such as a vessel dismantling facility operating in J since 1976, and the Seixalinho harbor in I, provided protection to the saltmarsh located upstream. H is the sector that showed the least change between periods. The median shows a similar trend as the mean in all sectors, except C and H, where the mean decreased but the median slightly increased (Fig. 3b).  (Table 4). In general, expansion was higher in the west, reaching a maximum value of 7.54 m/year in the northern bank (Fig. 2a). From 1995 to 2018, the trend shifted to the contraction in both banks in the areas west of and around the newly installed Seixalinho harbor (Figs. 2b and 4). Mean EPR values shifted to −1.49 m/ year in Nb, and −2.43 m/year in the Sb (Table 4). However, east of the harbor, the mean rates remained positive, with 0.58 m/year in Nb, and 0.29 m/year in Sb, while westwards, the mean rates were −2.83 m/year in Nb, and −4.22 m/year in Sb ( Fig. 4 and Table 4). The median shows the same trend as the mean, except in the north bank east of the Seixalinho harbor, where the median slightly increased. The maximum contraction value in this period occurred west of the Seixalinho harbor with −11.78 m/year, which evidences extremely high contraction of the tidal flats. Length-Based and Depth-Based Froude Numbers for the Vessels Differences of F nl and F nh within catamaran types (An), as well as within cacilheiros (Bn), are negligible (Table 2). Only A3 presents a higher F nl due to the smaller length of the vessel. All cacilheiros present F nl = 0.31, while all catamarans present F nl > 0.50. The F nh values are the same between all the An and Bn types, given their transit of the same area, at the same speeds. However, a significant difference is noticeable between catamarans and cacilheiros. The former always present much higher values, showing ci. 45% more in F nl and 54% in F nh . The different depths of the water column in the channel, essentially ranging between 5 and 10 m, have a noticeable influence on the depth-based parameter. Catamarans reach values of F nh up to 1.62 when h = 5 m and 1.14 when h = 10 m. Cacilheiros attain values of 0.73 when h = 5 m, and of 0.52 when h = 10 m (Table 2).

Land Use and Land Cover Changes: Transformation Patterns in Sheltered Saltmarshes
Saltmarshes were mapped for 2018 using the WorldView-3 scene, resulting in a Kappa Index of 0.9 in the validation, which correlates to a high accuracy. The analysis of the land use and land cover changes demonstrates that the saltmarsh area has grown in land previously used for salt harvesting, which was, meanwhile, abandoned. After abandonment, the salt pans changed into tidal flats, a process followed by the colonization of saltmarsh vegetation. As of 2018, saltmarshes occupied both the tidal flats and the dikes, with other non-marsh vegetation in the latter. This resulted in a colonization of 41% of the 1958 salt pan area and represents 50% of the current saltmarsh area in the Moita-Montijo Bay (Fig. 1).

Discussion
The results show that contraction was predominant in both periods analyzed. The contraction was larger in areas where the saltmarsh is highly exposed. From all the sectors, only the mean EPR of sector F increased since 1995, decreasing in all the others. This was the sector which had the highest contraction rates in the first period. With the modification of the harbor location, contraction rates decreased, although they remain very high (Fig. 3b). The closest saltmarsh to the catamarans route (sector A) stands out as the sector where contraction increased the most (123%), which implies that catamaran wakes have contributed to saltmarsh contraction. This accelerated cliffing and retreating, similarly described in other marshes (Van der Wal and Pye 2004;Fontolan et al. 2012) (Fig. 5a).
Wake waves may have also contributed to a contraction in other saltmarsh areas in Moita-Montijo Bay; however, these impacts may have been small and were not detected.
From 1958 to 1995, the maximum rate of contraction (− 5.85 m/year) occurred due to the construction of a vessel dismantling facility infrastructure in J (Fig. 2), which disrupted a significant area of saltmarsh. However, some saltmarsh was able to recover behind the infrastructure from 1995 to 2018, resulting in a positive rate in that period. The construction of the Seixalinho harbor, despite disrupting some saltmarsh when being built, promoted shelter for the saltmarsh in the nearby sector I, increasing the saltmarsh expansion in that sector from 1995 to 2018.
The results show contraction in the tidal flats margin only after 1995, west of the Seixalinho harbor, while eastwards, no significant change was observed. As such, contraction in Table 4 Mean (μ) and standard deviation (σ) of saltmarsh and tidal flat EPR for each sector and bank

Estuarine system
Exposure/location Sector/bank 1958Sector/bank -1995Sector/bank 1977Sector/bank -1995Sector/bank 1995Sector/bank -2018 μ (m/yr) σ μ (m/yr) σ μ (m/yr) σ Saltmarsh ( the tidal flats coincides with the introduction of the catamarans and has occurred only in the channel banks along the vessel route. Catamarans transit at supercritical (F nh > 1) and close to near-critical speeds (F nh ~ 0.8-1.1), while cacilheiros only travel at subcritical speeds (F nh < 1). When the tide is low, catamarans will sail at essentially supercritical speeds (~ 1.14 < F nh < ~ 1.62). However, when the tide rises (around 3.2 m in spring tide, and 1.5 m in neap tide), F nh will decrease due to the increase in the water height. In the sectors where catamarans travel with an F nh ~ 1.14 (westward of the Montijo channel, where water height is larger), and with an F nh ~ 1.62 (closer to the Seixalinho harbor, where water height is smaller), F nh can decrease to ~ 1 (critical speed), and ~ 1.40 (supercritical speed), respectively. As the margins of the tidal flats are only impacted during low tide, these are mainly impacted by wakes created at supercritical speeds. Saltmarsh is impacted by wakes created at nearcritical speeds westward of the channel, and at supercritical speeds closer to the Seixalinho harbor, as it is only impacted during high tide.
Most catamarans transit with an F nl close to the "hump speed" and meet the criteria to produce wake waves of high energy and large height. The A3 types might produce smaller wakes, due to a higher F nl . Furthermore, water-jet propulsion that is used by catamarans (Table 2) can increase wave height by 5-20% (Varyani 2006).
No wave height records exist for the Moita-Montijo Bay. However, catamaran maximum wave height (H max ) has been estimated by Oliveira and Vargas (2009) for the Alfeite beach, 5 km from the study area (Fig. 1), where a channel with a similar depth to the Moita-Montijo Bay exists. According to these authors, catamaran waves near the sailing The upper whisker is limited by the maximum value below or equal to Q3 + 1.5(IQR), and the lower whisker is limited by the minimum value above or equal to Q1 − 1.5(IQR), where Q3 is the 3rd quartile, Q1 is the 1st quartile, and IQR is the interquartile range (IQR = Q3 − Q1) Fig. 4 Mean, median, and quartiles of EPR for the tidal flats, along the Montijo channel banks from 1977 to 1995, and 1995 to 2018. W-Nb and W-Sb are the north bank and south bank, respectively, west of the Seixalinho harbor. E-Nb and E-Sb are the north bank and south bank, respectively, east of the Seixalinho harbor. The upper whisker is limited by the maximum value below or equal to Q3 + 1.5(IQR), and the lower whisker is limited by the minimum value above or equal to Q1 − 1.5(IQR), where Q3 is the 3rd quartile, Q1 is the 1st quartile, and IQR is the interquartile range (IQR = Q3 − Q1) route can reach up to a H max of 3.23 m, when traveling at 10.29 m/s. With slower speeds, H max decreases. At 7.72 m/s, the H max is 1.50 m, and at 5.14 m/s the H max is 0.51 m. If the shallow water effect is negligible, the wave height decreases and stabilizes at around the distance of 500 m. At that distance, the H max , at a speed of 10.29 m/s, is close to 0.90 m. In practice, however, the wakes will shoal considerably over the tidal flats, increasing in height until they break in the shore, or over longer and shallow tidal flats, until the wave energy is dissipated.
The maximum wind wave heights observed in the Alfeite are about 0.84 m, and the mean wind wave height is about 0.11 m (Freire et al. 2009). Oliveira and Vargas (2009) state that when catamarans travel at 10.29 m/s, the average annual wake wave energy dissipated at the beach is 2.5 times higher than the average annual wind wave energy. However, the Alfeite beach faces the longest fetch in the Estuary, of about 25 km in SW-NE orientation, while the longest fetch in the Moita-Montijo Bay is about 7 km in W-E orientation. This means that, even in storm events, wind wave heights in the Moita-Montijo Bay will generally be lower than those at the Alfeite beach, and the average wake wave energy dissipated will be even higher, when compared to the average wind wave energy. As such, the data shows that catamarans generate large height wakes and significantly alter the wave regime in the bay. That is true for both tidal flats and saltmarshes close to the transit route. The impact is greater on the tidal flat banks, as these are very close to the sailing route, where the energy of the wakes is still very high. However, as the channel banks progressively contract, the energy impacted on them decreases, and the contraction rates are expected to progressively decrease. Despite that, tidal flats next to sector A are not long or shallow enough for wakes to dissipate energy or break before reaching the saltmarsh during high tide, and tidal flat extension is only getting smaller. In this sector, wakes shoal considerably and break in the saltmarsh, towards the west (Fig. 5b), where wakes can be generated close to critical speeds. Vessel wakes significantly contribute to shoreline change when sailing is regular, concentrated, and close to the shore (Houser 2010). The change rates in the tidal flat and in the saltmarsh at the sector A after 1995 are consistent with the characteristics of the modeled wakes along the channel.
Cacilheiros, on the other hand, generate weaker and smaller wakes. The variable that further differentiates catamarans and cacilheiros is the travel speed. Cacilheiros travel 55% slower at maximum speed than catamarans at service speed. When cacilheiros sailed in the Montijo channel, they only sailed at subcritical speeds (F nh < 1). Also, F nl was always much below the "hump speeds." Thus, they did not present the potential to reach high dimensionless Froude numbers. The wake height generated by the cacilheiros has also not been measured, but, despite the different characteristics of the vessels, such as tonnage, passenger loading, size, and hull type, the wakes must be similar to wakes generated by catamarans sailing at the same speed (5.14 m/s). These wakes are close to 0.51 m near the sailing line, according to an estimation by Oliveira and Vargas (2009). Furthermore, the results show that the cacilheiros did not cause contraction of tidal flat and saltmarsh.
Changes in traffic frequency are also relevant to assess the change rates. Data from 1988 to 1998 shows that the weekly trips increased by 77%, and even more in the following years (Fig. 6). Consequently, the number of wake waves created, also increased, which should have intensified bank contraction, as also shown by Bilkovic et al. (2017) for the Chesapeake Bay.
The effect of vessel wakes on increased erosion is supported by its spatial distribution and clear linkage to the more exposed saltmarsh and tidal flat channel sectors. However, other factors may have contributed to increasing this effect and may play a role in saltmarsh contraction: (1) the reduction of sediment availability due to the construction of dams in the Tagus river catchment in the twentieth century, (2) sea level rise, which was in the order of 1.9-3.1 mm/ year, (3) the potential increase in storm events under the current climate change scenario, but we have no data for these in the Tagus Estuary, and (4) small fast boats used for fishing and recreational purposes that were not accounted in this study but showed an impact in other estuaries (Bilkovic et al. 2017).
Regarding the current saltmarsh area within inactive salt pans, the recorded expansion shows the importance of these sheltered areas for the colonization by saltmarsh species, when compared to the other saltmarsh areas. However, salt pans are fixed infrastructures that limit the saltmarsh expansion within their dikes. If the ongoing scenario prevails, where urban expansion near saltmarsh and estuarine environments persists, and the highly exposed saltmarshes continue to erode at the current rate, the vegetation will be restricted to those sheltered areas, without the ability for further expansion. This, however, does not mean that salt pan dikes should be removed, because, as demonstrated, the dikes provide protection from waves and allow for sediment retention, and saltmarsh colonization and expansion. The high exposure saltmarshes should, instead, be protected from the wave action.

3
The observed trend for contraction in the saltmarsh of Moita-Montijo Bay contrasts with the trends found by Lopes et al. (2020) in the nearby marshes of Corroios and Pancas in the Tagus Estuary, where relative stability and expansion, respectively, have been observed since 1984. The characteristics of the Moita-Montijo Bay, however, differ from the Corroios and Pancas marshes. The Corroios marsh is more sheltered from wave action, with a shorter maximum fetch (about 2.5 km in N-S orientation) than the one of Moita-Montijo Bay (about 7 km in W-E orientation), and to the predominant wind directions in the Estuary, which result in wind wave directions contrary to the Corroios saltmarsh, located mainly in the N and NW parts of the Seixal Bay (Fig. 1). Pancas has a significantly higher maximum fetch than the Moita-Montijo Bay (> 20 km), but the marsh shows larger protection by tidal flats, which dissipate wave energy. It also shows higher sedimentation rates than the rest of the marshes sampled by Silva et al. (2013), which exceed the mean rate of sea level rise observed at the Cascais gauge at the mouth of the Tagus Estuary. In addition, no public transport vessels transit near those saltmarshes. Other authors have shown saltmarsh edge contraction due to vessel-generated waves: Castillo et al. (2000) for the Odiel Estuary (Spain), Bilkovic et al. (2017) for the Chesapeake Bay (USA), and Safty and Marsooli (2020) for the Jamaica Bay (USA).
At the Moita-Montijo Bay, the effects of the catamaraninduced waves, besides increasing contraction and landward migration of the tidal flats margin, also generated the formation of coarse lag deposits (Figs. 5c, 7, and 8). These features are associated with the transport of the finer sediments, leaving behind coarse sands, oyster, and other bivalve shells along the channel banks (Figs. 5c, 7, and 8). These deposits emerge at low tide but in some sectors stand above the water line even at high tide. The coarse accumulation shows that the backwash has lower energy than the wakes, sustaining the deposit. Such effects have caused erosion of oyster reefs in the banks of the Montijo channel, in a similar manner to what has been observed in the Mosquito Lagoon, USA (Grizzle et al. 2002). There, reef erosion has been observed since 1943, with similar deposits created along channel banks parallel to the vessel routes, and the oyster decline observed is associated with vessel traffic (Grizzle et al. 2002;Wall et al. 2005). In the Moita-Montijo Bay, incipient lag deposits are visible in the 1995 orthophotos. The introduction of the catamarans in 1994-1995 led to the expansion of the lag deposits, together with increased saltmarsh margin retreat (Fig. 8). However, they were not the cause for oyster mortality that occurred in the mid 1970s in the Tagus Estuary, due to organotin pollution from antifouling paint (Dias 1990).

Conclusions
The analysis of historical aerial photos, orthophotos, and very high-resolution satellite imagery, together with field observations allowed us to identify the main changes in the saltmarshes and tidal flats of the Moita-Montijo Bay since 1958. The results reveal the impacts associated with recent changes in vessel transit. Exposed saltmarshes have been contracting since 1958, with a mean rate of − 0.38 m/year until 1995. Since 1995, saltmarsh contraction intensified to a mean rate of − 0.44 m/year. The rate more than doubled in the areas bordering vessel transit routes, especially after the modernization from the small and slow cacilheiros, to the fast catamarans in 1995, resulting in stronger wake waves. From 1995 to 2018, tidal flats in the Montijo channel, parallel to vessel traffic (west of the Seixalinho harbor), presented contraction rates that reached −11.78 m/year in the south banks, while before 1995, no significant change occurred. The high contraction rates gave origin to the formation of coarse barrier lag deposits in the outer margin of the tidal flats. These consist mainly of oyster and other bivalve shells, corresponding to the gravelly fraction of the washed tidal flats.
While contraction prevailed in the exposed saltmarshes, saltmarsh expansion took place inside salt pans abandoned after 1958. In 2018, 50% of the total saltmarsh area in the Moita-Montijo Bay were saltmarshes recently formed inside former salt pans, as they provide an ideal sheltered environment. These new saltmarshes will only extend until they get restricted by the salt pan's dikes. Outside them, the saltmarsh is contracting and suffering from the "coastal squeeze" effect, unable to migrate inland as it is limited by the builtup environment (Boorman 2003;Doody 2004).
Due to the high contraction in both saltmarshes and tidal flats west of the Seixalinho harbor, as a result of catamaran wakes, strategies to mitigate or reverse this trend should be considered. For saltmarshes, special attention must be considered to sector A, where catamaran wakes show more impact. However, other areas of exposed saltmarsh not affected by catamaran wakes should also be protected, as rates of contraction remain high.
Froude number analysis indicates that catamarans transit at supercritical or near-critical speeds, as well as close to the "hump speed," and generate large height wakes. The transit speed of catamarans inside the Moita-Montijo Bay needs to be readjusted in order to reduce wake impacts on the tidal flats and saltmarshes. This measure would only need to be enforced inside the bay, from the Seixalinho harbor to the western limit of the bay, along around 4 km. Further physical modeling on wave dynamics would be recommended before implementing these measures, although observing the impacts of speed reduction and tuning the sailing speed accordingly would be an easy-to-apply approach. These measures should be accompanied by the implementation of nature-based solutions for saltmarsh and tidal flat regeneration. These should, in the case of saltmarsh, include "living shorelines," which act in a similar way to abandoned salt pans. "Living shoreline" projects could also benefit from reusing the deposited oyster shells (See Bilkovic et al. 2016;Moosavi 2017). These actions could be complemented by the application of protection measures to specific sectors of the Moita-Montijo Bay, limiting further land reclamation, and assuring the protection of the sheltered saltmarsh and tidal flats.