Effects of Beach Nourishment Project on Coastal Geomorphology and Mangrove Dynamics in Southern Louisiana, USA

: Relative sea-level (RSL) rise associated with decreased ﬂuvial sediment discharge and increased hurricane activity have contributed to the high rate of shoreline retreat and threatened coastal ecosystems in Port Fourchon, Louisiana, USA. This study, based on QuickBird/drone images (2004–2019) and LIDAR data (1998–2013), analyzed the impacts of shoreline dynamics on mangroves ( Avicennia germinans ) and marshes before and after the initiation of a beach nourishment project in 2013. The coastal barrier and dune crest migrated landward between 1998 and 2013. Meanwhile, the dune crest height increased between 1998 and 2001, then decreased in 2013, probably due to hurricane impacts. The total sediment volume along this sandy coastal barrier presented an overall trend of decline in the 1998–2013 period, resulting in a wetlands loss of ~15.6 ha along 4 km of coastline. This has led to a landward sand migration onto muddy tidal ﬂats occupied by Avicennia germinans (1.08 ha) and Spartina (14.52 ha). However, the beach nourishment project resulted in the advancement of the beach barrier from Nov/2012 to Jan/2015, followed by a relatively stable period between Jan/2015 and Mar/2019. Additionally, both the dune crest height and sediment volume increased between 2013 and 2019. This set of factors favored the establishment and expansion of mangroves (3.2 ha) and saltmarshes (25.4 ha) along the backbarrier environments after 2013, allowing the tidal ﬂats to keep pace with the RSL rise. However, waves and currents caused shoreline erosion following the beach nourishment project between Oct/2017 and Nov/2019, threatening wetlands by resuming the long-term process of shoreline retreat.


Introduction
The Mississippi River Delta has undergone significant geomorphological changes during the Holocene and Anthropocene, exerting impacts on the vegetation and economy of the Louisiana coastal zone [1][2][3][4][5][6]. Port Fourchon, located on the Caminada-Moreau headland at the lower Mississippi River Delta, is one of the most important logistical support centers for the offshore oil production industry in the U.S. This port is the only home-base for offshore oil terminals in the U.S., and serves as the intermodal support hub for~90% of the Gulf of Mexico drilling and~16% of the U.S. domestic oil and gas production [7]. However, the rapid acceleration of global sea-level rise (~3. 4 mm/year at the present with an expected rise of 65 cm by 2100) has jeopardized the stability of many coastal areas around the world [8], particularly along the Mississippi River Delta, where reduced sediment supply, hurricane activity, and relative sea-level (RSL) rise have been causing shoreline retreat at an alarming rate of~12-14 m/year [9][10][11][12][13][14][15]. Moreover, three-dimensional geospatial data indicate that the sand dunes at Caminada-Moreau Headlands were migrating inland at a pace of up to 4.4 m/year between 1998 and 2013 [12]suggesting that both the shoreline and the beach barrier were migrating landward as a response to sea-level rise and reduced sediment supply.
Recent studies investigating the effects of marine hazards, such as storm surges, on coastal areas have been mostly focused on the safety and protection of humans and assets without considering the relevance of coastal ecosystems [16]. In this context, the Caminada-Moreau headland is the home-base and boreal limit of the North American mangroves [17]. Along with being recognized as the most productive ecosystem in the subtropics, mangroves can also mitigate shoreline retreat by attenuating waves, trapping sediments, and accumulating peat soil. These characteristics make mangroves an essential coastal geomorphological component in withstanding sea-level rise and hurricanes [18][19][20][21][22][23][24]. However, a recent study suggests that mangroves cannot withstand a RSL rise exceeding 6.1 mm/year, which is far below the current rate of RSL rise (~9.3 mm/year) in the Port Fourchon area [25]. Hence, both natural and economic functions at the Mississippi River Delta are being threatened by the sea-level rise.
As part of the Louisiana Coastal Protection and Restoration Authority's coastal restoration effort to strengthen the beach barrier and mitigate the threats of shoreline retreat to the infrastructures and coastal wetlands at the lower Mississippi [26], a beach nourishment project was undertaken in 2013 to replenish 2.8 × 106 m 3 of sediments (dredged from offshore) along 9.5 km of coastlines near Port Fourchon [10]. In addition, sand fencing was installed in 2013 along the dunes to control the location and rate of aeolian erosion or sand deposition [27,28]. Dune vegetation (Panicum amarum and Uniola paniculata) was planted in some sectors of the supratidal zone in 2016 [29][30][31]. However, what are the ecological impacts of such abrupt coastal morphodynamic transformations? This concept is not wellunderstood. For example, the landward migration of beach barriers resulted in the burial of muddy tidal flats and degradation of mangrove vegetation in Brazil [20,23,32]. Mangrove ecotones on high intertidal flats or near the saltwater-freshwater interfaces may migrate landward to indirectly compensate for the loss of mangrove areas in lower intertidal flats, as the influence of saline waters increases during marine incursions [20,23,[33][34][35]. Large data gaps exist in the literature regarding the recent coastal transformation at the Mississippi River Delta, and additional research incorporating high-resolution spatial-temporal analysis is needed to provide a robust baseline dataset for Louisiana's coastal restoration projects. To fill these gaps, this study aims to investigate the interactions between the black mangrove (Avicennia germinans) dynamics and the shoreline retreat before and after the beach nourishment project in the Caminada-Moreau Headlands around Bay Champagne, Louisiana, USA ( Figure 1). We provide high-resolution spatial-temporal analysis based on planialtimetric data acquired from satellite, drone, and Lidar data to study the dynamics between the migration of beach barrier and mangrove encroachment from 1998 to 2019 in the light of sea-level rise.

Methods and Materials
The high-resolution spatial-temporal analysis of a beach barrier in front of Bay Champagne lagoon was based on the examination of planialtimetric data and oceanographic time-series data, following a pre-designed methodology flow chart consisting of three phases ( Figure 2): (1) spatial-temporal analysis based on satellite images and Lidar data; (2) development of digital elevation models based on photogrammetry of drone images with field validation; (3) data integration to evaluate the impacts of the beach nourishment project on backbarrier wetlands.

Methods and Materials
The high-resolution spatial-temporal analysis of a beach barrier in Champagne lagoon was based on the examination of planialtimet oceanographic time-series data, following a pre-designed methodolog consisting of three phases ( Figure 2): (1) spatial-temporal analysis base images and Lidar data; (2) development of digital elevation mod photogrammetry of drone images with field validation; (3) data integrati the impacts of the beach nourishment project on backbarrier wetlands.

Study Area
The Bay Champagne lagoon (between 29 • 09 -29 • 06 N, 90 • 11 -90 • 08 W, Figure 1) is part of the Lafourche subdelta lobe and situated at the southeast side of Port Fourchon [2]. The lagoon-tidal flat system is hyper saline (45 ) and flushed by diurnal microtides at a range of~0.3 m. The climate is humid subtropical, with annual monthly temperatures ranging between 6 • C and 30 • C, and an average annual accumulated precipitation at 1600 mm/year [36]. The cyclic switching of the Mississippi River Delta lobe, coupled with human activities, have caused marine transgressions at the Bay Champagne coast for over several hundred years [37]. The shoreline retreated~1550 m between 1887 and 2012, but the rate of shoreline erosion has fluctuated from~14.8 m/year (1887-1930) to~9.8 m/year  and~12 m/year   [10,38,39]. storm surges generated by landfalling hurricanes [10,40]. For instance, in 2002, Hurric Lili (category 2 in the Saffir-Simpson storm scale) produced a 3 m storm surge and in duced extensive overwash deposits to the east side of the Lagoon [41]. In addition, Hu canes Katrina (2005), Rita (2005), Gustav (2008), and Ike (2008) also generated ~5 m to of storm surges that affected this area [42].

Historical Oceangrophic Data
RSL data from Grand Isle  and Eugene Island , located 26 to the northeast and 120 km to the northwest of the study area, respectively, were acce at the National Oceanic and Atmospheric Administration website [36]. Fluvial disch data (1985-2017, monthly mean in ft3/s) for the Bayou Lafourche River at Thibodaux (29°47′52′′N, 90°49′21′′W), 99 km upstream of Bay Champagne, were acquired from United States Geological Survey [43]. Currently, a saltmarsh-mangrove ecotone consisting of Spartina alterniflora and stunted stands of Avicennia germinans exists on the landward side of Bay Champagne. Seaward of the lagoon, a narrow, sandy beach barrier (1-2 m high) blocks tides, waves, and longshore currents from entering the lagoon, but this barrier is regularly breached by storm surges generated by landfalling hurricanes [10,40]. For instance, in 2002, Hurricane Lili (category 2 in the Saffir-Simpson storm scale) produced a 3 m storm surge and introduced extensive overwash deposits to the east side of the Lagoon [41]. In addition, Hurricanes Katrina (2005), Rita (2005), Gustav (2008), and Ike (2008) also generated~5 m to 1 m of storm surges that affected this area [42].

Historical Oceangrophic Data
RSL data from Grand Isle  and Eugene Island , located 26 km to the northeast and 120 km to the northwest of the study area, respectively, were accessed at the National Oceanic and Atmospheric Administration website [36]. Fluvial discharge data (1985-2017, monthly mean in ft3/s) for the Bayou Lafourche River at Thibodaux, LA (29 • 47 52 N, 90 • 49 21 W), 99 km upstream of Bay Champagne, were acquired from the United States Geological Survey [43].

Remote Sensing
The planialtimetric data were acquired from drone and satellite images, as well as LIDAR data. QuickBird satellite images, downloaded from Google Earth Engine with a ground pixel resolution of 2.44 m (multispectral) and three bands (blue, green, red) [44], Fieldwork was completed during the drone survey to validate the topographic data, as well as vegetation types/heights and shoreline dynamics between Oct/2017 and Nov/2019. Topographic surveys allowed determining the intertidal and supratidal zones limits during that time interval. The planialtimetric data were acquired by a Trimble Catalyst receiver, supported by a differential Global Navigation Satellite System with a decimeter correction (precision ± 10 cm) and an electronic Self Leveling Horizontal Rotary Laser. These data were used as Ground Control Points (GCPs, 61 points) to evaluate and calibrate the digital elevation model (DEM) acquired by photogrammetry (Table S1, Supplementary Material).

Image Classification
The vegetation and geomorphological features were manually classified by photointerpretation using various tools in the Global Mapper version 18 software. Drone images (resolution of 2.6 cm) permitted the identification of Avicennia trees in the marshes and the beach barrier. These images and fieldwork data provided a consistent indicator to support the classification derived from the Quickbird images (Table S1, Supplementary Material). Ground Control Points (GCPs, Figure 1 and Table S1, Supplementary Material) reinforced the identification of each land cover type (black mangrove, marsh, and beach

Image Classification
The vegetation and geomorphological features were manually classified by photointerpretation using various tools in the Global Mapper version 18 software. Drone images (resolution of 2.6 cm) permitted the identification of Avicennia trees in the marshes and the beach barrier. These images and fieldwork data provided a consistent indicator to support the classification derived from the Quickbird images (Table S1, Supplementary Material).
Ground Control Points (GCPs, Figure 1 and Table S1, Supplementary Material) reinforced the identification of each land cover type (black mangrove, marsh, and beach barrier) according to the image features (multispectral digital numbers) related to the texture of drone orthoimages (spectral information between 380 and 710 nm). This cross-validated data produced a highly accurate classification of vegetation and geomorphological units. Drone panoramic aerial photos were also used to identify the vegetation and geomorphological units ( Figure 1e). Details about drone image processing may be found in [23,24].

Analyses of Spatial Data
Coordinates of infrastructure, such as houses and boat ramps, were used as stable reference lines for spatial-temporal analysis. The upper limit of the intertidal zone, evidenced by a wet-dry line in the images, was used as a reference to evaluate the shoreline dynamics. Distance measurements were obtained by Global Mapper version 18 software on georeferenced satellite and drone images. Considering the ground pixel resolution of 2.44 m of the QuickBird images, a margin of error of 2.5 m was estimated for the planimetric data obtained from these images. Planimetric transects perpendicular to the stable reference lines were used to measure the dynamics of the sandy coastal barrier. Linear regression was calculated between the shoreline dynamics and (1) the relative sea-level rise (mm), and (2) the highest and lowest monthly average values of the fluvial discharge (ft3/s) during the period of retreat/advance of shoreline ( Table 1). The linear regression was calculated between the retreat/advance of shoreline, using transects B, C, and D (m) as references, and the relative sea-level rise (mm). The long-term, linear RSL rise trend is relative to the most recent mean sea-level datum established by the Center for Operational Oceanographic Products and Services. The long-term linear trend is based on the monthly mean sea level, eliminating seasonal fluctuations related to coastal ocean temperatures, currents, winds, and atmospheric pressures. Global Mapper version 18 [45], Agisoft Metashape Professional version 1.6.2 [46], R programming language [47], and the R package 'Performance Analytics" [48] were used to analyze the spatial-temporal dataset. Details regarding the materials and methods are described in the Supplementary Information. Table 1. Retreat and advance of the distal and proximal limits of the shoreline (m) before and after the nourishment project with a margin of error of ±2.5 m. Data based on Quickbird images, using reference the transects A, B, C, and D (see location in Figure 4). Also included in this table are: the relative sea-level rises, data based on the relative sea-level trend of 9.08 ± 0.42 mm/year (1947-2018) recorded in Grand Isle, 26 km northeast of the study area; and the average highest and lowest monthly fluvial discharges.

D Models
The drone images, obtained by a digital 4 K/20 MP (RGB) camera, were treated using the Agisoft Metashape Professional 1.6.2. to generate 3D spatial data and orthomosaics with the support of planialtimetric GCPs (www.agisoft.com) (accessed on 7 May 2021).
( Figure 1). Orthomosaic images of 2017, 2018, and 2019 were used for the time series analysis. The software constructed a set of points in 3D space from all matched pairs between aligned photos. Erroneous points in the sparse point cloud were removed to improve the model final geometry. This cleaned sparse point cloud was used as a reference to reconstruct a more detailed set of geometries known as the dense point cloud [49]. This dense and accurate 3D point cloud with point spacings between 3 and 5 cm was manually classified. The contrasts of colors and elevations of point clouds enabled identification of points representing the vegetation cover and the sandy barrier surface. The points representing the sandy barrier surface were used to obtain a digital terrain model (DTM), which represents the substrate surface without the vegetation cover. A mesh of the sandy flat surface was then developed based only on the points representing the topographic surface of the terrain. This model was adjusted to the GCPs obtained by the field topographic survey (Table S1, Supplementary Material). A digital surface model (DSM) representing the natural (trees and herbs) and built (houses and streets) features was also produced [46]. The vertical differences between GCP and the DTM allowed a quantitative analysis of that model, following Equation (1), as suggested by [23]: where Z dif = the vertical differences, Z DEM = the Z value of the 3D dense point cloud, and Z grd = the Z value of the Ground Control Point. The vertical differences (Z dif ) were lower than 15 cm, indicating a vertical margin of error of ± 15 cm for the 3D models. The horizontal differences (latitude and longitude) were <0.71 m (Table S1, Supplementary Material). The differences between the latitude/longitude data obtained by the Trimble Catalyst GNSS receiver in the GCPs and drone surveying have been attributed to a lower accuracy of the drone GPS than the Catalyst GNSS receiver [50,51]. In contrast, elevations based on aerial photogrammetry of drones present high vertical accuracy [52,53]. The final digital terrain model was adjusted using the GCPs planimetric values. Considering the X dif , Y dif , and Z dif values, margins of error were estimated at ±0.076 m 3 and ±0.15 m 3 for the volume calculations based on drone and Lidar data (vertical and horizontal accuracy of 15 and 100 cm), respectively. An elevation grid for the ground was obtained based on the mean dense point cloud to minimize the effects of vegetation and seasonality on the drone and Lidar surveys. Vertical features were referenced to NAVD88. The shoreline position and dune crest were defined as the mean high tide water elevation (higher level of the intertidal zone) and the maximum surface elevation, respectively, identified by cross-shore profiles. Sediment volumes were measured according to the elevation grid generated for each drone and Lidar survey, relative to a baseline, defined as mean sea-level (0 m). Cut-and-fill volumes were calculated within a selected area using Global Mapper software version 18. Volumetric calculations were performed by dividing the area of interest up in to small rectangular pieces following a uniform grid and then calculating the sum volume of the small 3D rectangles (Volume = Height * Pixel Size) between terrain models and the cut surface [45]. Two fences along the dunes were used as a common reference in the drone images to delimit the target zones (Figures 1c and 6c): Zone 1-part of the supratidal and the sandy intertidal flat under the action of currents and waves; Zone 2-part of the supratidal area; and Zone 3-part of the supratidal and the intertidal flat behind the coastal barrier. In addition, Global Mapper generated a vertical profile along a specified path using loaded planialtimetric datasets [45]. A spatial and temporal sequence of these profiles shows the dune crest dynamics and the coastal morphology in three dimensions with the most pronounced vertical variations recorded along a beach barrier [54]. Eight planialtimetric profiles, identified by the red lines in the figures, were developed to record the coastal morphology changes in a temporal sequence. Seven profiles are cross-shore transects that start landward of the dune ridge and end at the shoreline (perpendicular to the shoreline). The locations chosen for each cross-shore transects have a wide spatial representation of the intertidal and supratidal zones of the studied coast (Figures 1, 3c, 6d and 7d). One longshore profile follows the top of the dune ridge line (parallel to the shoreline) of the year of each digital terrain model (Figure 5a,b). Planialtimetric cross-shore profiles were used for sediment volume analysis, while planimetric cross-shore transects were used for measuring shoreline and habitat changes. files, identified by the red lines in the figures, were developed to record th phology changes in a temporal sequence. Seven profiles are cross-shore tran landward of the dune ridge and end at the shoreline (perpendicular to the locations chosen for each cross-shore transects have a wide spatial repres intertidal and supratidal zones of the studied coast (Figures 1, 3c, 6d, 7d). profile follows the top of the dune ridge line (parallel to the shoreline) of t digital terrain model (Figure 5a,b). Planialtimetric cross-shore profiles wer iment volume analysis, while planimetric cross-shore transects were used shoreline and habitat changes.

Climatic and Oceanographic Data
Precipitation data from 1989 to 2018 revealed the highest values in J 368 mm; mean: 213 mm) and August (range 44-488 mm; mean: 192 mm), in October (range: 1-384 mm; mean: 104 mm) and November (range: 9-358 the dune crests, showed an increase in sediment volume of ~9116 m 3 . This trend als curred in zone 3 (part of the supratidal and the intertidal flat behind the coastal ba 11.47 ha), with a sediment gain of 16,040 m 3 (Figure 6), under the influence of mang and marshes (Figure 7).

Dynamics of the Backbarrier Wetlands
The muddy tidal flats exhibited topographic gradients between 0 and 47 cm a the mean sea-level (Figures 1 and 7). Tidal flats occupied by monospecific colonies o icennia germinans (<2.3 m tall) were 13-47 cm higher than those occupied by Sparti terniflora (0.2-12 cm), while the lowest elevations (~0 cm) had no vegetation cover sandy barriers (~1-3 m height) were occupied by herbaceous plants, primarily Pan

Dynamics of the Beach Barrier before the Nourishment Project
The overall sediment volume of the beach barrier in our study area oscillated between 1998 and 2013. In 1998, the overall volume and floor area of the studied barrier was 357,010 m 3 and 51 ha, respectively. These numbers increased to 554,073 m 3  Planialtimetric data obtained by drone between Oct/2017 and Nov/2019 also indicated no significant shoreline migration ( Figure 6). However, the volume of sediments along the intertidal zone, (i.e., between mean high tide and mean low tide), and supratidal zone (above high tide, flooded only at spring tide or during storms) changed during that period. The sandy intertidal flat in front of the coastal barrier was exposed to waves and currents and showed signs of erosion. By contrast, the backbarrier wetlands received sediment by aeolian and washover transport (Figure 3, vertical profiles g -h and i -j ). For instance, the intertidal to supratidal zone (Zone 1: 3.75 ha) showed a reduction in sediment volume of~10,049 m 3 between Oct/2017 (38,016 m 3 ) and Nov/2019 (27, 957 m 3 ). However, the supratidal zone (Zone 2: 5.27 ha), exposed to aeolian processes and with fences along the dune crests, showed an increase in sediment volume of~9116 m 3 . This trend also occurred in zone 3 (part of the supratidal and the intertidal flat behind the coastal barrier, 11.47 ha), with a sediment gain of 16,040 m 3 (Figure 6), under the influence of mangroves and marshes (Figure 7).

Dynamics of the Backbarrier Wetlands
The muddy tidal flats exhibited topographic gradients between 0 and 47 cm above the mean sea-level (Figures 1 and 7). Tidal flats occupied by monospecific colonies of Avicennia germinans (<2.3 m tall) were 13-47 cm higher than those occupied by Spartina alterniflora (0.2-12 cm), while the lowest elevations (~0 cm) had no vegetation cover. The sandy barriers (~1-3 m height) were occupied by herbaceous plants, primarily Panicum amarum and Spartina spartinae (Figure 1c,d). In the northeast and southwest ends of the Bay Champagne, the muddy tidal flats occupied by mangroves and saltmarshes had gradual sand accumulation during the coastal barrier retrogradation. This process (sand accumulation) was coincident with a loss of~15.  Table 2). Mangroves were established during this period and the height of Avicennia shrubs reached up to 1 m tall, preferentially occupying tidal flats with saltmarshes along the backbarrier environment (Figure 7).

The Effect of the 2013 Beach Nourishment Project on Coastal Morphodynamics
Fluctuations in RSL and fluvial sediment supply to coastal systems can affect the dynamics of beach barriers. Several hydro-sedimentary processes interacting within spatial and temporal scales contribute to changes in beach barrier sediment volumes. The rate of shoreline advance or retreat is generally controlled by the equilibrium between sediment accommodation caused by relative sea-level change and sediment supply [55,56]. If the lower shoreface is shallower than that required for the coastal equilibrium, sand is relocated to the upper shoreface and shoreline would shift seaward. By contrast, the coast shifts landward when the lower shoreface is too deep. These transgressive and regressive shoreline movements may occur regardless of the eustatic sea-level [57][58][59][60][61][62].
In the Mississippi River, suspended sediment load has decreased by~50% since the 1850s, primarily due to the construction of >50,000 dams along the Mississippi basin [9,[63][64][65][66][67]. According to the Quickbird images, the most significant inland migration of the beach barrier (185-164 m) occurred during the period of 2004-2012 ( Figure 4, Table 1), and, according to the LIDAR data, the maximum inland dune crest migration was~220 m along the northeastern shoreline between 1998 and 2013 (Figure 3a, vertical profile c -d and e -f ). The volume of the studied barrier oscillated between 357,010 and 554,073 m 3 from 1998 to 2013. However, construction of the stone and rubber bulkheads in front of the barrier prior to 1998 likely have attenuated the impact of waves and currents along the southwestern studied coast (Figure 1b), causing an increase in the sand sedimentation and a decrease in the rates of shoreline retreat in the transect A (Figures 4 and 7a, Table 1). The effects of these artificial structures are evident along the studied beach barrier, since the largest inland dune crest and beach barrier migration occurred along the northeastern shoreline, farthest away from the bulkheads effects ( Figure 1, Figure 3a, and Figure 7a). By contrast, an advance of the beach barrier was recorded after the beach nourishment project, mainly between Nov/2012 and Jan/2015 ( Figure 4 and Table 1). In addition, the dune height and volume of the studied barrier increased from 1.  Table 2). Mangroves were established during this period and the height of Avicennia shrubs reached up to 1 m tall, preferentially occupying tidal flats with saltmarshes along the backbarrier environment (Figure 7).  Despite the beach nourishment project's effort to stabilize the beach barrier, the Quickbird images, recorded between Jan/2015 and Mar/2019, revealed shoreline retreat in the transects A and B and stability in transects C and D (Figure 4 and Table 1). In addition, our planialtimetric data, obtained by drone between Oct/2017 and Nov/2019, revealed erosion of sandy intertidal flat and part of the supratidal deposits, under the action of waves ( Figure 6). The most likely scenario is that these processes will be intensified as a result of future RSL rise to continue the erosion of supratidal and intertidal flats, increasing the rate of shoreline retreat. This indicates that the renourishment might have partially stabilized the beach barrier.
The magnitude of beach barrier retreat reported in the Bay Champagne is broadly comparable to Dietz Figure 5), may have been caused by the removal of beach sediments during some of these events (see also [12]). This hypothesis is supported by the landward migration of overwash fans behind the beach barrier between 2004 and 2012 (Figures 1, 4 and 8), which is typical evidence for storm surge activities [14,41,75,76]. In addition, the near absence of the beach barriers in front of Bay Champagne and the intense erosion of inland mangrove islands after Hurricane Katrina and Rita (2005), as shown by satellite images, attest to the impact of these events on this coast ( Figure 8).

Dynamics of the Coastal Barrier and Relationship to the Wetlands
The establishment and development of mangroves are mainly controlled by the interactions between fluvial discharge and tidal water elevations, as well as the local hydrodynamics of waves and tidal and/or littoral currents, coastal topography, and rates of availability and types of sediment [78][79][80]. The disturbance caused by RSL rise may trigger changes in sediment vertical accretion rates and salinity gradients, favoring or preventing the preservation of mangroves. They may also determine the reestablishment of mangroves in new areas of the same or higher topographic elevations [24,79,[81][82][83][84][85]. Since mangrove forests can keep pace with a RSL rise up to~6.1 mm/year [25,86], they mitigate the impacts of marine transgressions [22,[86][87][88]. Mangroves may also attenuate the velocity of wind and waves to protect the shoreline from erosion [89].
In our study area, Avicennia roots should have caused topographic gradients in the study area occupied by saltmarshes (0.2-12 cm) and Avicennia shrubs (13-47 cm) ( Figure 1 and Table S1 in the Supplementary Material), suggesting a higher sediment vertical accretion in areas occupied by Avicennia. Mangrove establishment causes positive feedback, because their structures dissipate wave and currents energy, triggering a micro-turbulence, which results in flocculation processes and fluvial mud trapping, and in turn, increases the substrate elevation [90]. This situation is ideal for mangrove regeneration [21,90,91]. Under high rates of RSL rise, deltaic mangrove substrates may still aggrade due to a combination of allochthonous and autochthonous sediment accumulation [24,80,92], even if the sediment input is low [93]. Hence, mangrove forests in our study area are expected to mitigate the effect of the high RSL rise, especially if adequate fluvial sediments are available to coastal systems, which would contribute to the stability of the beach barrier.
However, surface accretion rate can only keep pace with high RSL rise 10-20 mm/year scenario (RCP 8.5) up to year 2070 and 2055 in basin and fringe mangrove settings, respectively [94]. In particular, mangrove forests in areas of low sediment supply and low tidal range are more susceptible to degradation [92,95]. Considering these factors [86,92,96], the Port Fourchon mangrove is expected to be susceptible to degradation because: (1) they are under the influence of a microtidal regime (~0.3 m; whereas the RSL has risen by 9 mm/year in the study area: [97]; (2) the fluvial sediment input has decreased during the past decades [9,66]; (3) a series of hurricanes have impacted this coast [10]; and (4) successive extreme freeze events have caused mangrove mortality in several areas along the Gulf of Mexico [97][98][99][100], mainly in the Louisiana coast [101][102][103][104][105]. However, despite these challenges, mangroves inhabiting the muddy flats (mainly between 13 and 47 cm above mean sea-level) at Port Fourchon, have encroached tidal flats previously occupied by Spartina located between 0.2 and 12 cm above the mean sea level. This process has increased the mangrove area from 23 to 124 ha between 2004 and 2018 (Cohen et al., unpublished data). Similar mangrove encroachment has also been recorded at many other areas along the Gulf of Mexico [100,[105][106][107], which attests to the high resilience of this ecosystem.
In our study area, mangroves also occur behind the beach barriers, where they are protected from waves, currents, and the effects of shoreline retreat. The landward migration of the coastal barrier between 2004 and 2012 prevented the establishment and development of backbarrier wetlands. This process covered part of the muddy tidal flats and degraded~15.6 ha of wetlands ( Figure 8). The seaward advance or stabilization of the coastal barriers between 2013 and 2019, however, allowed the establishment of backbarrier mangroves (3.2 ha) and saltmarshes (25.4 ha), consisting mainly of Avicennia and Spartina ( Figure 8). Therefore, it is likely that the 2013 beach nourishment project has contributed to a more stable coastal morphodynamic environment for mangrove development, at least temporarily. In turn, the expansion of mangrove and saltmarsh increased the sediment accumulation, decreased the landward dune migration, and retarded the process of shoreline retreat (Figures 3 and 7). Hence, this positive feedback between strengthened beach barrier and backbarrier wetland has stabilized the shoreline in our study area under the present RSL rise. More importantly, such positive feedback between coastal ecosystem and restoration effort has been documented in other areas around the globe, where the coupling between engineered structures (e.g., seawalls, groynes, breakwaters, and beach nourishment) and planting Spartina and mangrove afforestation was most effective in reducing coastal erosion [107][108][109][110] and decreasing the costs of coastal defenses [111]. In this context, restored environments and healthy ecosystems are more productive and support a sustainable coastal and sea economy [112,113].
However, within a few years after the beach nourishment project, the action of waves and currents, intensified by continuing RSL rise, continued to erode the sea-facing coastal barriers, as recorded between Oct/2017 and Nov/2019. Hence, we believe that such positive feedback is temporary in the face of a RSL rise rate of~9 mm/year in the study area [97], and predict that the shoreline retreat will likely resume, and the mangrove and saltmarshes will be degraded through sand accumulation on the backbarrier muddy tidal flats currently occupied by wetlands. In the event of another breaching of the beach barrier, especially in front of Bay Champagne, the wetlands substrate will be exposed to the direct action of waves and currents. Mangrove and saltmarsh substrates will be eroded, and wetland areas currently at the supratidal and intertidal zones will turn into open-water environments. Therefore, the resilience of the studied mangroves will be challenged by the long-term trends of rising RSL, diminishing sediment supply, and the impacts of future coastal barrier retreat. Overall, this positive feedback, triggered by the RSL rise, will lead to an increase in the vulnerability of this ecologically and economically important American coast.
As demonstrated by many previous studies, the coastal morphodynamics in Louisiana are closely related to the sediment supply from the Mississippi River [1,2,[136][137][138]. One possible solution for combating the shoreline retreat at Port Fourchon will be to divert more sediments from the Atchafalaya River to the current Mississippi River Delta via the Old River Control Structure near Red River Landing, Louisiana [139,140]. However, this action will reduce the sediment supply to the Atchafalaya-Wax Lake delta complex [141]. The 2017 Coastal Master Plan by the State of Louisiana [26] proposed to divert sediments to the East Terrebonne Parish from the lower Atchafalaya. This coastal restoration plan could further reduce riverine sediment supply to the Chenier Plains situated in southwestern Louisiana, while the sediment load from the four nearby rivers is already declining (in total: 3.43 × 105 tons per year) [142]. Hence, Louisiana's coastal restoration efforts are facing a dilemma of either losing both sides of the delta or saving one side. In this context, the methodological flow used in this work could be expanded to other beach barriers along the Gulf of Mexico to broaden the assessment of the nourishment project effects on shoreline and coastal ecosystems for adequate decision-making to face coastal retrogradation. contributed to tidal flats keeping pace with RSL rise and increased stability of the beach barrier. However, even after the beach nourishment project, the action of waves and currents in front of the coastal barriers eroded part of the sandy intertidal flats between Oct/2017 and Nov/2019. This process will lead to the resumption of the shoreline retreat and future loss of wetlands area, increasing the vulnerability of this biologically and economically significant coast of the USA.

Conclusions
Coastal engineering projects have become increasingly important in the world's coastal zones [143][144][145][146][147]. While in some cases the coastal morphodynamic impacts of these engineering projects may have been considered as part of these engineering plans, quite often the ecological impacts of these human activities are not adequately documented. This paper presents a study from a rapidly retreating part of the Louisiana and Gulf of Mexico coastline to evaluate the effectiveness of such coastal nourishment project and especially its ecological impacts on coastal wetland and mangrove ecosystems. This dataset and the assessments will be useful for decision-makers in developing sound policies and measures to combat land loss and coastal erosion.