The effect of coastal landform development on decadal-to millennial-scale longshore sediment ﬂ uxes: Evidence from the Holocene evolution of the central mid-Atlantic coast, USA

The behavior of siliciclastic coastal systems is largely controlled by the interplay between accommo- dation creation and in ﬁ lling. Factors responsible for altering sediment ﬂ uxes to and along open-ocean coasts include cross-shore mobilization of sediment primarily from tidal currents and storms as well as changes in alongshore transport rates moderated by changing wave conditions, river sediment inputs, arti ﬁ cial shoreline hardening and modi ﬁ cation, and natural sediment trapping in updrift coastal land- forms. This paper focuses on the latter relationships. To address understudied interactions between updrift coastal landforms and downdrift coastal behavior, we quantify the volume and ﬂ uxes of sediment trapped in the Assateague-Chincoteague-Wallops barrier-island complex along the Virginia, USA coast and relate these volumes to downdrift coastal-system behavior. During the last ca. 2250 years, these barriers trapped 216 million m 3 of sand through the growth of complex beach- and foredune-ridge systems. A period ( ca . 400 to 190 years ago) of reduced/no progradation on Chincoteague and Assa- teague islands corresponds with sediment sequestration in updrift ﬂ ood-tidal deltas. This ﬁ nding emphasizes the important control of tidal inlets on alongshore sediment ﬂ uxes on barrier-island coasts. Rapid historical spit elongation during the last 190 years has trapped an average of 681,000 m 3 yr (cid:1) 1 of sand; this occurred coincident with downdrift barrier-island erosion/migration at long-term rates of > 3 m yr (cid:1) 1 . Historical sand ﬂ uxes to the elongating spit on southern Assateague Island and progradational beach ridges on northernmost Wallops Islands are equivalent to at least 60% of estimated regional longshore transport rates. We therefore propose that sediment trapping and associated wave refraction are the primary drivers of downdrift barrier erosion, while storminess and sea-level rise are secondary forcings of change affecting equally the entire barrier-island chain. Global context is provided by a compilation of sediment trapping through growth of similar longshore sand sinks, which indicates the volume of sediment incorporated into the elongating spit end of Assateague Island is similar to sandy beach- and foredune-ridge plains (10 8 m 3 ), but average annual trapping at the spit is at least six times greater than those at most mainland-attached, progradational systems. However, Chincoteague and Wallops, two progradational barrier islands, incorporate sand at rates broadly similar to large strand- plains. Our ﬁ ndings emphasize the need to account for natural longshore sediment trapping in multi-decadal coastal management efforts on sandy, siliciclastic coasts. © 2021 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY (http://creativecommons.org/licenses/by/4.0/).

As demonstrated by these diverse studies, sediment trapping at groins, inlets, and ebb-tidal deltas is relatively well-understood. However, the role of the growth or erosion of natural updrift siliciclastic landforms, such as the filling of coastal embayments, erosion of sandy headlands, and progradation/elongation of barrier islands/spits remains understudied (see Hein and Ashton, 2020). Specifically, the formation and collapse of updrift sediment sinks can regulate downdrift sediment fluxes and play a predominant role in the behavior of sandy coastal landforms (Anthony, 1995(Anthony, , 2015Park and Wells, 2007;Fruergaard et al., 2019Fruergaard et al., , 2021Oliver et al., 2020), but the dynamics, volumes, and fluxes of sediment trapped and mobilized are poorly constrained.
What are the magnitudes of sediment trapping at progradational barrier islands and elongational spits and what are associated effects on longshore transport and the behavior of the downdrift coast? To answer this question we integrate millennial, centennial, and decadal records of sediment fluxes trapped in the progradational-elongational Assateague-Chincoteague-Wallops barrier-island system (northern Virginia, USA; Fig. 1) to reconstruct the multi-kilometer-scale development and morphodynamics of this sandy coastal sediment sink. Additionally, we integrate geospatial, sedimentological, and geochronological data to calculate volumetric storage and explore the drivers of time-varying fluxes of sediment trapped in the system. We place sediment storage at this site in context with other potential drivers of downdrift coastal erosion including sea-level rise, increased storm frequency, variations in framework geology, and an alongshore transport gradient. Finally, we compare the sediment stored in this system to estimates of sediment stored in similar globally distributed coastal depositional landforms to ascertain the global applicability of these findings.

Coastal setting
The Assateague-Chincoteague-Wallops barrier-island system is located on the central U.S. mid-Atlantic coast, surrounding the north-south-oriented Chincoteague Inlet (Fig. 1). Landward of these barriers is the Virginia portion of the Delmarva Peninsula, composed of unconsolidated coastal plain sediments deposited during at least five interglacial highstands (Krantz et al., 2016). The Assateague-Chincoteague-Wallops system is situated at the northto-south transition from long, linear wave-dominated barrier islands of southern Delaware and Maryland, to the shorter, mixedenergy Virginia Barrier Islands. Tropical and extratropical cyclones drive the southerly regional sediment transport system (Fenster et al., 2016). Regional extratropical cyclone frequency has increased from the late 1800se1990s (Hayden and Hayden, 2003) and hurricane-generated wave heights have increased since the 1970s (Komar and Allan, 2008). Based on records from 1978 to 2015 CE, two local tide gauges record mean tidal ranges of 1.23 m (Wachapreague, VA; 30 km south of the study site) and 0.64 m (Ocean City, MD; 50 km north of the study site; . Local wave data from 1980 to 2012 CE indicate dominant wave heights of 1.2 m and mean wave periods of 8.3 s near Assateague Island . Chincoteague Inlet, which separates southern Assateague Island and northern Wallops Island has an approximately north-south orientation and conveys water and sediment between the Atlantic Ocean and the southern portion of Chincoteague Bay and the northern backbarrier lagoon of Wallops Island (Beudin et al., 2017). The inlet is approximately 1.8 km wide, with an 8e12 m deep main channel located on the western margin of the inlet and a shallow (~2 m deep) platform/shoal located to the east of the main channel (McPherran et al., 2021). The inlet has a mean tidal range of 0.66 m, average significant wave height of 1.11 m, and a 44 million m 3 tidal prism (Jarrett, 1976). During strong wind events, changes in wind direction and intensity drive exchange between Chincoteague Bay and the Atlantic Ocean; during weak wind events, tidal forcing dominates exchange processes (Kang et al., 2017). For example, inflow to Chincoteague Bay through Chincoteague Inlet increases during periods of strong southwest winds, while outflow increases with northwesterly winds (Kang et al., 2017). Following major storms, freshwater exits Chincoteague Bay through the inlet, with velocities capable of mobilizing sediment (McPherran et al., 2021).

Morphology of Assateague, Chincoteague, and Wallops islands
The present-day morphologies of the Assateague-Chincoteague-Wallops barrier-island system attest to the spatial and temporal complexity of processes operating along this reach. Despite this complexity, the presence of beach-and foredune-ridge plains across islands provides a means for unraveling the evolutionary history of this region, determining barrier-system morphodynamics, and inferring causative mechanisms for individual landform formation.
At 58 km in length, the wave-dominated Assateague Island is among the longest islands on the U.S. East Coast. The series of recurves which comprise the southern portion of the island are marked by variably-oriented beach and foredune ridges (generally 1e3 m high; Fig. 2). The tallest such ridge ("Lighthouse Ridge") exceeds 8 m in height and is the location of the ca. 1830 CE shoreline, where a coastal navigation light was originally constructed in 1833 CE. Since around 1830 CE the island has elongated to the south nearly 7 km as multiple beach-and foredune-ridge sets, including "Fishing Point" (or "Tom's Cove Hook") joined to Assateague proper by the narrow (~500 m wide) Tom's Cove Fig. 1. Study area and field data collected from the Assateague-Chincoteague-Wallops barrier-island system A) The southern Maryland and northern Virginia barrier-island coast. Assateague Island is~58 km long, with the southern terminus consisting of progradational beach-and-foredune ridges. South (downdrift) of southern Assateague is a shoreline offset, described locally as an "arc of erosion". Imagery from Esri. B) Study site showing locations of~34 km of ground-penetrating radar tracklines, 24 direct-push Geoprobe (~20e24 m long) cores (this study), 15 auger and hydrologic test drill cores (Halsey, 1978;Goettle, 1978), seven vibracores (0.6e5 m long; this study), optically stimulated luminescence sampling sites (hand augers; this study), and primary stratigraphic cross-sections. C and D) Panels show additional detail and core labels. Where human disturbance is minimal, distinct ridge and swale topography is evident on Chincoteague, Assateague, and Wallops islands. The southern terminus of Assateague is marked by curvilinear beach and foredune ridge features formed by spit elongation. DEM from lidar (USGS, 2016).
Isthmus. The southern spit-end grew at a long-term (1800se2000) rate of 21.5 m yr À1 and an average near-term (1970e2000) rate of 40.3 m yr À1 (Hapke et al., 2010). The overall system morphodynamics are more complicated and include areas of erosion and variable alongshore progradation rates . Today, the U.S. Fish and Wildlife Service manages the southern portion of the island as a wildlife refuge with light recreational uses permitted. Because of the interruption of longshore sediment transport by the Ocean City Inlet jetties to the north, the north-ernmost~10 km of Assateague Island (located > 40þ km from our study site) has been nourished four times since 1998 CE (Campbell and Benedet, 2006;ASBPA, 2020;Elko et al., 2021). By contrast, Ocean City, Marylanddlocated immediately north of Assateague and Ocean City Inlet dhas been nourished fourteen times since 1963 CE with seven of those nourishment projects occurring since 1998 CE (Campbell and Benedet, 2006;ASBPA, 2020;Elko et al., 2021).
Chincoteague Island, today located entirely landward of Assateague Island, was once an open-ocean barrier. The Pocomoke Native Americans originally hunted and fished on the island, which was later occupied by Europeans beginning in the mid-17th century. Eventually, Chincoteague evolved into a fishing village by the late 19th century, and today houses a growing tourist economy. A series of variably oriented (WSW-ENE and SW-NE) arcuate beach and foredune ridges (<1e2 m high) exist on Chincoteague Inlet in areas undisturbed by development ( Figs. 1 and 2). The island varies in width from~500 to 1500þ m. Immediately to the east, Chincoteague is fronted by Piney Island, which is either a former part of Chincoteague Island or a separated recurve of Assateague Island (see Goettle, 1978).
South and west of Chincoteague Inlet, the more than 15 km long Wallops Island possesses a bulbous northern end (~1500 m wide) and progressively narrower (~300 m wide) southern terminus. The northernmost end contains a series of arcuate beach and foredune ridges oriented SW-NE, most of which formed during the late 20th and early 21st centuries. An inlet separating Wallops Island from Assawoman Island (immediately to the south) closed in the mid-1980s in response to an engineered reduction in tidal prism . Since 1945 CE, the National Advisory Committee for Aeronautics and its successor, the National Aeronautics and Space Administration (NASA), has owned and managed Wallops Island. Consequently, the majority of Wallops Island has experienced heavy engineering and human alterations, including most recently construction of a seawall, multiple episodes of beach nourishment (years: 2012 CE, 2014 CE ), and the in-progress (2021 CE) construction of shore-attached breakwaters backfilled with sediment removed from the northern end of the island. The need to protect infrastructure from the complex hydrodynamic and sediment transport processes at Chincoteague Inlet and the refraction of waves around the southernmost portion of Assateague Island drives the human management of this system. This wave refraction creates a nodal zone which moves sand both southward and northward at a maximum flux of~46,000 m 3 yr À1 (King et al., 2011), and feeds the rapidly growing northern end of the island (up to 16 m yr À1 ; Hapke et al., 2010). The sediment transported to the south of this nodal zone plays a major role in supplying the southern Virginia Barrier Islands with beach-quality sand. However, modeling results (Generalized Model for Simulating Shoreline Change) indicate that net sediment transport fluxes decrease rapidly to the south from the nodal point and that the southerly flux gradients correspond closely with shoreline retreat rates (King et al., 2011).

Holocene development and sea-level history
The Holocene age barrier-backbarrier systems along the Virginia and Maryland coasts developed atop Pleistocene-aged estuarine sediments, likely deposited during Marine Isotope Stage 5 (Goettle, 1978). The Holocene system consists of silty lagoonal deposits and sandy marine shoal/shoreface, barrier, and dune-ridge sediments (Goettle, 1978). Using 12 auger borings, three uncalibrated radiocarbon dates, and three amino acid racemization age estimates, Goettle (1978) proposed that the Assateague-Chincoteague-Wallops barrier system developed first with Chincoteague Island as an open-ocean barrier ca. 2000 years ago, followed by southerly growth of Assateague Island seaward of Chincoteague in a series of seven distinct phases, as exemplified by the development of associated ridgesets. Seminack and McBride (2015a) refined the model to incorporate the opening and closure of multiple inlets along the entire~58 km length of Assateague.
Relative sea level on the Virginia coast rose 8.5 m over the last 5000 years, at an average rate of~1.5 mm yr À1 between 5000 and 4000 years ago until it slowed moderately to~1.3 mm yr À1 from 4000 years ago to 1900 CE (Engelhart and Horton, 2012). Preservation of backbarrier deposits and the development of the Holocene barrier-island system along the Virginia coast coincided with this rise in sea level over the middle to late Holocene (Finklestein and Ferland, 1987;Raff et al., 2018). Over the late 20th century to present, regional sea level rose at a rate of 3.5e5 mm yr À1 (Boon and Mitchell, 2015).

Paleo-shoreline mapping
Historical shoreline-change data from southern Assateague Island are derived from Hein et al. (2019b). We mapped pre-historical shoreline positions from beach and foredune ridge morphology using modern satellite imagery and USGS (2016) topobathy lidar. We obtained additional geochronology from either historical records (georeferenced historical maps and NOAA t-sheets) or optically stimulated luminescence (OSL) dating.
We collected four samples for OSL analysis from wave-built lithofacies using methods described by Oliver et al. (2015). These were analyzed at the OSL dating laboratory at the University of Illinois, Urbana-Champaign ( Table 2). The PVC pipe cores were opened, and the mineral extraction was conducted in a subdued orange light environment. Ten centimeters of sediment was removed from the bottom (to avoid any sediment unintentionally exposed to sunlight during collection). The actual OSL sample was extracted about 15 cm from the bottom of the core and treated with 10% hydrochloric acid (HCl) and 20% hydrogen peroxide (H 2 O 2 ) to remove carbonate and organic material. The sediment was then dried and sieved to separate grains in the~2.7e2.0 phi size range (150e250 mm). Quartz grains from these samples were extracted and none showed any significant contamination from feldspar. For the equivalent dose (De) measurements, we relied on an automated Lexsyg Smart system (Richter et al., 2015), and measurements were carried out with a single-aliquot regenerative dose (SAR) protocol Wintle, 2000, 2003). To obtain the dose rate, sediments from the bottom section of each core were dried, a representative portion was encapsulated in petri dishes (~20 g) and sealed with two coatings of epoxy gel. The specific activities (Bq/kg) were measured with a broad-energy high-purity germanium (BEGe) detector, in a planar configuration. Further details can be found in the supplementary materials.

Ground-penetrating radar
A total of~34 km of predominantly cross-shore and shoreparallel ground-penetrating radar (GPR) surveys were conducted across all three islands using a 250 MHz MALA Geosciences shielded antenna. GPR lines were topographically corrected using elevation profiles derived from USGS (2016) topobathy lidar. All GPR data were post-processed following the methods of Carruthers et al. (2013) using site-specific filtering, migration, and variable gain control and time-depth converted using a migration-derived radar velocity of 7 cm/ns using DECO-Geophysical Co. Ltd.'s RadExplorer software program. The radar velocity value was selected from established values for saturated sandy barrier sediments (e.g., Oliver et al., 2019a) given the presence of a shallow (<50 cm) fresh groundwater table at all sites.

Onshore and offshore stratigraphy
We collected 24 direct-push (Geoprobe) sediment cores (20e24 m deep) across all three islands ( Fig. 1) to characterize their pre-Holocene and Holocene stratigraphy. Additionally, we collected seven vibracores (0.6e5 m deep) in Tom's Cove, the shallow lagoon landward of the southern Assateague isthmus. All cores were opened, photographed, and described using visual standards for sediment grain size, texture, mineralogy, and color (Munsell). Grain-size distributions of select core samples (based on facies descriptions) were determined using a Beckman-Coulter Laser Diffraction Particle Size Analyzer. Samples were run in triplicate, and data are reported as averages of the triplicates. Geochronological control of stratigraphic layers is provided by radiocarbon samples from various depths in direct-push cores. Accelerator mass spectrometer radiocarbon analyses of 11 shell, peat, and terrestrial root samples were performed at the National Ocean Sciences Accelerator Mass Spectrometry Facility (NOSAMS; Woods Hole, MA, USA). All radiocarbon ages, including two additional ages from prior work (Halsey, 1978;Goettle, 1978), were calibrated using OxCal 4.4 (Bronk Ramsey, 2009), which includes a reservoir correction. Terrestrial samples (peat, roots) were calibrated with Intcal20  calibration curves and marine samples (all mollusks) were calibrated using Marine20 , corrected to a DR of À33 ± 40 years (Rick et al., 2012).
To investigate the offshore stratigraphy and characterize the mobile shoreface sediments, 38 km of high-resolution (submeter) seismic data were collected at~0.5e1.0 km line spacings using an Applied Acoustics AA300 boomer (operated between 150 and 300 J) seismic system with a CSP300 seismic energy source and 4.5 m long Applied Acoustics hydrophone streamer with 8 elements. A Trimble DSM 132 dGPS marine positioning receiver and antenna enabled merging real-time digital geographic positions to digital SEG-Y seismic data. This system provided submeter horizontal positional accuracy and up to 60 m of penetration based on an assumed sound velocity of 1500 m/s. Chesapeake Technologies' SonarWiz software version 7 was used for data acquisition, processing, and interpretation. Seismic facies were interpreted based on a seismic facies analysis (Sangree and Widmier, 1979) and seismic facies correlation with previous studies from this same region (e.g., Toscano et al., 1989;Wikel, 2008;Brothers et al., 2020).

Sediment volumes and fluxes
Sediment volumes were determined for Assateague, Chincoteague, and Wallops islands by multiplying the area of each island by an average thickness of the barrier island. Modern areas were mapped using recent basemap imagery in ArcGIS 10.6 (Esri) and historical areas were mapped using shoreline and area data compiled from previous publications (e.g., Deaton et al., 2017;Hein et al., 2019b) and additional historical maps and charts. To calculate barrier thickness, the average height of undisturbed ridges and swales above North American Vertical Datum of 1988 (NAVD 88) was measured on each island from a high-resolution lidar DEM and an estimated average depth below NAVD 88 was determined from sediment cores. For each paleo-shoreline a conservative estimate of foreshore volume of was calculated by multiplying the shoreline distance by the area of a simplified right-triangle-shaped wedge of foreshore sand. Estimates of foreshore maximum thickness (5.5 m Assateague, 5.3 m Chincoteague, 6.5 m Fishing Point, and~3 m Wallops) and cross-shore length (75 m for all) were determined from modern foreshore bathymetry (USGS, 2016) and are consistent with sediment core data from this study. We acknowledge that a singular thickness value for the barrier and foreshore does not capture potential variability in ridge and swale topography or in subaqueous depth, but, by far, the largest factors controlling island volume estimates are island area and shoreline length, respectively. Fluxes were calculated by dividing total barrier and foreshore volumes by the period of deposition. Additionally, volumetric estimates of mixed mud and sand shoreface deposits were calculated using the same approach.

Shoreline geochronology and progradation
A combined dataset of OSL dates (Table 1; Fig. 3) and historical shoreline data (Fig. 3) provide geochronological control for the physical evolution of Assateague, Chincoteague, and Wallops islands. The westernmost ridge on Chincoteague Island formed ca. 2250 years ago, while the eastern-most ridge formed ca. 400 years ago (~1620 CE; Fig. 3; Table 1). An OSL date from a ridge on Assateague Island, at the northernmost limit of our study site, returns an age of ca. 120 years ago (~1900 CE), approximately coincident with the closing of an updrift inlet  and, to the south, growth of Lighthouse Ridge adjacent to the ca. 1830 CE shoreline.
Assateague Island prograded in a generally southeast direction from 1830 CE to 1859 CE but began a south to southwesterly elongation in the late 1800s, which has continued to present (Fig. 3). The northern end of Wallops Island prograded in a net easterly to northeasterly direction between the late 1800s and present. Central Wallops largely eroded or remained stable during that same period (Fig. 3).
Shoreline progradation is evident not only in historical maps, but also in GPR profiles from Chincoteague and central Assateague islands (Fig. 4). On Chincoteague, a combination of seawarddipping (~1.5e3.0 ) subsurface reflections imaged in radar profiles and ridge ages that are progressively younger to the east indicate that the island built seaward from ca. 2250 to ca. 400 years ago. A similar set of observations from Assateague Island, albeit with a much younger set of shorelines (1830 CE to present), likewise reveals the progradational development of that system.

Stratigraphic units and interpretations
Sedimentological data (Figs. 5e7), radiocarbon dates (Table 2), and seismic facies and interpretations ( Fig. 8) together inform the primary stratigraphic units within the Assateague-Chincoteague-Wallops barrier-island system and the proximal offshore region (Tables 3 and 4). Seismic facies interpretations and seismic unit identification are based on newly obtained nearshore seismic data, ground-truthed with onshore sediment cores obtained in this study and correlated to existing offshore seismic data of Brothers et al. (2020).

Pleistocene shallow marine
The lowermost unit is composed of a shelly, glauconitic, greenish-gray (Munsell color: Gley 1 5GY 5/1) sandy clay to silty or clayey very fine sand (median grain size: 0.02e0.19 mm; Table 3). The clay content is highly variable (sorting: 34) and shell hash is abundant within this facies. This unit is encountered in cores from Chincoteague Island and the central and northern portion of the study site on Assateague Island (depth: À15 m MSL and deeper) but was not observed underlying Wallops Island and Fishing Point.
This facies most likely formed in a shallow marine depositional environment. Radiocarbon dates (one >45,000 calibrated years before present [cal. Yrs BP]) located stratigraphically above this unit indicate a Pleistocene or later age for this unit. Goettle (1978) interpreted this unit as being the mouth of an estuary or an inlet possibly built during one or more sea-level highstands.
The equivalent seismic unit is characterized by both gently seaward (0.4e0.9 ) and southerly (0.06e0.15 ) dipping reflections but is reflection free in places (Fig. 8). The unit is likely equivalent to unit Qpp observed and described by Brothers et al. (2020) as a Pleistocene transgressive system tract which fills the Persimmon Point paleochannel (active during Marine Isotope Stage 22, 16, or 12; Table 4). This interpretation is consistent with the seaward and southerly dipping to absent internal reflections seen in our seismic data and the likely marine origin indicated by our sediment-core data.

Pleistocene estuarine
A fining-upward facies (~À10 to À15 m MSL) directly overlies the lower-most unit observed in these cores. Under Wallops and Chincoteague islands, this is the deepest unit observed (Figs. 6 and 7). This unit transitions from a basal coarse sand with pebbles, some shell fragments, and some whole shells to a laminated silty clay and very fine sand. The upper portion of the unit alternates between sections of flaser and lenticular bedding (average sorting: 24) ( Table 3). The coarse-grained basal section contains abundant pebbles (>2 mm); the finer than 2 mm subsample has a median grain size of 0.25 mm. The finer-grained uppermost section has a median grain size of 0.01e0.20 mm. Color ranges from a grayish brown (coarse section; Munsell color: 10 YR 5/2) to dark gray (fine section; Gley 1 4/N).
Mottling and oxidation at the topmost unit (i.e., evidence of possible subaerial exposure) as well as an "infinite" radiocarbon age (>45,000 cal Yrs BP) of a sample from the top of this unit indicate a Pleistocene age for the facies. Therefore, we interpret this unit as representing a Pleistocene estuary which transitions up-section from greater terrestrial/fluvial influence (coarse grained) to more marine/tidal influence (fine grained; laminated).
The corresponding seismic unit is low-amplitude to reflectionfree ( Fig. 8; Table 4). The base of the seismic unit is a highamplitude, continuous gently-seaward-dipping (0.03e0.05 ) to planar (i.e., 'flat') reflection at a depth of 15e20 m below MSL ( Fig. 8; Table 4). This seismic facies corresponds to the Q2 unit described by Brothers et al. (2020) as a highstand systems tract of estuarine and marine origin, while the lower contact of the unit corresponds to the Pleistocene-age U7 unconformity (transgressive

Transitional peat/gyttja/soil
Situated stratigraphically above the fining-upward unit is a black to very dark grayish brown (Munsell colors: 10 YR 2/1 and 10 YR 3/2) organic-rich silty soil to peat up to 50 cm thick, but typically observed as <25 cm thick with a median grain size of 0.035 mm (Table 3). This unit is laterally discontinuous from the mainland to the ocean on Chincoteague Island and in the central and northern portions of our Assateague Island study area (Fig. 6a). It is entirely absent in the cores from Wallops Island and Fishing Point (Fig. 6b). Radiocarbon ages of samples from the top (6400e6200 cal Yrs BP) and the bottom (>45,000 cal Yrs. BP) of this unit indicate that it is likely a transitional peat or gyttja formed at the leading edge of Holocene transgression.

Holocene lagoon
This unit, which unconformably overlies the transitional and/or Pleistocene deposits under Chincoteague and Wallops islands and landward of the isthmus on southern Assateague, is composed of dark gray (Munsell color: 10 YR 4/1) clayey silt to silty clay with very fine sand lenses; snail shells (Urosalpinx cinereal) and shell hash are rarely observed (Table 3). The median grain size of samples from this facies ranges from 0.01 to 0.03 mm and sorting is~24. In Tom's Cove (landward of the narrow isthmus on southern Assateague) a series of shallow (0.6e5 m) vibracores reveal the uppermost stratigraphy of this unit (see Fig. 1 for locations). Proximal to the isthmus, the sediment is dominantly fine to medium sand with lenses of silt, silty sand, and/or coarse sand. Distal from the isthmus, towards the center of the cove, the sediment is dominantly clayey silt with lenses of very fine to fine sand (À5.2 to À2.5 m msl), while towards the modern inlet (western portion of Tom's Cove) the sediment in this unit coarsens upward from clayey silt (À5 to À3.5 m MSL) to silty sand and sandy silt (À3.5 to À2 m MSL). Halsey (1978) observed this unit landward of Chincoteague Island, where it is up to 10 m thick and contains saltmarsh peat in the uppermost meter (Fig. 5). The unit is up to 5 m thick underlying the landward-most portions of Chincoteague and Wallops islands but thins seaward and disappears entirely under easternmost Chincoteague and central Fishing Point (Fig. 6).
Radiocarbon dates of samples collected from within this unit (4300e2200 cal Yrs. BP) and at its base (6400e6200 cal Yrs. BP) indicate a middle Holocene to modern age. Thus, we interpret the sediments in this unit as deposited in a Holocene backbarrier lagoon, with the uppermost portions of this unit landward of modern Chincoteague, Wallops, and Tom's Cove Isthmus continuing to accumulate sediment to the present time.

Holocene shoreface and foreshore
Overlying and seaward of the dominantly muddy Holocene lagoon is a heterogenous mixed sand and mud unit. Specifically, this unit contains gray (Munsell color: 10 YR 5/1) dominantly fine to very fine sand (median grain size: 0.11e0.30 mm) with some sandy silt, lenses of pure silt and rarely observed clay with abundant dwarf clam shells (Mulinia lateralis) and shell hash ( Table 3). The unit generally coarsens upward from muddy very fine sand to muddy to clean fine sand in all cores. Samples from the facies are moderately to poorly sorted (average sorting: 0.94). However, the unit is laterally variable with the shallowest (À5 m MSL or shallower) and thinnest deposits (3 m or less) observed on westernmost Chincoteague and northern Wallops Island, and the deepest (À15 m MSL) and thickest (6e7 m) observed under Fishing Point/   Fig. 1 for location), modified from Halsey (1978). Backbarrier lagoon and saltmarsh sediments (together up to 10 m thick) are located landward of Chincoteague Island, a relict progradational barrier island. Assateague Island, an elongational barrier island/spit, is located seaward of Chincoteague Island. Note multiple changes in orientation given by the "breaks" along the profile.
Tom's Cove Hook (Figs. 6 and 7). The uppermost sections of this unit are resolved in vibracores from central and western Tom's Cove at depths of À5.1 m MSL and deeper.
Radiocarbon dates of shell samples collected from this unit under Fishing Point (modern to 300 cal Yrs. BP, beneath a subaerial beach formed within the last 100 years) indicate recent deposition and likely reworking of the sediment. When paired with sediment texture data and the coarsening-upward trend, this indicates a shoreface depositional environment possibly characterized by migrating sand ridges, such as those found offshore of modern southern Assateague (see Pendleton et al., 2017). The coarseningupward trend commonly reflects a transition from lower to upper shoreface depositional environments (see Rodriguez et al., 2001;Timmons et al., 2010;Hollis et al., 2019).
This unit extends offshore as the modern Holocene cover of shoals (i.e., shoreface ridges) and sand sheets (Fig. 8, Table 4). In seismic profiles, the unit is marked by medium amplitude, medium frequency, parallel, sheet-like reflections that generally conform to the seafloor morphology and vary from 5 to 15 m in depth and 5e10 m in thickness. It corresponds to the Qmn unit of Brothers et al. (2020), described as a highstand systems tract composed of the modern sandy shoreface. The thickness of this unit increases to the south and east near the southernmost tip of modern Assateague Island (Fig. 8). This spatial variation in thickness is consistent with a Holocene sand thickness map from Wikel (2008) as well as historical bathymetric data which indicate the presence of high-relief shoals offshore of southern Assateague (Fig. 9).

Holocene channel and inlet
This unit contains medium amplitude, low frequency, concave up reflections ( Fig. 8; Table 4). While this unit is not encountered in cores from this study, it is consistent in depth (À20 m MSL and deeper) and seismic character with unit Qcch described by Brothers et al. (2020) as channel-fill deposited since the Last Glacial Maximum. As such, it is likely Holocene-age channel and/or inlet fill, possibly (but not definitively) related to the inlet observed by Halsey (1978) underlying Assateague Island near Lighthouse Ridge (see Fig. 5).

Holocene barrier island and dune
The uppermost stratigraphic unit found across all three islands is composed of yellowish brown (Munsell color: 10 YR 5/4) fine to medium sand (median grain size: 0.19e0.52 mm; average sorting: 0.64). Radargrams collected from central Chincoteague and Assateague islands show evidence of abundant seaward-dipping (1 e 5 to the southwest) clinoforms in this unit. On Assateague, the presence of chaotic internal GPR reflections differentiate aeolian deposition (i.e., reworking and homogenization of material by aeolian processes) from the underlying beach subunit. However, on Chincoteague, no chaotic internal GPR reflections are observed, as anthropogenic road fill has disturbed aeolian sediments at sites where GPR data were collected (Fig. 4). Data from GPR, shoreline chronology, and sediment cores together indicate that this sandy unit was formed as a series of progradational beach and foredune ridges which comprise the relict (Chincoteague) and modern (Assateague and Wallops) open-ocean barrier islands.

Sediment volumes and fluxes
Paired analysis of time-varying barrier-island aerial change (OSL dates, shoreline analysis) and barrier thickness (sediment core data) allow for calculation of both fluxes and total volumes of beach and foredune ridge sand captured in this multi-island system, as well as total volumes of the mixed sand and mud shoreface Holocene barrier island and dune fine to coarse sand; subunits of variable organic content, texture, and grain size include beach, washover, dunes, and interdune swales; swales contain organic-rich silt and wetland species À7.5 to surface shoreface, beach, dune, washover, and inlet   Island cross-section is oriented entirely shore-parallel, while the Assateague Island cross-section changes orientation from shore-normal (0e~3.5 km) to shore-parallel (~3.5e10.5 km). Note: Internal barrier-island bedding is artistically inferred from available groundpenetrating radar profiles.
deposits (Table 5). Chincoteague, Assateague, and Wallops contain 81 million, 127 million, and 7 million m 3 of sand preserved through beach and foredune ridge progradation, respectively. Average sand fluxes trapped during active progradation range from 40,000 m 3 yr À1 (Chincoteague and Wallops) to 681,000 m 3 yr À1 (Assateague). On Chincoteague, the relict shoreface unit contains 69 million m 3 of sediment (sand and mud), while Assateague and Wallops preserve 108 million and 4 million m 3 of shoreface sediment, respectively. Over the last ca. 120 years, Fishing Point trapped dominantly fine to medium sand with some beds of coarse to very coarse sand (see Supplemental Materials Fig. S1). Progradation of the barrier deposits preserved underlying shoreface deposits which contain dominantly fine sand but also very fine sand, silt, and clay of varying content (see Supplemental Materials Fig. S1).

Quaternary deposition at Assateague, Chincoteague, and Wallops islands
We interpret the pre-Holocene and middle Holocene to modern depositional history of the Assateague-Chincoteague-Wallops barrier-island system using an integrated analysis of geochronology, sediment-core data, seismic stratigraphic data, and volume and flux reconstructions (Fig. 10).

Pre-Holocene deposition at Assateague, Chincoteague, and Wallops islands
The Quaternary geologic evolution of the Delmarva Peninsula and proximal continental shelf is marked by a series of sea-level highstands and lowstands which together resulted in a sequence of welded transgressive and regressive coastal barrier, nearshore, estuarine, and fluvial deposits (e.g., Mixon, 1985;Ramsey, 2010;Brothers et al., 2020). The Persimmon Point paleochannel underlies the modern Assateague-Chincoteague-Wallops barrier system and likely formed during Marine Isotope Stage 22 (~866 ka), 16 (~676 ka), or 12 (~478 ka) by the flow of the Susquehanna and Potomac rivers (Brothers et al., 2020). While data from this study do not resolve this subaerial unconformity, our seismic data and cores do indicate that infilling of this paleochannel occurred during the Pleistocene, likely in the form of a transgressive systems tract (Brothers et al., 2020) composed of primarily marine and estuarine deposits. A Pleistocene-age transgressive ravinement surface overlies this feature (U7 per Brothers et al., 2020) and the base of the overlying unit contains coarse, reworked sediment, possibly of fluvial origin. The youngest Pleistocene deposits underlying the Assateague-Chincoteague-Wallops barrier system consist of estuarine sediments and represent a highstand systems tract.

Stage I e Migration of Chincoteague/Wallops: Wave erosion of backbarrier deposits
Starting 6000 years ago, initial backbarrier deposition occurred in the proto-Wallops and Chincoteague backbarriers (Fig. 10a), consistent with data from other barrier-backbarrier systems across the Maryland and Virginia coasts (e.g., Finkelstein and Ferland, 1987;Raff et al., 2018;Shawler et al., 2021). Preservation of backbarrier deposits is coincident with an apparent deceleration in relative sea-level rise from~2 to~1.5 mm yr À1 (Engelhart and Horton, 2012;Raff et al., 2018). If barriers existed earlier, they likely were situated much farther offshore, and the shelf would contain the only preserved deposits associated with proto-and incipient barriers (Swift, 1975;Finkelstein and Ferland, 1987).
Transgression dominated the net behavior of the barrier system from ca. 6000 to 2250 years ago. Tidal inlet and overwash processes led to net barrier rollover. While periods of barrier stabilization or even progradation may have occurred, they are not recorded in Table 5 Field site volumes and fluxes.   In all panels, the darkest gray features are oldest, and the lightest gray features are youngest. A) Stage I: Prior to~2.25 ka, net landward migration of the barrier system occurred in a regime of slow relative sea-level rise (~1 mm yr À1 ); B) Stage II: Around 2.25 ka Wallops was located offshore of its present location, and the initial progradation of Chincoteague was driven by high alongshore and onshore sediment fluxes, primarily via onshore migration of bars. A nodal zone of diverging longshore transport directions developed on Chincoteague, controlled by wave refraction around the these data. As the barriers migrated landward, waves and currents eroded backbarrier sediments exposed on the shoreface, leading to preservation of relict backbarrier deposits which pinch out in a seaward direction. Given the net landward migration of the barrier systems, only these backbarrier deposits are preserved.

Stage II e Stabilization of Chincoteague Island: Sediment supply as a control on barrier state changes
About 2250 years ago, Chincoteague Island stabilized at or near the location of its western-most ridge and Wallops Island was offshore of its present location (Fig. 10b). Myriad factors likely contributed to the timing and location of the stabilization of Chincoteague Island. Complex antecedent morphology, as explored in depth as a factor in barrier-island migration by Shawler et al. (2021), is an unlikely control here, as the antecedent surface underlying Chincoteague and Wallops islands is largely planar and lacks the elevated surfaces upon which barrier islands to the south were pinned. However, shoreface sediment fluxes may have played an important role. For example, many of Virginia's barrier islands are particularly sensitive to subtle changes in sediment delivery rates, which may control state changes between landward migration and stabilization/progradation . A range of factors can control variations in cross-shore and longshore sand fluxes along this coast, including ebb-tidal delta configuration (Fenster et al., 2016), inlet sediment bypassing (Fenster and Dolan, 1996), backbarrier sand trapping McBride, 2015b, 2019), excavation of antecedent sediment (Raff et al., 2018;Shawler et al., 2021) by ephemeral tidal inlets, and modified wave refraction patterns and sediment trapping associated with the growth of updrift spits and ebb-tidal deltas (Fenster et al., 2016;Jones, 2016). An unknown combination of such processes likely temporarily increased sediment fluxes to Chincoteague Island around 2250 years ago and led to the associated transition from retrogradation to progradation of the island. Additionally, this period of island stabilization corresponds to a decrease in storminess and likely associated reduction in overwash fluxes and island dissection/ breaching, as observed at other U.S. East Coast barrier sites (e.g., Mallinson et al., 2011).

Stage III e Progradation of Chincoteague Island: Formation of a large Holocene sand deposit
Initial progradation of Chincoteague Island began ca. 2250 years ago and lasted until ca. 400 years ago (Fig. 10c). While the net dip directions of reflections in shore-normal radargrams along Chincoteague Island are seaward, there is also evidence of shallow, landward-dipping reflections (see Fig. 4), indicative of progradation through onshore migration and welding of swash bars (e.g., Hine, 1979;Carter, 1986;Nooren et al., 2017). We suggest that progradation of Chincoteague Island was driven by ebb-deltaassociated sediment bypassing at tidal inlets immediately north and south of the island, as observed along inlet-adjacent shorelines throughout the Virginia Barrier Islands (e.g., (Fenster and Dolan, 1996). To the north of Chincoteague Island, wave refraction around what was then the southern terminus of Assateague Island and an ebb-tidal delta associated with the southeast-northwestoriented inlet separating Chincoteague and Assateague islands likely created a nodal zone (region of divergent longshore transport) on central Chincoteague. Sediment was directed away from this region, both to the north and south along the Chincoteague shoreline. To the north, this sediment entered the semi-protected coastal reach landward of southern-most Assateague Island, and accumulated in a wave-shadow zone in the form of beach and foredune ridges. In this manner, Chincoteague Island underwent a period of widening similar to that observed for modern Wallops Island, where a nodal zonedlocated along the central part of the island and caused by wave refraction around southern Assateaguedleads to diverging sediment fluxes, net northerly transport along the northern end of the island, and the eastward growth of beach and foredune ridges along the northern depocenter .
The growing shoreface and beach and foredune ridges of Chincoteague Island incorporated approximately 81 million m 3 of sand which otherwise likely would have been transported to downdrift barriers. However, the annual sand trapping flux varied with time. From ca. 2250 to 1300 years ago the barrier volume grew by an average of~44,000 m 3 yr À1 and~46,000 m 3 yr À1 between ca. 1300 to 400 years ago (Table 5; Fig. 11). This suggests that sediment retention rates within Chincoteague Island did not substantially change with time and the impact of Chincoteague's widening on downdrift sand fluxes likely remained constant. However, the sand supply to individual downdrift barriers was also dependent on other local processes, such as changes in ebb-tidal delta sediment bypassing, cross-shore sediment fluxes, and tidal and wave ravinement of antecedent geology (e.g., Fenster et al., 2016;Raff et al., 2018;Shawler et al., 2019b;Shawler et al., 2021).

Stage IV e Elongation of Assateague Island: Importance of updrift inlets and barrier connectivity
A period of little to no deposition occurred in the Assateague-Chincoteague-Wallops system when tidal inlets became more prominent updrift on Assateague (Fig. 11). Between ca. 1620 CE and 1755 CE, Assateague Island elongated 4 km to the south and seaward of Chincoteague, forming an initial recurved spit. The opening of five ephemeral inlets along northern and central Assateague (~3e~28 km north of Lighthouse Ridge) between ca. 1755 CE and 1830 CE  temporarily stopped the elongation/progradation of Assateague Island. We estimate the volumes of sand trapped in the flood tidal deltas of Cherry Tree (~3 km updrift) and Green Run (~14 km updrift) inlets using aerial mapping of inlet-associated features identified by , paired with average regional flood tidal delta thickness values (~3 m thick) from Seminack and McBride (2019). The estimates indicate that the fluxes of sediment into growing flood-tidal deltas of these two inlets possibly reduced longshore transport to the Assateague-Chincoteague-Wallops system by as much as~222,000 m 3 yr À1 (20% of post-1830 CE fluxes; >100% of pre-1830 CE fluxes) during the period of active inlet activity. This resulted in limited downdrift progradation between ca. 400 and 190 years ago (Fig. 10d). We hypothesize that continued fluxes of fine sand from the beach and nearshore system to the adjacent dunes during this period allowed for aggradation of the 8þ m tall aeolian ridge (Lighthouse Ridge). Conceptual (Psuty, 2004) and morphodynamic  models link similar periods of dune-building with slowed shoreline progradation, an observation confirmed by local (e.g., Parramore Island, Virginia; Raff et al., 2018), regional (e.g.,Cape Lookout, North Carolina; Elliott et al., southern terminus of Assateague Island; C) Stage III: Progradation of Chincoteague continued through progressive welding off bars by inlet sediment bypassing and subsequent reworking of sand by longshore transport processes; the southward elongation of Assateague is inferred from the shifting nodal zone evident in the ridge orientation on Chincoteague; D) Stage IV: Assateague Island began to elongate seaward of Chincoteague Island. The exact timing of this multi-stage development is unknown, but it likely occurred between ca. 1620 CE and 1755 CE. The immediately updrift Cherry Tree Inlet was ephemerally open between 1755 CE and 1827 CE, thereby halting or greatly slowing downdrift progradation and, instead, allowing for vertical dune building of Lighthouse Ridge (8þ m tall).; E) Stage V: The progradation (net direction NW to SE) of southern Assateague was driven by cross-shore sediment fluxes combined with increased alongshore sand fluxes from closure of updrift inlets, which previously functioned as a sediment trap. F) Stage VI: The modern system continues to elongate/prograde and is marked by increased human infrastructure. 2015), and global (e.g., Pedro Beach, Australia; Oliver et al., 2019b) field studies.

Stage V e Progradation/elongation of Assateague and Wallops islands: Shoals as a framework control on spit development
Southerly elongation of Assateague Island resumed in ca. 1830 CE and lasted to 1910 CE, during which time the barrier trapped~87 million m 3 of sand through the accelerated formation of a beachand foredune-ridge plain and the development of a narrow isthmus. Fluxes to the system were around 1 million m 3 per year during this phase (Fig. 11). The onset of this period of rapid progradation coincided with the closure of both the proximal Cherry Tree Inlet and the more distal ephemeral Sinepuxent Inlet (~24 km north of Lighthouse Ridge). Alongshore fluxes were likely further enhanced by the closure of North Beach (~28 km north) and Green Run (~14 km north) inlets in 1840 CE and 1880 CE, respectively (Goettle, 1978;. With the closure of both inlets, sand delivery to backbarrier flood-tidal deltas ceased and ebb-tidal deltas collapsed, together increasing downdrift sand fluxes to southern Assateague Island. Furthermore, the growth of Assateague during this time coincides with state shifts of islands south (downdrift) of Assateague, including landward migration and rapid thinning (by over 200 m) of Metompkin Island (Byrnes, 1988;Deaton et al., 2017), accelerated migration of Cedar Island (starting ca. 1852 CE; Shawler et al., 2019b), and rapid erosion of southern Parramore Island (starting ca. 1870 CE; Raff et al., 2018). Taken together, these observations indicate the possible importance of sand fluxes between updrift inlets and downdrift spits, and, by extension, interconnections between elongating spits and the behavior of downdrift barrier islands.
In the late 1800s, Assateague Island continued to elongate in a southwesterly to southerly direction (Fig. 10e). An isthmus formed south of the 1859 CE shoreline and by 1910 CE extended over a series of shoals/shoreface ridges (Fig. 9b). We propose that the offshore and shoreface-attached shoals south of Assateague Island acted as a framework control on accommodation by providing a platform upon which the spit could rapidly prograde, eventually determining the subaerial morphology of the Tom's Cove Isthmus. Stratigraphic data from this study indicate that the isthmus, as well as the entire recurved spit (formed from 1910 CE to present), is underlain by shoreface sediment of a consistent depth (À5 to À6 m MSL). The growth of the isthmus and recurved spit created the shallow Tom's Cove in which fine sediments could settle in the quiet-water embayment atop the underlying shoreface sediment. The shoals provided comparatively uniform, shallow accommodation which differs from the deeper (below À10 m MSL) adjacent shoreface. Recent shoreface mapping (Wikel, 2008, Fig. 9c) and seismic stratigraphy from this study (Fig. 8) indicate the presence of thick (up to 7 m) shoals offshore of modern Assateague that may provide a shallow platform for continued/future spit progradation.
Since 1851 CE, the northernmost portion of Wallops Island has prograded to the north and east. During this time, northern  [Campbell and Benedet, 2006;ASBPA, 2020;Elko et al., 2021]). The data indicate that natural sediment trapping through barrier-island progradation and spit elongation overwhelms any major contribution from alongshore transport of artificially placed updrift sediment. Note the change in horizontal scale starting at 1820 CE. B) Time-varying average fluxes of sand trapped in each depositional feature through time. A period of no/little deposition in the Assateague-Chincoteague-Wallops system corresponds to a period of heightened updrift inlet activity from 1740 to 1830 CE. Estimates of fluxes into two of those inlets are calculated from aerial mapping of flood tidal delta area, supplemented by representative sediment core data from Seminack and McBride (2019). Note the change in horizontal axis scale starting at 1740 CE. Wallops trapped 7 million m 3 of sand at an average rate of 42,000 m 3 yr À1 (Table 5; Fig. 11). An additional 4 million m 3 of mixed sand and mud was preserved as the barrier grew over the shoreface. Since 1910 CE, the elongation of southernmost Assateague Island has trapped 43 million m 3 of sand at a rate of over 400,000 m 3 yr À1 (Table 5; Fig. 11). During this time, the downdrift Metompkin, Cedar, and Parramore islands have rapidly eroded and/ or migrated landward at long-term (1850e2010 CE) average rates of !4 m yr À1 and short-term (1980e2010 CE) average rates of !10 m yr À1 (Deaton et al., 2017). Today, both southern Assateague and northern Wallops islands function as sediment sinks through continued growth of beach and foredune ridges.
5.1.7. Stage VI e Modern system: human intervention in the sediment transport system Modern coastal change along the Assateague-Chincoteague-Wallops barrier-island complex occurs in response to both natural and human influences. The recurved spit at the terminus of Assateague Island continues to grow southward, and northernmost Wallops Island continues to widen to the east and elongate to the north. By contrast, the narrow isthmus on southern Assateague occasionally breaches and is now located~700 m landward of its late 19th and early 20th century position. This landward migration is likely to continue as the isthmus is frequently overwashed and/or breached by storms, which leads to sand deposition in the barrierproximal (i.e., eastern) portion of Tom's Cove. Progradation of the entire system has occurred since ca. 2250 years ago through the natural influx of sand through longshore transport. However, since ca. 1961, the system has likely received additional inputs supplied from the reworking of updrift beach nourishment sands (Fig. 11). Yet, this artificial, short-term nourishment fluxdeven in the highly unlikely case that 100% of it were transferred to downdrift reservoirsdpales in comparison to natural (i.e., pre-nourishment) sand trapping in this system; over the last 60 years, updrift nourishment volumes along the beaches of Ocean City, Maryland (~60 km north) and northern Assateague Island (~55 km north) combine for onlỹ 10 million m 3 of sand, as compared with the~30 million m 3 of sand captured by Fishing Point/Wallops Island over this same time.
On Wallops Island, recent anthropogenic modifications include beach nourishment (since 2010; 3 million m 3 of sand in total) and ongoing efforts to stabilize a seawall that has existed since 1945 CE . Additionally, the youngest beach/ foredune ridge on northern Wallops is being mined as a source of nourishment sand for an ongoing (as of July 2021 CE) central Wallops Island breakwater project (NASA, 2019). Predictions of future trajectories for this portion of the system must account for both the natural changes and human dynamics, such as anthropogenic shoreline-control and sediment-management efforts (see Miselis and Lorenzo-Trueba, 2017;Armstrong and Lazarus, 2019;Lazarus and Goldstein, 2019). 5.2. Implications for regional sediment transport

Sand trapping at southern Assateague Island
Net shoreline erosion along the Virginia Barrier Islands downdrift of Wallops Island has increased from a system-wide long-term rate of 5 m yr À1 (1870e2010 CE) to 7 m yr À1 in the period between 1980 CE and 2010 CE (Deaton et al., 2017, Fig. 12a). Assawoman, Metompkin, and Cedar islands have experienced rapid short-term historical shoreline transgression (4e11 m yr À1 on average between 1980 and 2010 CE) largely resulting from landward island migration, and resulting in a geomorphic feature termed the 'arc of erosion', which extends over 35 km south of southernmost Assateague Island (e.g., Rice and Leatherman, 1983;Kraus and Galgano, 2001;Nebel et al., 2012;Fenster et al., 2016;Deaton et al., 2017).
South of Cedar Island, Parramore Island is undergoing accelerating erosion, and may enter this same low-relief, washover-dominated regime in the coming decades (Raff et al., 2018).
Four mechanisms may individually or (more likely) collectively explain this observed acceleration and southerly extension in island erosion and landward migration over the historical period: 1) increased storm frequency (e.g., Hayden and Hayden, 2003;Komar and Allan, 2008; 2) increased rates of sea-level rise along the Virginia coast and attendant tidal prism changes (Fenster et al., 2011;Deaton et al., 2017;FitzGerald et al., 2018); 3) changes to the alongshore transport gradient from north to south caused by changing wave refraction patterns around the southern end of Assateague Island (Jones, 2016); and 4) sediment trapping at the recurved spit of southern Assateague (Fenster et al., 2016).
Here, we quantify sediment volumes at southern Assateague Island to constrain the role of growth of these updrift siliciclastic landforms in modifying downdrift barrier behavior. Since the 19th century, Assateague and Wallops islands have together trapped 134 million m 3 of sand at a rate of 720,000 m 3 yr À1 . Yet, these trapping magnitudes are unequal: the average flux of sand trapped at Assateague is ten times greater than that trapped at Wallops Island. The grain-size distributions of this sand are roughly equivalent to those found along the beaches of Assawoman, Metompkin, Cedar, and Parramore islands (Fenster et al., 2016). This finding suggests that the sediment trapped on southern Assateague Island consists of an appropriate texture to contribute to the downdrift barrier beaches (Fig. 12b). Additionally, the subaqueous shoals and/or shoreface deposits over which Assateague and Wallops islands prograded represent an additional sediment reservoir of 113 million m 3 . However, much of this material consists of heterogeneous sand and mud and therefore is too fine to function as "beach quality" sand for downdrift barrier islands. While we account for sediment trapping through barrier-island progradation and the attendant burial of shoreface sediment, sediment transport and/or trapping mechanisms on the modern shoreface of southern Assateague is/are poorly understood. Landward and downdrift movement of offshore sand ridges is evident and possibly linked with shoreline behavior on Assateague Island (Wikel, 2008;Pendleton et al., 2017). However, the exact sediment transport pathways, mechanisms, and timescales remain unclear. Future work must account the influence of shoreface sand ridges and other framework geologic features on hydrodynamic and sediment transport/ storage processes.
A compilation of seven engineering reports and studies conducted between 1956 and 2005 demonstrates that longshore transport along the Assateague Island coast ranges from 115,000 to 1,100,000 m 3 yr À1 , with all but one value clustering between 115,000 and 460,000 m 3 yr À1 . The one outlier (1,100,000 m 3 yr À1 ) is based on wave refraction using wave gauge data from~125 km south of Lighthouse Ridge (Headland et al., 1987). Based on these estimates, we conclude that Assateague and Wallops islands annually trap a sand volume equivalent to at least 60% of that transported along the Virginia coast north of Assateague Island; removing the Headland et al. (1987) value, trapping fluxes exceed longshore transport rates by as much as a factor of six. This suggests that true transport fluxes are closer to the upper-end estimate (!1 million m 3 yr À1 ) and/or that significant volumes of sediment are contributed from the inner shelf. Further complicating this, the floodand ebb-tidal delta complexes at Chincoteague Inlet represent an additional, but unquantified, sediment/sand sink.
From this analysis, it is clear that growth of the Assateague-Chincoteague-Wallops barrier-island complex represents a very significant longshore sediment sink along the northern Virginia coast. Furthermore, we propose that, in addition to changing patterns of wave refraction and associated alongshore north-tosouth transport gradients (e.g., Jones, 2016), this process is predominantly responsible for net erosional conditions experienced from Assawoman to Parramore islands. The downdrift erosion observed along the most vulnerable reaches is amplified by a sandpoor shoreface and sediment sequestration within large ebb-tidal deltas south of Assateague (e.g., Fenster et al., 2011;Fenster et al., 2016). These results are consistent with observations along the Alabama-Mississippi coast, which indicate the important role that sediment deficits, in addition to storms and sea-level rise, play in barrier island land loss (Morton, 2008;Eisemann et al., 2018;Hollis et al., 2019;Gal et al., 2021). Similar alongshore couplings exist along the Danish Wadden Sea coast, where an alongshore transport gradient from updrift (sediment loss) to downdrift (high sediment supply) results in progradation of the downdrift coast (Fruergaard et al., 2019).
In contrast to sand trapping and alongshore wave gradients, factors which are often recognized as responsible for barrier-island change (e.g., storms, sea-level rise) are likely to affect the Virginia Barrier Islands uniformly. These therefore cannot explain the longshore gradients in barrier behavior. While differential responses to storm events can lead to intra-island variations in barrier morphology (e.g., Houser et al., 2008), a single storm will typically affect a large region (i.e., the full Virginia Barrier Island chain), eroding barrier islands and moving sand alongshore, temporarily offshore, or permanently landward via overwash. Likewise, relative sea-level rise is similar across the southern Delmarva Peninsula, and will therefore impact all barriers in the chain in a similar manner; in this case, by creating additional accommodation and forcing complex barrier behavior including continued retreat (Moore et al., 2010;Lorenzo-Trueba and Ashton, 2014). Beyond the "arc of erosion", additional local factors which may control differential future behavior of individual islands include variable underlying stratigraphy (e.g., Brenner et al., 2015), complex antecedent morphology (e.g., Shawler et al., 2021), ecogeomorphic feedbacks between islands and their associated dunes and barrier-backbarrier marshes and lagoons (e.g., Walters et al., 2014;Dur an Vinent and Moore, 2015;Reeves et al., 2020), and current and future human modifications such as shoreattached breakwaters being installed along central Wallops Island and beach nourishment on Assateague and Wallops islands.

Sediment trapping at tidal inlets
A growing body of literature emphasizes the importance of tidal inlets in controlling barrier-island behavior. For example, in their "runaway transgression hypothesis" FitzGerald et al. (2008FitzGerald et al. ( , 2018 posit that 1) sea-level rise drives backbarrier marsh loss and thus increases backbarrier tidal prism; 2) increased tidal prism enlarges inlet cross-sectional area and modifies transport patterns, leading to the growth of floodand ebb-tidal delta shoals; and 3) sand sequestering in depositional landforms (i.e., shoals) eventually contributes to barrier-island disintegration and/or rapid landward migration. However, a recent test of this hypothesis on the Virginia coast indicates that landward barrier migration more than compensates for backbarrier marsh loss, inhibiting tidal prism changes at most inlets (Deaton et al., 2017). However, at least Wachapreague Inlet (~42 km south of Lighthouse Ridge) has experienced a modest increase in tidal prism through time (Fenster et al., 2011) and its ebb-tidal delta may function as a trap for longshore sediment to the southernmost Virginia barrier islands (Fenster et al., 2016). This latter study highlights the role of tidal inlets as potential sinks for alongshore-transported sand. Further, in addition to ebb deltas, the formation and growth of flood-tidal deltas has also been recognized as a significant coastal sand sink (e.g., Nienhuis and Ashton, 2016). Flood-tidal delta deposition can provide a platform for barrier migration (e.g., Mallinson et al., 2010) and increase barrier-island resilience to drowning in response to sea-level rise (e.g., Nienhuis and Lorenzo-Trueba, 2019) but can also trap sand and thus impact local barrier sediment supply (e.g., Shawler et al., 2019b;Z ainescu et al., 2019).
Our data and these prior studies all highlight that inlets, through sequestration of sand in both ebb-and flood-tidal deltas, function as important controls on sand transport on barrier coasts, though the precise tidal-inlet morphodynamic response to sea-level rise likely depends on several site-specific variables. These processes are likely to be more active during periods of more frequent and/or intense storms, such as during the Medieval Climate Anomaly (ca. 950e1250 CE) and the Little Ice Age (ca. 1400e1900 CE), when both the Virginia Barrier Islands and the Outer Banks of North Carolina experienced increased inlet formation (Culver et al., 2007;Timmons et al., 2010;Mallinson et al., 2011Mallinson et al., , 2018Raff et al., 2018). Given the likelihood of enhanced 21st century tropical and extratropical cyclone activity along the U.S. mid-Atlantic (Bender et al., 2010;Michaelis et al., 2017;Paerl et al., 2019), the formation of new inlets or re-breaching at former inlet locations is a realistic scenario along the Virginia coast. For example, the extratropical cyclonic Ash Wednesday (1962) storm caused breaches on northern Assateague Island (~48 km updrift Lighthouse Ridge) and recent extratropical storms have caused regular ephemeral breaching of Tom's Cove Isthmus (Seminack and McBride, 2015b; see also Fig. 2b). Data from this study indicate that the barrier-proximal portion of Tom's Cove contains potentially erodible sand and that the central portion of the embayment offers accommodation that may allow for the development of flood-tidal delta deposits. Such a scenariodsimilar to that associated with the formation of Cherry Tree Inlet along a narrow section of Assateague during the late 18th and early 19th centuriesdwould not only impact the local dynamics at Chincoteague Inlet but may also affect local (1 km) to regional (10þ km) sediment fluxes through the formation of an additional longshore transport sink. Future research and planning in Virginia and along other barrier coasts must emphasize the role and timescales of inlet formation and the growth of both ebb and flood tidal deltas on interconnectivity between updrift and downdrift barriers.

Global importance of large longshore sediment sinks
Coastal landforms such as spits, beach/foredune-ridge plains, wave-dominated deltas, strandplains, and chenier plains function as sediment traps which modify rates of alongshore sediment bypassing to downdrift coasts. Yet, the behavior (e.g., progradation, erosion, stability) of these systems is also dependent on sediment supply from updrift and cross-shore sources. In other words, coastal landform formation is both a product of sediment inputs and a control on downdrift sediment supply. For example, updrift sandy headland erosion provides sufficient sediment to promote downdrift barrier elongation/progradation on the Louisiana coast (Torres et al., 2020). Likewise, the formation and stability of beach ridges on the West African coast is driven by changes in both sand source and transport patterns (Anthony, 1995). In southeastern Australia, time-varying embayment interconnectivity drives changing progradational shoreline response (Oliver et al., 2020). Similarly, sandy shoreline behavior along Santa Catarina Island in Brazil is controlled by time-varying sediment fluxes resulting from the complex interplay of headland bypassing, dune overpassing, longshore transport between and past embayed beaches, and the growth of strandplains and spits which starve downdrift coasts (Vieira da Silva, 2016a; Vieira da Silva, 2016b;Hein et al., 2019b). On the barrier-island coast of the Danish Wadden Sea, updrift coastal reconfiguration resulted in collapse of the downdrift barrier-island chain (Fruergaard et al., 2021). Along the North Carolina coast in the United States, spit growth at the Power Squadron Spit trapped 30% of longshore transport and reduced wave energy at an immediately downdrift barrier island, leading to an accelerated rate of shoreline retreat (Park and Wells, 2007); this is similar to our observations along the Virginia coast. Likewise, deposition at North Island, a spit on the South Carolina (USA) coast, captures an estimated 15% of alongshore and cross-shore sediment inputs to its coastal compartment (Wright et al., 2017). These studies and our own results emphasize the dynamic competition between longshore transport and temporary to permanent storage in complex coastal depositional landforms.
We compile examples of coastal depositional landforms and place Assateague, Chincoteague, and Wallops islands in context with these other systems (Table 6). While this list is not necessarily comprehensive, it does provide global context for this case study. The volume of sand trapped in southern Assateague (10 8 m 3 ) since ca. 1830 CE is equivalent to the large subtropical beach-ridge plains along the coasts of Brazil and Australia. Yet, the annual sand flux into Assateague is at least six times greater, an observation that largely reflects that rates of progradation are one-to-two orders of magnitude higher on most shore-parallel elongating spits, most especially southern Assateague. By contrast, cross-shore prograding barrier islands such as Chincoteague and Wallops accumulate sediment at rates similar to strandplains (10 4 m 3 yr À1 ) and, thus, the fluxes of sediment trapped are lower than those on Assateague. The compilation demonstrates that progradational barrier islands generally trap sediment over long timescales (centuries and millennia) at annual rates similar to strandplains; notably both grow predominantly cross-shore and seaward, into generally deeper water. By contrast, the examples of spits, which largely elongate parallel to shore and thus capture sediment along their relatively low-energy recurved ends, are much more dynamic and vary greatly in their total volumes, fluxes, and progradation rates. This observation aligns well with numerical modeling work indicating that spit morphodynamics depend on complex interactions between wave climate, updrift shoreline behavior and sediment fluxes, alongshore sediment transport gradients, and instability of the possibly highly-autogenic spit hook .
These insights emphasize the importance of accounting for highly local controls (i.e., sediment trapping, framework geology, changing hydrodynamics) on coastal behavior in the context of coastal management. For example, given the alongshore couplings indicated by our results, a system-wide approach to managing sediment is necessary along the Virginia coast and, by extension, other coasts worldwide. This has also been proposed for Louisiana (USA), where barrier-island restoration and management efforts have historically focused on individual islands rather than multiple barrier-backbarrier systems along the same chain (Khalil et al., 2015). Foundational science which explores the framework geology of the Virginia coast (e.g., Raff et al., 2018;Brothers et al., 2020;Shawler et al., 2021) is an important first step towards improving coastal management. Indeed, similar efforts to map the shoreface in New South Wales, Australia demonstrate that increased knowledge of framework geology and improved shoreline change models together enhance coastal management and planning .

Conclusions
This study emphasizes the important role of natural coastal sediment sinks in the regulation of longshore transport and behavior of downdrift siliciclastic coastal systems. Through the integration of millennial-, centennial-, and decadal-scale records of coastal change recorded in progradational ridges and backbarrier deposits of the Assateague-Chincoteague-Wallops system, we demonstrate: The Assateague-Chincoteague-Wallops barrier-island system formed offshore of its present position >6000 years ago and underwent a multi-centennial period of net landward migration until ca. 2250 years ago. Chincoteague Island stabilized and prograded from ca. 2250 to 400 years ago; the rapid elongation and progradation of southern Assateague and northern Wallops began in ca. 1830 CE (~190 years ago). A phase of reduced/no progradation at southern Assateague Island from ca. 400 to 190 years ago coincided with updrift barrier breaching and sand sequestration in flood-tidal delta shoals. Chincoteague and Wallops, two progradational barrier islands, trap total volumes of sand (~10 6 e10 7 m 3 ) at rates (~10 4 m 3 yr À1 ) broadly similar to global examples of mainland-attached beachridge plains. Southern Assateague Island, a southerly elongating spit, traps a similar volume of sand (10 8 m 3 ) as compared with strandplains, but with sediment fluxes that are up to six times greater. Over the last 120 years, the growth of southern Assateague and Wallops islands together trapped at least 60% of longshore sediment, a process that likely starved downdrift barriers of 'beach-quality' sand and corresponds with the rapid erosion/ migration (!3 m yr À1 ) of downdrift barrier islands. Regional coastal planning must account for the impacts of natural coastal sediment sinks such as progradational landforms and tidal inlets thatdin conjunction with forcings such as sealevel rise and storminessdestablish the baseline conditions that control the behavior of sandy coastal systems.
These findings stand in direct contrast to recent work which predicts the wholesale collapse of sandy coastal features in response to sea-level rise (e.g., Vousdoukas et al., 2020). Instead, this work emphasizes the importance of regional controls on sediment transport as a determinant of changes between regressive and transgressive coastal behavior.

Data availability
Sediment core descriptions, grain size analysis results, and editable radiocarbon and OSL tables from this study are available at: https://doi.org/10.25773/53bv-4p15.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.