Vertical accretion trends project doughnut-like fragmentation of saltmarshes

The study investigates how coastal saltmarshes in the Georgia Bight (South Carolina, Georgia, Florida) maintain vertical elevation relative to rising sea levels, focusing on the balance between belowground organic production and surficial deposition of inorganic sediment. While models and field studies suggest that marshes with abundant organic production or sediment supply can keep pace with relative sea-level rise (RSLR), responses are heterogeneous because sediment availability is often limited and warming, storms, and other stressors can undermine elevation gains. Large-scale compilations show some resilience via accelerating accretion with rising sea levels, but they often miss local, sub-kilometer processes that control sediment delivery to the marsh surface. The research question centers on whether barrier-associated, backbarrier marshes exhibit spatially variable resilience linked to local sediment supply and exposure to hydrodynamic forcing, and whether marsh edges near large water bodies accrete faster than interior marshes, potentially leading to a fragmented, doughnut-like pattern of marsh persistence around drowning interiors.

Prior work indicates marsh accretion capacity depends on organic productivity, sediment trapping under increased inundation, nutrient delivery during storms, and pore-space dynamics in organic-rich soils. However, limited mineral sediment (notably at higher latitudes), warming-induced decomposition, and storm-related damage can reduce resilience. Global and latitudinal syntheses and models have documented marsh capacity to adjust to accelerating sea-level rise, sometimes showing accelerating accretion, but they often do not capture fine-scale sediment-delivery processes. Along the US South Atlantic, backbarrier marshes differ from microtidal, deltaic systems: they are protected by wide barriers, experience constrained storm surge through inlets, and are fronted by limited-fetch waves except near inlets. Previous Georgia Bight accretion estimates largely came from marsh interiors, distal from major sediment sources, potentially underrepresenting edge and inlet-proximal dynamics.

Field sampling and study area: Sixteen sediment cores were collected in December 2017 from backbarrier marsh platforms behind four barrier systems in the Georgia Bight: Cape Romain (SC), Hilton Head (SC), Sapelo Island (GA), and Amelia Island (FL). Stations were placed along mainland-to-barrier transects at approximately one-third and two-thirds of the backbarrier width, ~10 m inboard of the marsh edge and landward of levees, at the boundary between short and tall forms of Spartina alterniflora. Cores (15.3 cm diameter; up to ~80 cm length) were extracted with a metal corer into PVC liners; elevations were surveyed with RTK-GPS (m NAVD88, ±2 cm), converted to mean tidal level (MTL), and normalized to local tidal range to compute Z_MHW.

Core processing and measurements: Cores were split, described, photographed, and sectioned at 1 cm (0–6 cm), 2 cm (6–30 cm), and 5 cm intervals below 30 cm. For each sample, wet/dry bulk density and loss-on-ignition (LOI; 650 °C, 14 h) were measured. Gamma spectrometry quantified 210Pb (total and supported) and 137Cs activities after sealing samples for ≥30 days to allow secular equilibrium. Excess 210Pb (210Pbxs) was calculated by subtracting supported from total 210Pb.

Accretion rate calculations: Accretion rates were computed using the Constant Flux and Constant Sedimentation (CFCS) model (S = λ/m; λ = 0.03114 yr−1; m = slope of depth vs ln(210Pb activity)) from the surface to the depth where 210Pbxs ≈ 0, omitting outliers and surface mixed layers. Uncertainties were estimated via sensitivity analysis by sequentially removing points and recalculating rates. Independent 137Cs-based accretion rates used the depth of first occurrence or peak (1963–1964) with uncertainties reflecting sample thickness.

Normalization to sea-level rise: For each core, the contemporaneous RSLR over the accretion record (RSLRC) was derived by linear regression of monthly sea level from the nearest tide gauge(s) for the time window corresponding to the 210Pb-derived record; for Sapelo Island and northern Cape Romain-adjacent sites, rates from two equidistant gauges were averaged. AEC was defined as 210PbCFCS/RSLRC. To account for potential response lags, the 50-year RSLR preceding sampling (RSLR50) was also computed, and AE50 = 210PbCFCS/RSLR50. For cores with disturbed upper 4–10 cm, the calculated accretion rate below the mixed layer was propagated to the surface for RSLR calculations, providing a conservative estimate.

Regional database: A Georgia Bight database updated prior compilations and added the 16 new cores, including published 210Pb, 137Cs, Surface Elevation Table (SET), and Marker Horizon (MH) measurements. When not provided, uncertainties of ±3% (210Pb) and ±6% (137Cs) or ±0.10 mm yr−1 (SET/MH) were assumed. Z_MHW, RSLRC, AEC, and AE50 were computed uniformly where possible; some records lacking down-core 210Pb data could not be assigned AEC/RSLRC. Very short records (<5 years) or those with problematic mixed layers were excluded or partially excluded as documented.

• New accretion rates: 210PbCFCS accretion ranged from 1.91 ± 0.07 to 13.91 ± 0.07 mm yr−1 (mean 6.29 ± 0.18 mm yr−1). Within-system variability (avg range ~7.8 mm yr−1) exceeded between-system differences (range of system means ~2.73 mm yr−1).
• Agreement between methods: After excluding one outlier (AI-T1S3), 210PbCFCS and 137Cs rates agreed moderately strongly (r2 = 0.81; p < 0.01).
• Accretion relative to sea-level rise: 14/16 stations had AEC > 1; mean AEC across stations was 1.9 ± 0.3 (i.e., nearly twice contemporaneous RSLR). Mean AE50 across stations was 2.0 ± 0.1; only two stations had AEC and AE50 < 1 (HH-T2S4, CR-T1S4).
• Elevation relationship: Tide-range-normalized elevation (Z_MHW) explained ~20% of variance in accretion metrics (weak positive correlations; p ≤ 0.1), contrary to the commonly assumed negative elevation–accretion relation, likely due to narrow platform elevation range (~68 cm).
• Composition: Mean bulk density (BDpb-xs) and LOI (LOIpb-xs) were 0.42 ± 0.03 g cm−3 and 19.3 ± 0.9%, respectively, with no significant correlations between AEC/AE50 and BD or LOI.
• Exposure controls: Highest accretion rates occurred at stations with greatest exposure to large bays and tidal inlets (e.g., SI-T2S2: 13.43 ± 0.43 mm yr−1 near Sapelo Sound/Inlet; HH-T2S2 and HH-T1S2 facing Calibogue Sound). Removing the single most exposed station per barrier reduced mean AEC and AE50 from 1.89 and 2.01 to 1.48 and 1.49, highlighting the outsized influence of highly exposed edges.
• Regional synthesis: Across the Georgia Bight database, accretion more closely matched RSLR50 than contemporaneous RSLR, reflecting timescale biases. Mean AE50 for all data (including new cores) was 0.94 ± 0.16 (near parity with 50-year RSLR). Excluding the 16 new edge-proximal cores, mean AE50 dropped to 0.72 ± 0.17, and ~75% of measurements had AE50 < 1, indicating many marshes, especially interiors, are not keeping pace.
• Method differences: Shorter-record methods (SET, MH, 137Cs) showed lower AEC but higher AE50 relative to AEC, consistent with response lags; longer 210Pb records showed similar AEC and AE50 on average.
• Conceptual outcome: The combined results indicate a spatial dichotomy—edge and inlet-proximal marshes accrete rapidly due to enhanced allochthonous mineral inputs, whereas interior and low-exposure marshes often lag RSLR, implying a trend toward doughnut-like fragmentation (persistent peripheries, drowning interiors).

The findings address the core hypothesis that local sediment supply and exposure strongly modulate marsh resilience to sea-level rise within barrier-associated backbarrier settings. Edge-proximal and inlet/bay-facing marshes benefit from wave- and surge-driven suspension and delivery of mineral sediments from adjacent channels, flats, lagoon floors, and storm inputs, enabling vertical growth at rates that outpace contemporaneous and 50-year RSLR. Conversely, interior marshes, distant from energetic sediment sources and with limited inundation-driven delivery, often accrete more slowly and risk submergence.

The weak positive correlation between elevation and accretion likely reflects constrained platform elevation variability and possible recycling processes (e.g., marsh-edge soil creep) that maintain sediment redistribution near edges. Methodological comparisons show that accretion–RSLR relationships depend on observation timescales; AE50 often exceeds AEC for short records, reinforcing the importance of lagged responses and longer-term context when assessing resilience. Regionally, despite instances of rapid edge accretion, most marshes—especially interiors—appear to be falling behind RSLR when new edge-proximal measurements are excluded, underscoring widespread vulnerability.

These insights emphasize that spatial heterogeneity at fine scales (10s to 100s of meters) governs resilience, with implications for monitoring and management: targeting edge zones alone can overestimate system-wide resilience, whereas interior-focused assessments may understate the stabilizing contributions of edges to sediment budgets and elevation capital.

This study demonstrates that in Georgia Bight barrier-backed marshes, vertical accretion is highly spatially variable: well-exposed, edge-proximal marshes near inlets and large bays commonly accrete at 1.5–2.0× RSLR (and up to >4× at one site), while protected interior/platform marshes often lag sea-level rise. The results reconcile apparent discrepancies in regional resilience assessments by highlighting strong local controls on sediment delivery and by accounting for response lags via 50-year RSLR comparisons. Collectively, the system appears to be trending toward doughnut-like fragmentation—persisting perimeters with drowning interiors—which threatens the integrity of backbarrier ecosystems and the stability of adjacent barrier islands.

Future research should: (1) intensify spatially explicit sampling across exposure gradients and distances from edges to quantify sediment-delivery decay; (2) integrate hydrodynamic–morphodynamic models with storm climatology to project exposure-driven accretion under changing storm regimes; (3) monitor interior marsh elevation dynamics and sediment budgets with SET/MH networks and radionuclides; and (4) assess management interventions (e.g., thin-layer placement, sediment diversions, channel/flat restoration) to enhance interior sediment supply.

• Methodological assumptions: The 210Pb CFCS model assumes constant 210Pb flux and sedimentation; deviations (e.g., event layers, bioturbation, mixed layers) can bias rates. Mixed surface layers required omission and extrapolation to the surface, introducing uncertainty (conservative estimates).
• Outlier/core-specific uncertainties: One core (AI-T1S3) exhibited anomalously high 210PbCFCS relative to 137Cs, suggesting reworking/event deposition; it was treated as an outlier in comparisons.
• Timescale bias: Shorter observation methods (SET/MH, 137Cs) are sensitive to recent acceleration in RSLR; comparisons with contemporaneous RSLR vs 50-year RSLR can yield different interpretations of resilience.
• Spatial sampling bias: New cores were collected near platform edges (~10 m inboard) and may overrepresent high-exposure settings relative to interior marsh conditions; regional literature data are skewed toward interiors, complicating uniform comparisons.
• Data gaps in regional compilation: Many published 210Pb records lacked down-core profiles, preventing determination of RSLRC/AEC and limiting some analyses; very short records were excluded.
• Generalizability: Findings pertain to mesotidal, barrier-associated systems of the Georgia Bight with sandier sediments and specific inlet/bay morphologies; extrapolation to other settings should consider differing sediment supplies and hydrodynamics.