Ocean cavity regime shift reversed West Antarctic grounding line retreat in the late Holocene

The study addresses why the Siple Coast grounding line of the West Antarctic Ice Sheet retreated far inland during the mid–late Holocene and subsequently readvanced within the last ~1.7 ka. Two hypotheses are evaluated: (1) climate-driven changes in sub-ice-shelf ocean conditions (a transition from a warm to cold ocean cavity) and (2) isostatic rebound-driven regrounding due to delayed Earth deformation. Contextually, WAIS holds ~5.3 m sea-level equivalent and is a key concern for future sea-level rise. Geological evidence (radiocarbon-dated subglacial tills and sediment proxies) indicates substantial mid-Holocene retreat (7.5–5.3 ka BP) and late Holocene readvance of multiple Siple Coast ice streams (Whillans, Mercer, Kamb, Bindschadler). The purpose is to test which mechanism best explains the timing and magnitude of these grounding line migrations and to quantify the interaction between ice dynamics, solid Earth response, and ocean forcing.

Prior work suggests that WAIS underwent substantial Holocene changes, with mid-Holocene retreat inferred from subglacial tills and marine proxies in the Ross Embayment. A regrounding mechanism via delayed isostatic rebound was proposed for Antarctic embayments and explored using simplified Earth deformation models sensitive to mantle viscosity; only a narrow viscosity range yielded Holocene regrounding in earlier studies, often occurring earlier than new age constraints. Regional observations indicate modern Ross Embayment stability (positive mass balance) associated with a cold ocean cavity flushed by High Salinity Shelf Water from polynyas, which limits modified Circumpolar Deep Water (mCDW) intrusion compared to the Amundsen sector. Paleo-proxy evidence (e.g., biomarkers, diatoms, MSA, elephant seal presence) points to mid-Holocene warm conditions and reduced polynya efficiency, with later Neoglacial cooling and strengthening of polynyas. Climate model-derived ocean forcings (e.g., TraCE-21ka) have biases and do not resolve cavities, motivating sensitivity experiments with prescribed Holocene cavity warming. Previous modeling and GIA theory highlight gravitational, rotational, and deformational (GRD) sea-level effects that can stabilize or destabilize grounding lines, underscoring the need for self-consistent GIA treatment.

• Ice sheet model (ISM): Parallel Ice Sheet Model (PISM v2.0.3), hybrid stress balance, 20 km resolution with subgrid grounding line scheme, iceberg calving from strain rates plus a minimum-thickness criterion. Long spin-up: 20-year SIA smoothing; 130 kyr thermal equilibration with EPICA Dome C anomalies; 130 kyr full-physics; 20 kyr constant glacial maximum climate before transient runs.
• Forcing: Surface climate (temperature and precipitation) anomalies from WAIS Divide ice core applied to modern climatology using a PDD melt scheme (60% refreezing). Ocean thermal/salinity fields from CCSM3 TraCE-21ka at 500 m depth for basal melt via a thermodynamic parameterization tuned to reproduce modern cold-cavity and high-melt settings. Global mean sea-level forcing from deglacial proxy compilations.
• Ocean forcing sensitivity: To represent unmodeled cavity processes, imposed Ross Sea Holocene ocean temperature anomalies up to about +0.6°C (main analyses +0.3 to +0.5°C), typically from 7 ka to 1.6 ka BP (some tests to 1.0 ka BP) to emulate a warm cavity period based on regional proxies for mCDW intrusion, then reverting to standard TraCE-21ka forcing (relative cooling) to emulate transition to a cold cavity.
• Earth deformation in ISM: One-dimensional viscoelastic Earth model (elastic plate over viscous half-space) with parameters varied in an ensemble: mantle viscosity 1e19–5e21 Pa s (focus on 5e20–1e21 Pa s for Holocene retreat/readvance), lithosphere flexural rigidity 1e23–5e25 N m, mantle density 3300–4500 kg m−3. Ensemble size n≈270 exploring ice dynamics and forcing parameters.
• Global GIA modeling (GRD effects): A 1D radially varying, gravitationally self-consistent sea-level model with shoreline migration, solving normal-mode Earth response. Two end-member Earth structures: (i) GIAWE (weak Earth): lithosphere 60 km, upper mantle 1e18–1e19 Pa s, lower mantle 1e22 Pa s (Amundsen-like); (ii) GIASE (strong Earth): lithosphere 90 km, upper mantle 5e20 Pa s, lower mantle 5e21 Pa s (Siple Coast-like). Time window algorithm for efficiency; updates every 100 years with Antarctic ice thickness histories from ISM simulations.
• Iterative coupling tests: Additional advance-phase experiments starting near minimum extent (~2.0 ka BP) with bed topography held fixed vs applying bed changes from GIAWE and GIASE to assess the role of GIA-driven bed evolution during readvance.
• Model-data comparison: Grounding line trajectories compared to age constraints from radiocarbon-dated subglacial/grounding-zone sediments and input–decay modeling for WIS, MIS, BIS, KIS; relative sea-level reconstructions from Scott Coast and Terra Nova Bay used to assess GRD realism.

• The reversal from mid-Holocene retreat to late Holocene readvance is best explained by a transition from a warm to cold ocean cavity; GIA modulates sensitivity but is not the primary driver.
• Only simulations with mid-Holocene ocean temperature anomalies of at least about +0.5°C (averaged over 100-year intervals) produce retreat distances reaching WIS, MIS, BIS, and KIS sites, consistent with proxy ages. Lower anomalies (+0.3–0.4°C) yield more limited retreat and/or earlier timing mismatches.
• Upon removal of anomalous warming (relative cooling), grounding lines readvance rapidly: WIS, MIS, and KIS return near modern positions within ~400 years; BIS lags by ~200 years, indicating lower sensitivity at BIS relative to WIS/MIS.
• Delayed rebound alone is inconsistent with geological age constraints: with no or low Holocene warming, reproducing the observed retreat requires relatively high mantle viscosity, produces retreat too early, and yields gradual readvance over millennia rather than centuries.
• GRD effects evaluated with global GIA models show that during advance the local bed tends to subside upstream of the grounding line (due to increased ice load), contradicting the notion that prolonged delayed uplift drives readvance. No models predict delayed uplift dominating the advance phase.
• Bed response and relative sea-level histories are better captured by global GIA with GRD than by the ISM’s simplified deformation; GIAWE matches Scott Coast/Terra Nova Bay relative sea-level trends better than the simplified ISM but yields unrealistically large uplift that would ground parts of the Ross Ice Shelf, suggesting Siple Coast Earth structure is stronger (GIASE-like).
• Basal melt changes dominate mass balance: retreat-phase basal melt rates are <50% of Amundsen Sea sector bulk rates (i.e., generally <~10 m yr−1 versus up to ~20 m yr−1 there); cooling reduces basal melt while ice flux remains high, thickening shelves and promoting regrounding. Surface mass balance variations (<0.2 m yr−1 at WAIS Divide) are an order of magnitude smaller and cannot account for the advance without implausible increases.
• Proxy synthesis supports the regime shift: mid-Holocene warm conditions, reduced polynya efficiency, and likely mCDW intrusions (6–2.8 ka BP; 1.6–0.7 ka BP) coincide with retreat; late Holocene Neoglacial cooling and strengthening polynyas favor a cold cavity and readvance, with centennial variability (e.g., elephant seal presence ~2.3–1.8 ka BP) aligning with transient warm episodes and delayed advances at BIS/KIS.

The modeling tests directly address the competing hypotheses. A warm-to-cold cavity regime shift reproduces both the magnitude and rapid timing of late Holocene readvance inferred from radiocarbon constraints, whereas delayed isostatic rebound alone fails to match the proxy-constrained chronology and dynamics. GRD-informed GIA modeling shows subsidence during readvance and no prolonged delayed uplift signal, further arguing against rebound-driven regrounding as the primary mechanism. Earth rheology still matters by shaping bed evolution and thus modulating sensitivity to ocean forcing; a relatively strong Siple Coast Earth structure (higher viscosity, thicker lithosphere) is more consistent with observed grounding line behavior than a weak Earth structure that would unrealistically ground the shelf. Paleoclimate proxies (MSA, diatoms, biomarkers, elephant seals, temperature reconstructions) coherently indicate mid-Holocene warmer oceans with weak polynyas transitioning to late Holocene cooler conditions with stronger katabatic winds and polynya efficiency, aligning with a shift from warm to cold cavity states. Bed topography changes may also have indirectly influenced sub-ice-shelf circulation by modifying the pathways for mCDW, suggesting an interplay between long-term rebound and ocean circulation state changes. The findings imply that the Siple Coast grounding line is highly sensitive to sub-ice-shelf ocean heat flux and circulation regime, underscoring vulnerability under future reductions in sea ice and changes in Southern Ocean overturning driven by anthropogenic forcing.

The study demonstrates that a Holocene reversal of the Siple Coast grounding line—from significant mid-Holocene inland retreat to late Holocene readvance—is most plausibly driven by an ocean cavity regime shift from warm (mCDW-influenced) to cold (polynya-fed HSSW) conditions. Glacioisostatic adjustment modulates the ice-sheet response but does not primarily drive the readvance; GRD-consistent models show subsidence during advance and no delayed uplift dominating the process. Model–proxy consistency requires modest mid-Holocene cavity warming (≈+0.5°C) and yields rapid readvance on centennial timescales once cooling occurs. These insights emphasize the critical role of sub-ice-shelf ocean conditions for WAIS stability in the Ross Embayment. Future work should include cavity-resolving regional ocean models coupled to ice sheets to disentangle the contributions of bed topography versus climate-driven circulation changes, and incorporate laterally varying (3D) Earth structure for improved GRD realism. Given projected changes in Southern Ocean circulation and Antarctic sea ice, the Siple Coast grounding line remains vulnerable to increased ocean thermal forcing.

• Ocean forcing relies on coarse-resolution CCSM3 TraCE-21ka fields that do not resolve ice-shelf cavities; imposed Holocene temperature anomalies approximate, rather than simulate, cavity processes and mCDW intrusions.
• ISM resolution (20 km) and simplified parameterizations (e.g., calving law, PDD scheme) may limit process fidelity; some ensemble members still under-represent retreat extent indicated by geological data.
• Earth structure is treated as 1D in the GIA model and simplified in the ISM; strong lateral heterogeneity across the Ross Embayment is not fully captured, affecting GRD predictions and bed evolution.
• Limited direct observational constraints (e.g., few bedrock or RSL records near the Siple Coast, across a tectonic divide) hinder rigorous validation of uplift/subsidence histories.
• Timing and magnitude of imposed warming/cooling intervals and amplitudes involve uncertainties; centennial-scale variability is not explicitly resolved in all experiments.
• Differences among ice streams (e.g., BIS lag) may reflect unmodeled variations in water mass properties, cavity geometry, or circulation not represented without cavity-resolving ocean models.