Climate-controlled submarine landslides on the Antarctic continental margin

Submarine landslides are global geohazards capable of mobilizing volumes of sediment far exceeding terrestrial landslides and generating tsunamis with significant socio-economic impacts. Historic examples include the 1929 Grand Banks and 1998 Papua New Guinea events, and the Holocene Storegga slide. Around Antarctica, observations and models show tsunamis from distant sources can reach its margins, implying that tsunamis generated on Antarctic margins could likewise impact distant Southern Hemisphere coasts. However, few submarine landslides are documented around Antarctica, leaving uncertainty about their frequency and hazard potential, especially as subsea telecommunications to Antarctica are being considered. Glaciated continental margins, comprising ~20% of global continental margins, are sensitive to climate change. On high-latitude margins, multiple mechanisms have been implicated in slope failure, including changes in sedimentation and lithology, gas hydrate dissociation, ice-proximal dynamics, glacioisostatic seismicity, and relative sea-level changes. Rapid glacigenic sedimentation can elevate pore pressures and reduce effective stress, promoting failure. Because many triggers are climate-sensitive, warming may increase landslide likelihood. On glacial–interglacial timescales, climate-driven changes in sedimentation can form weak layers, notably diatom oozes deposited during interglacials within seasonal sea-ice zones under contouritic, hemipelagic, and plumitic regimes. Variations in sea ice and ocean temperatures influence diatom production, and failure along weak layers can result from rapid loading by glacigenic sediments. Yet identifying weak layers is challenging due to data limitations and dating uncertainties. The Iselin Bank on the eastern Ross Sea slope overlies a faulted basin within the passive West Antarctic Rift System. The region records repeated glacial advances since the early Miocene, prominent contourite deposition, and present interactions among Antarctic Slope Current, Circumpolar Deep Water, and Ross Sea Bottom Water. Here, we present a high-resolution, multidisciplinary analysis integrating downhole logs, lithology, chronology, and seismic data from IODP Expedition 374 to identify weak layers beneath a large submarine landslide complex on the Iselin Bank, constrain minimum landslide ages, and assess climate’s role in preconditioning and triggering Antarctic submarine landslides.

Prior studies on high-latitude margins document that submarine landslides can be triggered by climate-sensitive processes: rapid changes in sedimentation and lithology, gas hydrate dissociation, glacioisostatic rebound–related seismicity, sea-level fluctuations, and ice-proximal dynamics. Interglacial conditions can enhance diatom productivity and deposition, forming diatom oozes that alter sediment geotechnical properties (higher water and opal content, increased compressibility and permeability), potentially creating weak layers. Loading by overlying glacigenic sediments with low permeability can trap pore fluids, increase overpressure, and reduce shear strength, preconditioning slopes to fail. Numerical and physical models emphasize the importance of sediment compressibility, sedimentation rates, and stratigraphic contrasts in slope stability, particularly on low-gradient, contourite-influenced margins. In Antarctica, regional seismic stratigraphy (Ross Sea Sequences and Unconformities) and prior drilling (e.g., ANDRILL, IODP Sites) show Neogene to Quaternary variability in ice extent, sea ice, and bottom currents, with intervals of enhanced diatom deposition during warm periods (Miocene Climate Optimum; mid-Pliocene Warm Period) and increased glacigenic sedimentation during cooling phases. Large Holocene failures on Northern Hemisphere glaciated margins (e.g., Storegga) underscore the potential for climate-forced preconditioning and seismic triggering, but the Antarctic record remains sparse due to limited data coverage.

The study integrates geomorphology, seismic stratigraphy, coring, downhole logging, and biostratigraphic chronology. Geophysical data: Multibeam echosounder (12 kHz Reson SeaBat 7150/8111) acquired during 2017 ANTSSS and ODYSSEA cruises were processed to 30 m grids; slope gradients were computed in ArcGIS. High-resolution single-channel seismic (IT17RS301/302/303/315/316) used two 210 cu in GI guns, 9.5 m mini-streamer, with ~3 m vertical resolution after processing (bandpass filtering, deconvolution, stacking, time migration). Legacy multichannel seismic lines (BGR80-009A, IT94A127A) were reprocessed (Seismic Unix; Paradigm Echos/Geodepth) achieving ~17 m and ~13 m vertical resolution, respectively. A sound velocity of 1600 m s−1 converted two-way travel time to depth. Submarine landslide volumes were estimated by interpolating a pre-failure surface (Topo to Raster) from headwall-adjacent depths and differencing with modern bathymetry (Cut/Fill) at 30 m grid resolution, yielding minimum evacuation volumes. Sediment and log data: IODP Site U1523 (828 m water depth) on the outer Ross Sea shelf included five closely spaced holes (A–E). High-resolution shipboard physical properties (bulk density, MS, P-wave velocity, spectral gamma, color) and post-cruise core-scanning XRF (1–2 cm) provided proxies for grain size and opal content (Zr/Rb, Ti/Al, In(Si/Al)). Grain size of the <1 mm fraction was measured every 15 cm (Mastersizer 2000); coarse fraction (>1 mm) by dry sieving. Undrained shear strengths (Torvane) and MAD-derived porosity/densities were measured on discrete samples. Downhole logs (Hole U1523D) included modified triple combo and FMS-sonic; porosity (SPHI) derived from sonic; Vp/Vs from compressional and shear velocities; FMS images characterized resistivity textures. Core–log–seismic integration used P-wave and sonic velocities to build a depth–time relationship (Petrel) and trace weak layers and post-failure reflectors between seismic profiles and Site U1523. Chronology: A revised age model combined dense diatom biostratigraphy (20–40 cm spacing), radiolarian and dinoflagellate events, and two magnetic polarity reversals. Composite depth scale across holes incorporated core gaps and recovery percentages. Two linear accumulation segments constrained 0–1.78 Ma (0–21.7 m CSF-A; 1.219 cm kyr−1) and 1.78–2.58 Ma (21.7–48.52 m; 3.352 cm kyr−1). Minimum ages of weak layers were obtained by tracing horizons directly beneath scarps into Site U1523; minimum submarine landslide ages were derived from post-failure reflectors traced into Site U1523. Regional seismic ties to Ross Sea Sequences/Unconformities constrained ages of mass-transport deposits on the rise. Uncertainties reflect biostratigraphic sampling, reworking, core gaps, and seismic resolution.

• A >6000 km² submarine landslide complex extends >100 km along the Iselin Bank upper slope (average slope ~6.5°), with multiple scarps >100 m high. Southern region main scarps: S1 (volume ~19 km³; water depth ~1116 m; area ~141.5 km²) and S2 (~13 km³; ~1500 m; ~106 km²); northern region hosts a crescent-shaped failure >70 km³ (>370 km², ~1300 m water depth). Four to five large scarps also occur on the continental rise. • Seismic profiles identify three laterally continuous weak layers beneath scarps: WL1 and WL1b beneath S1 and S1b, and WL2 beneath S2; weak layers traceable >15 km along-slope. • Mass-transport deposits on the rise include the Iselin MTD (>960 km², max thickness ~0.3 s TWT ≈240 m, volume ~230 km³) and a buried MTD (~50 km², ~0.17 s TWT ≈136 m thick, min. volume ~7 km³). • Core and log data show weak layers consist of diatom oozes/diatom-rich muds with high porosity and low density, overlain by coarser glacigenic packages (sandy muds, gravels, diamicts) with higher density and shear strength. Examples: WL1 (66.2–70.2 mbsf) muddy diatom ooze with dispersed clasts; porosity ~63%, bulk density ~1.6 g cm⁻³, shear strength ~1.65 kg cm⁻²; overlying package shows increased grain size (D4,3 ~174 µm), density (~2.0 g cm⁻³), MS (~128×10⁻⁵ SI), shear strength (~3.5 kg cm⁻²), and decreased porosity (~52%). WL1b (56.6–60.9 mbsf) exhibits similar contrasts. WL2 (~267 mbsf) shows moderate velocity (~1780 m s⁻¹), moderate MS (~2070 counts), higher porosity (~42%), low resistivity with interbedded diatom-rich muds; overlying package is diamictic with higher velocity (~1880 m s⁻¹), MS (~2260), and lower porosity (~41%). • Chronology: Minimum weak layer ages—WL1b: ~2.93–2.82 Ma; WL1: ~3.21–3.07 Ma; WL2 likely >13 Ma, extrapolated ~14.8 Ma. Minimum landslide ages based on post-failure reflectors—S1: <0.4 Ma; S1b: ~1.72 Ma; S2: ~12.14 Ma. Rise MTDs constrained to ~11–14.5 Ma (buried MTD) and <2.5 Ma (Iselin MTD). • Paleoenvironmental interpretation links WL1 and WL1b deposition to late Pliocene warm intervals (mPWP; MIS G17/G11) with reduced sea ice, warm-water incursions, and enhanced diatom productivity; WL2 to the end of the Miocene Climate Optimum with warm Southern Ocean and reduced ice. Overlying diamicts reflect subsequent cooling, enhanced ice proximity/IRD, and stronger bottom-current winnowing. • Gas hydrate dissociation is unlikely as a trigger (no seismic gas indicators near scarps; stability likely at these depths). Excess pore pressures from diatomaceous weak layers alone are insufficient to cause failure at observed low sedimentation rates; an external trigger is required. • Proposed trigger: seismicity associated with glacioisostatic adjustment following ice loading/unloading during major climate transitions; contourite mound geometry and current scouring may have contributed to over-steepening and local undercutting. • Hazard implication: Landslide and MTD volumes (up to ~230 km³) are sufficient to generate tsunamis analogous to other high-latitude failures; teleconnections could impact Southern Hemisphere coasts and Antarctic infrastructure.

The study demonstrates that Antarctic continental slope failures on the Iselin Bank are controlled by climate-driven alternations in sedimentation that create mechanically weak, diatomaceous layers overlain by dense, low-permeability glacigenic units. This lithologic juxtaposition promotes pore-pressure buildup and reduced shear strength, preconditioning the slope. The timing of weak layer formation aligns with globally warm intervals (mPWP and MCO), while failure ages follow subsequent cooling or ice-sheet reorganizations, consistent with a scenario where seismicity from glacioisostatic readjustment acts as the proximal trigger. Contourite mound construction and strong bottom currents further modulate slope gradients and potential undercutting, enhancing instability. These findings address the central question of how climate influences preconditioning and triggering of Antarctic submarine landslides by linking depositional regimes, oceanographic variability, and ice-sheet dynamics to weak layer development and failure timing. Given projected warming, retreat, and potential increases in glacioisostatic seismicity, similar Antarctic margin settings may experience heightened landslide and tsunami risk, with implications for submarine cables and coastal hazards.

Recurrent Neogene slope failures along >100 km of the Iselin Bank are tied to weak layers of water-rich diatom oozes deposited during warm intervals (MCO, mPWP) and subsequently overlain by dense glacigenic diamicts during cooler phases. Core–log–seismic integration constrains weak layer ages (WL1b ~2.93–2.82 Ma, WL1 ~3.21–3.07 Ma, WL2 ~14.8 Ma) and minimum landslide ages (S1 <0.4 Ma, S1b ~1.72 Ma, S2 ~12.14 Ma). While climate warmth fosters weak layer formation, glacioisostatic seismicity during ice retreat is the most likely failure trigger. The volumes involved (up to ~230 km³ MTDs) indicate a credible tsunami hazard from Antarctic margins. Future work should prioritize mapping and dating weak layers around Antarctica, improving age models where core recovery is low, acquiring higher-resolution seismic data downdip of headwalls, and quantifying in situ geotechnical properties (permeability, compressibility) to refine stability models and tsunami hazard assessments, particularly in regions targeted for submarine cable routes.

• Chronologic uncertainties arise from biostratigraphic sample spacing, reworking, core recovery gaps (e.g., 52% below 120 mbsf; extremely low recovery below ~155 mbsf), and reliance on linear age–depth segments; WL2 age is extrapolated beyond the calibrated depth range. • Seismic resolution limits (single-channel ~3 m; MCS ~13–17 m) and limited down-slope data reduce precision in correlating scarps to deposits and MTD extents. • Direct measurements of permeability and comprehensive geotechnical testing were not performed; interpretations rely on analog relationships for similar sediments. • Gas presence/hydrate influences cannot be entirely excluded despite lack of seismic indicators near scarps. • Volume estimates are minima due to incomplete bathymetric coverage down-slope of headwalls.