Description of submarine landforms

Multibeam-bathymetric data, together with sub-bottom and seismic-reflection profiles, show the distribution of seafloor and sub-seafloor landforms and sediments on the Yermak Plateau (Figs 1, 2, 3). The Yermak Plateau, NW of Spitsbergen, is mainly 600–1000 m below present sea level (Fig. 1a, b). It also encompasses several areas shallower than 600 m, including a southerly ridge crest, the Sverdrup Bank, and NE and western topographic basement highs (Fig. 1a).

Seafloor landforms on the Yermak Plateau include elongate ridges, linear to curvilinear depressions and a moat (Figs 1, 2, 3). Linear and slightly curvilinear elongate ridges are identified in three areas: on the southern ridge crest in water depths generally <600 m (Fig. 1c); in a small area around 80° N, 10° E; and to the north and west of Sverdrup Bank in water depths of 725–850 m (Fig. 1e) (Dowdeswell et al. 2010; Gebhardt et al. 2011). The elongate ridges on the southern ridge crest and at 80° N, 10° E are orientated SSE–NNW (Fig. 1c), with a mean crest-to-trough height of 7–10 m and wavelength of around 500 m (Fig. 1d). The elongate ridges to the north and west of Sverdrup Bank are orientated NNE–SSW, with a crest-to-trough depth of up to 25 m and a wavelength of several hundred metres (Fig. 1e, f).

Two types of linear to curvilinear depressions are found (Fig. 2a–d). The first are wide depressions that have parallel to sub-parallel conformity and are ≤6 km long, 600 m wide and 20 m deep (Fig. 2a, b) (Dowdeswell et al. 2010). A second type occurs as narrow, isolated features with irregular orientations (Figs 2c, 3) (Vogt et al. 1994; Dowdeswell et al. 2010; Gebhardt et al. 2011). They are generally less than 1 km long, 100 m wide and 10 m deep.

A curved moat, 2 km wide, is a distinctive landform on the Plateau (Figs 1a, 2e) (Vogt et al. 1994; Dowdeswell et al. 2010). The flat seafloor of the moat reaches 1200 m deep and is featureless at the resolution of our data (around 40 m); its north and west margin is characterized by a 400 m high slope of about 25° that is dissected by gullies (Fig. 2e).

Interpretation of submarine landforms

The elongate ridges on the Yermak Plateau crests could be interpreted as large lineations formed by either: (a) a grounded ice sheet from the south; (b) the keels of an ice-shelf remnant from the north; or (c) deep-keeled icebergs trapped within multi-year sea ice (Dowdeswell et al. 2010; Gebhardt et al. 2011). The lineations are similar to mega-scale glacial lineations (MSGLs), which are produced in deforming subglacial sediments (Clark 1993). However, it is unlikely that the large lineations were formed by the northward flow of a grounded ice sheet from Svalbard due to the lack of overconsolidation observed in sediments retrieved from this unit at ODP Site 910 (Fig. 1a) (Dowdeswell et al. 2010). Sediment overconsolidation is expected after long ice-sheet occupation. If the lineations were produced by a grounded ice sheet, it probably persisted for a relatively short period and/or was characterized by relatively thin ice (Dowdeswell et al. 2010). The large lineations (Fig. 1c–f) are also compatible with a floating ice-shelf remnant or deep-keeled icebergs trapped within multi-year sea ice impinging on the Yermak Plateau, potentially derived from the Arctic Basin to the north (Dowdeswell et al. 2010; Jakobsson et al. 2010; Gebhardt et al. 2011) or a larger coherent ice shelf covering the entire central Arctic Ocean (Jakobsson et al. 2016).

The dimensions and geometry of the two types of linear to curvilinear depression on the Yermak Plateau suggest that they are iceberg ploughmarks (Dowdeswell et al. 1993). The wide, parallel to sub-parallel ploughmarks (Fig. 2d) are probably formed by the keels of several icebergs that were trapped within multi-year sea ice, which provided a uniform pattern of iceberg drift. The narrow, irregular depressions (Fig. 2c) are interpreted as iceberg ploughmarks of single icebergs (Vogt et al. 1994; Dowdeswell et al. 2010; Gebhardt et al. 2011). Some narrow ploughmarks cross-cut the large lineations and the wide, parallel ploughmarks (Fig. 2c), implying a younger relative age.

The moat (Fig. 2e) is interpreted as a current-related overflow channel. It was probably formed by the NE-flowing West Spitsbergen Current, which was established in the early Holocene and is consistent with the water-flow direction around Svalbard today (Vogt et al. 1994; Dowdeswell et al. 2010).

Description of acoustic facies and reflectors

Three acoustic facies are described from the shallow stratigraphy of the Yermak Plateau: Facies A, B and C are acoustically stratified, impenetrable/chaotic and transparent, respectively (Figs 1, 2).

Facies Ai overlies the large lineations; it has a draping geometry and a variable thickness of 5–25 m (Fig. 1g, h, j) (Gebhardt et al. 2011). To the NE of Sverdrup Bank, the lineations are overlain by up to 25 m of Facies Ai. To the NW and west of the western basement high, lineations are overlain by 10–15 m and 5 m of Facies Ai, respectively (Fig. 1g–j). Facies Aii is present on the Yermak Plateau flanks and has smooth lower and upper reflections (Fig. 2f).

Facies B is acoustically impenetrable and chaotic on sub-bottom and seismic-reflection profiles, respectively (Fig. 1i, j). Facies Bi is up to 50 m thick and its upper reflection, R1, is the hummocky surface into which the large lineations are incised. Its lower reflection is a prominent erosional unconformity at 70–90 m below the seafloor (Fig. 1i) (Kristoffersen et al. 2004). This unconformity is traced to depths of 850 m below present sea level and extends to c. 82° N (Gebhardt et al. 2011). Facies Bii occurs in association with iceberg-ploughed areas of the seafloor. It has an irregular upper reflection and is overlain by a thin sediment veneer (Fig. 2d).

Facies C forms acoustically transparent lenses at several different depths within well-stratified sediment (Fig. 2g). It is present on the SW side of the southern ridge crest (Dowdeswell et al. 2010) and along the foot of Sverdrup Bank and the western basement high (Gebhardt et al. 2011).

Interpretation of acoustic facies and reflectors

Facies Ai (Fig. 1g, h) is interpreted as hemipelagic sediment draping the large lineations. Sediment cores from the Yermak Plateau suggest that the age of this sediment extends to at least MIS 5 (Dowdeswell et al. 2010). The thicker sediment overlying the lineations to the NE of Sverdrup Bank is probably a result of higher sedimentation rates due to bottom-current activity (Gebhardt et al. 2011). Facies Aii is interpreted as hemipelagic sediment draping the flanks of the Yermak Plateau (Fig. 2f) (Gebhardt et al. 2011).

Facies Bi (Fig. 1g, h) is inferred to be subglacial till deposited by a grounded ice sheet (Gebhardt et al. 2011). This is supported by its chaotic character on seismic-reflection profiles and its position directly overlying a prominent erosional unconformity (Fig. 1i). The unconformity corresponds with the top of an overconsolidated unit in ODP Site 910, which has been attributed to the grounding of a marine-based ice sheet around the MIS 19/20 boundary (Kristoffersen et al. 2004; Knies et al. 2007).

The highly irregular upper surface of Facies Bii (Fig. 2d) results from the ploughing of seafloor sediments by iceberg keels. The thin sediment drape (up to 5 m) overlying the ploughmarks suggests that they are younger than the large lineations, which are overlain by 5–25 m of sediment (Gebhardt et al. 2011).

The transparent acoustic character of Facies C (Fig. 2g), together with its location at the foot of several topographic highs, suggests that it represents mass-transport deposits. The presence of mass-transport deposits at different depths within well-layered sediments indicates that the downslope movement of debris occurred during a number of different events, possibly as floating ice contacted the floor of the Yermak Plateau, bulldozing sediment down the western flanks of its topographic highs (Dowdeswell et al. 2010; Gebhardt et al. 2011).

Discussion: evidence of ice grounding on the Yermak Plateau

The assemblage of landforms and sediments preserved on the Yermak Plateau (Fig. 3) provides evidence for at least three major ice-grounding events. (a) An early ice-sheet grounding event around MIS 19/20 (Dowdeswell et al. 2010; Gebhardt et al. 2011). (b) At least one subsequent ice-sheet-, ice-shelf- or iceberg-grounding event, inferred from the large lineations close to the seafloor (Figs 1c–f, 3). Although the lineations have not been dated directly, an MIS 6 age has been proposed (Dowdeswell et al. 2010; Jakobsson et al. 2010). (c) Iceberg grounding during the Late Pleistocene is inferred from iceberg ploughmarks in water depths of 400–800 m (Fig. 2a–d) (Vogt et al. 1994; Dowdeswell et al. 2010).

The landform assemblage on the Yermak Plateau (Figs 1, 2, 3) is unusual compared with assemblages that are typically formed by the advance and retreat of grounded ice, which generally include MSGLs and grounding-zone wedges (e.g. Ottesen et al. 2005). The location of the Yermak Plateau, seaward of the shelf break off NW Spitsbergen at the extreme NW margin of past Eurasian ice sheets and probably offshore of inter-ice stream ice-sheet drainage basins, provides an explanation of why it was only covered by ice sheets during particularly extensive full-glacial periods. During some other glacial periods, the Yermak Plateau was probably subjected to ice-grounding events from extensive Arctic ice shelves.

Similar assemblages of glacial lineations and iceberg ploughmarks, also dated to MIS 6, have been identified in water depths down to 1000 m on a number of bathymetric highs in the Arctic Ocean, including the Chukchi Borderlands and Lomosonov Ridge (Kristoffersen et al. 2004; Polyak et al. 2007; Jakobsson et al. 2008; Jakobsson 2016). These data provide evidence for a marine-based ice sheet with an extensive ice shelf on the northern Eurasian shelf and in the Arctic Ocean during MIS 6 (Jakobsson et al. 2014, 2016; Jakobsson 2016). This pattern of mapped glacial landforms in the Arctic Ocean gives new impetus to the concept of an ice shelf covering the entire central Arctic Ocean (Jakobsson et al. 2016).