Cretaceous–Paleogene tectonic reconstructions of the South Scotia Ridge and implications for the initiation of subduction in the Scotia Sea

The Cenozoic separation of the Antarctic Peninsula and South America, and the accompanying opening of Drake Passage, was driven by the creation of the Scotia plate and the westwards subduction of the South American plate. The growth of the Scotia plate led to crustal block dispersal along the North and South Scotia ridges (Fig. 1). Today, these ridges form prominent submarine topographic highs and have a strong influence on the pathway of the Antarctic circumpolar current. The pre-translation position of the elevated banks and crustal blocks of the Scotia ridges are the source of considerable debate (e.g. Dalziel et al. 2021; Riley et al. 2022; Schellart et al. 2023), with only limited geological control on the origin of the submerged banks.

Detrital zircon provenance analysis on the crustal blocks that form the North Scotia Ridge point to a geological affinity with the Fuegian Andes (e.g. Carter et al. 2014; Riley et al. 2019; Dalziel et al. 2021), whereas Riley et al. (2022) established connections between parts of the South Scotia Ridge and the Antarctic plate. We present here the first direct geochronology results for mafic magmatic rocks from Bruce Bank and Jane Bank of the South Scotia Ridge and examine potential correlations with the Antarctic plate, the timing of the onset of subduction in the Scotia Sea and post-translation magmatism.

The Scotia Sea is located between the Antarctic Peninsula and the South Atlantic Ocean and forms a key component of the circumpolar Southern Ocean. The Scotia Sea is bound by the submerged mountainous belts of the North and South Scotia ridges, which form extensive ridges up to 1500 km in length and extend from the Shackleton Fracture Zone in the west to the East Scotia Ridge in the east (Fig. 1). Prior to the development of the Scotia plate, the Magallanes and Central Scotia plates formed as a result of plate reorganization and later (Oligocene–Miocene) seafloor spreading on the West Scotia Ridge (Eagles and Jokat 2014; Maldonado et al. 2014) and also spreading on the East Scotia Ridge from c. 15 Ma to the present day (Larter et al. 2003). The central zone of the Scotia Sea is less well constrained and consists of oceanic crust and potentially extended continental crust of uncertain age (Eagles 2010). Proposed oceanic crustal ages for the central Scotia Sea range from 21 to 7 Ma (Barker 2001), Miocene (Vérard et al. 2012), Eocene to mid-Miocene (Beniest and Schellart 2020), Early Cretaceous (Dalziel et al. 2013b) and Late Jurassic to mid-Cretaceous (Eagles 2010; Riley et al. 2019).

The creation of the Scotia plate from the Late Paleogene to the Neogene developed alongside the westwards-directed subduction of the South American plate (Fig. 1). Subduction continues today at the South Sandwich Islands, which form an intra-oceanic island arc chain of volcanoes on the Sandwich plate (Fig. 1). Prior evidence for subduction of the South American plate beneath the Central Scotia plate during the Late Cenozoic has been identified in the central Scotia Sea, where Dalziel et al. (2013a) described a chain of eroded volcanoes that have been dated from the Oligocene to the Miocene. Dalziel et al. (2013a) interpreted the chain of eroded volcanoes as an ancestral island arc from the Late Cenozoic that predates the present-day South Sandwich island arc. The ancestral South Sandwich arc (ASSA; Fig. 2a) had a duration of at least 20 myr (c. 31–10 Ma; Dalziel et al. 2013a; Riley et al. 2021) and has geochemical characteristics indicating an intra-oceanic island arc setting (Pearce et al. 2014; Riley et al. 2021).

Other studies have indicated a much longer geological history for the ASSA, with subduction of the Antarctic plate in the NW Weddell Sea (Fig. 1) potentially initiating during the Late Cretaceous (Eagles 2010; Vérard et al. 2012; van de Lagemaat et al. 2021) or Early Paleogene (Maldonado et al. 2014), but with no determined age to support an older history. The onset of subduction was interpreted to be the result of a shift in relative motion between the continental blocks of South America and the Antarctic Peninsula (Barker 2001; Eagles 2010; van de Lagemaat et al. 2021). This change in relative plate motion would have led to convergence at a plate boundary in the northwestern sector of the Weddell Sea (the Endurance Collision Zone; Fig. 1) during the mid-Cretaceous (Barker et al. 1991; van de Lagemaat et al. 2021), leading to the early initiation of subduction.

Schellart et al. (2023) proposed a different mechanism for the initiation of subduction in the Scotia Sea as a result of a subduction invasion polarity switch, with stresses from the long-lived eastwards-directed subduction beneath the Antarctic–South American margin leading to the initiation of westwards subduction, potentially in the mid-Cretaceous. The model suggests that compressive stresses develop in the overriding plate as a result of strong trenchwards basal drag induced by subduction-driven whole-mantle poloidal return flow.

A factor in understanding the mid-Cretaceous tectonics of the proto-Pacific margin prior to the development of the Scotia Sea is the closure of the Rocas Verdes back-arc basin (Fig. 1) at c. 100 Ma associated with underthrusting of the Rocas Verdes oceanic crust prior to sedimentation (Muller et al. 2021).

The South Scotia Ridge is a prominent east–west-trending array of submarine crustal fragments (the Terror, Pirie, Bruce, Discovery and Herdman banks) separated by steep-sided ocean basins (the Ona, Protector, Dove and Scan (Discovery) basins; Fig. 2) that extends from the Shackleton Fracture Zone in the west to the Sandwich plate in the east (Fig. 1). The South Scotia Ridge is cut by a sinistral transform fault that separates the Scotia and Antarctic plates (Fig. 1). Separate from the continental fragments of the South Scotia Ridge is the larger crustal block of the South Orkney microcontinent, which is exposed at its northern margin, forming the South Orkney Islands (Fig. 2). The South Orkney microcontinent was contiguous with the northern Antarctic Peninsula, but separated during the Late Eocene–Oligocene rifting that led to the development of the Powell Basin (Eagles and Livermore 2002), the Scan Basin (Eagles et al. 2006) and presumably other rifted basins of the South Scotia Ridge (Pérez et al. 2019). Its geology is predominantly Permian–Triassic turbidites forming part of an accretionary complex (Flowerdew et al. 2011; Riley et al. 2023).

The South Orkney Islands succession is cut by a suite of dykes that were provisionally dated as Early Cenozoic (referred to in King and Barker 1988) and were correlated to Cenozoic arc magmatism of the Antarctic Peninsula. Basaltic rocks with an arc composition have also been dredged (D.93, D.95, D.96; Fig. 2) from the western and eastern margins of the South Orkney microcontinent and have been dated in the interval 85–68 Ma (K–Ar whole-rock analysis; Barber et al. 1991), whereas basalts recovered from the southern margin of Powell Basin (D.64; Fig. 2) have been dated in the interval 49–47 Ma (Barber et al. 1991).

The geology of the crustal blocks of the South Scotia Ridge are largely unknown (e.g. Lodolo et al. 2010; Civile et al. 2012), but are generally considered to be thinned continental crust on the basis of seismic reflection profiles that are not characteristic of basaltic crust (Galindo-Zaldívar et al. 2002). Udintsev et al. (2012) reported dredged lithologies of paragneiss, granitoid and sedimentary rocks that they interpreted to be in situ and may correlate with the geology of the South Orkney Islands. The geology of the southern margin of Bruce Bank has been examined by Riley et al. (2022), who investigated the provenance of metasedimentary lithologies dredged from depths of c. 850 m. Detrital zircon geochronology of the metasedimentary units from Bruce Bank are consistent with a direct correlation to the Mesozoic successions of the South Orkney Islands and also share a geological affinity with the northern Antarctic Peninsula and components of the Cordillera Darwin metamorphic complex of Tierra del Fuego (Fig. 1).

To the east of Bruce Bank, Discovery Bank and the more southerly Jane Bank (Fig. 2a) are interpreted to have a geological history that is distinct from the crustal blocks further west. Parts of the Discovery and Jane banks are interpreted to preserve Oligocene–Miocene volcanism that developed in an intra-oceanic island arc setting and represent the southern sector of the ASSA (Barker et al. 1984; Pearce et al. 2014; Riley et al. 2021).

Several researchers have considered the Eocene configuration of the crustal blocks of the South Scotia Ridge prior to the development of the Scotia Sea and the opening of Drake Passage (e.g. King and Barker 1988; Vérard et al. 2012; Dalziel et al. 2013b, 2021; Eagles and Jokat 2014; Galindo-Zaldívar et al. 2014; Pérez et al. 2019; van de Lagemaat et al. 2021; Riley et al. 2022; Schellart et al. 2023). The broad consensus is that the continental fragments of the South Scotia Ridge, including the South Orkney microcontinent, originated close to the southern tip of the Fuegian Andes (Fig. 1) based on their Eocene palaeolocations. This was developed further by Riley et al. (2022), who used zircon age provenance combined with kinematic modelling to determine that the crustal blocks of the South Scotia Ridge had a close relationship with the Antarctic plate and the Cordillera Darwin metamorphic complex in the Late Mesozoic.

The topographic highs of the Bruce and Jane banks (Fig. 2a) are explored here to investigate their relationship with the geological successions of the South Orkney Islands, the other crustal blocks of the South Scotia Ridge and the northern Antarctic Peninsula/southern Patagonia. We investigate the mafic magmatic rocks of the central Bruce Bank to determine if they are related to the opening of Scan Basin (Pérez et al. 2014) or whether they have a pre-translation geological history. We also examine the age of Jane Bank volcanism and whether it forms an extension of the Oligocene–Miocene ASSA (cf. Pearce et al. 2014) or forms part of an older magmatic basement to the crustal block.

Bruce Bank is c. 120 km wide and is bounded by discrete basins on three sides: Scan Basin to the east, Dove Basin to the west and the deeply incised (c. 6000 m depth) Bruce Deep to the south, which separates Bruce Bank from the adjacent South Orkney microcontinental block (Fig. 2b). Seismic data from Scan Basin, east of Bruce Bank (Fig. 2a), indicates that the eastern margin of Bruce Bank has facies of continental or a transitional nature, in contrast with the widespread mafic intrusions and oceanic crust that characterize the seafloor of Scan Basin (Pérez et al. 2014). Eagles et al. (2006) referred to Scan Basin as the ‘Discovery Basin’ and suggested it opened prior to 42 Ma or at 36–33 Ma, which is in close agreement with the interpretation of Schreider et al. (2017), who determined that oceanic spreading took place in the northern part of the Scan Basin along a NW–SE-trending spreading centre during the interval 35–29 Ma. These estimates also overlap with an age of 30.2 Ma for basin formation derived from heat flow analysis (Barker et al. 2013), although Pérez et al. (2014) suggested a later (Oligocene) opening for Scan Basin.

Jane Bank is an elongate crustal block separated from the South Orkney microcontinent by Jane Basin (Fig. 2b). Jane Bank was considered by Pearce et al. (2014) to form a Miocene segment of the ASSA based on a general younging trend (from the available geochronology) from the north (Oligocene) to the south (Miocene) of the ancestral arc. No geochronology is available for Jane Bank, but geochemical analysis (Barker et al. 1984; Pearce et al. 2014) of samples collected at two dredge sites (Barker et al. 1984) were interpreted to infer an intra-oceanic arc setting akin to the neighbouring Discovery Bank (Fig. 2a). Jane Basin to the NW of Jane Bank (Fig. 2a) has been interpreted to represent a back-arc basin (Barker et al. 1984), although a post-subduction origin has also been proposed (Bohoyo et al. 2002).

Geological samples were dredged from several sites across Bruce Bank during British Antarctic Survey cruise DY088 on the RRS Discovery (March–April 2018) and sites from Jane Bank were dredged in March 1981 (cruise SHACK80) on the RRS Shackleton. Sampling sites with steep topography were selected to reduce the risk of sampling glacial dropstones and to increase the likelihood of recovering in situ lithologies.

Site selection on Bruce Bank was determined using the onboard EM-122 multibeam echosounder, combined with available GEBCO data (www.gebco.net/) and an SBP-120 echosounder for sub-bottom profile information to identify sites with only minor draped sediments. The bathymetry data indicated that the eastern and southern margins of the Bruce Bank would provide the most suitable sites for rock dredging because they were characterized by steeper slopes (>25°). The dredge sites on Jane Bank were located on east-facing steep (>25°) scarp slopes (Barker et al. 1984). The locations of the dredge sites are shown in Figure 2b–d.

Samples were dredged from five sites on the eastern and southern margins of Bruce Bank where the steepest scarp slopes (>25°) were identified following multibeam and sub-bottom profile surveys. Although the sediment cover was anticipated to be minimal on the steepest slopes, three sites (DR.223, DR.224 and DR.226; Fig. 2d) were characterized by the high recovery (>70%) of a poorly lithified, fine-grained sedimentary unit. This unit was examined by Riley et al. (2022), who interpreted an Eocene age and cool water deposition for the cover rocks based on their nannofossil assemblage. Alongside the poorly lithified units, coarse-grained, in situ sedimentary lithologies were recovered from site DR.223 and were interpreted to represent the bedrock beneath the more recent, poorly lithified cover sediments. The metasedimentary units were examined by Riley et al. (2022) and interpreted to be closely related to the Late Mesozoic successions exposed in the South Orkney Islands.

The two other dredge sites (DR.225 and DR.227; Fig. 2d) on Bruce Bank are characterized by mafic magmatic lithologies, including volcanic units, possible dyke rocks and coarser grained lithologies (gabbro and ultramafic units). The recovered material from the dredge sites of eastern Bruce Bank supports the interpretations of both Civile et al. (2012) and Pérez et al. (2014), who suggested that the margins of Scan Basin and eastern Bruce Bank are likely be dominated by mafic intrusions and volcanic units. Two samples from site DR.225 (c. 950 m depth) are examined here for their ⁴⁰Ar/³⁹Ar geochronology and provide the first direct geochronology results for igneous rocks from Bruce Bank. Sample DR.225.14 is from a mafic porphyritic lithology interpreted as a probable hypabyssal unit and DR.225.27 is a fine-grained basaltic lava with phenocrysts of olivine and pyroxene.

Four sites (DR.80, DR.81, DR.84 and DR.85; Fig. 2c) were selected for rock dredging on the SE margin of Jane Bank at depths in the range 5300–1700 m. The locations were identified based on slope angle analysis, with sites >25° deemed the most suitable for recovering in situ material with no sediment cover (Barker et al. 1984). The dredge sites yielded a significant return of mafic–silicic volcanic rocks and volcaniclastic lithologies characterized by a thin (few millimetres) Fe–Mn crust. A fine-grained basaltic lava (DR.84.8) from dredge site DR.84 (c. 1900 m depth) is dated here. The geochemistry of volcanic rocks from neighbouring dredge sites DR.80 and DR.85 was presented in Pearce et al. (2014) and is discussed further here.

⁴⁰Ar/³⁹Ar dating was performed at the Department of Earth Sciences, The Open University (Milton Keynes, UK). The samples were crushed using a pestle and mortar and the crushate was sieved and washed repeatedly in deionized water to remove dust and clay particles from the surfaces of all the size fractions. Using a binocular microscope, whole-rock pieces that were free from alteration were selected for further analysis. The picked separates were cleaned ultrasonically in acetone and deionized water, dried using a hot-plate and then packaged in aluminium foil packets c. 10 mm × 10 mm in size prior to irradiation.

Samples were irradiated at the McMaster Nuclear Reactor (McMaster University, Hamilton, Ontario, Canada) in reactor position 8E for 120 h using Cd shielding. The neutron flux was monitored using biotite mineral standard GA1550, which has an age of 99.738 ± 0.104 Ma (Renne et al. 2010). Standards were packed either side of the unknown samples for irradiation and analysed using the single-grain fusion method with a 1059 nm CSI fibre laser and an MAP215-50 mass spectrometer. The J-values were calculated by linear extrapolation between the two measured J-values. The values for each sample are shown in the data table and a 0.5% error on J-value was used.

The irradiated samples were loaded into an ultra-high-vacuum system and a 1059 nm CSI fibre laser was focused into the sample chamber and used to step-heat the basalt samples. After passing through a liquid nitrogen trap, the extracted gases were cleaned for 5 min using two SAES AP-10 getters (one at 450°C and one at room temperature), following which the gases were introduced into an MAP 215-50 mass spectrometer for analysis. The mass discrimination value was measured at 283 for ⁴⁰Ar/³⁶Ar (using a calibration noble gas mixture of known composition). System blanks were measured before and after every one or two sample analyses. The gas clean-up and inlet was fully automated, with measurements of ⁴⁰Ar, ³⁹Ar, ³⁸Ar, ³⁷Ar and ³⁶Ar, each for ten scans, and the final measurements are extrapolations back to the inlet time.

The system blanks measured before and after every one or two sample analyses were subtracted from the raw sample data. The results were corrected for ³⁷Ar and ³⁹Ar decay and neutron-induced interference reactions. We used correction factors of (³⁹Ar/³⁷Ar)Ca = 0.00065 ± 0.00000325, (³⁶Ar/³⁷Ar)Ca = 0.000265 ± 0.000001325 and (⁴⁰Ar/³⁹Ar)K = 0.0085 ± 0.0000425 based on analyses of Ca and K salts. Ages were calculated using the atmospheric ⁴⁰Ar/³⁶Ar ratio of 298.56 (Lee et al. 2006) and the decay constants of Renne et al. (2010). All data corrections were carried out using an Excel macro and the ages were calculated using Isoplot 4.15 (Ludwig 2012). All ages are reported at the 2$σ$ level and include a 0.5% error on the J-value. Plateau criteria of at least 50% of the ³⁹Ar release in at least three consecutive steps were used. The ⁴⁰Ar/³⁹Ar data are presented in Supplementary Table 1.

For step-heating experiments, data are usually presented as a step-heating release spectrum and a plateau age can be calculated where the plateau criteria are met. The same data can also be plotted on an inverse isochron correlation plot and the intercepts on this plot allow a calculation of the age of the sample (from the ³⁹Ar/⁴⁰Ar intercept) and the ⁴⁰Ar/³⁶Ar ratio of the intercept. The isochron correlation plot allows the ⁴⁰Ar/³⁶Ar ratio for the sample to be calculated rather than assuming that the ratio is atmospheric, as is the case in the plateau age (⁴⁰Ar/³⁶Ar of atmospheric Ar 298.56) (Lee et al. 2006). Ideally, the agreement between both calculated ages and an atmospheric ratio for the intercept would provide the most confidence that the age calculated is reliable.

Three samples from the eastern margin of Bruce Bank and southern Jane Bank were selected for ⁴⁰Ar/³⁹Ar geochronology. Sample DR.225.14 (dolerite dyke, Bruce Bank) gave a plateau age of 52.9 ± 1.5 Ma (2$σ$⁠) (Fig. 3a), containing 100% of the ³⁹Ar. The inverse isochron plot for the same sample (Fig. 3b) yielded an age of 52.4 ± 3.0 Ma (2$σ$⁠) and a ⁴⁰Ar/³⁶Ar intercept of 302 ± 12, which is within error of the atmospheric value (298.56 determined by Lee et al. 2006) and appears to be tightly constrained. Both the plateau and inverse isochron ages overlap at the 2$σ$ level and an ⁴⁰Ar/³⁶Ar intercept of atmospheric composition suggests that this sample has not been affected by excess Ar.

Sample DR.225.27 (basaltic lava, Bruce Bank) produced a more disturbed release spectra compared with sample DR.225.14 and gave a plateau age of 79.2 ± 2.3 Ma (2$σ$⁠; MSWD = 1.07) (Fig. 3c), containing 61.7% of the ³⁹Ar. The inverse isochron plot for sample DR.225.27 (Fig. 3d) yielded an age of 78.2 ± 5.2 Ma (2$σ$⁠) and a ⁴⁰Ar/³⁶Ar intercept of 295 ± 14, which is within error of the atmospheric value (298.56 determined by Lee et al. 2006). Although this seems to be tightly constrained, it is accompanied by a large MSWD (2.1), but an ⁴⁰Ar/³⁶Ar intercept of atmospheric composition suggests that this sample has not been affected by excess Ar.

Sample DR.84.8 (basaltic lava, Jane Bank) produced a coherent plateau despite some structure and yielded an ⁴⁰Ar/³⁹Ar plateau age of 97.2 ± 1.1 Ma (2$σ$⁠; MSWD = 1.03) (Fig. 3e), containing 99.68% of the ³⁹Ar. The inverse isochron plot for sample DR.84.8 (Fig. 3f) yielded an age of 97.4 ± 1.0 Ma (2$σ$⁠; MSWD = 0.93) and a ⁴⁰Ar/³⁶Ar intercept of 280 ± 23, which is within error of the atmospheric value (298.56 determined by Lee et al. 2006), suggesting that this sample has not been affected by excess Ar.

Four samples (DR.225.13, DR.225.14, DR.225.27 and DR.225.30) from site DR.225 on Bruce Bank (Fig. 2b) were examined for their geochemistry (Riley et al. 2021). The porphyritic (plagioclase and hornblende) mafic dykes (DR.225.13 and DR.225.14) are calc-alkaline basaltic andesites with Mg values of c. 56, weakly enriched (La_N/Yb_N = 2–4.5) and have been dated (this study) at 52.9 ± 1.5 Ma (DR.225.14). The fine-grained basaltic lava (DR.225.27) from Bruce Bank (79.2 ± 2.3 Ma) is olivine and pyroxene phyric, is very similar in composition to DR.225.14 and is basaltic andesite (calc-alkaline) in composition. Sample DR.225.30 is a coarse-grained calc-alkaline gabbro dominated by plagioclase and pyroxene phenocrysts. The gabbro is characterized by high MgO (29.3 wt%) and Cr contents >1000 ppm and Ni contents of 330 ppm and may be a cumulate.

Pearce et al. (2014) analysed four samples from Jane Bank, which were initially described by Barker et al. (1984) as intermediate–acidic island arc tholeiitic volcanic rocks. The lavas are plagioclase phyric and plagioclase–clinopyroxene phyric. Sample DR.80.2 is a basaltic andesite and two samples (DR.81.1 and DR.81.2) from site DR.81 (Fig. 2c) are dacite–rhyolite in composition; a basaltic andesite lava from site DR.85 (adjacent to site DR.84) was also analysed (DR.85.5). Based on their relatively depleted geochemistry, Pearce et al. (2014) interpreted the Jane Bank tholeiitic rocks to represent the southern sector of the Oligocene–Miocene ASSA and correlated them with the basaltic lavas of Discovery Bank.

Basaltic rocks were also dredged from the neighbouring South Orkney microcontinent (Fig. 2). Their geochemistry has been described previously by Barber et al. (1991). The rocks were dated (K–Ar) as Late Cretaceous (84.6 ± 2.5 Ma to 68.3 ± 4.1 Ma) and Barber et al. (1991) interpreted them as calc-alkaline, arc-related basaltic andesites based on their depletion in the high field strength elements. They were interpreted to predate the Oligocene separation of the South Orkney microcontinent from the Antarctic Peninsula and, based on their subduction-related geochemical signature and Late Cretaceous age, the dredged rocks were considered to be an easterly extension of the magmatism of the Antarctic Peninsula (King and Barker 1988).

Mafic rocks from the South Scotia Ridge region are plotted in Figure 4 to aid our understanding of their likely tectonic setting using one of the most useful diagrams (Th/Yb v. Nb/Yb) to interpret contributions from the subducting slab and crustal components in an arc setting. The Bruce Bank and Jane Bank basalts are plotted alongside the mid-ocean ridge basalt–ocean island basalt array and the field of the South Sandwich intra-oceanic arc lavas, which are interpreted to have formed from an asthenospheric source because no pre-existing continental lithosphere is known to exist beneath most of the arc, although older crust may underlie its southern part (Leat et al. 2002, 2016). Also plotted are basaltic rocks from several localities across the Scotia Sea region, including basalts from Discovery Bank (Pearce et al. 2014; Riley et al. 2021), the Eötvös Escarpment (Fig. 2; Dalziel et al. 2013a), the North Scotia Ridge (Riley et al. 2019) and the field of continental arc basalts from Livingston Island (South Shetland Islands; Leat and Riley 2021).

The basalts from the Oligocene–Miocene ASSA exposed at Discovery Bank broadly fall within the field of the present-day South Sandwich island arc lavas, which is representative of an intra-oceanic island arc (Fig. 4). Basaltic rocks from the Eötvös Escarpment (Dalziel et al. 2013a), which also forms part of the ASSA, plot at slightly higher Nb/Yb values, but are generally depleted. There is a significant range within the composition of the Discovery Bank basalts, from depleted to more enriched compositions, that overlap with those from the North Scotia Ridge (Riley et al. 2019) and Bruce Bank (Pearce et al. 2014; Riley et al. 2021). Two basaltic/basaltic andesite samples (DR.80.2 and DR.85.5) from Jane Bank (Barker et al. 1984; Pearce et al. 2014) are also plotted and are close in composition to the basalts from the Eötvös Escarpment (Dalziel et al. 2013a) and a subset of the basalts from Discovery Bank (Pearce et al. 2014; Riley et al. 2021), suggesting a depleted source. Two basalts from the eastern flank of Bruce Bank (DR.225.30, DR.225.13), one sample from close to the Eötvös Escarpment (Dalziel et al. 2013a) and two samples from the North Scotia Ridge (Riley et al. 2019) all plot at more enriched compositions, with a more significant contribution from the continental lithosphere. Unfortunately, the dredged basaltic samples from the South Orkney microcontinent are not included in Figure 4 as the data tables were not available in Barber et al. (1991).

The South Scotia Ridge is an array of submerged and subaerial crustal blocks that are interpreted to have rifted from the Antarctic Peninsula and Fuegian Andes during the opening of Drake Passage (e.g. Eagles and Jokat 2014). The current configuration of crustal blocks and basins (Fig. 1) is the result of spreading on the West Scotia Ridge, subduction in the NW Weddell Sea, ridge–trench collisions at the South Orkney microcontinent and Jane Bank (Fig. 5), and extension and oceanic spreading forming basins between the crustal blocks (e.g. Maldonado et al. 2014; Pérez et al. 2019).

Many researchers favour a close affinity between the South Scotia Ridge and the geology of the Fuegian Andes of southern South America (e.g. King and Barker 1988; Eagles and Jokat 2014; Dalziel et al. 2021), whereas Riley et al. (2022) have proposed a closer link to the Antarctic plate and, potentially, the Cordillera Darwin metamorphic complex of southern Patagonia (Fig. 5), which includes a metasedimentary basement analogous to the Permian sedimentary rocks of the South Orkney Islands (Hervé et al. 2010). Detrital zircon provenance analysis also highlights the contrast between the lithologies of the North and South Scotia ridges, indicating separate geological histories (Riley et al. 2019, 2022).

New geochronological data presented here from the basaltic rocks of Bruce Bank and Jane Bank add valuable information to our understanding of the geological and tectonic history of the crustal blocks of the South Scotia Ridge, particularly in the light of recent tectonic models (e.g. Morales-Ocaña et al. 2023; Schellart et al. 2023). Late Cretaceous–Paleogene mafic magmatism has been determined from the eastern Bruce Bank and southeastern Jane Bank, indicating a geological history that is inconsistent with that suggested by Riley et al. (2021) and Pearce et al. (2014), respectively. The geochemistry of the Bruce Bank mafic rocks was described by Riley et al. (2021), who adopted the interpretation of Pérez et al. (2014) that they were related to magmatism associated with the Oligocene opening of the Scan and Dove basins. Magmatism on Jane Bank was interpreted by Pearce et al. (2014) as likely to be Miocene in age, forming part of the ASSA and a continuation of the Miocene magmatism identified from Discovery Bank (Pearce et al. 2014; Riley et al. 2021).

Understanding the timing of initiation of the ASSA is crucial in resolving the tectonic development of the Scotia Sea. Several workers have suggested a far longer geological history for the ASSA, with Barker et al. (1991), Eagles (2010) and van de Lagemaat et al. (2021) proposing a history extending to the Late Cretaceous, whereas Pearce et al. (2014) and Riley et al. (2021) suggest an Oligocene–Miocene history based on the available geochronology. In the plate reconstructions of Schellart et al. (2023), the initiation of Scotia Sea subduction is considered to be independent of any relative plate motion between the Antarctic Peninsula and South America, as proposed by Barker et al. (1991) and van de Lagemaat et al. (2021). Instead, they consider that Scotia Sea subduction is only dependent on the dynamics of subduction along the Pacific margin. This subduction zone needs to be at a mature stage of its evolution, with the subducted slab sinking into the lower mantle to generate a whole-mantle poloidal return flow cell, which can generate significant compressive stresses in the overriding plate (e.g. Schellart 2017). Schellart et al. (2023) suggested that a precise age for subduction initiation could not be determined, with the maximum age of subduction initiation constrained by the timing of Rocas Verdes basin closure (c. 100 Ma). However, the minimum age is constrained by the age of the oldest arc magmatism resulting from the ASSA (c. 31 Ma; Pearce et al. 2014; Riley et al. 2021). A mid- to Late Cretaceous age is also consistent with convergence transitioning to subduction in the northwestern Weddell Sea sector as a consequence of South American–Antarctic plate motion and rotation of the divergence vector (Eagles 2016). This model suggests that early NE-trending transform faults in the NW Weddell Sea would have transitioned to transpression as the divergence vector rotated towards north–south at c. 107 Ma, which would fit well with subduction-related magmatism at Jane Bank by c. 97 Ma. If westwards-directed subduction initiated in the Late Cretaceous, then it may represent a separate phase of subduction-related magmatism to the Oligocene–Miocene ASSA, particularly given the absence of ages from the Eocene.

Late Cretaceous–Paleogene plate reconstructions determined by Riley et al. (2022) from the proto-Scotia Sea region places the crustal blocks of the South Scotia Ridge adjacent to the South Orkney microcontinent and close to the proto-Pacific margin and the South Shetland Islands (Fig. 5). Late Cretaceous–Paleogene magmatism is also recorded from the South Orkney Islands and the South Orkney microcontinent (King and Barker 1988; Barber et al. 1991). Mafic magmatism from this interval is also widespread in the South Shetland Islands, particularly King George, Livingston and Robert islands (Fig. 2a; e.g. Leat and Riley 2021). The dominant magmatic rock type is calc-alkaline basalt/basaltic andesite, akin to the mafic rocks from the eastern Bruce Bank. This phase of Cretaceous–Paleogene magmatism is related to subduction along the proto-Pacific (Phoenix plate) margin (Leat and Riley 2021) and may also correlate with Cretaceous arc magmatism from the North Scotia Ridge (Riley et al. 2019) and South Georgia (Tanner and Rex 1979).

Sandstone recovered from dredge site DR.225 was examined by Riley et al. (2022) for its detrital zircon age profile (Fig. 6). Sample DR.225.28 was interpreted as in situ and is characterized by a very different age profile to the metasedimentary rocks from southern Bruce Bank (DR.223; Fig. 2b). The sandstone has prominent age peaks at c. 128 Ma and c. 185 Ma (Fig. 6) and a maximum likely depositional age of mid-Cretaceous. The detrital zircon age profile of the sandstone is also consistent with metasedimentary successions from the northern and western Antarctic Peninsula (Riley et al. 2023) and Patagonia (e.g. Pankhurst et al. 2000), which have prominent mid-Cretaceous and Early Jurassic age peaks reflecting enhanced volcanism during these intervals (Pankhurst et al. 2000; Riley et al. 2018). This indicates a likely palaeoposition of Bruce Bank proximal to the northern Antarctic Peninsula/southern South America during the mid- to Late Cretaceous.

Based on our new geochronology and recent plate reconstruction models for the Scotia Sea, we propose three possible tectonic scenarios for the development of Cretaceous–Paleogene magmatism of the South Scotia Ridge that incorporate eastwards- (Phoenix plate) and/or westwards- (Scotia Sea) directed subduction.

1. Late Cretaceous–Paleogene magmatism at Bruce Bank and Jane Bank is related to eastwards-directed subduction of the Phoenix plate beneath the Antarctic Peninsula and can be correlated to the convergent margin magmatism widely exposed across the South Shetland Islands, the South Orkney microcontinent, and also the arc magmatism of the North Scotia Ridge and South Georgia. The pre-translation position of the crustal blocks of the South Scotia Ridge are consistent with an Antarctic plate origin and a palaeoposition adjacent to the northern Antarctic Peninsula.

2. The magmatism exposed across Bruce Bank, Jane Bank and the South Orkney microcontinent is related to an early phase of westwards-directed subduction that initiated in the Scotia Sea to the east of the South Orkney microcontinent in the Late Cretaceous. This phase of westwards-directed subduction developed as a result of South America–Antarctic plate rotation, leading to transform fault development in the NW Weddell Sea that transitioned to subduction in the Late Cretaceous. An alternative model is that Jane Bank subduction was triggered by a subduction invasion polarity switch in the Scotia Sea and is likely to be a separate phase of arc magmatism from the Oligocene–Miocene ASSA.

3. Our favoured interpretation is that magmatism from Jane Bank (c. 97 Ma) represents the initiation of westwards-directed subduction in the Late Cretaceous in an intra-oceanic arc setting and shares geochemical characteristics with the Oligocene–Miocene ASSA. However, the magmatism of Bruce Bank and the South Orkney microcontinent (Early Cenozoic) is more closely associated with the Antarctic Peninsula convergent margin and South Shetland Islands magmatism that developed prior to crustal block translation in the Eocene. The detrital zircon properties of adjacent sandstones from eastern Bruce Bank support a pre-translation history closely aligned to the Antarctic Peninsula.

Dredge site DR.225 of the eastern Bruce Bank (Fig. 2) is also characterized by the recovery of abundant in situ samples of gabbro. The exact relationship between the Cretaceous basaltic lavas and dolerite dykes of site DR.225 with the gabbroic rocks is uncertain, but they were recovered from the same dredge site. It is tempting to correlate the gabbroic rocks of eastern Bruce Bank with the extensive Pacific margin anomaly (PMA) of the Antarctic Peninsula (Fig. 5; 100 Ma panel). The PMA is a long-wavelength positive magnetic anomaly that has been modelled as a 20 km thick magnetite-rich composite magmatic arc batholith of probable Cretaceous age (Garrett 1990; Johnson 1999). Garrett (1990) placed the PMA to the west of the Antarctic Peninsula and suggested that it extended into the Scotia Sea region. Martos et al. (2014) examined a regional compilation of magnetic anomalies from the Scotia Sea and traced the PMA into the crustal blocks of the South Scotia Ridge and potentially parts of the North Scotia Ridge. Martos et al. (2014) mapped the extent of the PMA through the South Scotia Ridge from the South Orkney microcontinent, southern Discovery Bank and possibly into Herdman Bank (Fig. 5). This was refined by Morales-Ocaña et al. (2023) to show the detailed extent of the PMA across the southern part of the South Orkney microcontinent. Magnetic anomalies from Bruce Bank match those of the southern South Orkney crustal block and Discovery Bank and may indicate that the PMA also extends into the eastern Bruce Bank. The Cretaceous magmatic ages of the eastern Bruce Bank reported here lend support to the gabbroic rocks forming a distal segment of the Late Cretaceous PMA.

Direct geochronology is absent from almost all the submerged banks of the South Scotia Ridge, which hampers geological and tectonic interpretations of the configuration of the continental fragments prior to the opening of the Scotia Sea and also how they evolved through the Cenozoic and controlled the development of the Antarctic circumpolar current. Our new ⁴⁰Ar/³⁹Ar geochronology results from mafic magmatic rocks of the crustal blocks of the South Scotia Ridge allow us to place firmer constraints on the tectonic history of the Scotia Sea. We have dated samples from the eastern Bruce Bank and southern Jane Bank, which yield Cretaceous–Paleogene ages that are inconsistent with current models. Bruce Bank magmatism was interpreted to be related to the post-translation volcanism associated with the Oligocene opening of Scan Basin, whereas Jane Bank was interpreted as a distal segment of the Miocene ASSA.

Based on our new data, combined with recently updated plate reconstruction models for the Scotia Sea, our preferred interpretation is that magmatism from Jane Bank (c. 97 Ma) represents the initiation of westwards-directed subduction in the Late Cretaceous in an intra-oceanic arc setting and shares geochemical characteristics with the Oligocene–Miocene ASSA. We consider it unlikely that Jane Bank originated as a product of eastward-directed subduction at the Pacific margin. In such a setting, Jane Bank would have been distal from the trench and separated from it by the South Orkney microcontinent and its accretionary complex. Therefore Jane Bank would have been in a rear-arc position relative to an eastwards-directed subduction zone, which is inconsistent with its depleted, intra-oceanic arc chemistry indicating closer proximity to a trench. Jane Bank is separated from the ASSA by the South Scotia Ridge transform fault and has a separate geological history from the Oligocene–Miocene arc.

Jane Basin could represent a back-arc rifted basin associated with mid-Cretaceous subduction. The magmatism of Bruce Bank and the South Orkney microcontinent (Early Cenozoic) is more closely associated with the Antarctic Peninsula–Patagonia convergent margin and the South Shetland Islands magmatism that developed prior to crustal block translation in the Eocene. The detrital zircon properties of adjacent sandstones from eastern Bruce Bank support a pre-translation history closely aligned to the Antarctic Peninsula/Patagonia. This is further supported by the occurrence of gabbroic rocks from Bruce Bank, which are tentatively correlated to the widespread Cretaceous PMA that extends from the western Antarctic Peninsula into the crustal blocks of the South Scotia Ridge.

The officers and crew of the RRS Discovery are thanked for their support. Camilla Wilkinson of The Open University provided the ⁴⁰Ar/³⁹Ar data and Mark Evans (British Antarctic Survey) carried out the sample preparation. This paper has benefited from the thorough reviews of Graeme Eagles and an anonymous referee.

TR: conceptualization (lead), investigation (lead), methodology (lead), project administration (lead), writing – original draft (lead), writing – review and editing (lead); AB: investigation (supporting), writing – original draft (supporting); KAH: investigation (supporting), writing – original draft (supporting); AC: writing – original draft (supporting); PTL: investigation (supporting), writing – original draft (supporting).

This study is part of the British Antarctic Survey Polar Science for Planet Earth programme, funded by the Natural Environmental Research Council (National Capability Funding).

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.

The data that support this research are all available as Supplementary files linked to this article. Full datasets are also hosted at the British Antarctic Survey's Polar Data Centre at https://doi.org/10.5285/417A7F70-4FF2-428E-829D-0423772F66B4