Revisiting the Australian‐Antarctic Ocean‐Continent Transition Zone Using Petrological and Geophysical Characterization of Exhumed Subcontinental Mantle

The final lithospheric breakup of the Australian‐Antarctic rift system remains controversial due to sparse geological constraints on the nature of the basement along the ocean‐continent transition (OCT) zones. We present new interpretations of multichannel seismic reflection transects and new petrological data of dredged mantle rocks along the East Antarctic margin (Seamount B, offshore Terre Adélie). By combining both data sets, we show that a 50–100 km wide domain of cold and fertile subcontinental mantle was exhumed along the magma‐poor Antarctic margin. This study represents only the second locality, along with the Iberia‐Newfoundland margins, where the importance of exhumed mantle domains along OCTs can be clearly identified. The dredged peridotites preserve characteristics similar to mantle xenoliths found in syn‐ to post‐rift volcanism at the eastern end of the Australian margin (Victoria and Tasmania), indicating the exhumation of fertile subcontinental mantle during rifting between Australia and Antarctica. Seamount B represents the initial stages of exhumation of cold subcontinental lithosphere along an OCT during rifting. This thick mantle domain was likely affected by melt impregnation at high pressure (8 kbar), leading to the formation of plagioclase‐pyroxenites. The combination of continental rifted blocks, a wide domain of volcanic‐poor subcontinental mantle and (ultra‐) slow spreading is analogous to OCTs from the Jurassic Western Tethys and Iberia‐Newfoundland rifted margins. Additionally, evidence of melt stagnation at high pressure suggests that magmatism along the Australian‐Antarctic rifted margins was sufficient to form magnetic anomalies that can be used as isochrons despite their formation in lithosphere other than mature, steady‐state ocean crust.


Introduction
The transition from continental rifting to steady-state ocean spreading is generally characterized by two contrasting end-member models of rifting, corresponding to a magma-poor system with exhumation of subcontinental mantle (Iberia-Newfoundland and Jurassic Western Tethys, e.g., Manatschal, 2004;Whitmarsh et al., 2001), a magma-rich model characterized by a rapid transition from rifting to ocean spreading and extensive magmatism (e.g., South Atlantic; Franke, 2013) with an intermediate model characterized by the rapid production of MORB magmatism during crustal extension (e.g., South China Sea; Larsen et al., 2018).
The Australian-Antarctic conjugate margins represent a locality where the transition from rifting to spreading has been characterized as a magma-poor system (e.g., Gillard et al., 2015;Sayers et al., 2001, Figure 1). In contrast with the Iberia-Newfoundland conjugate margins, the Australian-Antarctic margins lack drill sites, with the result that the width and nature of the ocean-continent transition (OCT) zone is poorly characterized (e.g., Beslier et al., 2004;Nicholls et al., 1981;Niida & Yuasa, 1995). Consequently, two end-member hypotheses have proposed that the OCT is either (i) composed of a wide domain of highly stretched continental crust interspersed with magmatic intrusions and only minor exhumed mantle (e.g., Colwell et al., 2006;Sayers et al., 2001) or (ii) a wide domain of exhumed subcontinental mantle followed by non-steady-state proto-oceanic lithosphere (e.g., Beslier et al., 2004;Gillard et al., 2015).
Complicating matters, a symmetrical series of >31,000 km linear magnetic anomalies (C34 to C20) have been interpreted in the OCT zones of the Australian and Antarctic margins (e.g., Tikku & Candes, 1999;Tikku & Cande, 2000;Whittaker et al., 2007;Figure 1). A petrological characterization of the OCT is crucial in order to describe initial magnetic anomalies defining lithospheric breakup and the formation of steady-state oceanic lithosphere (e.g., Matthews & Bath, 1967;Pitman & Heirtzler, 1966;Vine & Matthews, 1963;Williams et al., 2019). The complexity of OCTs implies that defining the first magnetic anomaly related to steady-state oceanic accretion is controversial (e.g., Gillard et al., 2015;Nirrengarten et al., 2018). This includes the magnetic anomalies of the early rift to drift transition along the OCT of the Australian-Antarctic margin (magnetic anomalies C34n and C33n; e.g., Gillard et al., 2015;Tikku & Direen, 2008;Whittaker et al., 2007). In this particular case, the interpretation of these magnetic anomalies as "true" isochrons has been interpreted to imply the onset of steady-state oceanic lithosphere or formed via protracted magmatism and serpentinization during rifting and hyperextension and therefore not "true" isochrons. However, the conflation of linear magnetic anomalies and the production of steady-state oceanic lithosphere are perhaps misleading, given the observation of exhumed lithosphere with linear magnetic anomalies via thin and sparse volcanism at the Southwest Indian Ridge (SWIR; Bronner et al., 2014). Thus, constraining the petrological and structural characteristics of the basement along OCTs is crucial.
We combine multichannel seismic reflection data along Terre Adélie, East Antarctica, in order to define the architecture of the OCT and domains of exhumed mantle with new petrological results from ultramafic rocks dredged from a bathymetric high along the Antarctic margin (Seamount B, Terre Adélie; Niida & Yuasa, 1995;Yuasa et al., 1997). We compare them to mantle xenoliths from syn-to post-rift volcanism in southeast Australia and preserved mantle domains emplaced along OCTs from the Western Tethys ( Figure 1). We show that the Australia-Antarctic OCT is comprosed of a 50-100 km wide zone of cold, fertile subcontinental mantle and records similar rift-related magmatic processes as subcontinental mantle domains exhumed along other OCTs and (ultra-) slow spreading systems. This combined geophysical and Figure 1. (a) Overview of the Australian-Antarctic margins, using a regional 1-min gravity-derived bathymetric map (ETOPO-1; Amante & Eakins, 2009) and interpreted magnetic anomalies (after Seton et al., 2014). Seamount B dredging locations are shown (red diamond) as well as locations for the Tasmanian, Victorian, and Diamantina zone samples (black diamonds) which are included in this study. Locations of seismic reflection profiles used in this study (red lines) and profiles presented in Gillard et al. (2015Gillard et al. ( , 2016 are shown. The thin white line defines the Australian-Antarctic Discordance (Whittaker et al., 2010). (b) A detailed view of Seamount B location (red diamond), seismic profiles with shot points (red lines), and the Adélie Rift Block (white shade) is shown, using ETOPO-1 (top) and the Antarctic magnetic anomaly map (bottom, ADMAP2, Golynsky et al., 2018). The zone of exhumed continental mantle determined with the seismic profiles (see Figure 2) is indicated (green line, with shot point [SP] numbers). Note that in the east of Seamount B, additional magnetic lineations have been identified (red lines).
petrological study highlights the importance of exhumed mantle domains along present-day ultra-slow spreading OCTs, which were previously uncovered only along the Iberia-Newfoundland rifted margins.

The Australian-Antarctic Rift System
Timing and mechanisms for lithospheric thinning and breakup between Southern Australia and conjugate East Antarctica are controversial (c.f. Eagles, 2019;Williams et al., 2019). The timing of transition from continental rifting to ocean spreading has been proposed to occur between the Late Jurassic to Early Cenozoic (e.g., Jacob & Dyment, 2014;Whittaker et al., 2007Whittaker et al., , 2013White et al., 2013;Williams et al., 2011Williams et al., , 2012. Initially, the interpretation of linear magnetic anomalies was used to support the initiation of seafloor spreading in the Eocene (Chron18; Le Pichon & Heirtzler, 1968), an age later revised to Chron22 (ca. 49 Ma), Chron34 (83.5 Ma), or 90-100 Ma (Cande & Mutter, 1982;Veevers, 1986;Weissel & Hayes, 1972), with a NNE-SSW migration of rifting between 160 and 95 Ma (e.g., Powell et al., 1988). Alternatively, Gillard et al. (2015) proposed that large domains of exhumed subcontinental mantle occur along the margins, implying that the first steady-state oceanic crust formed between 53.3 and 43.8 Ma (Chron 24o to 20o). However, the sparse sampled basement material along the Australian-Antarctic margins leaves the nature and width of OCTs ambiguous. Interpretations of geophysical data suggest up to 150 km of discrepancy when defining the initiation of unequivocal oceanic crust (e.g., Close et al., 2009;Colwell et al., 2006;Gillard et al., 2015;Leitchenkov et al., 2007). This leads to conflicting interpretations regarding the nature of magnetic anomalies, whether they can be used as isochrons for plate tectonic reconstructions, and the timing of the rift to drift transition (c.f. Williams et al., 2019; Figure 1).
The complexity of the Australian-Antarctic rift system is further compounded by the Australian-Antarctic Discordance (Figure 1), an unusually deep section of the global mid-ocean ridge system (e.g., Veevers, 1982) coinciding with cooler and/or depleted mantle conditions (Forsyth et al., 1987), chaotic bathymetry, and low magma supply rates (e.g., Christie et al., 1998). This feature has been interpreted either as a zone of downward convecting mantle at the boundary between a Pacific-and Indian-type mantle domain (e.g., Holmes et al., 2010;Veevers, 1982), restricted asthenospheric mantle flow as a result of adjacent thick subcontinental lithosphere and moderate ocean-spreading rates (Buck et al., 2009), or alternatively as resulting from cool depleted mantle domains related to the foundering of the ancient Gondwanaland subducting slab (e.g., Gurnis & Muller, 2003).

Geological Samples and Constraints Along the Australian-Antarctic Margins
Mantle rocks have been sporadically dredged along the Australian (Beslier et al., 2004;Nicholls et al., 1981) and Antarctic (e.g., Niida & Yuasa, 1995) margins. However, the overall abundance and distribution of mantle rocks along the Australian-Antarctic OCT and their geodynamic relevance has been difficult to assess. In part, this is related to the dynamics of rifting which are likely to vary along the Australian-Antarctic margins (e.g., Direen et al., 2013). Moreover, most parts of the Australian-Antarctic basin are either unusually deep (Australia-Antarctic Discordance; Figure 1) or covered by thick sequences of syn-to post-rift sediments, in particular along the Antarctic margin (e.g., Whittaker et al., 2013), leading to challenges in targeting basement lithologies for dredging and International Ocean Discovery Program (IODP) drilling expeditions.

Seamount B, Antarctica
The (ultra-) slow Australian-Antarctic oblique rift system along the Terre Adélie/George V Land and Sorell Basin, west of Tasmania (Figure 1), provides opportunities to constrain the magmatic and tectonic processes occurring during the final stage of continental breakup of Gondwana between ca. 145 and 35 Ma (e.g., Cande & Mutter, 1982;Stagg & Reading, 2007;Totterdell et al., 2000;Whittaker et al., 2007). This eastern most part of the Australia-Antarctic rift system is a magma-poor oblique rift system with breakup proposed to have occurred in the Maastrichtian (e.g., Direen et al., 2011Direen et al., , 2013Lavin, 1997;Stagg & Reading, 2007). Japanese dredging expeditions (TH91, TH95; Tanahashi et al., 1997;Yuasa et al., 1997) targeted the oceanic basement along this oblique rift system about 500 km offshore Terre Adélie and George V Land, East Antarctica. These expeditions dredged several seamounts, including Seamount B, a bathymetric feature elevated about 1,000 m above the surrounding seafloor (Site D1201; Niida & Yuasa, 1995;Yuasa et al., 1997) ( Figure 1). Preliminary results on several Seamount B samples were published in Yuasa et al. (1997) and Niida and Yuasa (1995), indicating that the majority of dredged samples were variably serpentinized ultramafic rocks, dominated by fertile lherzolites and pyroxenites. No volcanic rocks were dredged during this expedition, and dredged metamorphic rocks of unknown origin were interpreted as rafted sea-ice debris . The combination of a free-air gravity anomaly at Seamount B, coupled to the lack of dredged volcanic rocks, suggests that Seamount B represents a bathymetric high dominated by unaltered ultramafic rocks . Dredging of bathymetric highs alongside Seamount B did not recover mantle rocks but fragments of continental crust (granite, gneiss, and diorite) (Tanahashi et al., 1997), implying a complex geological environment with a close spatial associated of bathymetric highs dominated by mantle rocks and continental fragments.

Tasmania and Victoria
Mantle fragments or xenoliths sampled from volcanic eruptions (e.g., Nickel & Green, 1984) are a key tool enabling the study of the subcontinental mantle below the continental crust. The mantle xenoliths along the Australian southern margin in Tasmania and Victoria (Beyer, 2002;Frey & Green, 1974;Nickel & Green, 1984;Norman, 1998) are related to volcanism in the Paleocene-Miocene (Tasmania, Beyer, 2002) to Holocene (Southern Victoria, Norman, 1998; Figure 1). These xenoliths range from fertile, amphibole (±phlogopite)-bearing lherzolites to depleted harzburgites with mineral composition and abundances controlled by partial melting and refertilization events (e.g., Beyer, 2002). Overall, these xenoliths indicate the sampling of compositionally heterogeneous, yet generally fertile, cold (<950°C) subcontinental lithosphere along the Australian margin during syn-to post-rifting volcanism.

Diamantina Zone, Western Australia
In 1977 and 1998, variably serpentinized spinel ± plagioclase-peridotites and pyroxenites were dredged in the Diamantina Zone, along the southwestern Australian margin (Beslier et al., 2004;Nicholls et al., 1981; -survey MD110; Figure 1). Although published data from these rocks are sparse (Chatin et al., 1998;Nicholls et al., 1981), these dredging expeditions indicate that subcontinental mantle rocks are exhumed along the western part of the southern Australian margin. The low equilibrium temperatures of these peridotites (T°of 900°C to 1060°C, Nicholls et al., 1981) and fertile compositions combined with spatially associated alkaline magmatism unlike MORB magmatism formed at an oceanic spreading center led Chatin et al. (1998) and Beslier et al. (2004) to suggest that these mantle rocks were emplaced at the ocean floor following continental rifting.

Magmatic Constraints From Ophiolites and OCTs
Peridotite massifs preserved in orogenic belts (e.g., Bodinier & Godard, 2003;Menzies & Dupuy, 1991) have been used as to constrain magmatic processes occurring in the mantle, including the Lizard (e.g., Green, 1964), Tinaquillo (e.g., Green, 1963), and Ronda (e.g., Masaaki, 1980) massifs, although the origins of these divers peridotite massifs have been debated (c.f. Bodinier & Godard, 2003). In several cases (e.g., Green, 1963Green, , 1964, distinctive peridotites with coarse-grained lherzolite to harzburgite with high-alumina orthopyroxene, clinopyroxene, and spinel are variably overprinted by penetrative foliation and local mylonitization. Remnant pyroxenes porphyroclasts show low-alumina rims matching the compositions of fine-grained, recrystallized matrix mineral and orthopyroxene porphyroclasts show exsolved clinopyroxene lamellae, indicative of decreasing P-T conditions (between 1.5 and 0.5 Gpa). The overprinting of a previous paragenesis can be further recorded by spinel rimmed by plagioclase as well as with olivine, pyroxenes, and Cr-rich spinel and pargasitic amphibole present as a minor or major phase in the foliated matrix, with the appearance of plagioclase implying low pressures.
In terms of magmatism, ophiolites preserved in the Western and Central Alps represent a unique environment allowing mapping of the evolution of petrological characteristics of mantle domains from rifting to ocean spreading (e.g., Picazo et al., 2016). The initial stages of mantle exhumation during rifting along the Western Tethys are accommodated with only sparse to no magmatism. This is likely a consequence of the initial thickness of the cold subcontinental lithospheric mantle leading to the crystallization of melts at higher pressures as well as initial limited mantle melting due to minor lithospheric thinning and asthenospheric upwelling (e.g., Müntener et al., 2010). Thus, an initial "inherited mantle domain" found along hyperextended continental crust is formed of cold (850°C to 950°C), heterogeneous and generally fertile, spinel-bearing subcontinental lithospheric mantle (e.g., Picazo et al., 2016). This inherited mantle domain transitions oceanward into a hybrid "refertilized mantle" domain, dominated by warmer (1000°C to 1100°C) peridotites equilibrated in the plagioclase-stability field and preserving remnant inherited mantle fragments (e.g., Müntener et al., 2010;Picazo et al., 2016;Piccardo et al., 2007). This petrological evolution of mantle rocks along (ultra-) slow OCTs is related to increasing oceanward refertilization of exhumed mantle domains by MORB-type melts during progressive rifting (Picazo et al., 2016). This magmatic evolution can be tracked by the composition of clinopyroxene and spinel as a result of equilibration of MORB melts in the plagioclase-stability field during rifting and transition to (ultra-) slow spreading (Müntener et al., 2010). Thus, mineral compositions in exhumed mantle rocks can be used as fingerprints of magmatic processes occurring prior to and during rifting.

Seismic Reflection Data
Two multichannel seismic reflection profiles crossing the Antarctic continental margin, that pass closely by each side of Seamount B (Figure 1), have been stratigraphically and structurally analyzed and interpreted (Figures 2 and 3; Russian data set RAE53, source: SCAR's Seismic Data Library System, Wardell et al., 2007). The data were collected in 2008 by the Russian Polar Marine Geosurvey Expedition, using a 352-channel streamer and airguns with a volume of 30-47 L (Leitchenkov et al., 2015). Our seismic interpretation follows the classification and criteria after Gillard et al. (2015Gillard et al. ( , 2016.

Seismic Stratigraphy
Our seismo-stratigraphic interpretation indicate the presence of three domains along Seamount B, based on their characteristic structural and stratigraphic patterns: hyperextended continental domain, a zone of exhumed continental mantle (ZECM), and a "proto-oceanic" domain that also likely contains exhumed continental mantle with variable amounts of magmatism.
The landward hyperextended continental domain of the Adélie Rift Block shows characteristic closely spaced fault structures, partly domino faults which extend across the entire pre-rift/early rift sedimentary unit (Figures 2 and 3a). "Sag-type" basins are observed within this part of the sedimentary unit, which is a characteristic stratigraphic feature for extended rift sediments. In both seismic profiles, a continental dipping, high-amplitude reflector is interpreted as the Moho which is detectable around 10 s two-way travel time (TWT) and rises up to 8 s TWT toward the ocean floor where it disappears (Figures 2 and 3a). Such a sharp shallowing of the Moho is characteristic for domains where exhumation of the subcontinental mantle to the seafloor is observed (e.g., Gillard et al., 2016). Syn-rift sediments onlap onto the exhumed mantle basement, which contains characteristic high-angle fault structures ( Figure 3). The imaged basement and overlying sedimentary units along the interpreted zone of exhumed subcontinental mantle are both affected by major deformation structures, which is interpreted as an indicator of exhumed continental mantle, as similar structures are not typically observed in domains of steady-state oceanic crust (e.g., Beslier et al., 1994;Gillard et al., 2015). A prominent debris flow at the oceanward end of the ZECM is detected (Figure 2b), which indicates sediment mass wasting caused by the exhumation of mantle material (after Gillard et al., 2016).
The basement structure changes slightly toward the north; a shallow horizontal high-amplitude reflection pattern is observed which forms a basal boundary layer from which several high-angle faults extend oceanward ( Figure 2b and 3b). Such structures have been described as characteristic features for "proto-oceanic" domains (Gillard et al., 2015).
The seismic profiles indicate that the lateral extent of the interpreted ZECM is about 50-100 km wide ( Figure 2b). They show a lateral consistency on a regional scale, indicating that Seamount B is located about 30 km oceanward from the hyperextended continental crust into the domain of exhumed subcontinental mantle ( Figure 2, Figure 3).
anorthoclase USNM133868 (Na), and microcline (K). Counting times were 10 s peak-counting times for Na, Mn, Ti, K ka, Ca ka, and P ka; 30 s peak-counting times for Si and Ni; and 40 s peak-counting times for Al, Fe, Cr, and Mg. Off-peak-counting times were 10 to 20 s on each side of the peak.
In order to determine the pre-exsolution composition of the pyroxenes, "bulk" pyroxene compositions were also acquired as maps on the Cameca SX100 electron microprobe in the central science laboratory at University of Tasmania. Elements were acquired using Kα lines and analyzing crystals LIF for Mn, Fe, Ni, and Cr and LPET for Ti and Ca. The elements Na, Mg, Si, and Al were analyzed on TAP. The calibration standards were San Carlos Olivine (Mg), Plagioclase Lake County (Al), Anorthoclase Kakanui (Na), apecularite (Fe), rutile (Ti), bustamite (Mn), wollastonite (Ca), clinopyroxene (Si), chromite (Cr), and nickel silicide (Ni). The background was corrected using the mean atomic number method (Donovan & Tingle, 1996). Peak times were 400 ms per pixel with a pixel size of 5 × 5 μm and 32 × 32 pixel per map, resulting in a scanned area of 160 × 160 μm on the cores of the pyroxenes. Oxygen was calculated by cation stoichiometry and included in the matrix correction. The matrix correction algorithm utilized was Armstrong/Love Scott (Armstrong, 1988). An average pyroxene composition was calculated for each map and can be found in Supporting Information S1.
Mineral trace element abundances were determined using an Agilent 7500cs quadrupole inductively coupled plasma mass spectrometer (ICP-MS) coupled to a RESOlution HR laser ablation system equipped with a 193 nm Coherent COMPex Pro ArF Excimer Laser and a S155 ablation cell with constant geometry design at the University of Tasmania. A flow of He carrier gas at a rate of 0.35 L/min carried aerosols ablated by the laser out of the chamber to be mixed with Ar gas and carried to the plasma torch. Quantification was performed using NIST612 as the calibration reference material (Jochum & Stoll, 2008) and Ca as the internal standard element. Both BCR-2g and GSD-1g (Jochum & Stoll, 2008) were used as secondary standards for quality monitoring (Supporting Information S1). Both primary and secondary standards were analyzed at the beginning and end of each batch to correct for instrument drift. Data quantification was performed using an in-house data reduction spreadsheet.

Clinopyroxene
Clinopyroxene in lherzolites is found either as large porphyroclasts or as smaller grains (neoblasts) associated with plagioclase and olivine. Clinopyroxene shows elevated Cr 2 O 3 abundances (0.45 -1.16 wt%) and Mg# Fetot of 90-93.55 ( Figure 5). Al 2 O 3 and Na 2 O abundances decrease with increasing Mg#, from~8 to 5 wt% and 2.3 to 1 wt%, respectively. Clinopyroxene porphyroclasts show a decrease in Al 2 O 3 and Na 2 O and increase in Mg# from the core toward the rims, with smaller clinopyroxene having similar composition as clinopyroxene porphyroclast rims (Figures 5 and 6). Clinopyroxene from residual spinel-lherzolites show slightly lower Na 2 O, TiO 2 , and Al 2 O 3 abundances ( Figure 5).
Clinopyroxene from pyroxenites show more differentiated compositions compared to clinopyroxene from lherzolites, with Mg# Fetot between 85.5 and 87.5 and lower abundances in compatible elements such as

10.1029/2020GC009040
Geochemistry, Geophysics, Geosystems ( Figures 5 and 7), coupled to a slight enrichment in La (Figure 7a). The pyroxenites show higher abundances in REE and depletion in europium implying cocrystallization with plagioclase.

Plagioclase
Fresh plagioclase is found in lherzolites and pyroxenites. Plagioclase has andesine composition in lherzolites (An30-An42) and labradorite compositions in pyroxenites (An45-An60). REE patterns in plagioclase from pyroxenites show similar L-REE enrichment as oceanic gabbros and abyssal pyroxenites (Figure 7d). Plagioclase is an interstitial phase found rimming spinel or associated with clinopyroxene ± olivine in

Amphibole
Interstitial amphibole is found in lherzolites and pyroxenites and classified as a pargasite with some high-TiO 2 grains grading toward kaersutite according to the classification of Leake et al. (1997). These amphiboles have high Mg# Fetot between 86.5 and 90.2 and Cr 2 O 3 contents of 0.97 -1.64 wt%. Na 2 O wt% contents remain near constant (3.12 -3.76 wt%), whereas both TiO 2 and K 2 O abundances vary significantly, between 1.7 -5.5 and 0.02 -0.41 wt%, respectively.

Thermometry and Barometry of Seamount B Mantle Rocks
Core to rim zonation of minerals and textural relationships between minerals indicate that bulk equilibrium of the peridotites was not attained. However, pyroxene cores preserve fertile mantle compositions and have been used to calculate equilibrium temperatures (Figure 9a). Equilibrium temperatures for Seamount B mantle rocks were calculated using orthopyroxene major elements and the thermometer of T°C a-in-opx (Brey & Köhler, 1990). As this thermometer is pressure dependent, a pressure of 15 kbar was used is this samples initially equilibrated within the spinel stability field. Average clinopyroxene and orthopyroxene core compositions were used to estimate temperature using the two-pyroxene thermometer (T BKN ) of Brey and Köhler (1990) as well as the pyroxene REE thermometer (T°R EE ) of Liang et al. (2013). Equilibrium temperatures range between 892°C (±50°C, 1σ) and 815°C to 860°C for major element thermometers, respectively, whereas T°calculated using REE vary between 882°C and 1116°C (Table 1 and Figure 9). Bulk analysis of orthopyroxene porphyroblast cores give us temperatures of 1091 ± 17 and 1266 ± 54°C (1σ, T°C a-in-opx on samples D1302-1 and D1302-6) as a consequence of the inclusion of clinopyroxene exsolution lamellae and indicative of cooling from high-T°conditions.
The composition of plagioclase is correlated with pressure, with anorthite content of plagioclase decreasing with pressure, and a barometer for plagioclase equilibrium in mantle rocks has been experimentally determined (Borghini et al., 2011;Chalot-Prat et al., 2013;Fumagalli et al., 2017). We have combined interstitial plagioclase and interstitial clinopyroxene pairs as well as rims of clinopyroxene and orthopyroxene porphyroclasts. The application of this barometer using the equations of Fumagalli et al. (2017) in combination with  (Brey & Köhler, 1990). Note that the plagioclase-out reaction is a function of bulk composition (e.g., Fumagalli et al., 2017). Error bars for pressure are 1σ.

Geochemistry, Geophysics, Geosystems
T°B KN shows high-pressure equilibrium conditions of 8-9 kbar within the upper limit of plagioclase stability at~860°C to 960°C. These pressure estimates are consistent with the low, and homogenous, anorthite content (An32-An40) of interstitial plagioclase in peridotites ( Figure 9). As no olivine was observed in thin section, a prerequisite for the application of this barometer (Fumagalli et al., 2017), we did not calculate pressures for pyroxenites. However, plagioclase in pyroxenites shows overall similar compositions, suggesting similar P-T conditions as the peridotites.

Fertile Subcontinental Mantle Along the Australian-Antarctic OCT
The composition of clinopyroxene and Cr-spinel are proxies for partial melting and melt-rock reaction processes (e.g., Dick, 1989;Johnson et al., 1990). Abyssal (oceanic) peridotites show lower Na 2 O, and Al 2 O 3 and increasing Cr 2 O 3 in clinopyroxene and increasing Cr# and decreasing Mg# with low TiO 2 in spinel as a result of adiabatic decompression melting at a mid-ocean ridge (Figures 5 and 8; compilation from Warren, 2016). Seamount B peridotites are distinct from the field of abyssal peridotites, with the cores of clinopyroxene showing Na 2 O abundances above 1.5 wt% and Al 2 O 3 above 6 wt% whereas spinel Cr# remains <20 and high Mg# (>66) (Figures 5 and 8), implying fertile mantle compositions. A global compilation of trace element abundances in clinopyroxene from abyssal peridotites (Warren, 2016) shows fractionated REE patterns resulting from variable degrees of partial melting (Figure 7a). The trace element abundances of the majority of Seamount B peridotite clinopyroxene indicate only minor partial melting as illustrated in Figures 7a and 7c with lherzolites showing higher trace element abundances (e.g., Zr) than depleted MORB mantle clinopyroxene (Workman & Hart, 2005). Two spinel-peridotites (D1201-1 and D1201-4) have more depleted clinopyroxene compositions than other peridotites, suggesting higher partial melting (Figures 5  and 6). REE patterns of residual spinel-lherzolite D1201-1 indicate a partial melting trend after~5% melting in the spinel stability field, as illustrated by lower Zr abundances and lower Zr/Hf (Figure 7c) as well as light-REE depleted patterns (Figure 7a).
Calculated T°conditions for Seamount B peridotites from pyroxenes with well-developed exsolved lamellae are below 950°C. Such T°suggest a cold subcontinental mantle origin (Figure 9) whereas the higher temperatures calculated using trace element abundances between orthopyroxene and clinopyroxene (T°R EE , Figure 9) are consistent with slow cooling rates generally characteristic of subcontinental mantle peridotites (e.g., Dygert & Liang, 2015). In addition, Seamount B peridotites share petrological characteristics similar to subcontinental mantle xenoliths sampled along the Australian margin during syn-to post-rift volcanism (Figure 1a; Nickel & Green, 1984, Norman, 1998, Beyer, 2002. These include similar overlapping pressure-temperature conditions unlike those of abyssal peridotites (Figure 9). Clinopyroxene REE patterns from xenoliths reflect heterogeneous subcontinental mantle lithosphere underneath Tasmania which was affected by variable degrees of partial melting as well as local enrichment in light-REE patterns ascribed to fluid/melt-rock interaction (e.g., Beyer, 2002). REE patterns and Zr abundances of Seamount B clinopyroxene show a distinct overlap with fertile Tasmania xenoliths (Figures 7a and 7c). This is also consistent with overlapping elevated Na 2 O and Al 2 O 3 abundances of clinopyroxene and lower Cr# and higher Mg# in spinel between Seamount B and xenoliths from the Australian margin (Figures 5 and 8). In addition, positive La N /Ce N in depleted spinel-peridotites suggests cryptic infiltration of melt, enriching light-REE elements (Figure 7a) following partial melting, which is a common observation in subcontinental lithosphere below Tasmania (e.g., Beyer, 2002) (Figure 7a). Additionally, the presence of pargasites in these peridotites is indicative of local percolation of melt, or alternatively, minor amounts of fluids through these peridotites, and is a ubiquitous phase in subcontinental lithosphere and in plagioclase-peridotites having been affected by melt-rock interaction (e.g., Frey & Green, 1974;Kaczmarek & Müntener, 2008;Müntener et al., 2010;Vannucci et al., 1995) and represents a late phase crystallizing at temperatures below 1050°C (Niida & Green, 1999).
The combination of cold recorded temperatures, variably fertile mantle compositions and lower degrees of partial melting than is typical for abyssal peridotites (Figures 5, 7a, and 9), therefore suggests that Seamount B represents heterogeneous subcontinental lithosphere variably affected by small amounts of partial melting and local overprint by minor amounts of fluids or melts, as originally proposed by Niida and Yuasa (1995).
10.1029/2020GC009040 Figure 10. (a-c) Plate tectonic reconstructions showing rifting phase between 55 and 35 Ma (using the plate model of Whittaker et al., 2013), with reconstructed location of Seamount B (red diamond), seismic reflection profiles used in this study (red lines) and profiles presented in Gillard et al. (2015Gillard et al. ( , 2016. Local magmatism along the Australian-Antarctic margins (blue regions) and Tasmanian mantle xenoliths (Beyer, 2002; red stars) are presented. Interpreted domain of exhumed mantle along the Australian-Antarctic OCT (green regions, for simplicity only drawn in c) are based on data from Seamount B and interpretations from Gillard et al. (2015).

High-Pressure Melt Impregnation
Plate reconstructions of the rift-to drift transition between 55 and 35 Ma along the eastern section of Australian-Antarctic margins indicate an oblique opening of the Australian-Antarctic Basin along Tasmania (Stagg & Reading, 2007; Figure 10). Such a geometry suggests that the exhumation of Seamount B along the Antarctic OCT (Figures 2, 3, and 10) occurred about 600 km northwest of Tasmania along the Victoria margin. Thus, this indicates that two distinct mechanisms, syn-to post-rift volcanism and (ultra-) slow rifting, were tapping a similar domain of fertile subcontinental mantle with overlapping thermal and petrological characteristics beneath a broad area of the Australian-Antarctic rift system. However, although Seamount B reflects the exhumation of inherited subcontinental mantle to the seafloor, these mantle rocks were likely overprinted by melt impregnation and deformation during exhumation along the OCT.
The low temperatures of Seamount B plagioclase-peridotites ( Figure 9) unlike other examples of refertilized plagioclase-peridotites (Müntener et al., 2010) as well as the lack of clinopyroxene ( Figure 4) and spinel ( Figure 8) compositions indicating melt-impregnation events suggest that plagioclase has a metamorphic origin. As such, small core-rim zonation patterns in pyroxene ( Figure 5) might well reflect a failure to reach equilibrium during subsolidus decompression. Alternatively, the lack of textural evidence of plagioclaseolivine symplectites recording a metamorphic breakdown of spinel during decompression as well as the presence of interstitial plagioclase rimming spinel and cocrystallization with clinopyroxene (Figures 4a and 4b) suggests a magmatic origin for plagioclase. In addition, subtle core to rim zoning patterns in clinopyroxene ( Figure 5) and spinel (Figure 4b), as well as a lack of textural equilibrium in peridotites, could also be indicative of a local refertilization event in the plagioclase-stability field. Indeed, clinopyroxene rims and finer grained clinopyroxene associated with plagioclase have lower Al 2 O 3 and Na 2 O abundances than clinopyroxene cores, consistent with equilibration of clinopyroxene with plagioclase by percolating MORB melts (e.g., Müntener et al., 2010).
Seamount B pyroxenites show mylonite textures with rounded to strongly elongated porphyroclasts of orthopyroxene rimmed by neoblasts of clinopyroxene + plagioclase and lack olivine (Figures 4d-4f). These pyroxenites do not show any textural evidence of preexisting garnet, a phase generally present in high-pressure pyroxenites in subcontinental lithospheric mantle (e.g., Downes, 2007). Such textural evidence would include the development of symplectites as a consequence of the destabilization of garnet during decompression (e.g., Montanini et al., 2006;Müntener et al., 2010). In terms of mineral composition, the pyroxenites have aluminum-rich spinel (hercynite) (>60 wt% Al 2 O 3 ) unlike spinel from mantle rocks (e.g., Figure 8) as well as clinopyroxene with lower Mg# Fetot (84-88), Cr 2 O 3 abundances (<0.26 wt%), and higher TiO 2 abundances (0.83-1.37 wt%). These pyroxenites are more differentiated than pyroxenites found in abyssal peridotites and along OCTs (e.g., Basch, Rampone, Crispini, et al., 2019) and more consistent with clinopyroxene from troctolites and gabbros for mid-ocean ridge settings (e.g., Drouin et al., 2009;Sanfilippo et al., 2013;Suhr et al., 2008). Thus, this suggests that these pyroxenites formed from melts that were not buffered by the host peridotite (e.g., Basch, Rampone, Crispini, et al., 2019) and that the evolution of these minerals was in part controlled by crystal fractionation upon cooling in the plagioclase-stability field. Thus, whereas peridotites might record either a metamorphic origin of plagioclase or a diffuse event of melt-percolation buffering the mineral assemblage to mantle values, plagioclase-pyroxenites seem to record an enhanced stage of melt impregnation and differentiation in cold subcontinental lithosphere.
The plagioclase barometer of Fumagalli et al. (2017) indicates that peridotites equilibrated at 8-9 kbar (Figure 9b), with pyroxenites likely indicating similar crystallization pressures. Such pressures are significantly higher than generally recorded during melt percolation in mantle rocks exhumed at the seafloor and within the limits of the upper stability of plagioclase in mantle rocks (e.g., Borghini et al., 2009;Fumagalli et al., 2017). Pressure-dependent solid solutions such as jadeite in clinopyroxene (e.g., Borghini et al., 2009;Villiger et al., 2007) are significantly more elevated than for oceanic gabbros and troctolites 10.1029/2020GC009040 Geochemistry, Geophysics, Geosystems and overlap clinopyroxene from peridotites ( Figure 6). This suggest that crystallization of an assemblage of cpx + opx + pl + hc occurred at such high pressures (Figure 9). The lack of erupted volcanic products at Seamount B as well as the evolved composition of these minerals implies that we are not able to fully constrain the source of the ascending melt. In addition, enrichment of mid-REE to heavy-REE and Zr (Figure 7b) in clinopyroxene in these pyroxenites compared to troctolites, gabbros, and abyssal pyroxenites might reflect additional post-cumulus crystallization of trapped melts leading to a significant increase in mid-REE to heavy-REE (e.g., . Nevertheless, experimental studies on the crystallization of MORB have highlighted how pressure controls the abundance of individual phases and mineral compositions. Whereas low-pressure crystallization will lead to the formation of troctolites (olivine-plagioclase) and gabbros (olivine-plagioclase-clinopyroxene) (e.g., Green & Ringwood, 1967;Grove et al., 1992), high-pressure crystallization of slightly differentiated MORB melts leads to the crystallization of a mineral assemblage of clinopyroxene-orthopyroxene-Al-spinel-plagioclase (at 0.7-1 GPa, Villiger et al., 2007). The low anorthite content of plagioclase (<60An) in these pyroxenites thus reflects both the slightly evolved nature of the melt, the elevated pressure of crystallization, and the lack of H 2 O in the melt (e.g., Feig et al., 2006;Husen et al., 2016;Panjasawatwong et al., 1995;Villiger et al., 2007).
The presence of mylonite textures and melt impregnation at Seamount B overprinting the subcontinental lithosphere is not unlike other mantle domains emplaced along OCTs, which can show a juxtaposition between melt impregnation and deformation during mantle exhumation (e.g., Kaczmarek & Müntener, 2008). Indeed, subcontinental lithosphere sampled as xenoliths in southeast Australia show primarily granoblastic to pophyroclastic textures (e.g., Beyer, 2002;Frey & Green, 1974;Nickel & Green, 1984) and do not show any textural evidence of plagioclase-bearing mylonites. However, to the contrary of most OCTs (e.g., Fumagalli et al., 2017), Seamount B records evidence of significantly higher melt impregnation and melt differentiation at pressures ≥8 kbar. As Seamount B is located <50 km from the continental margin, these high pressures likely result from the elevated thickness of the subcontinental lithosphere during the initial stages of rifting, preventing partial melts originating from the upwelling asthenosphere from ascending to shallow depth. Thus, although syn-rift magmatism is likely recorded at Seamount B, this exhumed lithospheric mantle acts as a cold high-pressure "lithospheric sponge" (Müntener et al., 2010) storing ascending MORB-type melts and thereby explaining both the lack of dredged gabbros and lavas at Seamount B and the lack of magnetic anomaly atop Seamount B (Figure 1) or linear magnetic anomalies in the ZECM.

Cold Exhumation Paths and Similarities With Western Tethys Ophiolites
The characteristics of Seamount B include (i) bathymetric features west of Seamount B of continental origin (e.g., Tanahashi et al., 1997) and seismic transects locating Seamount B 20-50 km distant from hyperextended continental crust ( Figure 2); (ii) no evidence of volcanic products nor shallow intrusives; (iii) low equilibrium temperatures ( Figure 9); (iv) fertile clinopyroxene and spinel compositions implying limited of partial melting (Figures 5-8); (v) low-temperature decompression paths during rifting ( Figure 9); and (vi) high-pressure crystallization of MORB-type melts.
The petrology of Seamount B mantle rocks is consistent with interpretations of seismic lines which indicate that extension during late-stage rifting was accommodated by a broad 50-100 km wide zone of exhumed subcontinental mantle between hyperextended continental crust and a proto-oceanic domain (Figures 2 and 3; e.g., Direen et al., 2011Direen et al., , 2013. Seamount B mantle rocks are therefore similar to inherited mantle domains exhumed along an OCT during the initial stages of mantle exhumation in the Western Tethys (Picazo et al., 2016). In addition, the low equilibrium temperatures of these pyroxenites suggest a protracted history of local melt refertilization at high pressure during the initial stage of rifting followed by conductive cooling, consistent with a prolonged period of rifting and crustal thinning along the Australian-Antarctic margins (e.g., Powell et al., 1988).
The continental margin along Seamount B is itself interpreted as a~200 km wide rifted continental block (Adélie Rift Block) separated from the Antarctic margin by exhumed subcontinental mantle (e.g., Colwell et al., 2006;Direen et al., 2013). Similar characteristics are interpreted along the western Tethys and Iberia margins, where rifted continent blocks are separated by exhumed subcontinental mantle domains (e.g., Manatschal & Müntener, 2009;Manzotti et al., 2014;Whitmarsh et al., 2001). The width of the Western Tethys, where extension was primarily accommodated by crustal thinning and mantle 10.1029/2020GC009040 Geochemistry, Geophysics, Geosystems exhumation is estimated to have reached a maximum width of ≤700 km (e.g., Manatschal & Müntener, 2009;Vissers et al., 2013), with the main basin floored by exhumed mantle (Piemonte-Liguria basin) estimated to have reached a width of 200-300 km (e.g., Manzotti et al., 2014). In addition, Chenin et al. (2017) compiled and assessed the dimensions and architecture of present-day rifted margins and estimated the width of exhumed mantle domains along such OCTs to 20-110 km (median of 70 km). Thus, present-day examples of OCTs, as well as an estimated width for sections of the Western Tethys, suggest that our seismic interpretation of a 50-100 km wide domain of exhumed subcontinental mantle along the Antarctic rifted margin along the Adélie Rift Block (Figures 2 and 3) are well within the range of exhumed mantle domains of OCTs. This estimated width of exhumed subcontinental mantle is also within the range proposed by Gillard et al. (2015) for the central part of the Australian-Antarctic margins ( Figure 10) and estimated by Beslier et al. (2004) along the Diamantina Zone of western Australia.

Magnetic Anomalies Along the Australian-Antarctic OCT
Seamount B is located within a magnetically quiet zone (e.g., Seton et al., 2014;Figure 1c), approximately 50 km inboard of a (proto-) ocean crust based on our interpretation of seismic reflection and magnetic data (Figures 1  and 2). This is consistent with Seamount B being composed of serpentinized mantle rocks and a lack of volcanic products (Tanahashi et al., 1997), as serpentinized peridotites are unlikely to hold a coherent remnant magnetization capable of producing marine magnetic anomalies (Bronner et al., 2014;Maffione et al., 2014). Based on the case study of an OCT in the Western Tethys, Epin et al. (2019) offer further support that OCTs record a transition from no magmatic addition during the initial stages of mantle exhumation to increasing magmatism (shallow gabbroic intrusions and volcanism) oceanward (Müntener et al., 2010). Thus, evidence of melt impregnation at high pressure at Seamount B likely highlights the presence of melt generation during rifting. Melt generation upon rifting is also consistent with compositionally diverse (olivine melilitite to alkali olivine basalts) and discontinuous volcanism along the conjugate margin to Seamount B during rifting (Figure 10), indicating a variably fertile mantle source at depth (Frey et al., 1978). The gradual thinning of the subcontinental mantle and upwelling asthenosphere would then lead to the emplacement of gabbros at shallow depth and local volcanism further oceanward of Seamount B, coinciding with the appearance of a first (C24n.3n) magnetic anomaly ( Figure 1). Additionally, geophysical observations along the Southwestern Indian Ridge (Cannat et al., 2019) indicate that (ultra-) slow spreading systems transition from a nearly amagmatic spreading controlled by low-angle detachment faults and exhumation of oceanic core complexes to more magma-rich segments over less than 30 km. Thus, as a consequence of the thermal state and rheology of the lithosphere (Cannat et al., 2019), melt production might vary significantly along strike, with ascending melts either being erupted along volcanic-rich segments or forming impregnated plagioclase-peridotites along magma-poor segments. In both the Bight Basin and conjugate Wilkes Land margin, linear magnetic anomalies (e.g., Golynsky et al., 2012) extending for >1,000 km are observed across regions which have been interpreted as continental crust, exhumed continental mantle, and "proto-oceanic" crust (e.g., Close et al., 2009;Colwell et al., 2006;Gillard et al., 2015;Leitchenkov et al., 2007). Evidence of magmatism during mantle exhumation along Seamount B thus suggests that many magnetic anomalies, including C34y and C33o, but also extending to C24o and C20y (Tikku & Candes, 1999;Whittaker et al., 2007) might result from sparse volcanic and shallow magmatic additions to the exhumed (subcontinental or oceanic) mantle domain (e.g., Bronner et al., 2014) as a consequence of alternating magmatic and nearly amagmatic segments (e.g., Cannat et al., 2019).

Elucidating the Nature of Mantle Domains Along the Australia-Antarctic Margins: The Case for Further Shipboard Expeditions
Bathymetric features similar to Seamount B are ubiquitous throughout the Australian-Antarctic margins, and variously interpreted as volcanic edifices, peridotite ridges, or crustal blocks, depending on recovered dredged samples, geophysical properties, and distance to the continental shelf (e.g., Gillard et al., 2015Gillard et al., , 2016. The presence of bathymetric features with continental rocks in close proximity to Seamount B (Tanahashi et al., 1997) as well as the segmented architecture of (ultra-) slow spreading lithosphere (Cannat et al., 2019) underscores the structural complexity of (ultra-) slow spreading systems. Therefore, the Australian Southern Ocean represents a unique location where targeting bathymetric features along the Australian and Antarctic OCTs will help resolve magmatic and tectonic controversies and test mechanisms of rifting and spreading along (ultra-) slow spreading systems initially developed along Iberia-Newfoundland OCTs and the Western Tethys. Here we highlight some key petrological and tectonic questions that could be resolved by future IODP drilling expeditions.

Geochemistry, Geophysics, Geosystems
Tectonic subsidence curves modeled from industry well site data in sedimentary basins along the Australian OCT record an abnormal lack of subsidence during the early post-rift phase (~85 to~50 Ma; Brown et al., 2003;Totterdell et al., 2000). Although some minor subsidence during this time period can be observed at the drill sites in the more oceanward sedimentary basins, their rates are much lower than the expected McKenzie (1978) thermal subsidence of extended rift basins. Subsidence of passive margins is expected to occur during the rift to drift transition due to crustal thinning, cooling of the lithosphere, and sedimentary loading (e.g., Bott, 1992). However, the exhumation of cold subcontinental lithosphere to the sea floor coupled with melt impregnation upon exhumation might lead to a retardation of basin subsidence. In the Australian-Antarctic context, this implies mantle exhumation until approximately 50 Ma, consistent with the interpretation of Gillard et al. (2015) for the transition to steady-state oceanic spreading. Western Tethys OCTs indicate that extensive percolation of MORB melt forming plagioclase-peridotites can lead to a storage of 12% of ascending melt (Müntener et al., 2010). Extensive heating of cold subcontinental lithosphere will lead to the formation of more buoyant lithosphere (e.g., Chenin et al., 2017). For example, the western Antarctic rift system is characterized by anomalous high heat flow interpreted as resulting from syn-rift melt percolation leading to the formation of refertilized, plagioclase-peridotites incorporating up to 8% MORB-type melt (e.g., Martin et al., 2014). In addition, numerical modeling simulating hyperextension of subcontinental lithosphere indicates that the transition from spinel-bearing peridotites to plagioclase-bearing peridotites leads to a significant decrease in mantle density, thereby limiting subsidence (Kaus et al., 2005;Minakov & Podladchikov, 2012;Simon & Podladchikov, 2008). Seamount B represents the initial stages of exhumation of inherited, cold subcontinental mantle which might record early stages of high-pressure melt impregnation. Thus, extensive melt percolation is likely to occur oceanward. The Australian-Antarctic OCT therefore represents a unique environment to test whether retardation of basin subsidence might be controlled by the mechanism of extraction (volcanism) or storage ("lithospheric sponge") of MORB melt along an (ultra-) slow spreading system (e.g., Müntener et al., 2010).
Moreover, the low equilibrium T°of these mantle rocks, the fertile nature of Seamount B samples, evidence of melt impregnation, and the lack of dredged MORB volcanism along the Antarctic margin  is consistent with a lack of partial melting of Seamount B mantle rocks upon exhumation. This implies that the partial melting event recorded in the depleted spinel-lherzolites likely reveal "ancient" magmatic processes unrelated to rifting and mantle exhumation (e.g., McCarthy & Müntener, 2015;Rampone et al., 1998). Depleted mantle domains along OCTs and (ultra-) slow spreading localities are interpreted as preserving "ancient" melting events, such as (i) Proterozoic melting events (Macquarie Island, Southern Ocean, Dijkstra et al., 2010), (ii) Permian mantle-derived magmatism (Western Tethys, e.g., McCarthy & Müntener, 2015, 2019, (iii) subarc mantle during Pan-African Orogeny found in the Zabargad ophiolite (Red Sea, Brooker et al., 2004), and (v) subarc mantle along Newfoundland margin (ODP Site 1277; Müntener & Manatschal, 2006) and western Iberia/Morocco margin (volcanic xenoliths; Merle et al., 2012). The presence of a relic subduction zone along eastern Gondwanaland is suggested to be responsible of the formation of the Australian-Antarctic Discordance (Gurnis & Muller, 2003). Seamount B mantle rocks and Tasmanian and Victorian xenoliths indicate the presence of a wide domain of fertile subcontinental mantle along southeast Australia to the east of the Australian-Antarctic Discordance (Figures 5-9). Thus, in accordance with Gurnis and Muller (2003), exhumed mantle domains closer to the Australia-Antarctic Discordance should reflect characteristics of subduction zone processes, such as the more depleted nature of the mantle and fluid/melt-rock reactions as a consequence of (ancient) slab dehydration and/or melting (e.g., Bryant et al., 2007). The Australian-Antarctic spreading system therefore represents a key locality to study the effect of magmatic inheritance and the origin(s) of mantle heterogeneity in (ultra-) slow spreading systems (e.g., McCarthy & Müntener, 2015;Rampone & Hofmann, 2012;Sanfilippo et al., 2019;Warren et al., 2009) by targeting (i) the heterogeneity of abyssal peridotites along the active Southeast Indian Ridge, (ii) the evolution of mantle rocks along the Australia-Antarctic discordance, and (iii) the oceanward compositional evolution of mantle domains at OCTs.

Conclusions
We combine multichannel seismic reflection data and petrological measurements of dredged mantle rocks along Terre Adélie (East Antarctica) and show the presence of a 50-100 km wide domain of exhumed mantle which formed along the Terre Adélie/George V Land and Sorell Basin along the Australian-Antarctic rift 10.1029/2020GC009040 Geochemistry, Geophysics, Geosystems system. These mantle rocks represent fertile, cold (900°C) subcontinental mantle exhumed at the ocean floor during rifting. Seamount B mantle rocks show similar petrological characteristics as mantle xenoliths from the Australian margin, indicating that both (ultra-) slow spreading and volcanism sampled similar underlying subcontinental lithosphere. Syn-rift magmatism and melt differentiation are recorded by MORB melt crystallization at ≥8 kbar, leading to formation of plagioclase-pyroxenites. The initial stages of mantle exhumation during rifting likely inhibited volcanism and shallow melt intrusions due to the thickness of the cold subcontinental lithosphere. Thus, evidence of melt percolation at Seamount B suggests that syn-rift magmatism was prevalent, which suggests that magnetic anomalies result from thin, sparse volcanic/magmatic additions to the exhumed mantle domain during lithospheric extension and the transition toward steady-state oceanic lithosphere. Finally, we propose that the magmatic processes recorded during (ultra-) slow spreading along the Australian-Antarctic OCT might represent a recent analogue to Jurassic Western Tethys OCTs and represent a key target for future drilling and dredging expeditions.

Data Availability Statement
The seismic reflection data are freely available through the Seismic Data Library System of SCAR (https:// www.scar.org/sdls/), and geochemical analyses of minerals are freely available on the Institute of Marine and Antarctic Sciences (IMAS, University of Tasmania) Metadata Catalogue website (doi: 10.25959/ 5e6810710ac85). We thank S. Gilbert, J. Thompson, and S. Feig for their help with obtaining geochemical analyses. We are grateful for discussions with O. Müntener and reviews by R. Buck and an anonymous reviewer as well as editorial handling by Whitney Behr which helped clarify the manuscript.