Abstract
The development of the Transantarctic Mountains was initiated with the rifting of Rodinia and the formation of a late Neoproterozoic passive continental margin. In Cambrian time this rift setting evolved into an active margin with batholith emplacement into deformed and lightly metamorphosed upper Neoproterozoic–Cambrian strata, creating the Ross Orogen. Denudation and erosion of the Ross Orogen led to the formation of the pre-Devonian Kukri Erosion Surface on which Devonian quartzose sandstones accumulated in a continental setting. Palaeozoic magmatic arcs were intermittently active along the distal Panthalassan margin. Intra-cratonic basins developed in Permian time, one of which evolved into a foreland basin clearly related to a Permo-Triassic magmatic arc. The Palaeozoic–early Mesozoic arcs can be traced into both Australasia and South America. In Early Jurassic time the margin migrated outboard simultaneously with the advent of proximal silicic volcanism, emplacement of the Ferrar Large Igneous Province and Gondwana break-up. These events marked the onset of plate margin reorganization, and with it the early uplift of the Transantarctic Mountains. During Cretaceous and later time episodic uplift of the Transantarctic Mountains was accompanied by formation of a major crustal and lithospheric boundary marking the edge of the East Antarctic craton and the regions of crustal attenuation in the Ross (West Antarctic Rift System) and Weddell embayments.
The frontal scarp of the Transantarctic Mountains forms a geographic divide across the Antarctic interior, separating East and West Antarctica (Fig. 1). The geological record of the mountains themselves spans the Archean to Late Cenozoic, although in outcrop it is limited prior to the late Neoproterozoic and after Early Jurassic time. In the broadest outline, the Transantarctic Mountains comprise a pre-Devonian basement overlain by a Devonian–Jurassic sedimentary sequence, which in Early Jurassic time was overlain by extrusive basaltic rocks and intruded by sills. Younger rocks consist of Upper Cenozoic alkaline basaltic rocks and glacigenic strata.
Transantarctic Mountains region of Antarctica, showing the distribution of basement rocks (Proterozoic–early Palaeozoic) and regional cover beds (Devonian–Jurassic). Inset shows map area.
This review, from the perspective of the central Transantarctic Mountains and south Victoria Land, will address primarily the mid-Palaeozoic and younger tectonic events, concentrating on the Ross Sea sector and on tectonic relations with the broader Gondwana margin. Through Cenozoic times the evolution of the Transantarctic Mountains has been intimately tied to development of a continental-scale rift system in West Antarctica, today the site of Earth’s only marine ice sheet. The history of the mountain range is also significant for understanding the history of Antarctic glaciation and climate.
Archean to Neoproterozoic time: windows into the cratonic margin
The oldest rocks, restricted to the region of the upper Nimrod Glacier (Fig. 1), are Archean gneisses and schists intruded by 2.5 Ga granites (Goodge & Fanning 1999). These rocks were reworked by Palaeoproterozoic metamorphism and syn-tectonic magmatism at 1.7 Ga, and again during the Cambrian–earliest Ordovician Ross Orogeny (Goodge et al. 2001; Goodge & Fanning 2002). Glacial erratics in the Miller Range region indicate the presence also of Grenville-age (c. 1 Ga) crust in the hinterland (Goodge et al. 2010). Palaeoproterozoic to lower Cambrian igneous, metamorphic and sedimentary rocks crop out in the Shackleton Range (Fig. 1), forming three east–west structural terranes bounded by thrust fault systems (Buggisch & Kleinschmidt 2007), which detailed geochemical and isotopic data (Will et al. 2009, 2010) demonstrate have markedly different histories. Fossiliferous siliciclastic Ordovician strata, the Blaiklock Glacier Group, unconformably overlie older rocks in the Shackleton Range (Clarkson et al. 1995). The Miller and Shackleton ranges form part of the ring of Precambrian outcrops around East Antarctica that include pan-African age belts, Grenville age provinces, Proterozoic domains and Archean cratons (Fitzsimons 2003; Will et al. 2010; Boger 2011).
Late Neoproterozoic to early Palaeozoic (Cambrian) time: passive margin to subduction
Excluding the Miller and Shackleton ranges, basement rocks in the Transantarctic Mountains (Fig. 2) consist of Neoproterozoic strata and lower Cambrian to Ordovician(?) metasedimentary and metavolcanic rocks, all affected by deformation and granitoid emplacement associated with the Cambrian–earliest Ordovician Ross Orogeny (Rowell et al. 2001; Goodge et al. 2004a, b). Neoproterozoic siliciclastic strata crop out in the Pensacola Mountains (Hannah Ridge Formation) and the central Transantarctic Mountains (Beardmore Group). These rocks, which have yielded detrital zircon ages of latest Neoproterozoic (c. 556 Ma), form passive margin sequences deposited subsequent to the break-up of Rodinia. More widespread between the Pensacola Mountains and the Byrd Glacier are lower Cambrian to Ordovician siliciclastic and carbonate strata (Patuxent Formation in the Pensacola Mountains region and Byrd Group in the central Transantarctic Mountains). Cambrian silicic volcanic rocks are common in the Queen Maud and Thiel mountains (Pankhurst et al. 1988; Wareham et al. 2001). Deformation of these rocks was accompanied by intrusion of both pre- and syn-tectonic granitoids, the age of which ranges from about 560 Ma to, in isolated cases, 480 Ma (Borg et al. 1990; van Schmus et al. 1997; Rowell et al. 2001; Paulsen et al. 2007). There is evidence for early Cambrian deformation of the Hannah Ridge Formation, and a well-documented second episode of folding and low-grade metamorphism that is no younger than about 505 Ma. Deformation was followed by intrusion of post-tectonic granitoids. A basement discontinuity in the Byrd Glacier region (Stump et al. 2004; Carosi et al. 2007) marks a significant change in the metamorphic grade of basement rocks and a possible terrane boundary.
Distribution of Neoproterozoic and Ross Orogen rocks along the Transantarctic Mountains.
To the north, in south Victoria Land, the basement is dominated by medium- to high-grade metamorphic rocks of the Horney Formation (Borg et al. 1989) and the Koettlitz and Skelton groups (Goodge et al. 2004a; Cooper et al. 2011), and is intruded by a Neoproterozoic to Cambrian alkaline suite (Read et al. 2002) that pre-dates the more typical calc-alkaline granitoids of the Ross Orogen (Allibone & Wysoczanski 2002). The alkaline magma suite was emplaced between about 550 and 530 Ma, whereas the Ross granitoids are mainly younger than about 530 Ma, range to about 480 Ma, and again show both syn- and post-tectonic phases. The Skelton Group has been correlated with Neoproterozoic strata in the central Transantarctic Mountains (Goodge et al. 2004a; Cooper et al. 2011).
North Victoria Land, with a more complex Neoproterozoic to Ordovician history, comprises three fault-bounded lithotectonic terranes (Laird 1991). The Wilson terrane metamorphic rocks are intruded by Cambrian–lowest Ordovician granitoids with ages ranging from about 545 to 480 Ma (Bomparola et al. 2007; Giacomini et al. 2007; Federico et al. 2009) and form the continuation of the Ross Orogen in north Victoria Land. The narrow Cambrian-age allochthonous Bowers terrane, bounded by the Lanterman and Leap Year faults, lies outboard of the Wilson terrane (Bradshaw et al. 2009). It consists of a thick sequence (−10 km) of fossiliferous marine siliciclastic and volcaniclastic strata together with mafic volcanic rocks, overlain by conglomeratic strata, and comprises the Bowers Supergroup. The Robertson Bay Group, outboard of the Bowers terrane, is a Cambrian(?) to Ordovician turbidite sequence, which, similar to the Bowers, is strongly folded but only slightly metamorphosed, as indicated by illite crystallinity (Buggisch & Kleinschmidt 1991). A siliciclastic sequence of similar age and composition, the Swanson Formation (Bradshaw et al. 1983), crops out in Marie Byrd Land and is probably a correlative deep-sea fan sequence.
The nature of the north Victoria Land fault boundaries and the tectonic history remain controversial. Adjacent to the Lanterman Fault and forming the contact rocks of the Wilson terrane there are bodies of high pressure and ultra-high pressure rocks (Palmeri et al. 2003). The Leap Year Fault includes a zone of schists incorporating slices of both Bowers and Robertson Bay terrane rocks, and thus may represent a zone of imbricate thrusting. The contrasts in stratigraphic successions and metamorphic grade between the terranes have led to alternative tectonic hypotheses. In general, one hypothesis postulates major along-strike translation of the two outboard terranes and the other accretion and shortening of the terranes by major east-directed overthrusts, or some combination of both (Tessensohn & Henjes-Kunst 2005 and references therein), and with all events completed in early Palaeozoic time (Early Ordovician). More recent hypotheses are laid out by Bracciali et al. (2009) and Rocchi et al. (2011), who propose a more complex scenario of subduction, island arc formation in an extensional setting, and subsequent accretion and closure.
A Ross age syn-orogenic clastic sequence, the Douglas Conglomerate of the Byrd Group (Laird 1991) in the central Transantarctic Mountains, has been described, but no post-orogenic strata are known except for the Robertson Bay Group in north Victoria Land. Erosion of the Ross Orogen commenced in late Cambrian time and denudation continued through to the Late Silurian or Early Devonian, creating the widespread and distinctive Kukri Erosion Surface (Isbell 1999).
The Ross Orogen, including the Wilson terrane, reflects subduction of Panthalassan oceanic crust beneath the edge of the East Antarctic basement (Borg & DePaolo 1991), and clearly is correlative with the Delamerian Orogen of Australia (Finn et al. 1999; Glen 2005; Foden et al. 2006; Bradshaw 2007). The Bowers terrane has similarities to Cambrian basaltic volcanic rocks and siliciclastic sequences of the Delamerian Orogen, and also has been correlated with the Takaka terrane in New Zealand (Bradshaw et al. 2009). A single Ross-age granitoid is known from New Zealand (Allibone et al. 2009). Further afield, the Ross Orogen can be correlated with the Saldanian Orogen in South Africa (Scheppers & Armstrong 2002).
The Cambrian Surgeon Island granite (c. 512 Ma), which lies within the Robertson Bay terrane, but is not seen in outcrop to intrude those rocks (Fioretti et al. 2005), is possibly part of a terrane accreted in late Ross–Delamerian time (Cayley 2011). A rhyolite dredged from the Iselin Bank and having a late Proterozoic/early Cambrian age (Mortimer et al. 2011), and Cambrian orthogneiss and paragneisses in Marie Byrd Land (Pankhurst et al. 1998; Mukasa & Dalziel 2000), indicate that rocks with the same age as the Ross Orogen are present in West Antarctica. The Robertson Bay terrane, these isolated Ross-age rocks and the complexities of correlations between north Victoria Land, New Zealand and south-eastern Australia are discussed further in the next section
Ordovician to Late Triassic time: sedimentation and subduction
Devonian strata
The Kukri Erosion Surface is the defining contact in the Transantarctic Mountains, separating Ross-age granitoids and deformed rocks from essentially undeformed Devonian and younger rocks (Fig. 3). Although initially described as a peneplain (Gunn & Warren 1962), it is not a planar surface, but has topography as shown by local Devonian strata thickness variations of up to 700 m (Isbell 1999; Isbell et al. 2008). In south Victoria Land (Fig. 4), the surface is overlain by the 1500 m-thick quartzose strata of the Taylor Group (Bradshaw, this volume, in press). Sedimentary structures and trace fossils have been used to propose both marine and non-marine depositional environments, with only the Aztec Siltstone being clearly terrestrial. The age, based on palynomorphs, ranges from Early to Late Devonian (Kyle 1977). However, fish faunas from the Aztec Siltstone suggest an age no younger than late Middle–early Late Devonian (Young & Long 2005). Palaeocurrent directions are quite variable (Barrett & Kohn 1975), but the quartz-rich composition implies derivation from granitic rocks. The Devonian rocks assigned to the Taylor Group south of Byrd Glacier (Alexandra Formation and, in the Starshot Glacier region, the underlying Castle Crags Formation) have significantly lesser thicknesses and lack finer-grained beds except for the Castle Crags Formation. The age of the Castle Crags and Alexandra formations is unconstrained palaeontologically, and thus a Carboniferous age for the upper part of the Alexandra cannot be excluded. Beyond the most southerly outcrop of the Alexandra Formation at the Ramsey Glacier, Devonian rocks are present in the Ohio Range (Bradshaw 2013), where they form a 56 m-thick, fossiliferous, dominantly marine, siliciclastic sequence (Horlick Formation) with a palaeontological age of Emsian (Early Devonian; Boucot et al. 1963; Bradshaw & McCartan 1991). Siliciclastic strata with similar depositional environments and an inferred Devonian age are present in the Pensacola Mountains (Dover Sandstone; Cathcart & Schmidt 1977). The Crashsite Quartzite in the Ellsworth Mountains is Devonian in its upper part (Sporli 1992). However, at that time the mountains were located in Gondwana reconstructions between Africa and Antarctica (Watts & Bramall 1981; Grunow et al. 1987), and the geology is much more comparable with that of South Africa (Curtis & Storey 1996). Nevertheless, the Ohio Range and Ellsworth Mountains Devonian sequences carry the Malvinokafric fauna and hence can be linked to South Africa, the Falkland Islands and South America, each representing part of a widespread shallow marine environment. Further, there are links with Devonian marine faunas in New Zealand (Bradshaw & McCartan 1991).
Simplified stratigraphic columns for Devonian to Jurassic rocks overlying the Kukri Erosion Surface in the Transantarctic Mountains. Sources: Taylor and Victoria Groups in the central Transantarctic Mountains and south Victoria Land, Barrett (1991); Falla Formation, Elliot (1996); Hanson Formation, Elliot et al. (2007); Prebble Formation, Hanson & Elliot (1996); unnamed strata, south Victoria Land, Elliot & Grimes (2011); Mawson Formation, Ross et al. (2008); north Victoria Land glacial beds and Takrouna Formation, Collinson et al. (1986); Section Peak and ‘Shafer Peak’ formations, Schöner et al. (2007, 2011 respectively); Exposure Hill Formation, Viereck-Goette et al. (2007); Kirkpatrick Basalt, Elliot & Fleming (2008).
Present-day distribution of Devonian sedimentary rocks and Silurian–Devonian–Carboniferous granitoids and gneisses in Marie Byrd Land and the Antarctic Peninsula. Note that during Palaeozoic time, (a) the Ellsworth–Whitmore Mountains block and Haag Nunataks (heavy dash outlines) were located between Africa and Antarctica in the proto-Weddell Sea region; (b) the Filchner-Ronne Ice Shelf region is highly extended; (c) Thurston Island was displaced relative to the Antarctic Peninsula and the Ellsworth–Whitmore Mountains block; (d) Marie Byrd Land, prior to crustal stretching in the Cretaceous and Cenozoic, was located closer to the Transantarctic Mountains; and (e) the base of the Antarctic Peninsula would also have been closer to the Transantarctic Mountains.
The quartz-rich Devonian succession in the Transantarctic Mountains (Fig. 4), although principally derived from granitic sources located mainly in the East Antarctic, did include West Antarctic sources (Barrett & Kohn 1975; McCartan & Bradshaw 1987), and was deposited in a passively subsiding basin floored by Ross Orogen rocks. This setting is in marked contrast to the mid-Palaeozoic record of Marie Byrd Land, which is characterized by Devonian and Carboniferous plutons and metamorphic rocks (see the next section).
Early Palaeozoic (Ordovician to Early Carboniferous) plate margin
The Devonian Taylor Group and the late Devonian–Carboniferous magmatic belt of north Victoria Land and Marie Byrd Land need to be placed in the broader temporal and spatial context of the Pacific margin of Gondwana. This margin is better documented in the Lachlan Orogen of southeastern Australasia, which was widely developed outboard of the Delamerian Orogen that previously formed the Panthalassan margin (Gray & Foster 2004; Glen 2005), and in the Patagonian region of South America (Vaughan & Pankhurst 2008). Along the length of the margin from Australia/New Zealand to South America, there is a record of pre-Late Devonian events.
The Robertson Bay Group (north Victoria Land), the Swanson Group of Marie Byrd Land and the Greenland Group in New Zealand (Wandres & Bradshaw 2005) all have similarities to the Ordovician quartzose turbidite sequences in the Lachlan Orogen, and all represent erosion of continental sources in the Ross–Delamerian Orogen. All were deformed in an early phase of the Lachlan Orogen (Gray & Foster 2004; Glen 2005) and its inferred extension into West Antarctica; it is also recorded in lower Palaeozoic schists on the Campbell Plateau (Adams 2008). There is a single (poorly constrained) Ordovician age (Rb–Sr) from orthogneisses at Clark Island, eastern Marie Byrd Land (Fig. 4; Pankhurst et al. 1998). Granite clasts in conglomerates from the Antarctic Peninsula, one from the Marguerite Bay area (Loske et al. 1997) and several from View Point (Pankhurst et al. 2003; Bradshaw et al. 2012) have yielded Late Ordovician zircon ages. Ordovician ages are also reported for the Cordillera Darwin Metamorphic Complex and from a granite clast in overlying Jurassic strata (Hervé et al. 2010b). In the Deseado Massif region of Patagonia, granitic cobbles in a Permian sequence also suggest Ordovician magmatism (Pankhurst et al. 2003), which is better expressed in the Famatinian belt farther north (Pankhurst et al. 2001).
Silurian magmatism is documented in the Lachlan Orogen (Gray & Foster 2004; Glen 2005), but again evidence of the orogen in Antarctica is fragmentary. A calc-silicate gneiss at Deep Sea Drilling Project Site 270 in the Ross Sea yielded a titanite metamorphic age of 437 Ma (Mortimer et al. 2011), and a Silurian gneiss crops out in Marie Byrd Land (Pankhurst et al. 1988). In the Antarctic Peninsula an orthogneiss in northern Palmer Land has a protolith Silurian age (Millar et al. 2002) and granite cobbles from Marguerite Bay give Silurian ages (Tangeman et al. 1996; Loske et al. 1997). Late Silurian magmatism is also recorded in the Deseado Massif, Patagonia (Pankhurst et al. 2003).
The Ordovician through Middle Devonian record in West Antarctica and the Antarctic Peninsula is meagre at best, but it does link Australasia and South America, and suggests they are all part of one, episodically active, Panthalassan margin. The lack of granitoids and rarity of gneisses older than Late Devonian in north Victoria Land and Marie Byrd Land may reflect ice cover and tectonic history.
The Late Devonian–Carboniferous record of Marie Byrd Land (Fig. 5) is characterized by I-type Devonian plutons with ages of 375±5 Ma, and Carboniferous plutonic (with both I- and S-type characteristics) and metamorphic rocks dated at about 330–340 Ma (Weaver et al. 1991; Pankhurst et al. 1998; Mukasa & Dalziel 2000). These suggest that a subduction-related magmatic arc bordered the earlier depositional system of the Taylor Group and its correlatives. This arc is expressed in north Victoria Land by the Devonian Admiralty Intrusives and the Carboniferous Gallipoli Porphyries and Salamander Granite complex (Henjes-Kuntz & Kreuzer 2003 and references therein), which were emplaced across all three north Victoria Land terranes and thus constrain the age of any movement on the Leap Year and Lanterman faults to be pre-Late Devonian. However, the granitoids have S- and I-type characteristics, and their spatial relations have been interpreted to suggest NE-directed subduction (Borg & DePaolo 1991), which would imply that the Robertson Bay, Bowers and the eastern part of the Wilson terranes are allochthonous and were amalgamated with the western Wilson terrane in post-Devonian time. This is incompatible with strong evidence for early Palaeozoic amalgamation of the north Victoria Land terranes.
Gondwana plate margin in mid-Palaeozoic time, showing the distribution of Devonian strata in the Transantarctic Mountains, and granitoids and gneisses in the Devonian to Early Carboniferous orogen in West Antarctica, the Antarctic Peninsula and adjacent regions. The arrangement of the cratonic regions and the size and shape of many of the continental blocks are taken from a reconstruction provided by the PLATES Project at the Institute of Geophysics at the University of Texas. Palaeomagnetic studies (Grunow et al. 1991; DiVenere et al. 1996; see also Dalziel & Lawver 2001) provide constraints on the disposition of some of the Antarctic continental blocks along the Gondwana plate margin. Palaeomagnetic data for New Zealand are limited to two Permian poles (Haston et al. 1989) and one Late Triassic–Early Jurassic pole (Grindley et al. 1980), both from southern South Island and east of the Alpine Fault. The possibility that the Haag Nunataks may form a separate entity distinct from the Ellsworth–Whitmore Mountains block is not considered here. The Agulhas Plateau and Maurice Ewing Bank, both of African geological affinities, are also not considered. Continuity of geological provinces, as discussed by Vaughan & Storey (2000) and others, is assumed. The possibility of strike-parallel movements within the plate margin belt (e.g. Adams et al. 1998) makes precise locations for the plate margin terranes tenuous. Likewise, the possibility of pre-Jurassic accretion of continental blocks, continental ribbons and island arc terranes (e.g. Haston et al. 1989; Mortimer 2004) introduces further uncertainty. The reconstruction shown in this figure, and in Figures 7 and 9, is therefore necessarily schematic. A detailed discussion of the constraints and uncertainties in the reconstruction of the Gondwana plate margin is beyond the scope of this paper.
There is no indication of volcanic detritus in the Taylor Group, nor is there deformation that might be interpreted as far-field effects of subduction. The former suggests that initiation of the Devonian magmatic arc in Marie Byrd Land and north Victoria Land post-dated deposition of the Taylor Group, and the latter that either the arc was far distant or the arc was allochthonous/para-autochthonous and accreted in post-Devonian time, which again is in conflict with evidence for early Palaeozoic amalgamation of north Victoria Land.
The Late Devonian–Early Carboniferous magmatic arc in Marie Byrd Land extended into the Western Province of New Zealand, with granitoids present in the Buller Terrane and forming parts of the Median Batholith (Mortimer 2004; Wandres & Bradshaw 2005; Allibone et al. 2009). The arc continued into the southern part of the Lachlan Orogen in Tasmania and southeastern Australia (Gray & Foster 2004; Glen 2005). The Devonian–Carboniferous arc in the Antarctic Peninsula is represented only by orthogneisses (Millar et al. 2002). However, the active Gondwana margin continued through to Patagonia, where there is a fragmentary record of Devonian–Carboniferous magmatism (both S- and I-type) in the Deseado Massif and Cordillera Darwin, and also further north in the San Rafael block and the Sierras Pampeanas (Pankhurst et al. 2003; Hervé et al. 2010b).
The Lachlan Orogen, which persisted into the Carboniferous, has a well-documented complex history, involving subduction, rifting, accretion of arc terranes and incorporation of slices of continental crust and oceanic basalt. Gray & Foster (2004) have drawn analogies between the Lachlan Orogen and the present-day western Pacific. Along-strike variations in tectonic and magmatic events in orogenic belts are normal, and hence difficulties in correlation between Antarctica, New Zealand and Australia are not unexpected. The occurrence of Cambrian and Ordovician gneisses in Marie Byrd Land suggests an older crustal history in a belt far outboard of the Ross Orogen. Further, Nd model ages and U–Pb systematics of zircons from Marie Byrd Land granitoids (Pankhurst et al. 1998; Mukasa & Dalziel 2000), and Re–Os studies on Cenozoic spinel peridotite xenoliths (Handler et al. 2003), all suggest Proterozoic basement. Thus the possibility of autochthonous or para-autochthonous continental ribbons cannot be excluded. Accretion of continental blocks and arc terranes, often oblique, has been proposed for the Lachlan Orogen (Glen 2005; Cayley 2011) and for the upper Palaeozoic of Patagonia (Pankhurst et al. 2006). Rifting and later closure might explain the disjunct between the undeformed Lower to Middle Devonian quartzose Taylor Group and its correlative sequences, and the Gondwana margin, along which magmatic and compressional events were episodic throughout the early Palaeozoic, as well-documented in the Lachlan Orogen.
The Permo-Triassic Gondwana sequence
The Carboniferous time interval is represented in the Transantarctic Mountains by the Maya Erosion Surface, which is superimposed on Devonian beds but also merges with the older Kukri Erosion Surface where Devonian strata were stripped. Like the Kukri, the Maya surface has broad scale relief. Although subduction-related Carboniferous magmatism continued along the Panthalassan margin, most of the Transantarctic Mountains, as in the Devonian, must have been sufficiently distant to remain unaffected by any associated deformation and sedimentation. If any sediment had accumulated, whether derived from the craton or the inferred magmatic arc, it must have been removed by subsequent Late Carboniferous to Permian glaciation.
The Permian to Triassic Victoria Group, the Antarctic Gondwana sequence (Figs 3 & 6), occurs along the length of the Transantarctic Mountains and in at least three depositional basins. In both north and south Victoria Land, Victoria Group strata were deposited in intra-cratonic basins (Collinson et al. 1994). From the Byrd Glacier to the Ohio Range the Victoria Group accumulated in a cratonic basin that evolved in mid-Permian time into a foreland basin (Collinson et al. 1994). It is probable that by Late Triassic time the central Transantarctic Mountains basin was connected to the south Victoria Land basin but only peripherally, if at all, to north Victoria Land. The Permian Gondwana rocks from the Thiel Mountains to the Theron Mountains and beyond are geographically isolated platform sequences. Those in the Theron Mountains were derived mainly from a cratonic source to the SW (Brook 1972), that is, outboard of East Antarctica.
Present-day distribution of Permo-Triassic Gondwana rocks in the Transantarctic Mountains and the Ellsworth Mountains, and Upper Carboniferous to Triassic granitoids and gneisses in West Antarctica and the Antarctic Peninsula. See caption for Figure 4 regarding the allochthonous Ellsworth–Whitmore Mountains block, etc.
The north Victoria Land basin (Collinson et al. 1986) was probably narrow and possibly rift-related, and has only the eastern flank preserved. During Permian time a thick diamictite of probable glacial origin accumulated and was overlain by a fluvial sequence (Takrouna Formation) that changed to coal-bearing westwards away from the rift margin. Younger beds, the quartzose sandstones of the Section Peak Formation of Triassic–Jurassic age (Schöner et al. 2011), are geographically separated from the Permian rocks, overlie Wilson terrane granitic and metamorphic basement, and constitute a platform sequence. The Triassic Dicroidium flora has been found at one small isolated outcrop in north Victoria Land (Tessensohn & Mädler 1987), but its relationship to the Section Peak Formation is not known. In south Victoria Land, Victoria Group strata crop out between the David and Darwin glaciers, although it is only from Allan Hills southwards that correlations are clear. The thin glacial strata, rather than being related to a massive ice sheet, are now interpreted as glaciomarine and associated with two separate small ice sheets (Isbell 2010). These beds are overlain by Permian coal-bearing beds and then by Triassic fluvial strata that are in part carbonaceous. Clastic detritus was derived from flanking granitic sources to the east and west (Gondwana margin and East Antarctic basement, respectively) during Permian time. In Early Triassic time, there was an abrupt change in palaeocurrent direction to northerly, but it was not until Late Triassic time that the provenance changed with a significant influx of silicic volcanic detritus. The shift in direction of palaeoflow and the incoming of volcanic detritus has suggested a palaeodrainage link between the basin south of Byrd Glacier and that in south Victoria Land, and to possible over-topping of the Ross High inferred for the Byrd Glacier region (Collinson et al. 1994).
The Permo-Triassic Gondwana sequence crops out from south of the Byrd Glacier to the Ohio Range and beyond, but is thicker and more complete in the Beardmore–Shackleton glacier region (central Transantarctic Mountains) than anywhere else (Fig. 3). The glacial strata, formerly thought of as terrestrial and deposited from a single large ice sheet, are now interpreted as glaciomarine, deposited in distinct basins, and to represent much smaller-scale ice sheets (Isbell et al. 2008). These re-interpretations of the Permian glacial beds lead to a significantly different view of Gondwana glaciation in Antarctica (Isbell et al. 2008; Isbell 2010). The glacial strata in the central Transantarctic Mountains are overlain by post-glacial shales and fine-sandstones (Mackellar Formation) deposited in a large post-glacial lake or inland sea (Miller & Isbell 2010). Towards the Ohio Range, equivalent strata were deposited in environments ranging from freshwater/brackish to marine (Collinson et al. 1994), and in an inland sea that extended to South Africa and South America. Mackellar beds are overlain by a deltaic to fluvial sequence (Fairchild Formation), which then passes up into fluvial coal-bearing strata of the Buckley Formation. Palaeoflow was generally southeastward along the mountain range until the middle Permian, when there was a swing to a northwesterly direction. Correlative Permian strata are exposed through to the Ohio Range, and carbonaceous Glossopteris-bearing Permian beds occur at isolated nunataks as far as the Theron Mountains. The Permian–Triassic boundary, with one documented exception in the Shackleton Glacier region, appears to be a local disconformity (Collinson et al. 2006 and references therein).
Triassic strata in the central Transantarctic Mountains crop out from the Queen Alexandra Range to the Nilsen Plateau (Long et al. 2009), but are absent farther along the mountain range. The Triassic beds (Fremouw and Falla formations) are fluvial and in the upper part carbonaceous with thin coals. Palaeoflow was northwesterly. The approximate 180° reversal in uniform flow directions suggests that the extant record in the central Transantarctic Mountains mainly represents the axial part of the depositional basin (e.g. Barrett 1991). It is only in the Shackleton Glacier region that palaeoflow directly indicates derivation of detrital grains from the basin (arc) flank (Collinson et al. 1994), although north of the Beardmore Glacier there is some indication of flow from the craton during the Permian (Barrett et al. 1986).
Sedimentary petrology indicates a change in provenance in mid-Permian time from one dominated by basement igneous and metamorphic rocks to one with a significant volcanic input (Barrett et al. 1986). Palaeocurrent data from the Shackleton Glacier region demonstrate that the source of the intermediate to silicic detritus lay in West Antarctica. Detrital zircon studies (Elliot & Fanning 2008) on volcaniclastic upper Buckley sandstones revealed an abundance of primary igneous grains, and within the limitations of the method, suggest that magmatism may have started in the Early Permian. Although the overwhelming volcanic character of the upper Buckley in the Shackleton Glacier region is not repeated in the overlying Triassic strata, young zircon grains are present in those beds and show that magmatism continued through Late Triassic time (Elliot & Fanning 2011).
Late Carboniferous to Triassic plate margin
An upper Carboniferous orthogneiss (309 Ma) from Thurston Island and a granitoid from eastern Marie Byrd Land (Pankhurst et al. 1993) may represent an early phase of the younger Permo-Triassic arc of the Gondwana Panthalassan margin (Fig. 7). The Permo-Triassic magmatic arc is documented sparsely through West Antarctica by granitoids (Pankhurst et al. 1993, 1998; Mukasa & Dalziel 2000) and magmatism is supported by detrital zircon geochronology of Buckley, Fremouw and Falla sandstones (Elliot & Fanning 2008, 2011).
Gondwana plate margin during the late Palaeozoic–early Mesozoic, showing the distribution of Permian-Triassic strata in the Transantarctic Mountains and Late Carboniferous to Triassic granitoids and gneisses in Marie Byrd Land, the Antarctic Peninsula and adjacent regions. See caption for Figure 5 for information on the reconstruction.
The arc and associated environments are documented in New Zealand by the Permian Brook Street terrane of oceanic volcanic rocks, Upper Permian to Triassic volcaniclastic strata of the Murihiku, Maitai and Caples volcanic arc-related terranes, by the Rakaia terrane, which differs in that it was derived from a continental magmatic arc (Mortimer 2004; Wandres & Bradshaw 2005), and by a Permian to Triassic intra-oceanic I-type batholith complex (Price et al. 2011). Farther afield, it is documented in the New England Orogen of eastern Australia where initiation of the arc, with both intrusive and extrusive activity, was in Late Carboniferous time and extended through the Permian Period (Glen 2005).
Magmatic activity in the Antarctic Peninsula is recorded by granitoids (many S-type) and orthogneisses, mainly of Triassic age (Vaughan & Storey 2000; Millar et al. 2002), and also by detrital zircons, dominantly Permian, in the fore-arc Trinity Peninsula Group (Barbeau et al. 2010; Bradshaw et al. 2012). In South America magmatic rocks are found in the coastal ranges of Chile (Hervé et al. 1988), and northern Patagonia (Pankhurst et al. 2006), where they are part of the Carboniferous–Triassic record of subduction and accretion. A Permian metamorphic event is also documented in basement rock from Tierra del Fuego, Patagonia (Hervé et al. 2010a). Detrital igneous zircons of Permian age occur in the Duque de York complex on the Pacific Coast (Sepúlveda et al. 2010) and in a quartzite from the Cordillera Darwin (Hervé et al. 2010b).
In the Antarctic Peninsula, fore-arc assemblages are represented by the Upper Carboniferous and younger accretionary complex of the LeMay Group of Alexander Island (Doubleday et al. 1994; Kelly et al. 2001) and the Carboniferous(?)–Triassic upper plate Trinity Peninsula Group (see Bradshaw et al. 2012). In southern Chile the Late Carboniferous to Permian Madre de Dios Terrane includes lower plate accretionary complexes as well as upper plate strata derived from continental sources (Sepúlveda et al. 2008, 2010).
Gondwanide deformation, located well inboard of the arc, is documented from the Sierra de la Ventana of northern Patagonia, to the Cape fold belt in South Africa, the Falkland Islands, and the Ellsworth and Pensacola mountains. Located in the Weddell Sea sector of a Gondwana reconstruction, the Permian Polarstar Formation of the Ellsworth Mountains is marked by folded strata, volcanic detritus in sandstones and sparse tuff beds (Collinson et al. 1992; Curtis 2001). Isolated small outcrops of dipping volcaniclastic strata in the southernmost Antarctic Peninsula (Erehwon Nunatak region) contain Glossopteris (Laudon 1991) and represent beds deposited proximal to the orogen. Deformation of Permian beds in Antarctica is displayed in the Ellsworth Mountains and to a significantly lesser extent in the Pensacola Mountains where the Permian diamictites of the Gale Mudstone form open folds, which suggests that the latter region was near the inboard limit of Gondwanide deformation. The timing of deformation in Antarctica is poorly constrained to a post-Permian age (Curtis 2001), although in South Africa, deformation is inferred to have been episodic between 278 and 230 Ma (Hälbich et al. 1983). How far towards the Ross Sea sector deformation continued is unknown, but Permian–Triassic deformation is well documented in the New England Orogen of eastern Australia (Glen 2005). Nevertheless, the major influx of volcanic detritus into the central Transantarctic Mountains Victoria Group basin in mid-Permian time in the Shackleton Glacier region and the evolution of the foreland basin suggest continuing tectonism and magmatism related to the plate margin (Elliot & Fanning 2011).
The extant part of the foreland basin in the central Transantarctic Mountains is the axial region and the thickness of the whole sequence at its margins suggests that the original basin was much wider, extending towards the orogen on one flank and to a forebulge on the basement flank. The axial drainage migrated across the foreland, reflecting the interplay between accommodation space, sediment input and tectonism (Collinson et al. 1994).
Structure contours on the Kukri Erosion Surface in the central Transantarctic Mountains suggest a gently southward plunging syncline (Barrett 1969). Although it might be interpreted in terms of the foreland basin, the syncline appears to have a relatively steep limb on the craton flank, opens out to the SE, and its axial trace is not coincident with the dominant flow directions particularly for Triassic strata. Thus the syncline is more likely to represent mainly Jurassic and/or younger warping associated with uplift of, and faulting within, the Transantarctic Mountains.
The central Transantarctic Mountains Permo-Triassic Victoria Group strata reflect sedimentation and tectonism associated with the Gondwanide orogen. The magmatic arc itself is documented in Marie Byrd Land by granitoids, by igneous zircons in the Victoria Group, and by igneous and metamorphic rocks in the Antarctic Peninsula. A fold-and-thrust belt in the Weddell Sea sector may have extended into the Ross Sea sector. The changes in zircon age provinces recorded in the Victoria Group sandstones may reflect thrusting events bringing rocks of differing ages to the surface and/or changing sediment sources (Elliot & Fanning 2008). The orogen played a major role in the development of the foreland basin, in contrast to the intra-cratonic basins of north and south Victoria Land, which were not directly influenced. The influx of volcanic detritus in the Triassic of south Victoria Land may simply record the integration of drainage with that of the central Transantarctic Mountains, which originated in a region of volcaniclastic sediment input.
Early Jurassic time: rifting
Lower Jurassic strata
The far-field effects of subduction along the Gondwana active margin, which had influenced the evolution of the Transantarctic Mountains during Permian and Triassic time, were brought to a close with the events leading up to the basaltic magmatism which formed the precursor to Gondwana break-up. The pre-basalt Lower Jurassic stratigraphic record, characterized by silicic volcaniclastic beds, is sparse and largely confined to those areas where the Ferrar extrusive rocks are still present (Fig. 8).
Present-day distribution of Lower Jurassic silicic and basaltic rocks in the Transantarctic Mountains and Jurassic magmatic rocks in West Antarctica and the Antarctic Peninsula.
These rocks are exposed primarily in the central Transantarctic Mountains (Hanson Formation; Elliot 1996) and in north Victoria Land (Shafer Peak Formation; Schöner et al. 2007). In the central Transantarctic Mountains the sequence, disconformable on the Falla Formation, consists of tuffaceous sandstones and finer-grained beds and tuffs. Arkosic sandstones with coarse K-feldspar clasts are interbedded in the lower part of the sequence, indicate a proximal basement source and suggest the possibility of deposition in a rift (Elliot & Larsen 1993).
In south Victoria Land there is only one location known so far where silicic shard-bearing sedimentary rocks overlie the Triassic Lashly Formation, and that is at Coombs Hills (Elliot & Grimes 2011). The section is no more than a few tens of metres thick, and the upper contact is missing. However, silicic tuff clasts in the overlying Mawson Formation basaltic pyroclastic rocks suggest a formerly more widespread occurrence of silicic strata, now removed by erosion.
By contrast, a more extensive record is present north of the Priestley Glacier in north Victoria Land (Schöner et al. 2007; Viereck-Goette et al. 2007). There, the siliciclastic Section Peak Formation passes up into a silicic shard-bearing sandstone sequence informally named the Shafer Peak Formation. This sequence consists of tuffaceous sandstones and siltstones, and is reported to be interbedded with Ferrar basaltic pyroclastic rocks of the Exposure Hill Formation (redefined by Viereck-Goette et al. 2007).
The silicic shard-bearing sequences are overlain by basaltic pyroclastic rocks and then flood lavas of the Ferrar Large Igneous Province (Elliot & Fleming 2008). The pyroclastic rocks are principally lahars, tuff breccias and tuffs. In the Queen Alexandra Range, central Transantarctic Mountains, they form stratigraphic sequences up to c. 200 m thick, whereas in the Prince Albert Mountains they are only a few tens of metres thick, although the basal and top contacts are not exposed. At Allan Hills (south Victoria Land) stratified pyroclastic rocks are cut by diatremes (Elliot et al. 2005; Ross et al. 2008). In contrast, in both north Victoria Land and at Coombs Hills (south Victoria Land), and probably also at Otway Massif (central Transantarctic Mountains), the pyroclastic rocks fill shallow-level phreatic vents or diatremes in which the vertical extent of the deposit may be several hundred metres (Elliot & Hanson 2001; White & McClintock 2001; Viereck-Goette et al. 2007). Thin sequences of pyroclastic rocks also occur, both as rafts in the vents and overlying the diatremes.
With the few exceptions where breccias and lavas are interbedded, the pyroclastic activity was abruptly replaced by effusion of massive lava flows that form sequences as much as 900 m thick (Elliot & Fleming 2008). Thin sedimentary interbeds occur low in the lava succession and beneath the capping lava flow. Many of the flows are more than 100 m thick and locally attain as much as 230 m at the Otway Massif. In the Prince Albert Mountains a massive pillow palagonite succession (c. 100 m thick) occurs at the base of the lava pile and together with the great thickness of many lava flows in the central Transantarctic Mountains and north Victoria Land suggest confining topography at the time of emplacement. In support of a rift setting is the occurrence of monoclinal structures in the central Transantarctic Mountains, one of which, west-facing, offsets the Hanson Formation and has pyroclastic breccia of the Prebble Formation injected into the fault plane (Elliot & Larsen 1993), and another of which displaces Kirkpatrick lavas. These data are interpreted to suggest that the Jurassic rocks were emplaced in a rift that was tectonically active through much of the Early Jurassic. A west-facing faulted monocline displaces Permian beds east of the Marsh Glacier (Barrett et al. 1970), a NW-facing normal fault offsets Buckley and Fremouw beds at Coalsack Bluff (Collinson & Elliot 1984), and an uninvestigated east-facing faulted monocline is also visible on the west face of the Dominion Range, offsetting Victoria Group beds (Elliot et al. 1974). Further, a small graben structure offsets Victoria Group beds near the head of the Beardmore Glacier (Barrett & Elliot 1973). These undated structures are attributed to Early Jurassic tectonism because they occur some distance back from the Cenozoic escarpment (discussed later).
Sills ranging in thickness up to 700 m, but more commonly 100–200 m, were emplaced into the underlying Beacon sequence throughout the Transantarctic Mountains, and locally into basement rocks (Elliot & Fleming 2004). In complete Beacon sections, the cumulative thickness of sills may be as much as 1500 m, suggesting that the sills are the dominant component of the Ferrar Large Igneous Province.
The source, or sources, for the Ferrar magmas, and their dispersal, remains uncertain. Ferrar magmas are characterized by enriched initial Sr and Nd isotope ratios and by trace element characteristics with a strong crustal imprint (Mensing et al. 1984; Hergt et al. 1989a; Fleming et al. 1995, 1997; Molzahn et al. 1996). Almost all show great geochemical coherence. The exceptions are the capping lavas together with some of the sills in the Theron Mountains and Whichaway Nunataks, which have a single similar but very distinct, coherent and evolved composition (Elliot et al. 1999; Leat et al. 2006). The geochemical characteristics of the latter have been interpreted to demonstrate derivation from a single source that was located in the region of Gondwana break-up between Antarctica and Africa and from which magmas were transported at depth throughout the Transantarctic Mountains (Elliot et al. 1999). This origin has been extended to the Ferrar as a whole, which includes southeastern Australasia and lava in South Africa of possible Ferrar composition (Elliot & Fleming 2000, 2004). In this scenario, the Ferrar magmas acquired their enriched crustal signature at the point of generation, probably a mantle plume (Storey 1995; Storey & Kyle 1997; Storey et al. 2001). Subsequent evolution of the majority of magmas was by fractionation with a small amount of superimposed crustal contamination (Fleming et al. 1995, 1997). Although sourcing of the magmas from the Dufek Massif and transport through supracrustal sills has been proposed (Storey & Kyle 1997; Ferris et al. 2003), it seems unlikely given the single unique composition of the capping lava flow(s), which are spread over 1600 km, the contrasting geochemical differences (e.g. MgO content) in the majority of the lavas from any one region compared with the next (central Transantarctic Mountains v. south Victoria Land v. north Victoria Land), and the presence of basement highs (Ross High, and between south Victoria Land and north Victoria Land; Elliot & Fleming 2008). Alternatively, rather than a plume origin, derivation of the magmas from a linear source in the subjacent heterogeneous lithospheric mantle, possibly enriched by subducted sediment during the Palaeozoic and Early Mesozoic, has been proposed (Kyle et al. 1983; Cox 1988; Hergt et al. 1989b, 1991; Molzahn et al. 1996). However, given the heterogeneous mantle along a length of 3000 km and the varied crustal provinces through which the magmas would then have ascended, the coherence of the geochemical characteristics over such a distance seems to make this distributed source a problematic origin. Magma mixing and crustal anatexis have also been proposed as important mechanisms for developing the Ferrar geochemistry (e.g. Mensing et al. 1984, 1991).
The rift setting proposed for the Ferrar rocks and the underlying volcaniclastic beds suggests that the setting might have been a series of linked rift basins (Elliot & Fleming 2008). The rift basins are now marked by phreatomagmatic deposits that preceded eruption of the lavas, and these sites formed the foci for subsequent major sill and lava emplacement. The magmas themselves were transported long distances at deep crustal levels or at the crust–mantle boundary prior to vertical migration into and through the upper crust and supracrustal rocks. Major dyke swarms for magma transport have not been recognized in outcrop, although they have been inferred from magnetic data near the Dufek Massif (Ferris et al. 2003) and in the central Transantarctic Mountains (Goodge & Finn 2010). Bounding faults for this proposed rift system have not been identified, although the Hanson Formation arkoses and quartzose pebbly beds indicate that at least one fault could not have been too far distant; given the local thickness of Victoria Group strata, that fault could have had a cumulative displacement of as much as 2 km. It is possible that the major discontinuity, the Marsh Fault, trending north–south along the Marsh Glacier (central Transantarctic Mountains) and juxtaposing the Miller Range Proterozoic rocks with the Victoria Group to the east (Goodge & Finn 2010), represents such a bounding fault. Faults, defining a small graben, also cut Victoria Group beds at the head of the Beardmore Glacier, and lie along-strike from the Marsh Fault, but the offsets are at least an order of magnitude less.
The Ferrar Large Igneous Province, as defined by its geochemical characteristics, extends into Tasmania and southeastern Australia (Hergt et al. 1991), New Zealand (Mortimer et al. 1995), and possibly into South Africa (Elliot & Fleming 2000). The precise temporal and source region relationships with the Karoo Large Igneous Province remain uncertain, and compared with the Ferrar province, Karoo rocks have a wide range of distinct magma compositions (Marsh et al. 1997).
The relationship between the silicic and basaltic magmatism remains unclear. Silicic magmatism was initiated with the onset of Hanson Formation deposition, and although a hiatus is inferred at the contact of the Falla and Hanson formations, the timing is not adequately constrained. Zircons from a tuff at Mt Kirkpatrick, about 100 m below the Hanson/Prebble contact, gave a super high resolution ion microprobe U–Pb age of about 194 Ma, and from a tuff 50 m below the contact yielded an age of about 186 Ma (Elliot & Fanning unpublished data). A super high resolution ion microprobe U–Pb zircon age of 182.7 ± 1.8 Ma on a silicic tuff clast in the basaltic Prebble Formation at the Otway Massif (Elliot et al. 2007) together with silicic shards in the upper interbed (palaeosol) at Mount Bumstead, central Transantarctic Mountains (Elliot et al. 1991), suggest that silicic volcanism continued into and during Ferrar time. At the type section of the Hanson Formation, the upper tuff unit includes accretionary lapilli and other evidence for proximal volcanism. Thus the older volcanism recorded in the Hanson Formation could have been distal plate margin related, whereas the youngest volcanism was proximal and possibly then anatectic and related to emplacement of Ferrar magmas, which have a best estimate U–Pb zircon and baddeleyite thermal ionization mass spectrometry age of 183.6 ± 1.8 Ma (Encarnación et al. 1996). The ages of the Ferrar rocks are under further investigation (Burgess et al. 2011).
Early Jurassic plate margin
The long-standing active margin of Gondwana continued into the Jurassic Period (Fig. 9). The record is clear in New Zealand, where it is documented by forearc volcaniclastic strata of the Murihiku and Pahau terranes (Adams et al. 2002; Mortimer 2004; Wandres & Bradshaw 2005), and a single Lower Jurassic granite from the remote Bounty Island (Wandres & Bradshaw 2005; Adams 2008), but it is very sparse in West Antarctica where there are but two igneous rocks of Early Jurassic age in the Thurston Island block (Pankhurst et al. 1993). In the Antarctic Peninsula the record is more substantial, with both intrusive (Leat et al. 1995, 2009; Vaughan & Storey 2000) and extrusive rocks (Pankhurst et al. 2000; Riley et al. 2001; Hunter et al. 2006) demonstrating igneous activity in the Early Jurassic, although magmatism was much more extensive in later Jurassic time. The arc continued into southern South America, where it is recorded by I-type granites in the sub-cordilleran belt of Patagonia (Rapela et al. 2005), although the principal igneous activity of Early Jurassic age was eruption of the silicic Marifil Province in northern Patagonia (Pankhurst et al. 2000).
Early Jurassic Gondwana reconstruction showing the distribution of magmatic rocks associated with the Karoo–Ferrar Large Igneous Province and the magmatic arc along the Gondwana margin. See caption for Figure 5 for information on the reconstruction.
The setting for the Late Triassic to Early Jurassic magmatism in the Antarctic Peninsula is postulated to have been extensional (Storey et al. 1996; Vaughan & Storey 2000; Vaughan et al. 2013). However, the magmatic rocks emplaced during this time interval are interpreted to be a mixture of subduction-related and intra-plate granitoids (Millar et al. 2001). If the extensional setting of the Antarctic Peninsula sector of the Gondwana margin continued into West Antarctica, then the rift setting inferred for the Ferrar Large Igneous Province may have been a function of related back-arc extension, resulting from trench rollback, which thus created the deep crustal pathway for dispersal of Ferrar basaltic magmas.
Mid-Jurassic to mid-Cretaceous time: plate margin reorganization
This is a crucial time interval in the tectonic history of Antarctica, leading from an intact Gondwana to the now-isolated continent with the Ellsworth Mountains–Whitmore Mountains tectonic block, Thurston Island block, and others such as the Falkland Islands, moved from their postulated locations in the proto-Weddell sea region to their present positions. Based on palaeomagnetic data, there was major reorganization and rotation of the Ellsworth–Whitmore block (c. 90°) and the Falkland Islands (c. 150°) by 175 Ma, whereas the oldest Thurston Island block rotation is constrained only to be earlier than 125 Ma (Grunow et al. 1991). By that time the Ellsworth–Whitmore and Thurston Island blocks and the Antarctic Peninsula formed a single entity referred to as Weddellia. Subsequently Weddellia moved essentially as a single unit, rotating to its present position with respect to East Antarctica by about 110 Ma (Grunow et al. 1987, 1991). Displacements and rotations of the fragmented Gondwana margin have been attributed to the arrival of the Ferrar–Karoo mantle plume (Dalziel et al. 2000; Storey et al. 2001). However, the actual mechanisms and paths for these early movements remain unclear, although an analogy has been drawn with the Afar region (Dalziel & Lawver 2001) and with back-arc spreading and subduction rollback (Martin 2007). Further, aerogeophysical data for the head of the Weddell Sea have been interpreted to support a rift–rift–rift triple junction above a rising plume (Ferris et al. 2000).
Plate margin rearrangement must have occurred during a short time interval, for which there is evidence of extension in the Antarctic Peninsula region and continuing, although diminished, subduction (Vaughan & Storey 2000). Despite extensive marine geophysical surveys in the southern South Atlantic and Weddell Sea (König & Jokat 2006), the early stages of break-up are still obscure and it is only after anomaly M20 (167 Ma), the oldest identified anomaly which is located at the head of the Weddell Sea, that data allow interpretation of seafloor evolution. That anomaly determines by when the movement of the Ellsworth–Whitmore block and formation of the thinned crust of the Weddell embayment must have been effectively completed. Nevertheless, following those major movements, several phases of tectonism have been recognized for the Antarctic Peninsula, including possible accretion of suspect terranes (Vaughan & Storey 2000).
All the major translations and rotations must have occurred in a 15 m.y. interval following Karoo–Ferrar magmatism at about 183 Ma, and must have entailed movements along major shear zones and rotations between the various blocks. Further, there must have been stretching of continental crust and development of small oceanic basins (Dalziel & Lawver 2001). Later events (167–125 Ma) must have resulted from continuing Weddell Sea opening and plate margin development.
Cretaceous to Cenozoic time: rifting and uplift/denudation
The original thickness of the Ferrar lava sequence is uncertain and it is possible that a substantial section has been stripped. Studies of zeolites in basalt sequences have led to the recognition of zonations that must be related to temperature and thickness of the volcanic pile, fluid composition and exothermic hydration reactions (e.g. Walker 1960; Neuhoff et al. 2000). Zeolite zonation has been used to infer original thicknesses of volcanic piles. In Ferrar extrusive rocks, zeolite minerals filling amygdales, vugs and cavities between pillow basalts consist principally of stilbite and heulandite, with lesser scolecite and mesolite, plus apophyllite (Fleming et al. 1999). The absence of laumontite suggests that the original thickness of Kirkpatrick lavas could have been as much as 2500 m, although it was more likely to be some 1500 m or less (Neuhoff et al. 2000). Further, there is the possibility that the overburden included a now-eroded Jurassic to Cretaceous sedimentary cover that accumulated within the postulated Ferrar rift.
Fission track studies throughout the Transantarctic Mountains (see Miller et al. 2010 and references therein) have shown several late Mesozoic episodes of denudation (Fig. 10), but the principal event was initiated in Eocene time. The oldest and only Jurassic (165–150 Ma) denudation is recorded in the Thiel Mountains (Fitzgerald & Baldwin 2007); elsewhere in the Transantarctic Mountains significant middle Cretaceous (ca. 125–90 Ma) denudation is recorded in the Scott Glacier region (Fitzgerald & Stump 1997), central Transantarctic Mountains (Fitzgerald 1994) and both south and north Victoria Land (Fitzgerald & Gleadow 1988; Fitzgerald 1992; Balestrieri et al. 1997; Lisker 2002; Lisker et al. 2006; Storti et al. 2008). In support of Cretaceous denudation are the Rb–Sr and 40Ar/39Ar age determinations on apophyllite in Ferrar basaltic rocks, which are interpreted to reflect uplift and reorganization of ground water systems with consequent mineral precipitation (Fleming et al. 1999; Molzahn et al. 1999). Further, in north Victoria Land, low-temperature alteration of Ferrar lavas in the Cretaceous is also recorded by Rb–Sr relationships, K–Ar dates and palaeomagnetic pole positions (summarized in Faure & Mensing 1993; Fleming et al. 1993). The Transantarctic Mountains are segmented geographically and topographically, and this is consistent with the fission-track data, which suggest a number of individual blocks, delineated by major outlet glaciers that have slightly different histories. Those outlet glaciers undoubtedly follow major faults, which are also suggested by offsets of the Kukri Erosion Surface. These offsets of a few hundred metres are small compared with the lateral persistence of this erosion surface over a distance of more than 2000 km, which implies a broadly similar uplift history along the length of the Transantarctic Mountains (Fitzgerald 1994, 2002). Apophyllite dates from replacement deposits in Permian sandstone provide further evidence of groundwater events, possibly related to uplift, at about 65 Ma (Elliot et al. 2004).
Late Cretaceous extension and indicators in the Transantarctic Mountains of the onset of uplift/denudation during Late Jurassic and Cretaceous time. The West Antarctic Rift System is located between the long dashed lines; inferred Cretaceous basins from Decesari et al. (2007). Fission track ages are in upright lettering; in italics are Rb–Sr and 40Ar/39Ar apophyllite ages; underlined are ages for Rb–Sr ‘errorchrons’ for Kirkpatrick Basalt lavas. See text for references.
The Cenozoic uplift/denudation is associated with faulting along the Transantarctic Mountains front, which coincides with the rapid change in crustal thickness (40–20 km) from the East Antarctic basement to the Ross embayment (Lawrence et al. 2006), that is, the West Antarctic Rift System (e.g. Behrendt 1999). Faults along the front have been inferred or documented in south Victoria Land and the central Transantarctic Mountains (Beardmore and Shackleton glacier areas) from offset Beacon strata or basement rock thermochronology (Fitzgerald 1992, 1994; Miller et al. 2010). Only at one locality, Cape Surprise at the mouth of the Shackleton Glacier, is there outcrop of Permian Victoria Group strata that enables a direct estimate of the offset (c. 2.5 km) of the Kukri Erosion Surface along the mountain front (Miller et al. 2010). However, an even larger offset (>3 km) has been estimated from the Cape Roberts Project drill site number three (CRP-3), 10 km off Cape Roberts, south Victoria Land, which encountered Taylor Group (Devonian) quartz sandstone c. 1100 m below sea-level beneath Cenozoic glacimarine strata (Cape Roberts Science Team 2000, Barrett 2007). Comparable faults doubtless occur along other segments of the Transantarctic Mountains.
The uplift of the Transantarctic Mountains is clearly linked in some manner to the change in crustal thickness and the extensional regime of the West Antarctic Rift System. Investigations in Marie Byrd Land have shown that the extensional regime was initiated in Late Cretaceous time at about 105 Ma, and is documented by a core complex that formed while subduction along the margin was still active. This is supported by a fission track age of c. 100 Ma, indicating uplift/denudation at that time (Siddoway 2008). Correlated erosion surfaces in Marie Byrd Land and the South Island of New Zealand have been linked to this denudation in Late Cretaceous time (LeMasurier & Landis 1996). Much of the extension, up to 600 km, in the West Antarctic rift system is attributed to these Cretaceous events, during which four sedimentary basins were formed and filled by up to 4 km of sediment (Decesari et al. 2007). Compared with the Cenozoic, Cretaceous tectonism on the Transantarctic Mountains flank of the extensional rift system was limited, although widespread, and is documented in the denudation inferred from fission track data (Fig. 10). The separation of Campbell Plateau SE of New Zealand from Antarctica (Marie Byrd Land) started at about 84 Ma, and is a separate event that is notable for cutting across pre-existing extensional structures.
Based on fission track data, major denudation and uplift of the Transantarctic Mountains were initiated during the Eocene at about 50–55 Ma (Fig. 11), and there is a suggestion that the denudation event becomes younger southward from north Victoria Land to the central Transantarctic Mountains (Fitzgerald 2002; Miller et al. 2010). The onset in north Victoria Land is somewhat earlier than the opening of the Adare Trough, which was active from 43 to 26 Ma (Cande & Stock 2006). Adare Trough opening is linked to further extension in the Ross Sea basins, contributing to the increased thicknesses (up to −14 km in the Terror Rift) of strata in these basins (Trey et al. 1999). In a detailed analysis of seismic and core data from the Cape Roberts/McMurdo Sound region, five stages of tectonic activity and sedimentation have been recognized for the Victoria Land Basin (Fielding et al. 2006). The Ross Sea basins, located in rifts with thin crust, as little as 4 km in the Terror Rift (Davey & Brancolini 1995), surely extend south beneath the Ross Ice Shelf, and have been inferred from depth to basement and gravity anomalies (Decesari et al. 2007). Nevertheless, a 140 km-long seismic transect from the Ross Ice Shelf into the mountain front between the Nimrod and Beardmore glaciers revealed neither a large offset in thick sedimentary sequences beneath the western margin of the Ross Ice Shelf, nor a graben that might be the extension of the Victoria Land Basin/Terror Rift system, just where it might be expected given the known offsets in the frontal part of the range in the central Transantarctic Mountains (ten Brink et al. 1993; Fitzgerald 1994). Another apparent anomaly is the relative youth of the oldest strata yet cored compared with the start of denudation of the Transantarctic Mountains: earliest Oligocene (34 Ma) for the Victoria Land Basin (Cape Roberts Project drill site number three, or CRP-3; Barrett 2007); and late Oligocene for the Eastern Basin (DSDP Site 270; Ford & Barrett 1975; Fig. 11). However, because sedimentation was occurring in a region of numerous faults, this anomaly may simply reflect where cores have been obtained.
Distribution of Cenozoic volcanic rocks, Sirius Group deposits in the Transantarctic Mountains, location of sedimentary basins and areas of extension in the Ross Sea, and interpreted basins (and extension) beneath the Ross Ice Shelf. Fission track ages are for onset of the principal episode of uplift/denudation in Eocene time, except for those in italics, which are for youngest ages given by fission track analysis. CB, Central Basin; EB, Eastern Basin; NB, Northern Basin; VLB, Victoria Land Basin. Light shading marks the extent of the inferred Cretaceous basins beneath the Ross Sea (see Fig. 10); heavy dashed lines with darker shading mark crust highly extended in Cenozoic time (Decesari et al. 2007) and the Cenozoic basins beneath the Ross Sea and by inference beneath the Ross Ice Shelf.
The early Cenozoic phase of extension was followed by the onset of alkaline basaltic magmatism of the McMurdo Volcanic Group (LeMasurier & Thomson 1987), which was initiated by 48 Ma, the age of an alkali gabbro in north Victoria Land (Rocchi et al. 2003), although the age of the oldest extrusive rock, an ash from the Cape Roberts Project drill site number three (CRP-3), is much younger (24 Ma; Barrett 2007). Alkaline volcanism extended as far south as Mt Morning in south Victoria Land, but with two outliers near the head of the Scott Glacier (Stump et al. 1980). Active volcanism continues today at Mt Erebus and fumarolic activity is present at Mt Melbourne. On the other West Antarctic Rift System flank (Marie Byrd Land) alkaline volcanism (LeMasurier & Thomson 1987) started earlier (c. 36 Ma; Wilch & McIntosh 2000) and continues to the present in the fumarolic activity at Mt Berlin (76°S 136°W). The relationship, if any, between the Marie Byrd Land and McMurdo volcanic provinces, separated by several hundred kilometres, remains uncertain, although both tap deep mantle plume sources (Hole & LeMasurier 1994). The relationships between onset of rapid uplift (c. 55 Ma), the apparent onset of sedimentation (c. 34 Ma) and the onset of major alkaline volcanism (c. 24 Ma) remain uncertain.
The Transantarctic Mountains have the appearance of a classic rift shoulder, but the mechanism for uplift remains controversial. The range clearly marks a major geological boundary with contrasting geological provinces in West Antarctica. However, the topographic expression is much subdued in the segment from the Ohio Range through to the Weddell Sea, where the Ellsworth–Whitmore Mountains block abuts the Transantarctic Mountains (Storey et al. 1988), and in the Pensacola Mountains region where a significant geological boundary (crustal thickness change and/or major faulting) is recognized geophysically (Behrendt et al. 1974).
Discussion of the origin of the high topography of the Transantarctic Mountains, the mechanism of its formation and the relationship to the abrupt change in crustal thickness has been centred on the Ross Sea sector. More data exist for the Ross embayment than for the Weddell Sea sector, where there is a comparable vast province of thin crust beneath the Filchner–Ronne Ice Shelf. However, it should be noted that the latter region must differ in origin, dating from the initial break-up of Gondwana and probably for the most part completed by mid-Cretaceous time. Further, Stern et al. (2005) proposed that as much as half of the peak elevation of the Transantarctic Mountains in the Ross Sea sector may be from isostatic rebound, resulting from a combination of glacial incision and permanently frozen high peaks for the last 14 Ma, rather than large-scale tectonic processes.
An origin for the Transantarctic Mountains in the Ross Sea sector as simple rift shoulder uplift, resulting from extension in the West Antarctic Rift System, has been the long-standing interpretation (Davey & Brancolini 1987), and in this interpretation Marie Byrd Land has been considered the opposing rift shoulder. Uplift has been attributed to flexural uplift at the boundary between the contrasting regimes of the cold thick East Antarctic basement and the hot thin West Antarctic Rift System (e.g. Stern & ten Brink 1989). Lack of data to support the theoretical models led ten Brink et al. (1997) to propose that transtensional motion during the Eocene between East and West Antarctica, and resulting from plate re-organization in the SW Pacific Ocean, was the cause. This motion promoted a crustal break that enabled uplift of isostatically uncompensated crust and created the Transantarctic Mountains escarpment. There is accumulating evidence for transtensional motion in the north Victoria Land–western Ross Sea region (e.g. Wilson 1995; Storti et al. 2008), linking the Adare Trough, NW–SE structures in north and south Victoria Land, and the north–south oriented Victoria Land Basin.
A different approach has been through numerical modelling, which suggested the possibility that West Antarctica was a region of thick crust and high elevation, and subsequent extensional collapse has left the Transantarctic Mountains as a plateau margin remnant (Studinger et al. 2004; Bialas et al. 2007). This model formed the starting point for more recent numerical modelling of the relations between the Transantarctic Mountains and the West Antarctic Rift System (van Wijk et al. 2008). In support of the plateau, remnants of earlier drainage patterns scattered through the Transantarctic Mountains, with flow away from West Antarctica, have been advocated (Huerta 2007). The plateau, presumably coincident with the future West Antarctic Rift System, is postulated to have existed for some time until the mid-Cretaceous (c. 110 Ma), when subduction on the Pacific margin ceased and extension began with core complex development in Marie Byrd Land (Siddoway 2008). However, from break-up of Gondwana at c. 183 Ma until mid-Cretaceous time, the plate margin was being reorganized. This involved major rotations and displacements of various continental blocks including Marie Byrd Land, which according to palaeomagnetic data underwent major translation in mid-Cretaceous time (DiVenere et al. 1996). It is unclear how a high plateau could have been maintained under those circumstances. Further, it is unclear how plateau drainage would be distinguished from that on the gently sloping back side of a rift shoulder uplift. Cretaceous strata with a West Antarctic provenance on the East Antarctic flank of the Transantarctic Mountains would be an argument for the plateau model.
Fission track dating indicates uplift/denudation episodes as young as 24 Ma between north and south Victoria Land (Storti et al. 2008) and 20 Ma in the Beardmore Glacier region (Fitzgerald 1994). However, the amount of uplift in the last 14 Ma is relatively modest. In the Shackleton Glacier region it is estimated to be 290–790 m (Miller et al. 2010). In south Victoria Land geological evidence in the form of fjord deposits indicates a similar uplift of about 600 m in the last 10 myr (Webb & Wrenn 1982), and terrestrial volcanic cones show a maximum uplift of <300 m (permissibly 0 m) in the last 2.6 myr (Wilch et al. 1993). Further, the elevation of Miocene near-shore marine sediments off the south Victoria Land coast is within a few hundred metres of sea-level in the Cenozoic Investigations in the western Ross Sea (CIROS-1) and Cape Roberts Project (CRP) drill sites in McMurdo Sound, and likewise suggest relative Late Cenozoic tectonic stability of the rift margin (Fitzgerald 2002; Barrett 2007).
Regardless of the mechanism of formation, the uplifted Transantarctic Mountains have been considered one of the centres for initiation of Antarctic glaciation, having become a well-defined feature of the continent by 34 Ma (Wilson et al. 2012). Glacial deposits at high elevations along the Transantarctic Mountains, and known as the Sirius Group, appear to represent early ice sheets warmer and more dynamic than today, reaching the continental margin while the range was still rising (Barrett 2013). The deposits occur in a variety of topographic settings many of which are unrelated to present topography and drainage, and at elevations ranging from 1100 to more than 3500 m above sea-level. The oldest of these Sirius deposits could date back to the early Oligocene and might exist in the region between the heads of the Beardmore and Shackleton glaciers or at the head of the Scott Glacier. In a few localities they have yielded a wide array of fossils suggesting tundra conditions. Interbedded ash has allowed dating of what are considered to be the youngest of such high-elevation glacial deposits in south Victoria Land at 14 Ma, and dates the transition from warm to cold glaciation (Lewis et al. 2007, 2008).
The development of the present-day Transantarctic Mountains was a multi-stage process, with initial rifting associated with Gondwana fragmentation and the emplacement of the Ferrar province. A Late Jurassic to Early Cretaceous episode is recorded by fission track ages from the continental interior to south Victoria Land, but there is no obvious link either to the plate margin, which was active during that time, or to a rifting event that is otherwise documented. The mid-Cretaceous marks the onset of widespread extension, documented in Marie Byrd Land by the core complex, fission track and other isotopic data from the Transantarctic Mountains, and formation of the West Antarctic Rift System. The final Cenozoic stage is recorded by further extension in the West Antarctic Rift System, fission track data from the Transantarctic Mountains, and the alkaline basaltic provinces in Marie Byrd Land and Victoria Land. The alkaline basaltic phreatomagmatic volcanic edifices at the head of the Scott Glacier, dated at about 20 Ma, suggest that other basaltic centres may be scattered along the length of the range and either were eroded or are concealed by ice, and of course they directly indicate that ice existed on a substantially uplifted Transantarctic Mountains during early Miocene time.
Summary
The Transantarctic Mountains constitute one of Earth’s great intra-continental mountain belts, extending for more than 3000 km across the continent. Their origins go back to the late Neoproterozoic break-up of Rodinia when a rifted margin formed along the site of the Transantarctic Mountains. This rifted margin has played a significant role in subsequent geological events, beginning with the Cambrian to earliest Ordovician Ross orogeny.
A period of major denudation was followed by widespread siliciclastic sedimentation in Early Devonian time, and then by another hiatus through Late Devonian and Carboniferous time, possibly in part a consequence of continental glaciation. Sedimentation resumed during the Permian and Triassic Periods in intra-cratonic basins, one of which in the central Transantarctic Mountains evolved into a foreland basin. From early Palaeozoic through Triassic time the Panthalassan margin had a complex history that included episodic magmatic activity, and possibly rifting, continental ribbon formation and its subsequent accretion, and island arc accretion. Regardless of the margin evolution, there was no clear impact on sedimentation until middle Permian time when the central part of the mountain range (central Transantarctic Mountains) received a major influx of volcanic detritus. The setting changed with the inception of early Jurassic silicic volcanism, probably distal initially but proximal later. Silicic magmatism associated with back-arc extension along the Panthalassan margin was followed by major rifting and emplacement of the Ferrar Large Igneous Province.
Subsequent evolution was dominated first by disruption and reorganization of the Gondwana plate margin, with the rotation and translation of the Ellsworth–Whitmore Mountains block and other continental fragments, and then by episodic Transantarctic Mountains uplift culminating in a widespread 55 Ma event and subsequent denudation of the range. The episodic uplift of the range is linked in some uncertain manner, across a major crustal discontinuity, to extension in the adjacent Ross embayment. That extension, initiated in late Cretaceous time, formed the West Antarctic Rift System and its associated continental shelf basins. The mismatch between the major early Cenozoic uplift/denudation and mid to late Cenozoic basin filling and volcanism has yet to be resolved.
Acknowledgments
Preparation of this manuscript has been supported by the United States National Science Foundation Grant ANT-0944662. The author also wishes to acknowledge significant research support over many years from the Office of Polar Programs, National Science Foundation. Reviews by J. Bradshaw, I. Dalziel and A. Vaughan, and comments by the editor, P. Barrett, are greatly appreciated and have significantly improved the manuscript. Byrd Polar Research Center contribution no C-1424.
- © The Geological Society of London 2013
References
The geological and tectonic evolution of the Transantarctic Mountains: a review
- Article
- Abstract
- Archean to Neoproterozoic time: windows into the cratonic margin
- Late Neoproterozoic to early Palaeozoic (Cambrian) time: passive margin to subduction
- Ordovician to Late Triassic time: sedimentation and subduction
- Early Jurassic time: rifting
- Mid-Jurassic to mid-Cretaceous time: plate margin reorganization
- Cretaceous to Cenozoic time: rifting and uplift/denudation
- Summary
- Acknowledgments
- References
- Figures & Data
- Info & Metrics