The 3,000-kilometer-long Transantarctic Mountains are a dominant feature of the Antarctic continent, yet up to now scientists have been unable to adequately explain how they formed. In a new study, geologists report that the mountains appear to be the remnant edge of a gigantic high plateau that began stretching and thinning some 105 million years ago, leaving the peaks curving along the edge of a great plain.

This study revolutionizes thinking about Antarctica’s evolution. Previous studies have discussed ways in which the mountains may have risen; the current study says they were already high long ago, and that the adjacent land sank. After the mountain chain was isolated, its topography, with summits up to 4.5 kilometers high, was accentuated by erosion caused by glaciers.

Several of the researchers did extensive field work in Antarctica to collect rock samples and geophysical data that back their ideas.

FULL:

The high elevation and considerable length of the Transantarctic Mountains have led to

speculation about their origin. To date, no model has been able to adequately reconcile the juxtaposition

of the high, curvilinear Transantarctic Mountains with the adjacent West Antarctic

Rift System, a broad region of thin extended continental crust exhibiting wide rift characteristics.

We present a fi rst-order investigation into the idea that the West Antarctic Rift System–

Transantarctic Mountains region was a high-elevation plateau with thicker than normal crust

before the onset of continental extension. With major Cretaceous extension, the rift underwent

a topographic reversal, and a plateau edge with thickened crust, representing the ancestral

Transantarctic Mountains, remained. In the Cenozoic, minor extension and major denudation

reduce the crustal root while simultaneously uplifting peak heights in the mountains.

The Cretaceous stage of this concept is investigated using two-dimensional numerical models

to determine under what conditions plateau collapse is plausible. Model results indicate that

elevation of a remnant plateau edge decreases with increasing initial Moho temperature. Very

cold initial Moho temperatures, 675 °C, under the plateau leave a thick plateau edge but do

not exhibit wide rifting. A cold to moderate initial thermal structure, Moho temperatures of

675–850 °C, is needed to retain the plateau edge and still exhibit wide rifting in the middle of

the plateau. We conclude that this plateau collapse concept is possible using these numerical

experiments, and that application of this idea to the West Antarctic Rift System–Transantarctic

Mountains system is also supported by geological and geophysical evidence.

Keywords: Transantarctic Mountains, West Antarctic Rift System, rifting, numerical modeling,

plateau collapse, Antarctica.

Figure 1. Topographic and bathymetry map of

Transantarctic Mountains and Ross Embayment;

reference location map is in upper

right. Ross Sea is characterized by a series

of basins and basement highs and elevated

heat fl ow. Ross Embayment—Ross Sea plus

Ross Ice Shelf. WARS—Ross Embayment

plus extended parts of West Antarctica.

688 GEOLOGY, August 2007

the Basin and Range. The time between the end

of Cretaceous and beginning of Cenozoic rifting

is insuffi cient to cool and thicken the lithosphere

under the mountain–rift system interface to the

level required for rift shoulder uplift.

As a way to reconcile the structures, distribution

of rifting, and the great elevation of the

mountains, we explore the possibility that the

West Antarctic Rift System was a high-elevation

plateau with thicker than normal crust before

extension related to plateau collapse. In traditional

rift shoulder uplift (Fig. 2A), narrow rifting

occurs in cold, thick lithosphere. The rift fl anks

are elevated due to thermal and mechanical

effects, and there is no crustal thickening below

the rift fl ank. In the plateau scenario (Fig. 2B),

extension of hot, thickened crust initiates rifting

and subsidence in the middle of the plateau. The

cooler edge undergoes minimal stretching and is

tectonically and erosionally denuded as a result

of lateral variations in extension.

We propose a plateau collapse scenario, with

the West Antarctic Rift System as the main body

of the plateau and Transantarctic Mountains as

the plateau edge, to explain the denudational and

extensional history of the mountains–rift system.

Wide rifting in the Cretaceous extends the rift

system, but leaves the Transantarctic plateau edge

relatively intact. During the Cenozoic, denudation

reduces the crustal root under the mountains

while simultaneously enhancing large peak uplift,

as described in Stern et al. (2005). Numerical

models of the plateau collapse scenario have been

previously applied to Basin and Range extension

(e.g., Harry et al., 1993). We use numerical

models to test the Cretaceous stage, the major

phase of extension, of this new hypothesis for the

formation of the Transtantarctic Mountains and

West Antarctic Rift System and to determine the

conditions for which retention of a plateau edge

with thick crust is plausible.

TECTONIC SETTING

The Transantarctic Mountains and West Antarctica

have been the site of repeated orogenies,

including Rodinian assembly, Rodinian breakup,

and a transition from a passive to an active margin

during Gondwana assembly (e.g., Goodge,

2002). During the Cambrian–Ordovician Ross

orogeny the region was part of the active margin

of Gondwana (Goodge, 2002, and references

therein). Subduction moved outboard to

the Pacifi c margin of Gondwana from at least

320 Ma to ca. 110 Ma (Mukasa and Dalziel,

2000). Reconstructions (e.g., Foster and Gray,

2000) indicate that the along-strike equivalent

to the region between Marie Byrd Land

and the Transantarctic Mountains, what would

have been the majority of the unextended Ross

Embayment, from the Cambrian to the Middle

Devonian was the Lachlan Fold Belt (orogen) of

southeastern Australia, then adjacent to Antarctica.

This correlation implies high paleotopography

and thickened crust in the West Antarctic

Rift System into the Devonian.

Paleocurrent orientations, lithofacies, and

stratigraphic relationships of the Devonian–

Triassic Beacon Supergroup in the central

Transantarctic Mountains suggest that deposition

began within two intermontane or successor

basins after postorogenic uplift of the Ross

terrain (e.g., Isbell, 1999). These data and the

presence of sub-Beacon relief in the central

Transantarctic Mountains suggest high paleotopography

around and between these basins.

In southern Victoria Land, the Beacon shows a

progression from marine sediment deposition,

to coal measures, to alluvial plain sediments

(Barrett, 1980, 1991). This history implies the

presence of high topography both outboard

(Ross Plateau) and inboard (East Antarctica

hinterland) of the present-day Transantarctic

Mountains and rising paleotopography within

the basin, suggesting crustal thickening under

the Transantarctic basin during this period.

Voluminous Jurassic tholeiitic magmatism

along the Transantarctic Mountains marked

the onset of Gondwana breakup (Elliot, 1992).

While Jurassic rifting is believed to have

occurred within the present-day Trans ant arctic

Mountains–West Antarctic Rift System, no

major rift-bounding faults in the mountains or

the rift system have been located (e.g., Elliot and

Fleming, 2004). The presence of voluminous

Jurassic magmatic products along the mountain

front in the form of sills and lava fl ows suggests

that this event is the most likely candidate for

further thickening of the Transantarctic crust.

Subduction ca. 105 Ma off the Marie Byrd

Land active margin ceased (e.g., Weaver et al.,

1994, and references therein), transferring New

Zealand to the Pacifi c plate, and marking the

onset of major extension in the West Antarctic

Rift System (Lawver and Gahagan, 1995). Wide

rifting (~400 km) ended with the 84–79 Ma separation

of New Zealand and the Campbell Plateau

from Antarctica (Stock and Cande, 2002). Denudation

(1–2 km magnitude) during the Cretaceous

synchronous with this major extension

is recorded along the Transantarctic Mountains

(e.g., Fitzgerald, 2002, and references therein).

The major phase of denudation, 4-9 km along

the Transantarctic front, began in the Eocene,

spatially correlative with but temporally preceding

minor extension localized in the Ross Sea

adjacent to the mountains (e.g., Fitzgerald, 2002,

and references therein). The nature and extent of

Cenozoic rifting as well as its relationship to the

thermochronology data are still uncertain. Narrow

rifting in the Terror Rift, transtension in Victoria

Land (Wilson, 1995; Rossetti et al. 2006),

Adare Trough spreading projected into the Ross

Sea (Davey et al., 2006, and references therein),

and minimal extension with mostly climatecontrolled

denudation (Karner et al., 2005) have

all been proposed as key mechanisms active in the

Cenozoic. Cenozoic alkaline magmatism associated

with Cenozoic rifting (e.g., LeMasurier and

Thomson, 1990) has a much smaller volume and

limited extent compared to Jurassic magmatism,

and is not a viable mechanism for thickening the

Transantarctic crust.

Studinger et al., (2004, 2006) demonstrated

that gravity anomalies across southern Victoria

Land and the Scott and Reedy Glacier region

are consistent with crustal thickening under the

Transantarctic Mountains relative to East Antarctica.

Seismic data on the thickness of the Transantarctic

crust indicate 40 ± 2 km crust under the

mountains and 35 ± 2 km crust under cratonic

East Antarctica (Lawrence et al., 2006).

NUMERICAL MODEL

We conducted regional scale two-dimensional

numerical models to investigate the conditions

conducive for retention of a plateau edge with

thickened continental crust after extension. In the

models, a viscoelastic-plastic, non-Newtonian

layer of thickened continental crust and adjacent

crust of normal thickness underlain by a mantle

layer is extended by pulling at the edges with

a velocity of 1 cm yr–1. Deformation is tracked

using an explicit fi nite-element method similar to

the FLAC (fast Lagrangian analysis of continua)

technique (Lavier et al., 2000, and references

therein). (A complete description of the numerical

model is in the GSA Data Repository.1)

In all models, the initial plateau is 296 km

wide, and total model width is 800 km at

Figure 2. Two classes of models of Transantarctic

Mountain uplift and Ross Embayment

subsidence. Dashed lines approximate

brittle-ductile transition. Top panel shows

pre-rift confi guration; bottom panels show

post-rift confi guration. Class A demonstrates

extension of cold, thick lithosphere. Rift fl ank

uplift raises the edges of the rift due to thermal

effects. Extension is characterized by

narrow rifting, and isotherms are shallowed

locally. Class B, proposed in this paper, demonstrates

extension of a high plateau underlain

by hot thick crust. Extension occurs over

a broad region in the former plateau, and rift

fl anks retain much of their elevation through

retention of thick crustal roots. Isotherms are

shallowed over a wide region.

1GSA Data Repository item 2007177, description

of the numerical model, is available online at www.

geosociety.org/pubs/ft2007.htm, or on request from

editing@geosociety.org or Documents Secretary,

GSA, P.O. Box 9140, Boulder, CO 80301, USA.

GEOLOGY, August 2007 689

time 0 m.y. Initial plateau crustal thickness

is 55 km, and adjacent crust is 32 km thick.

Mantle material underlies the crust to 80 km.

Initial plateau topography is a rectangle elevated

3 km above the background and is isostatically

compensated at depth.

MODEL RESULTS

A variety of extension styles was observed

over a suite of 37 numerical experiments, including

core complex formation, wide rifting, and

narrow rifting. Here we present one example in

which the edge of the plateau retains signifi cant

crustal thickness to remain an elevated highland

after extension. Initial Moho temperature under

the plateau, Tm, is 680 °C (Fig. 3A).

After 5 m.y. and 100 km of extension

(Fig. 3B), the plateau exhibits characteristics

of a wide rift. Extension is accommodated by

crustal thinning over a wide area in the lower

crust that exhibits the highest strain rates. In

the upper crust, three basins accommodate

most of the extension. Peak elevations have

been reduced several hundred meters, and little

extension occurs outside of the plateau area.

At 10 m.y. and 200 km of extension (Fig. 3C),

several small basins and ridges have developed,

but the plateau edges remain several hundred

meters above these. Crust has thinned the most

under the middle of the plateau, and the edges

have undergone the least thinning.

At 20 m.y. and 400 km of extension (Fig. 3D),

the plateau edges have retained elevations of

~1.5 km. The crust has been thinned evenly

across a broad area formerly occupied by the

plateau, and much of the extended region is below

sea level. The lower crust has been replaced by

strong upper mantle in highly thinned areas. Our

model space ends at 20 m.y., analogous to the

end of the major phase of extension and minor

denudation during the Cretaceous.

A summary of the model results (Fig. 4)

demonstrates the thermal conditions required to

leave a plateau edge with signifi cant topography

and crustal thickness in comparison to the surrounding

area. The maximum elevation of the

plateau edge is plotted against Tm at 10 m.y. and

20 m.y. There is a roughly linear decrease of

plateau elevation with increased Tm, related to

the inverse relationship of crustal viscosity and

strength with temperature. Nonlinear variation

in extensional style due to variations in model

thermal structure may be responsible for the

scatter of elevations with Tm. A very cold initial

Tm, below ~675 °C, allows for a thickened remnant

edge, but the extension is characteristic of a

narrow rift. An intermediate temperature profi le,

with Tm ~675–850 °C, exhibits wide rifting or

core complex–style rifting and retains a thick

edge as a highland, with edge thickness decreasing

with increased temperature. Above ~850 °C,

no thickened root is retained.

DISCUSSION

The necessary thermal conditions for a

plateau collapse scenario in the Transantarctic

Mountains–West Antarctic Rift System could

be achieved by low to average concentrations

of radiogenic elements, crustal rocks with high

thermal conductivities in the plateau, or some

combination thereof.

These models are not designed to precisely

emulate the lithospheric architecture across the

mountains–rift system, but simply test a concept.

All models shown (Fig. 3) are symmetrical; i.e.,

there is a plateau remnant on either side. While

the Transantarctic Mountains represent one

edge of a plateau collapse, the opposite edge of

the plateau was represented by the active margin

of Gondwana (New Zealand–Marie Byrd Land)

and was rifted away with plate reorganization

following the breakup of Gondwana.

At the end of the Cretaceous, the major

extensional phase, our conceptual model leaves

an extended West Antarctic Rift System and

relatively unextended Transantarctic Mountains

plateau edge with a crustal root of ~12–16 km

and elevations of 1–2 km. Signifi cant denudation,

~4–9 km, along the mountain front and

decreasing inland, has occurred since the start of

the Eocene. This denudation would reduce the

crustal thickness of the mountains, yet gravity

and seismic studies indicate the presence of an

~5 km root under the Transantarctic Mountains

today compared to cratonic East Antarctica. If

denudation relates directly to erosion, a crustal

root ~9–14 km thick would be required at the

end of Cretaceous extension, as is present in

our model. All mechanisms of Cenozoic extension,

whether Adare Trough spreading, transtention,

narrow rifting in Terror Rift, or climatecontrolled

denudation without major extension,

are feasible within the context of our conceptual

model. Synchronous with Cenozoic denudation

reducing the crustal root of the Transantarctic

Figure 3. Topography, crustal thickness, and viscosity time slices for 0, 5, 10, and 20 m.y.

At 0 m.y. the plateau is represented by a simple box of thickened crust in blue, underlain

by olivine mantle in purple. At 5 m.y., basins and ridges have developed in the topography

and wide rifting is observed. Lower viscosity regions in upper crust represent shear zones.

Lower mantle material has been added to accommodate the space made. At 10 m.y., both

lower and upper crust have thinned over a wide region, but plateau edges remain high. At

20 m.y., the region of weak lower crust has thinned signifi cantly, and extension is focused at

the rift edge, shown here in the viscosity profi le as the lower viscosity upper crust in yellow.

High fl ank edges are underlain by crustal roots.

Figure 4. Maximum topography at plateau

edge vs. initial Moho temperature under

plateau for 10 m.y. and 20 m.y. Both plots

show a rough linear trend of decreasing

topography with increased Moho temperature,

implying that a cold to moderate initial Moho

is needed to retain a high plateau edge.

690 GEOLOGY, August 2007

Mountains, the peak elevations are increased as

much as 50% by glacial incision (Stern et al.,

2005), leading to the peak heights of as much as

4.5 km observed today.

CONCLUSIONS

Our numerical experiments demonstrate that

rifting a plateau can leave a remnant edge at

its fl ank that retains greater crustal thickness.

Coupled with the Cenozoic denudation history

of the Transantarctic Mountains, this root

would be comparable to that seen under the

mountains today. Initially very cold conditions,

initial Moho temperatures Tm 675 °C, do not

produce distributed extension and crustal thinning

as observed in the Ross Embayment. A

moderate temperature profi le, Tm ~675–850 °C,

is needed to retain crustal thickness and elevation

at plateau fl anks and exhibit wide rift characteristics

in the extended plateau. Topography

and crustal thickness are not retained at plateau

fl anks under initially hot, Tm > 850 °C, conditions.

A plateau collapse scenario agrees with

the geological history of the region and erosional

studies attributing signifi cant peak height

increase to the effects of glacial incision.