The opening of Drake Passage between South America and the Antarctic Peninsula may have removed the last barrier for the wind-driven Antarctic Circumpolar Current (ACC), the largest current in the world’s oceans, encircling the Antarctic continent. For over 35 yr, it has been speculated that its onset caused glaciation through “thermal isolation” of Antarctica (Kennett, 1977). Today, declining atmospheric concentrations of greenhouse gases (De Conto and Pollard, 2003) are seen more widely as the primary driver of glaciation, but the ACC strongly influences the evolution of oceanic circulation and life (Katz et al., 2011).

The oldest oceanic lithosphere in the Drake Passage gateway (Fig. 1) is slightly older than magnetic anomaly C10, ca. 28 Ma, middle Oligocene (Barker and Burrell, 1977). Claims of older through-flow from the Pacific Ocean to the Atlantic Ocean are based on the identification of older magnetic anomalies on the south side of the west Scotia Sea spreading center (Schreider et al. 2012), the presence of Paleogene grabens in Tierra del Fuego (Ghiglione et al., 2008), and Nd-isotope geochemistry (Scher and Martin, 2006). However, the magnetic anomalies are equivocal, and through-flow via grabens or other shallow, narrow structural gaps in the Andean–Antarctic Peninsula Cordillera may have been sufficient to influence the chemistry of the South Atlantic Ocean without a deep ACC (Barker and Thomas, 2004).

Petrology and geochemistry of rocks dredged from the central and southeastern Scotia Sea and the South Sandwich forearc indicate the existence of a now-submerged Neogene arc. This “ancestral South Sandwich arc” may have formed a barrier to a deep ACC, even with a deep, open Drake Passage (Dalziel et al., 2013). The existence of a barrier east of the Drake Passage had been suggested by Barker (2001), pointing out that its influence on the developing ACC would have depended on the physiography of the tectonically evolving North Scotia Ridge (NSR), which the present ACC crosses through three narrow conduits (Fig. 1). A contourite drift north of the westernmost and shallowest of the conduits started to form at 24.5–20.5 Ma (Oligocene-Miocene), based on likely sedimentation rates (Koenitz et al., 2008), and a deep ACC could have formed only after opening of these conduits (Dalziel et al., 2013). In a paper in this issue of Geology, Carter et al. (2014, p. 283), in contrast, argue that the South Georgia microcontinent (SGM) along the NSR would not have been a barrier to a deep ACC until the late Miocene (10 Ma).

The SGM is the largest segment of the North Scotia Ridge. Geologic mapping, and provenance studies of its Lower Cretaceous turbidites strongly indicate that it is a continuation of the hinterland of the Fuegian Andes, the Rocas Verdes basin with its ophiolitic floor, and the Patagonian batholith, and that it was originally situated immediately east of Cape Horn, south of Staten Island and Burdwood Bank, where such a hinterland is now “missing” (Dalziel et al., 1975; Macdonald et al., 1987; Fig. 1). Difficulty in explaining the kinematics of displacement to its present location, however, led to the hypothesis that it originated along the southern margin of the Falkland Plateau, south of Maurice Ewing Bank (Eagles, 2010).

Carter et al. obtained U-Pb age spectra on suites of detrital zircons from the turbidites on South Georgia, and compared them with data from the Cordillera Darwin metamorphic complex of Tierra del Fuego, the eastern end of the Magallanes basin foredeep to the north (Barbeau et al., 2009), and Permian strata of the Falkland Islands. Their results show an unequivocal match with the Fuegian, not the Falkland Islands, rocks, confirming the Andean origin of the SGM, while leaving the question of kinematics unsolved.

Separation of the SGM from South America postdates the mid-Cretaceous compressional inversion of the Rocas Verdes basin and its sedimentary infill. The apatite fission-track ages of 90–105 Ma arc granitoids on the southeastern side of South Georgia (Carter et al., 2014) indicate rapid uplift to the near surface after emplacement, in agreement with detrital zircon evidence for the unroofing of the Patagonian batholith (Tierra del Fuego) by the Late Cretaceous (Barbeau et al., 2009), and the presence of clasts of radiolarian slate and foliated Jurassic volcanics in upper Campanian strata in the foredeep north of the original position of the SGM (Olivero and Malumián, 2008). South Georgia thus must have been topographically high in the Late Cretaceous, as acknowledged by Carter et al. .

Their apatite and zircon fission-track and thermochronometric data appear to give a younger age limit, close to 45 Ma (middle Eocene), for the separation of the SGM from the Andes, coeval with an uplift and cooling event in Tierra del Fuego recorded by a dramatic shift in zircon chronofacies (Barbeau et al., 2009) and by thermochronometric data (Gombosi et al., 2009). Subsidence of South Georgia after Eocene exhumation would not be surprising for a crustal block broken off from a continent, then embedded in oceanic crust. The Precambrian basement of the narrow eastern Falkland Plateau subsided after its Early Cretaceous separation from Africa (Barker et al., 1976). The critical question with respect to the ACC is, whether subsidence was sufficient for the microcontinent to permit overflow of an ocean current with the global climatic potential of the present ACC. Carter et al. argue that their data require 2–4 km of re-burial by a sedimentary cover, and that the shallow Campbell Plateau south of New Zealand deflects the ACC, so that the SGM would have had to sink even lower to accommodate the cover. The authors thus imply that the SGM must have sunk 2–4 km, so as not to impede the ACC.

There are two problems, however. First, the Cretaceous mountains had to sink to 2–4 km below sea level, then be elevated to their present 3 km above sea level. Isostatically, this seems rather unlikely, given the present depth of the ocean floor around the shelf, 3000 m. Second, after separation of the microcontinent from South America, there was no source for clastic sediment. The total thickness of Eocene-Miocene pelagic deposits on the continental Maurice Ewing Bank at the eastern end of the Falkland Plateau is less than 0.5 km (Wise, 1988). There are unconformities in this succession, but it seems unlikely that pelagic deposits of the same age range on the South Georgia microcontinent immediately east of the deep Drake Passage would have been significantly thicker. These problems can probably only be resolved with more thermochronologic data.

Carter et al. emphasize that the North Scotia Ridge holds the key to deep through-flow from the Pacific to the Atlantic Ocean. The conduit east of the SGM clearly opened after its late Miocene collision with the northeast Georgia Rise, and the westernmost conduit is shallow, but very little is known about the 1600 km of ridge between Tierra del Fuego and the SGM. Seismic refraction data indicate that Burdwood Bank is a continuation of the Andean foredeep and its Jurassic volcanic basement (Ludwig et al., 1968). Dredged rocks from Davis Bank and the Barker Plateau (new name) are claimed to be Andean arc rocks (Pandey et al., 2010), and greenschists comparable to forearc rocks around the Scotia arc were collected from the isolated, almost inaccessible Shag Rocks (Tanner, 1982). Notably, the timing of the opening of the critical Shag Rocks Passage between these features, the central ACC conduit, is unknown. Thus, despite the data provided by Carter et al., more work is necessary to understand the tectonic development of the North Scotia Ridge and the tortuous, enigmatic path of the ACC in space and time.