The Ellsworth-Whitmore Mountain terrane of central Antarctica was part of the early Paleozoic amalgamation of Gondwana, including a 13,000 m section of Cambrian–Permian sediments in the Ellsworth Mountains deposited on Grenville-age crust. The Jurassic breakup of Gondwana involved a regional, bimodal magmatic event during which the Ellsworth-Whitmore terrane was intruded by intraplate granites before translation of the terrane to its present location in central Antarctica. Five widely separated granitic plutons in the Ellsworth-Whitmore terrane were analyzed for their whole-rock geochemistry (X-ray fluorescence), Sr, Nd, and Pb isotopic compositions, and U-Pb zircon ages to investigate the origins of the terrane magmas and their relationships to mafic magmatism of the 183 Ma Karoo-Ferrar large igneous province (LIP). We report high-precision (±0.1 m.y.) isotope dilution–thermal ionization mass spectrometry (ID-TIMS) U-Pb zircon ages from granitic rocks from the Whitmore Mountains (208.0 Ma), Nash Hills (177.4–177.3 Ma), Linck Nunatak (175.3 Ma), Pagano Nunatak (174.8 Ma), and the Pirrit Hills (174.3–173.9 Ma), and U-Pb sensitive high-resolution ion microprobe (SHRIMP) ages from the Whitmore Mountains (200 ± 5 Ma), Linck Nunatak (180 ± 4 Ma), Pagano Nunatak (174 ± 4 Ma), and the Pirrit Hills (168 ± 4 Ma). We then compared these results with existing K-Ar ages and Nd model ages, and used initial Sr, Nd, and Pb isotope ratios, combined with xenocrystic zircon U-Pb inheritance, to infer characteristics of the source(s) of the parent magmas. We conclude that the Jurassic plutons were not derived exclusively from crustal melts, but rather they are hybridized magmas composed of convecting mantle, subcontinental lithospheric mantle, and lower continental crustal contributions. The mantle contributions to the granites share isotopic similarities to the sources of other Jurassic LIP mafic magmas, including radiogenic 87Sr/86Sr (0.706–0.708), unradiogenic 143Nd/144Nd (εNd < –5), and Pb isotopes consistent with a low-µ source (where μ = 238U/204Pb). Isotopes and zircon xenocrysts point toward a crustal end member of predominantly Proterozoic provenance (0.5–1.0 Ga; Grenville crust), extending the trends illustrated by Ferrar mafic intrusive rocks, but contrasting with the inferred Archean crustal and/or lithospheric mantle contributions to some basalts of the Karoo sector of the LIP. The Ellsworth-Whitmore terrane granites are the result of mafic rocks underplating the hydrous crust, causing crustal melting, hybridization, and fractionation to produce granitic magmas that were eventually emplaced as post-Ferrar, within-plate melts at higher crustal levels as the Ellsworth-Whitmore terrane rifted off Gondwana (47°S) before migrating to its current position (82°S) in central Antarctica.
Central Antarctica, the 720,000 km2 region between the Ellsworth and Transantarctic Mountains, is characterized by the exposure of a few widely separated nunataks within the West Antarctic Ice Sheet (Fig. 1). It has frequently been referred to as the “problem child” of Antarctic and Gondwana tectonics (Dalziel and Elliot, 1982; Grunow et al., 1987; Storey et al., 1988a, 1988b, 1994). This largely reflects the fact that although the folded Lower Paleozoic stratigraphic succession of the Ellsworth Mountains (Anderson et al., 1962; Webers et al., 1982; Curtis, 2001) is analogous to that of the Transantarctic Mountains, its present trend is orthogonal to that of the Transantarctic Mountains. This led to the idea that it represents a rotated Gondwana fragment, known as the Ellsworth-Whitmore block or terrane—one of four individual geological terranes that constitute West Antarctica (Dalziel et al., 1987; Storey et al., 1988a). Its boundaries are not well defined, but seismic profiles (Bentley et al., 1960) confirm one as the margin of Archean–Proterozoic cratonic East Antarctica (e.g., Harley and Kelley, 2007; Goodge and Fanning, 2010); the other is thought to lie to the present north of the Ellsworth Mountains, since aeromagnetic surveying shows a similar basement signature as far as the Haag Nunataks (Maslanyj and Storey, 1990; Fig. 1).
Because of its remoteness and lack of continuous exposure, this area has rarely been visited by geologists: U.S. geological parties worked there in the years 1959–1965, and a joint U.S.-UK expedition visited the area in the mid-1980s. It is characterized by the sparse exposure of widely separated granite nunataks dated as Jurassic (K-Ar whole-rock radiometric ages by Craddock, 1972; Rb-Sr whole-rock isochrons by Millar and Pankhurst, 1987). The structural discontinuity between (Permian) Gondwana-orogen rocks in the Ellsworth Mountains and the Pensacola Mountains (on the nearby margin of East Antarctica) helped define the Ellsworth-Whitmore terrane in central Antarctica, thereby posing a tectonic mystery: What is the age of the crust into which these granitic magmas intruded, what plate boundaries bordered the Ellsworth-Whitmore terrane, and how does this crustal fragment fit into Gondwana and older reconstructions?
In the Transantarctic Mountains along the western margin of cratonic East Antarctica, there are widespread Jurassic mafic intrusions, e.g., the Dufek gabbroic massif and equivalent extrusive and hypabyssal rocks of the Ferrar dolerite. Riley and Knight (2001) summarized geochronological data for these in the range 180–183 Ma, whereas Burgess et al. (2015) reported highly precise U-Pb chemical abrasion–isotope dilution–thermal ionization mass spectrometry (CA-ID-TIMS) data indicating a very short period of emplacement at 182.78 ± 0.03 Ma. These, as well as other igneous rocks in South Africa, Tasmania, the Antarctic Peninsula, and Patagonia appear to be broadly contemporaneous or slightly older than the granites in the Ellsworth-Whitmore terrane. The Ferrar-Karoo-Tasman mafic igneous suite represents a divergent rifting phase as Gondwana fragmented separating the Ellsworth-Whitmore terrane continental crust from cratonic Antarctica (Grunow, 1993; Elliot and Fleming, 2000; König and Jokat, 2006). Early Jurassic silicic volcanism in Patagonia and the Antarctic Peninsula has also been interpreted as due to the breakup and dispersal of SW Gondwana (Pankhurst, 1990; Pankhurst et al., 2000). The geochemistry of the Ellsworth-Whitmore terrane granites was studied by Storey et al. (1988b), who interpreted them as resulting from differentiation of mantle-derived Ferrar-type magma with varying amounts of crustal contamination. Granite ages (K-Ar) were first reported on the geologic map of Antarctica (Craddock, 1972), and these results were subsequently supported by Rb-Sr methods (Millar and Pankhurst, 1987; Pankhurst et al., 1991).
In this study, we analyzed zircons from nine samples from the collections of U.S. field geologists in 1959–1965 by separating U and Pb chemically and determining their concentrations and isotopic compositions by ID-TIMS (Table 1). Zircon was also separated from four samples from subsequent 1983–1984 U.S.-UK field work (Dalziel and Pankhurst, 1984) and analyzed directly by sensitive high-resolution ion microprobe (SHRIMP; Table 1). Our goals were: (1) to use the precise U-Pb TIMS zircon ages to refine the less-precise K-Ar and Rb-Sr granite intrusion ages, thereby testing the proposed links between the Jurassic mafic (i.e., Ferrar suite) and felsic magmatism; (2) to see if the zircon histories (zircon SHRIMP ages) would provide a constraint on the ages of the basement rocks from which the granites may have been derived; and (3) to use additional geochemical (X-ray fluorescence [XRF]) and isotopic (Sr, Nd, Pb) data to augment the sparse data set from the Ellsworth-Whitmore terrane granites to further constrain the magmatic sources and test the tectonomagmatic hypothesis of Storey et al. (1988b) regarding Gondwana dispersal.
The geology of Antarctica was first pieced together with the American Geographical Society’s compilation of the Antarctic Folio Series maps (Craddock, 1969) and geologic map of the continent (Craddock, 1972). The Ellsworth-Whitmore terrane sits between the Weddell Sea in the east and the Ross Sea in the west, where the nature of the crust is poorly known and seemingly little deformed (Huebscher et al., 1996), despite the 90° counterclockwise rotation of the Ellsworth-Whitmore terrane block out of Gondwana (see following). The Ross Sea is thought to be the location of active divergence, namely, the Terror rift system (Le Masurier, 1990; Salvini et al., 1997; Paulsen et al., 2014), and it is the location of ongoing drilling projects. To the north of the Ellsworth-Whitmore terrane, there is the Antarctic Peninsula (see Burton-Johnson and Riley, 2015), to the northwest is the Thurston Island terrane (see Pankhurst et al., 1993; Riley et al., 2016), and to the west is Marie Byrd Land (see Pankhurst et al., 1998). The Ellsworth-Whitmore terrane boundary is tectonically stationary and aseismic; the POLENET global positioning system (GPS) array has been in place since 2009 and may better resolve the plate velocities, if any, in central Antarctica (Dalziel, 2008) and help define the boundaries of the West Antarctic terranes.
The Haag Nunataks (Millar and Pankhurst, 1987; Storey et al., 1994) to the northeast of the Ellsworth Mountains are composed of Grenville-age gneisses dated at 990–1260 Ma by Rb-Sr and K-Ar methods and confirmed by unpublished U-Pb SHRIMP zircon data (this paper). These are the oldest rocks in West Antarctica. Nd model ages (TDM) here and elsewhere in the Ellsworth-Whitmore terrane (Storey et al., 1994; Curtis et al., 1999) and the presence of Grenville-age detrital zircons in the Ellsworth Mountains Paleozoic section suggest that much of the Ellsworth-Whitmore terrane is underlain by Grenville crust (Flowerdew et al., 2007; Craddock et al., 2008). The Ellsworth Mountains are the highest peaks in Antarctica (up to 4892 m) and contain an essentially continuous Cambrian–Permian section of 13,000 m (Webers et al., 1982; Flowerdew et al., 2007; Elliot et al., 2014) that was deformed as part of the Permian–Triassic Gondwanide orogeny (Craddock and Webers, 1964; Craddock et al., 1965; Craddock, 1966; Curtis, 1997, 2001). Four kilometers of exhumation in the Early Cretaceous (ca. 100 Ma) have been documented by fission-track studies from the Vinson Massif (4892 m) profile in the Ellsworth Mountains (Fitzgerald and Stump, 1991).
As Gondwana dispersed in the early Mesozoic, counterclockwise rotation of the Ellsworth Mountains (Watts and Bramall, 1981; Grunow et al., 1987; Randall and Mac Niocaill, 2004) was broadly contemporaneous with intrusion of granites in central Antarctica, the mafic suite of the Ferrar-Karoo-Tasman large igneous province (LIP), and much of the silicic volcanism of southern Patagonia (Pankhurst et al., 2000). Thus, the Ellsworth-Whitmore terrane granite suite is part of a larger Early-to-Middle Jurassic LIP that includes the Ferrar suite and represents a deep-seated thermal event driven by mantle-derived magmatism within the continental crust. This thermal disturbance, identified by Storey (1995) as a mantle “superplume,” played a fundamental role as a heating and triggering mechanism for crustal extension and the breakup of Gondwanaland.
Our contribution is based on geologic maps and field descriptions of many of the remote mountains and nunataks within the Ellsworth-Whitmore terrane (Appendices 2–51) and detailed whole-rock geochemistry, isotope studies (Sr, Pb, and Nd), and U-Pb zircon geochronology of the granites to further improve understanding of the geology and tectonic history of the Ellsworth-Whitmore terrane.
Detailed procedures are given in Appendix 1 (see footnote 1) and are only briefly summarized here.
Whole-rock powders were prepared at Macalester College, Minnesota. Geochemical analyses were performed by XRF spectrometry (Phillips PW-2400 spectrometer with a Rh target) at Macalester College following the methods of Vervoort et al. (2007). Major elements were determined after lithium metaborate/tetraborate fusion, and trace element analysis was conducted on pressed powder pellets.
Rb-Sr and Sm-Nd isotope analyses were performed by mass spectrometric isotope dilution following HCl-HNO3 dissolution and cation exchange chromatography, using an Isoprobe-T spectrometer in the Boise State University Isotope Geology Laboratory. Pb isotope compositions were determined on K-feldspar sequentially leached with HF following the method of Housh and Bowring (1991), with the least radiogenic isotopic composition being taken as representative of the magmatic Pb. Reproducibility data for isotope analyses are given in the tables.
Zircons for precise mass spectrometry chronology were separated using standard techniques at Carleton College, Minnesota, mounted in epoxy resin, and imaged by cathodoluminescence in a scanning electron microscope. Only simple igneous-zoned grains were extracted for U-Pb zircon chronology at Boise State University Isotope Geology Laboratory following modified chemical abrasion (Mattinson, 2005). U and Pb isotopic measurements were made on an IsotopX GV Isoprobe-T multicollector TIMS equipped with an ion-counting Daly detector. Internal errors on analyses of single grains are at 2σ, and errors on weighted mean dates are at the 95% confidence interval; internal 2σ errors were expanded by the square root of the mean square of weighted deviates (MSWD) and the Student’s t multiplier of n – 1 degrees of freedom.
Earlier U-Pb dating reported here was carried out using SHRIMP instruments at The Australian National University (Williams, 1998) on zircon concentrates prepared at the Natural Environment Research Council (NERC) Isotope Geosciences Laboratory (British Geological Survey, Keyworth, UK). Cathodoluminescence was used to select igneous-zoned areas for dating, and a few older cores were also analyzed in the case of the Linck Nunatak sample (this paper). The age uncertainties reported in Table 1 are 95% confidence limit estimates including counting statistics, the reproducibility of the standard (SL13: 572 Ma) during the analytical session, and an additional 1% (1σ) to make allowance for fact that this standard was subsequently found not to be ideally homogeneous in composition (see Ireland et al., 2008).
Calculations for all U-Pb data were performed using Isoplot 3.0 (Ludwig, 2003), but U-decay constant uncertainties were not taken into account.
The Ellsworth-Whitmore terrane contains a number of separate isolated rock outcrop areas above ice level (Figs. 1 and 2). The Ellsworth Mountains and Haag Nunataks were detailed in previous sections, and we present results here for the Pirrit Hills (north), Nash Hills, Martin Hills, Whitmore Mountains, Linck Nunatak, Hart Hills, and Pagano Nunatak (south). The outcrops of Mt. Johns, Moreland Nunatak, and Mt. Woollard were described by Storey and MacDonald (1987) and do not contain granites. Geologic maps of some field sites are the unpublished results of the field efforts of Cam Craddock and colleagues (see Acknowledgments) in the 1960s, and these materials are included in Appendices 2–5 (see footnote 1).
Sample sites can be found in Figure 1 and are described in the following section. A summary of all radiometric ages is given in Table 1. ID-TIMS results are in Figure 3 and Table 2; SHRIMP results are in Figure 4 and Table 3. Fifty of the 55 ID-TIMS analyses from five sites yielded concordant 206Pb/238U dates that are Triassic–Jurassic in age. Dates older than those used in the weighted mean calculations are thought to have inherited components, and one younger one is thought to have undergone Pb loss. SHRIMP analyses of zircons from three sites (Pirrit Hills, Pagano Nunatak, and Linck Nunatak) yielded Jurassic crystallization ages and some older, inherited ages.
The Pirrit Hills (81°17′S, 85°21′W, first positioned in 1958 and named for glaciologist John Pirrit) are a range of peaks 14 km in length, ∼110 km southwest of the southernmost point of the Ellsworth Mountains (Fig. 2A). The range is composed of pink granite that is locally pegmatitic with tourmaline, beryl, and muscovite; it is jointed and weakly foliated, and there are metasedimentary rocks exposed nearby that are folded and contain a steep foliation (Appendix 2 [see footnote 1]). Craddock (1972) reported a K-Ar age of 176 ± 6 Ma, Millar and Pankhurst (1987) reported a Rb-Sr age of 173 ± 3 Ma, and Storey et al. (1988b) reported a Nd model age of 1740 Ma (Table 1).
Four of the eight analyzed grains from Pirrit Hills granite sample 60-8-27 yielded ID-TIMS equivalent dates with a weighted mean 206Pb/238U age of 174.06 ± 0.16 Ma (MSWD = 3.0). Two other grains are slightly older (175 Ma), and two others are considerably older (178 Ma, 243 Ma). Five of the seven analyzed grains from granite sample 60-8 yielded equivalent dates with a weighted mean 206Pb/238U age of 174.01 ± 0.14 Ma (MSWD = 2.0). Again, one grain is slightly older (175 Ma), and another is considerably older (181 Ma). Overall, nine grains from the two samples yielded equivalent dates with a weighted mean 206Pb/238U age of 174.04 ± 0.08 Ma (MSWD = 2.3; Table 2; Fig. 3). Pirrit Hills granite (sample R.2243.4) yielded a SHRIMP age of 168 ± 4 Ma (n = 13, MSWD = 1.3) with evidence of inheritance back to ca. 900 Ma (Table 3; Fig. 4). More recently, Lee et al. (2012) presented a U-Pb zircon SHRIMP age of 164.5 ± 2.3 Ma for this granite.
The Nash Hills (81°53′S, 89°23′W, first surveyed in 1958 and named for U.S. naval officer A.R. Nash) are ∼110 km southwest of the Pirrit Hills and extend for ∼20 km. The smaller outcrop of the Martin Hills (82°04′S, 88°01′W, ∼30 km southeast of the Nash Hills) were named after L.R. Martin, scientific leader of Byrd Station in 1962. Due to the proximity of the two ranges and their similar geology, i.e., granites intruding ∼400 m of folded and foliated Nash Hills Formation metasediments (Appendices 3 and 4 [see footnote 1]), they are referred to collectively as the Nash-Martin Hills. Craddock (1972) reported six K-Ar ages on granites and a porphyritic andesite that range in age from 163 to 175 Ma. Millar and Pankhurst (1987) reported a Rb-Sr age of 175 ± 8 Ma, and Storey et al. (1988b) reported a Nd model age of 1270 Ma (Table 1).
Zircon from four samples was analyzed for U-Pb by ID-TIMS (Table 2). Four grains from sample 63-C-69 yielded consistent dates with a weighted mean 206Pb/238U age of 177.49 ± 0.11 Ma (MSWD = 1.9), and five grains from sample 63-C-67 yielded a weighted mean 206Pb/238U age of 177.46 ± 0.05 Ma (MSWD = 1.0). Six of the seven analyzed grains from sample 63-C-68 yielded equivalent dates with a weighted mean 206Pb/238U age of 177.42 ± 0.09 Ma (MSWD = 2.5); one grain was strongly discordant and considerably older (with a 207Pb/206Pb age of 1409 Ma). Four of the seven analyzed grains from sample 63-C-63 yielded equivalent dates with a weighted mean 206Pb/238U age of 177.38 ± 0.12 Ma (MSWD = 1.7). Another grain is slightly older, and two others are considerably older. Nineteen grains from the four samples from the Nash Hills yielded equivalent dates with a combined weighted mean 206Pb/238U age of 177.44 ± 0.04 Ma (MSWD = 2.1).
The Linck Nunatak (82°41′S, 104°12′W, first surveyed in 1959 and named for U.S. Geological Survey Branch Chief M. Kerwin Linck) consist of four small outcrops on the southwest side of the Whitmore Mountains, which are ∼220 km west-southwest of the Nash Hills. These outcrops are composed of gray leucogranite that crosscut the Mount Seelig granite (see later herein); many dikes of the Linck granite contain xenoliths of the older Mount Seelig granite (Webers et al., 1982). Pankhurst et al. (1991) reported a Rb-Sr age of 176 ± 5 Ma, and Storey et al. (1988b) reported a Nd model age of 1541 Ma (Table 1).
Three of the five analyzed zircon grains from Linck sample 65-W-80 yielded equivalent ID-TIMS dates with a weighted mean 206Pb/238U age of 174.82 ± 0.26 Ma (MSWD = 2.6). Two other grains are discordant and considerably older (Table 2; Fig. 3). Zircons recovered from a granite dike yielded a U-Pb SHRIMP age of 180 ± 4 Ma (four youngest grains only, MSWD = 1.1). The older apparent ages are interpreted as reflecting inherited zircon (Table 3; Fig. 4). They show a clear Triassic component (220 Ma, 230 Ma), as well as ca. 300 Ma and older inheritance (ca. 380–600 Ma and ca. 1040 Ma). This is not a comprehensive study of zircon provenance, since the main thrust of the work was to date crystallization, but it gives a brief impression of the crustal origin of magma components, not their relative importance. The 300–600 Ma ages should be treated with caution due to the paucity and, in some cases, discordance of the data.
Pagano Nunatak (83°41′24″S, 87°37′0″W; Fig. 2B) is an isolated, high-relief (1830 m) pillar of granite first observed by Ed Thiel (1959–60) and named for U.S. naval officer Gerald Pagano. It consists of a pink-gray granite crosscut by leucogranite dikes, both of which preserve two joint sets. A geologic map was provided in Webers et al. (1982). Millar and Pankhurst (1987) reported a Rb-Sr age of 175 ± 8 Ma, and Storey et al. (1988b) reported a Nd model age of 1410 Ma.
Four of the five analyzed grains yielded equivalent dates with a weighted mean 206Pb/238U age of 174.62 ± 0.16 Ma (MSWD = 2.0; Table 2; Fig. 3). Another grain is slightly younger. The Pagano Nunatak granite (R.2215.4) SHRIMP age is 174 ± 4 Ma (n = 13, MSWD = 1.3) Two grains had inherited ages (ca. 420 Ma, 580 Ma), and there is evidence of Pb loss to 160 Ma (Table 3; Fig. 4).
The Hart Hills (83°43′S, 89°05′W, first observed from the air in 1959 and named for Pembroke Hart, a geophysicist involved in International Geophysical Year exploration of Antarctica) are a series of low hills ∼13 km west of Pagano Nunatak, ∼210 m above the ice level, and ∼14 km2 in extent. The stratigraphic section includes ∼400 m of cleaved metasediments intruded by a 100-m-thick undated quartz gabbro (Webers et al., 1982). The gabbro has geochemical affinities to the Jurassic Ferrar dolerite (Vennum and Storey, 1987). A geologic map is provided in Appendix 5 (see footnote 1).
Webers et al. (1982) described the exploration history of this small range (82°35′S, 104°30′W) as well as its general geologic, geochemical, and geochronologic relations. The range is named for U.S. Geological Survey topographic engineer George D. Whitmore after being surveyed in 1959. The Mount Seelig granite (3022 m) is a coarse-grained intrusion with a K-Ar age between 174 ± 4 Ma and 190 ± 8 Ma (Craddock, 1972) and an Rb-Sr whole-rock age of 203 ± 8 Ma (Pankhurst et al., 1991). It intrudes a metasedimentary sequence of unknown age (Storey and MacDonald, 1987). The Mount Seelig granite has a poorly developed micaceous lineation and is crosscut by the fine-grained Linck Nunatak granite (see earlier section).
Four of the seven analyzed grains from Mount Seelig granite sample 65-W-44 yielded ID-TIMS equivalent dates with a weighted mean 206Pb/238U age of 207.96 ± 0.06 Ma (MSWD = 1.1). Two other grains are slightly older, and another is considerably older (Table 2; Fig. 3). Sample R.2226.1 yielded a SHRIMP age of 194 ± 2 Ma (n = 15, MSWD = 1.6) with a bimodal age suite (190 Ma and 200 Ma) and Pb loss to ca. 180 Ma (Table 3; Fig. 4). Thirty-nine zircon analyses using SHRIMP gave ages in the range 176–210 Ma (ignoring one very high-U grain). The pattern is clearly bimodal (Fig. 4) and is thought to reflect disturbance due to a major mid-Jurassic thermal event (see later herein); the 22 oldest ages give a weighted mean age of 200 ± 4 Ma (MSWD = 1.2), confirmed by the Sambridge and Compston (1994) unmixing algorithm.
The new major- and trace-element data obtained in this study were combined with the geochemical data set presented in Storey et al. (1988b) and data from Lee et al. (2012) for the Pirrit Hills. Overall, the Jurassic Ellsworth-Whitmore terrane granitic suite ranges from metaluminous granodioritic to peraluminous granitic intrusions (A/CNK = 0.74–1.27 [A/CNK: One important criterion for classification of granitoids involves the molecular ratio of Ab03/CaO + Na20 + K20]), but it is predominantly medium- to coarse-grained leucocratic biotite ± muscovite granite. Most samples of the Ellsworth-Whitmore terrane have enrichments in large ion lithophile elements (LILEs) and high field strength elements (HFSEs), with large negative Eu anomalies (Eu/Eu*). These characteristics indicate strong fractionation trends where feldspar is the dominant stable aluminous phase. Depletion of HFSEs in the most-evolved granitic samples can be attributed to fractionation of minor phases (zircon, sphene, apatite, and allanite), thus limiting source and genetic interpretations based on the HFSEs alone. In the Nb + Y versus Rb tectonic discrimination plot (Fig. 5), the combined Jurassic geochemical data spread well across the border of the syncollisional granite (syn-COLG) and within-plate granite (WPG) fields, with none falling unambiguously in the volcanic arc granite (VAG) field. Data for several of the widely separated field sites overlap, although the Pirrit Hills granites are markedly more enriched in Rb than the rest. The Linck Nunatak granites fall entirely within the syn-COLG field, whereas the single sample of the Martin Hills granite plots in the within-plate granite field (as does that of the Mount Seelig granite, which is ∼30 m.y. older).
Isotopes (Sr, Nd, Pb)
Our new results for the Jurassic granites exhibit a wide range of radiogenic initial 87Sr/86Sr (0.7096–0.7179) and much less variable initial εNd (–4.3 to –5.6). These are comparable to and within the slightly broader ranges of 0.7070–0.7232 and –4.5 to –5.9 for these parameters found by Millar and Pankhurst (1987) and Storey et al. (1988b); they show the same differences between the individual outcrops (Table 1). The initial 87Sr/86Sr value of the Whitmore Mountains granite (0.7094) is comparable to that of the Pirrit Hills granite, and the initial εNd of –2.4 agrees with one of the previously published values. The Sr and Nd isotope compositions are correlated, resulting in a subhorizontal array stretching from isotopic compositions similar to those of the enriched-mantle sources of ocean-island basalts (OIB) toward a highly enriched (time integrated high-Rb/Sr, low Sm/Nd) reservoir characteristic of Proterozoic continental crust (Fig. 6). This trend is essentially the same as that shown by Storey et al. (1988b). The greater variation in initial 87Sr/86Sr relative to143Nd/144Nd appears to be a robust signal in these granitoids, given that it is consistently shown by initial 87Sr/86Sr ratios from Rb-Sr whole-rock isochrons and low-Rb/Sr apatite analyses. The low 87Sr/86Sr end of the array overlaps with the fields of a variety of Jurassic mafic rocks of the Karoo-Ferrar and Parana LIPs, including low-Ti Parana basalts, Tasmanian dolerites, and Ferrar andesitic basalts (Mortimer et al., 1995; Antonini et al., 1999; Peate et al., 1999; Hawkesworth et al., 1986; Petrini et al., 1987).
In Sr-Pb space (Fig. 6), the moderately elevated 206Pb/204Pb (>18.47) value of the Ellsworth-Whitmore terrane granitoids highlights their similarity to Ferrar mafic rocks and low-Ti Parana basalts, but it distinguishes them from most Karoo magmas. The subvertical array defined by the Ellsworth-Whitmore terrane granites trends toward an isotopically enriched source with time-integrated high Rb/Sr and moderate U/Pb, typical of continental crust. In detail, the Ellsworth-Whitmore terrane granitoids define a short linear array in Pb isotope space (Fig. 6); the end members of this array can be reproduced with a variety of three-stage Pb isotope evolution models that vary the third-stage µ (238U/204Pb) between 9.6 and 13.6 at ca. 1 Ga. The radiogenic end of this array trends toward and past the enriched mantle II (EMII) OIB end member, toward compositions again typical of continental crust. The unradiogenic end of this array is remarkably similar to the Pb isotope compositions of both Tasmanian dolerites and Ferrar basaltic andesites (Hergt et al., 1989; Mortimer et al., 1995; Antonini et al., 1999), and somewhat similar to low-Ti Parana basalts, although the Parana field extends to less radiogenic Pb (Peate et al., 1999; Hawkesworth et al., 1986; Petrini et al., 1987).
The Ellsworth-Whitmore terrane Jurassic granite suite has fairly uniform Nd isotope signatures, with εNd at 175 Ma ranging from ∼–4.3 to –5.9 (Table 1), although Pankhurst et al. (1991) reported two more negative values of –7.9 (Pirrit Hills) and –7.1 (Linck Nunatak). The Mount Seelig granite sample analyzed here has an εNd at 208 Ma of –2.4. Negative values of εNd imply an old crustal contribution to the magmas. Model ages projecting the measured Sm/Nd ratios back in time before Jurassic crystallization (Fig. 7) give widespread model ages of ca. 0.6–1.9 Ga for separation from a chondritic source and ca. 1.4–3.2 Ma for a depleted mantle source. However, Nd model ages calculated by assuming a more uniform crustal source composition (De Paolo et al., 1991) are consistently Mesoproterozoic (almost entirely in the range 1.33–1.40 Ga for the Pirrit and Nash-Martin Hills and Pagano Nunatak).
Age of Ellsworth-Whitmore Terrane Granite Magmatism
The U-Pb geochronological data obtained on separated zircon in this study provide robust support for earlier K-Ar and Rb-Sr ages for the emplacement of the Ellsworth-Whitmore terrane granites (Table 1). The new ID-TIMS ages of 174.04 ± 0.08 Ma (Pirrit Hills), 177.44 ± 0.04 Ma (Nash-Martin Hills), 174.62 ± 0.16 Ma (Pagano Nunatak), and 174.82 ± 0.26 Ma (Linck Nunatak) are the most precise yet and may be taken as the best estimates for the ages of crystallization of the individual granite bodies. They overlap within errors with all previous Rb-Sr isochron ages, and mostly with the U-Pb SHRIMP ages. The main exception in this group is the slightly low SHRIMP age for the Pirrit Hills granite (168 ± 4 Ma), where we surmise that zircon may have undergone small amounts of Pb loss during the emplacement of pegmatites. The very high U contents of most of these zircon grains make them prone to radiation damage, which facilitates Pb loss, whereas the chemical leaching process used in the ID-TIMS analysis should effectively remove material affected in this way. A similar explanation may be suggested for the 165 ± 2 Ma SHRIMP age of Lee et al. (2012) for Pirrit Hills zircons, although in addition, our recalculation of the data in Table 2 of that paper suggests an older result of ca. 172 Ma, which would be compatible with both our TIMS age and the published Rb-Sr age. The SHRIMP age for the Linck Nunatak granite may be distorted by the abundance of inherited zircon apparent in the analysis. Our conclusion is that these granites were all emplaced within a rather short period between 174 and 177 Ma. This corresponds to upper Early Jurassic, essentially late Toarcian (International Commission on Stratigraphy, 2016).
The Whitmore Mountains granites present a more complex chronology. The new ID-TIMS age of 207.96 ± 0.06 Ma is within uncertainty of the Rb-Sr errorchron age (Table 1). They are taken as establishing a latest Triassic episode of Ellsworth-Whitmore terrane granite magmatism, although it is unknown whether this was more widely spread, either in space or time. It is clearly older than the 174, 176, and 190 Ma K-Ar ages presented by Craddock (1972) and the U-Pb SHRIMP age of 200 ± 2 Ma. The K-Ar ages imply that Jurassic magmatism ca. 175 Ma may have occurred here 30 m.y. after the Triassic event. Twelve of the individual grain SHRIMP dates for the Mount Seelig granite are younger than 190 Ma, and this could indicate that zircon in this sample was affected by Pb loss associated with this later magmatism.
Magma Genesis and Regional Implications
The Ellsworth-Whitmore terrane basement is presumed to be mostly Grenville-age crust as exposed at the Haag Nunataks (Millar and Pankhurst, 1987), with Grenville-age detrital zircons found in overlying Paleozoic sediments (Flowerdew et al., 2007; Craddock et al., 2008) and evidence for Grenville-age inheritance found by SHRIMP analyses of Mesozoic zircons in the granites reported in this study. The Ellsworth-Whitmore terrane is also characterized by the presence of Jurassic granites, exposed as widely separated nunataks, that have the same U-Pb zircon crystallization ages, and related bulk geochemistry (within-plate granite) and isotopic (Sr, Pb, Nd) signatures. The Ellsworth-Whitmore terrane is surrounded by crustal fragments and terranes that do not contain Grenville-orogen affinities (Fig. 1; Dalziel and Elliot, 1982; Storey et al., 1988a). The closest parts of Marie Byrd Land, Thurston Island, and the Antarctic Peninsula have largely Phanerozoic histories with only indirect evidence for older basement, an evolution broadly related to long-standing accretion on an active margin with subduction-related magmatism being the norm (i.e., the Terra Australis orogen of Cawood, 2005). On the other hand, as noted already, Early Jurassic volcanism in the Antarctic Peninsula and Patagonia is more easily explained as resulting from early rifting that gave rise to the eventual dispersal of crustal fragments that included the Ellsworth-Whitmore terrane and its ultimate migration to its present position in central Antarctica. The absence of a VAG geochemical signature in the granites studied here strongly suggests that they were also related to this rifting stage rather than to an active convergent margin. Moreover, no Jurassic collision has ever been postulated for West Antarctica and would not seem likely during a major crustal rifting stage. The regional granite distribution and geochemistry thus indicate a within-plate scenario for the Ellsworth-Whitmore terrane granites.
The petrogenesis of these within-plate granites almost certainly involved crustal melting, which was specifically proposed by Lee et al. (2012) for the Pirrit Hills granite. However, the need for a significant heat source, together with the voluminous mantle-derived mafic magmatism of the Karoo-Ferrar province, suggests that a more mixed origin may have occurred: Storey et al. (1988b) proposed a complex generation of crustal melts that mixed in substantial proportions with the more mafic Ferrar suite, as discussed in the following.
Temporal Relations with the Karoo-Ferrar LIP
The granites of the Whitmore Mountains (Mount Seelig) appear to be 20 or 30 m.y. older than the Ferrar mafic magmatism and the main Jurassic granite array in the Ellsworth-Whitmore terrane, respectively, and are thus interpreted as a separate magmatic event. Triassic granites in western Antarctica and southern Patagonia have been previously interpreted as arc magmas related to Andean-style subduction along the southern South American and West Antarctic corridor (e.g., Meneilly et al., 1987; Rapela and Pankhurst, 1996; Millar et al., 2001; Appendix 6 [see footnote 1]). The position of the Triassic subducting margin is problematical both in Patagonia and in the Ellsworth-Whitmore terrane, largely due to the lack of any clear control of the surrounding crustal geometry in prerifting times and, in the latter case, extensive ice cover.
The timing for the Karoo-Ferrar Jurassic LIP has been constrained by the high-precision dating of Burgess et al. (2015) to a very short-lived event of a few hundred thousand years essentially at 183 Ma (Karoo volcanism slightly predating that in Ferrar). Jurassic silicic LIP volcanism in Patagonia and the Antarctic Peninsula has been ascribed to three main episodes (Pankhurst et al., 2000): V1 (188–178 Ma), V2 (172–162 Ma), and V3 (157–153 Ma), with the first encompassing the Karoo-Ferrar mafic magmatism and the later two reflecting migration as rifting developed. The TIMS U-Pb ages presented here show that the Ellsworth-Whitmore terrane Jurassic granites were emplaced within the interval 178–174 Ma, starting some 5 m.y. after the Ferrar event. This requires a modification to the mixing model of Storey et al. (1988b), since it requires the survival or regeneration of a Ferrar mafic end member after this 5 m.y. delay. Given that Ferrar magmas have been interpreted as originating in an enriched mantle that had incorporated variable amounts of enriched subcontinental mantle lithosphere (e.g., Molzahn et al., 1996), it is feasible that a similar source would have still been available at the time of Ellsworth-Whitmore terrane granite genesis. Alternatively, Bryan et al. (2002) proposed that a mixing-assimilation-storage-hybridization (MASH) type model would easily explain the slightly younger ages of the Ellsworth-Whitmore terrane granites relative to the mafic Ferrar suite. This process is especially prominent with volatile-rich silicic magmas relative to anhydrous silicic magmas. It is a pattern observed in much younger rocks too, even within single eruptive centers, such as large-volume rhyolites in the Basin and Range Province that contain zircons that may be up to 8–10 m.y. older than the eruptive age. The implication is long-term storage of the silicic magmas in the middle-lower crust before emplacement in the upper crust (magma chamber).
Storey et al. (1988b) showed that closed-system fractionation of a Ferrar-type “parental” magma would produce compositions similar to the Pirrit Hills granite (Fig. 5). They proposed that variable mixing of magmas along the fractionation trend of the Ferrar suite with direct crustal melts could explain the range of magmatic compositions observed in the Nash Hills, Pagano Nunatak, and Linck Nunatak suites. It is worth noting that similar petrogenetic trends driven by input of enriched mantle–derived magmas and deep crustal melting have been proposed to explain Eocene–Oligocene magmatism in Cordilleran core complexes, where the thermal disturbance is intimately related to the high-temperature attenuation fabrics observed in metamorphic core complexes (e.g., Konstantinou et al., 2014).
The crustal end member in Ellsworth-Whitmore terrane granite petrogenesis would be highly radiogenic with respect to 87Sr/86Sr and moderately unradiogenic with respect to εNd, with radiogenic 207Pb/204Pb at moderate 206Pb/204Pb. The subhorizontal array in εNd versus 87Sr/86Sr data (Fig. 6) points toward a Proterozoic crustal end member, based on the moderately unradiogenic εNd (∼–2.4 to –5.6). This inference is consistent with the ages of inherited xenocrystic zircon cores in the granites, as well as three-stage Pb isotope evolution models, whereby the minimum age of crustal reservoir segregation, e.g., increase in µ required to produce the elevated 207Pb/204Pb at moderate 206Pb/204Pb, was found to be ca. 1.0 Ga (Grenville crust). The coherent trends of the Sr-Nd and Pb-Pb isotope arrays suggest that the isotope systematics are dominated by two-component mixing; however, a more diverse suite of enriched sources is suggested by the weaker correlation between Pb and Sr isotopes. The additional enrichment sources could be deeply buried Paleozoic sediments, as in the Ellsworth Mountains, remelted syn-COLG crust, or Grenville crust. This is not surprising given that zircon xenocrysts attest to ca. 2.5–0.5 Ga crustal contributions. Undoubtedly, the Paleozoic sediments of the Ellsworth-Whitmore terrane were themselves in large part derived from underlying Proterozoic basement.
On the whole, the isotopic characteristics and inferred Proterozoic crustal reservoir ages for the Ellsworth-Whitmore terrane are in contrast to the inferred lithospheric contaminants for Karoo lavas, which are clearly distinguished by unradiogenic Sr, Nd, and Pb. The latter are consistent with the nature of the Archean lithosphere through which the Karoo lavas were emplaced, in contrast with the dominantly Proterozoic basement of the Ellsworth Mountains (Millar and Pankhurst, 1987; Flowerdew et al., 2007).
The Pb-Pb isotope systematics are consistent with a range of Pb isotope evolution models that produce end members of the array through Mesoproterozoic fractionation of µ (U/Pb). This could be taken to suggest a common origin of the radiogenic (high-µ) and unradiogenic (low-µ) reservoirs, for example, their identities as complementary enriched crust and depleted subcontinental mantle portions of a lithospheric column formed during ca. 1 Ga tectonomagmatism.
In summary, simple isotopic and trace-element mixing models (Figs. 5 and 6) between magmas along the fractionation trend of the Ferrar suite and crustal melts can produce the magmatic compositions observed in the Nash Hills, Pagano Nunatak, and Linck Nunatak suites, and in this sense, our data support the initial interpretation of Storey et al. (1988b).
The results of this study provide evidence of the continuation of the Triassic West Antarctic magmatic arc based on the ID-TIMS U-Pb zircon age of 208 Ma in the Whitmore Mountains granite and Rb-Sr ages in the Deseado Massif of southern Patagonia (Rapela and Pankhurst, 1996; Fig. 8). The dynamics of the plate separation and widespread Karoo-Ferrar mafic event at 183 Ma indicate a brief event (Burgess et al., 2015) associated with dispersal of the Ellsworth-Whitmore terrane from Gondwana and initiation of the thermal and kinematic conditions that gave rise to the Subcordilleran batholith of western Patagonia (185–181 Ma; Rapela et al., 2005), the more widespread Jurassic andesite-rhyolite volcanism of Patagonia and the Antarctic Peninsula (188–153 Ma; Pankhurst et al., 2000), and the within-plate granites in the Ellsworth-Whitmore terrane at 178–174 Ma. The Jurassic magmas in the Ellsworth-Whitmore terrane indicate that during this event, the subcontinental lithosphere was thermally eroded, and the whole lithospheric column was probably weakened. While the geochemical data from the Jurassic Ellsworth-Whitmore terrane granites can be interpreted to reflect a regional back-arc extensional event, the intrusion of large volumes of basalt associated with the Karoo-Ferrar LIP implies a convective mantle-driven (plume or hotspot) event that was associated with a brief, well-dated thermal anomaly in the mantle. The primitive magmas associated with such a mantle-driven thermal event would have been significantly altered/contaminated in geochemistry by melting of the subcontinental lithosphere and the lower crust. Thus, at a lithospheric scale, this thermal disturbance played a fundamental role in thermally eroding/melting the lithospheric column and was the triggering mechanism for crustal extension and the breakup of Gondwana.
The Ellsworth-Whitmore terrane, with its Grenville-age crust, was part of Rodinia at ca. 1 Ga (Moores, 1991; Dalziel, 1992, 1997; Wareham et al., 1998) and may have been in proximity to Laurentia, since Keweenawan rift magmatic rocks have been identified in Coats Land (Loewy et al., 2011; U-Pb zircon age of 1112 Ma). Cambrian sedimentation is known, or inferred, across the Ellsworth-Whitmore terrane (Storey and MacDonald, 1987; Webers et al., 1982; Flowerdew et al., 2007) as Gondwana began to amalgamate in the early Paleozoic, although Cambrian sedimentation was local and rift-related, in a convergent tectonic setting (Curtis, 1997, 2001; Craddock et al., 2008). Sedimentation continued across southern Gondwana in the Paleozoic until the supercontinent formed in conjunction with the Permian Gondwanide orogen (DuToit, 1937; Halbich, 1992; DeWit and Ransome, 1992). As Gondwana began to fragment at ca. 210 Ma (Lawver et al., 1991), the Ellsworth-Whitmore terrane was located in the Natal Embayment, and all the now-dispersed portions of the Gondwanide thrust belt were continuous and aligned, including the Ellsworth Mountains (Dalziel and Grunow, 1992; Fig. 8A). Intrusion of the Mount Seelig granite (203–208 Ma) to form the Whitmore Mountains occurred before the breakup of Gondwana, so this intrusion was likely related to arc magmatism associated with Triassic subduction ∼1000 km to the west (Rapela et al., 1992; Rapela and Pankhurst, 1996; Fig. 8A). The Karoo-Ferrar LIP province (183 Ma; <1 m.y. duration) was related to the robust breakup of Gondwana as evidenced by a mafic igneous suite that is found along the ∼5000 km boundary between East Antarctica and the terranes of West Antarctica and Africa and that includes the Weddell and Limpopo triple junctions within the associated hotspot (Elliot and Fleming, 2000; Fig. 8B). Initial rifting is also recorded by the Early Jurassic rhyolites in northeastern Patagonia (V1 episode of Pankhurst et al., 2000) and the Subcordilleran batholith farther west (Rapela et al., 2005). Paleomagnetic results (Grunow et al., 1987; Dalziel and Grunow, 1992) suggest that the Ellsworth-Whitmore terrane was at 47°S when the bulk of the within-plate granites (177 Ma) were intruded and far from the remnants of Gondwana as the Weddell Sea opened (Grunow, 1993). This was also the time of the more widespread Chon Aike volcanism of Patagonia, which merges into subduction-related magmatism in the Andean domain (Pankhurst et al., 2000; Riley et al., 2001; Fig. 8C). The Ellsworth-Whitmore terrane is now at 82°S, but there are no paleomagnetic data to constrain the 35° of latitudinal motion of the Ellsworth-Whitmore terrane between the Late Jurassic and present. Outcrops in the Ellsworth-Whitmore terrane are widely separated and include a few nunataks and the highest peaks in the Ellsworth Mountains, the latter of which were uplifted ∼4 km at 120 Ma (Fitzgerald and Stump, 1991). The cause of this uplift is unknown, but the high-standing portions of the Ellsworth-Whitmore terrane are important in stabilizing the ice volumes in central Antarctica (Dalziel, 2008).
We conclude that the Triassic–Jurassic Ellsworth-Whitmore terrane granites are mixtures of crustal and mantle melts, specifically, Ferrar-type basalts that have undergone a significant degree of fractional crystallization with concomitant assimilation of Grenville-age crust and/or Paleozoic sedimentary rocks derived there from, resulting from the deep-seated Limpopo (Bouvet) hotspot thermal disturbance linked to the subsequent breakup of Gondwana (Storey et al., 1988 a, 1988b; Elliot and Fleming, 2000). Arc-related magmatism produced the Mount Seelig granite that intruded the Ellsworth-Whitmore terrane at 208 Ma, followed by regional Karoo-Ferrar (183 Ma) mafic intrusions and extrusions, and finally within-plate granitic intrusions at 174–178 Ma, which were intruded after the Ellsworth-Whitmore terrane rotated to its position in central Antarctica. Combined geochemical (XRF, isotopes) and geochronological studies (U-Pb SHRIMP and CA-ID-TIMS ages on zircon) provide multiple insights into the petrogenesis of simple granites (see also Bickford et al., 2006; Schmitz et al., 2006).
This project is a continuation of the early exploration and sample collection of J. Campbell Craddock (1959–1960, 1962–1963, 1964–1965 field seasons; deceased 2006), Ed Thiel (1959–1960, 1960–1961 field seasons; deceased 1961 in an Antarctic plane crash), and Gerald Webers (1964–1965 field season; deceased 2008) in central Antarctica. Details of the early exploration of central Antarctica can be found in the preface of Geological Society of America Memoir 170 (Webers et al., 1992). The original Mylar (2 × 3 ft) geologic maps in Appendices 2–5 (see footnote 1) have been archived at the Byrd Polar Institute, Ohio State University. John P. Craddock acquired these samples and materials in 2002. Anne Grunow contributed paleomagnetic cores (separated for zircons) from Pagano Nunatak. Mark Fanning is thanked for his direction of the sensitive high-resolution ion microprobe (SHRIMP) analyses. John Splettstoesser (deceased 2016), Staci Loewy, and an anonymous reviewer greatly improved the clarity of the manuscript.
↵§ Present address: U.S. Geological Survey, Reston, Virginia 20192, USA.
↵# Present address: Cyprus Hydrocarbons Company, CY-1066 Nicosia, Cyprus.
↵1GSA Data Repository item 2016236, Appendices 1–5: geologic maps of the Pirrit, Martin, Nash and Nash Hills, central Antartica, is available at http://www.geosociety.org/pubs/ft2016.htm or by request to .
Science Editor: David I. Schofield Associate Editor: Andrew Hynes
- Received 12 January 2016.
- Revision received 9 June 2016.
- Accepted 12 July 2016.
- © 2016 Geological Society of America