Granitoids associated with the Neoproterozoic–early Paleozoic Ross orogeny are extensively exposed in the Dry Valleys region of southern Victoria Land, Antarctica, affording an exceptional opportunity to gain insight into the temporal and spatial scales of continental arc magmatism. Samples spanning 150 km along strike and 50 km across strike were selected for isotopic and geochemical analysis. Zircon U-Pb geochronology and the first Hf isotope data for Dry Valleys granitoids, coupled with whole-rock elemental data, reveal mixing between enriched lithospheric mantle and Precambrian crustal components and indicate that the principal phase of magmatism in the Dry Valleys area was restricted to a period of 23 m.y., from ca. 515 to 492 Ma. This relatively short period of magmatism contrasts with other segments of the Ross orogen, in which magmatism spanned greater than 100 m.y. Most calc-alkaline intrusions spanned 515–500 Ma, while postkinematic granitoids with alkali-calcic geochemical signatures spanned 505–492 Ma, indicating a transitional shift to an overall extensional tectonic regime. Zircon εHf(i) values range between −0.3 and –7.2, with two-stage depleted-mantle model ages ranging from 1.5 to 1.9 Ga. Low εHf(i) values in mafic samples are consistent with derivation from an enriched subcontinental lithospheric mantle source, while the large-volume granitic intrusions show evidence for increasing assimilation of old crust over time. A broadening of the εHf(i) range to more negative values in the younger intrusions may reflect crustal thickening or underplating of fertile continental material into the source region of the arc.

Subduction zones control the generation and recycling of much of Earth’s crust. Petrogenetic and geochronologic studies of magmatism at convergent margins play an important role in understanding the rates and relative importance of such processes. The structure of active continental arcs can be studied using geophysical techniques, and igneous processes through volcanism, but information about the dynamics of magmatic processes at deeper crustal levels is largely absent. It is also impossible to study the entire sequence of magmatism, from initiation to termination, in active systems. Extensively exposed midcrustal igneous rocks in the Dry Valleys area of southern Victoria Land, Antarctica, yield insights into the timing, duration, and source of an exhumed subduction-related magmatic system.

The Ross orogen in the Transantarctic Mountains forms the boundary between the Proterozoic and Archean East Antarctic craton and the amalgamated terranes of West Antarctica (Fig. 1; Dalziel and Elliot, 1982; Borg and DePaolo, 1991; Goodge et al., 1993b; Stump, 1995). A geochemically diverse suite of igneous rocks in the southern Victoria Land segment of the orogen records a history of magmatism that resulted from westward subduction of paleo–Pacific oceanic lithosphere beneath the East Gondwanan margin during the Neoproterozoic–Ordovician (Borg et al., 1987; Borg and DePaolo, 1991; Dalziel, 1992; Smillie, 1992; Allibone et al., 1993a, 1993b; Stump, 1995; Encarnación and Grunow, 1996; Cox et al., 2000; Stump et al., 2006). Large volumes of plutonic rocks are especially well exposed in the Dry Valleys area of southern Victoria Land, which is the largest ice-free area on the continent.

Magmatism in the Dry Valleys is thought to have spanned ∼45 m.y., from 530 to 485 Ma (summarized in Allibone and Wysoczanski, 2002), but with few broad geochronology studies, ages for plutons are currently limited, and thus our understanding of the timing and duration of magmatism in the Dry Valleys remains incomplete. Here, we present zircon U-Pb ages, hafnium (Hf) isotope compositions and model ages, and whole-rock major- and trace-element geochemistry for 31 samples spanning 150 km along strike and 50 km across strike of the Ross orogen (Fig. 1), including the ∼1000 km2 Bonney Pluton. Specifically, we aim to place further constraints on the timing and duration of magmatism as well as potential magma source(s) and their evolution through time. Our new data indicate that most of this batholith-scale intrusive system was constructed in only ∼23 m.y. from 515 to 492 Ma. An enriched subcontinental lithospheric mantle may have been the primary source of the mafic intrusions, but recycling of Proterozoic continental crust played a significant role in the evolution of the larger granitic plutons.

The Ross orogen is a continental arc that is thought to have been constructed on the attenuated margin of the Archean and Proterozoic East Antarctic craton (Grindley and McDougall, 1969; Borg et al., 1990; Finn et al., 2006; Goodge and Finn, 2010), though cratonic rocks crop out only in the Miller and Geologist Ranges of the central Transantarctic Mountains (Borg et al., 1990; Goodge and Fanning, 1999; Goodge et al., 1991, 1992). Along the orogen, synrift and passive-margin sedimentary rocks deposited during and after the breakup of Rodinia are deformed and metamorphosed and host granitic (sensu lato) batholiths known collectively as the Granite Harbour intrusives (Gunn and Warren, 1962). Metamorphic and igneous rocks associated with the Ross orogen are unconformably overlain by clastic sedimentary rocks of the Devonian− Triassic Beacon Supergroup (McKelvey et al., 1970; summarized in Barrett, 1981), and both are intruded by Ferrar dolerite sills associated with the breakup of Gondwana in the Jurassic (Heimann et al., 1994; Encarnación et al., 1996). Ross-age crystalline basement is now exposed in the rift-shoulder of the Cenozoic extension-related Transantarctic Mountains (Gleadow and Fitzgerald, 1987).

Extensive geochronologic and isotopic studies of granitoids from northern Victoria Land and central Transantarctic Mountains have resulted in various tectonic models for the evolution of the Gondwanan margin in the Neoproterozoic and early Paleozoic (e.g., Borg et al., 1990; Borg and DePaolo, 1991; Rocchi et al., 1998; Goodge et al., 2012). The Sr and Nd isotopic compositions of plutonic rocks and their host metasedimentary rocks in southern Victoria Land have been reported by various workers (Borg and DePaolo, 1994; Hall et al., 1995; Cooper et al., 1997; Cox et al., 2000; Read et al., 2002; Mellish et al., 2002; Cottle and Cooper, 2006a, 2006b). Extensive exposures of metamorphic and granitic basement in southern Victoria Land provide an opportunity for a more detailed geochronologic and isotopic study.

Within the Dry Valleys area of southern Victoria Land, plutonic rocks of the Granite Harbour intrusives are hosted in metamorphic rocks of the Skelton Group. The Skelton Group in the Dry Valleys consists of multiply deformed marble, psammitic schist, calc-silicate, and lesser amounts of pelitic schist and amphibolite intercalated with layers of granitic orthogneiss (Findlay et al., 1984). These rocks were metamorphosed to uppermost amphibolite conditions, and psammitic and pelitic schists are migmatitic (Allibone, 1992; Allibone and Norris, 1992; Cox, 1992). Allibone and Norris (1992) concluded that relatively small volumes of anatectic melt (“microplutons” ≤ 50 m across) could not have contributed significantly to the larger-scale Granite Harbour plutons; however, it is possible that assimilation of anatectic melt(s) occurred at deeper, unexposed crustal levels.

Plutonic rocks constitute ∼70% of exposed pre-Devonian basement in the Dry Valleys (DV; Fig. 1). The granitoids, consisting of ∼40–50 plutons over an area greater than 5000 km2 (Fig. 1; Forsyth et al., 2002; Cox et al., 2012), have been divided into three magmatic suites based on intrusion characteristics, deformation style, and composition (Smillie, 1992; Allibone et al., 1993b). Based on field relationships and whole-rock geochemistry, these workers interpreted the DV1a and DV1b suites as subduction-related units, with the subsequent DV2 suite generated in an extensional regime after the cessation of subduction. Though the classification of all of the plutonic rocks in the Dry Valleys into three suites may be a simplification, the DV1a, DV1b, and DV2 suite divisions are a useful framework for the discussion of the samples from this study.

Limited geochronology suggests that the DV1a and DV1b suites overlap in age and were emplaced between ca. 530 Ma and 490 Ma, whereas the DV2 intrusions largely postdate the DV1 suites, with a youngest U-Pb zircon age of ca. 485 Ma (Encarnación and Grunow, 1996; Allibone and Wysoczanski, 2002). The largest DV1a pluton (Bonney Pluton) was emplaced at a midcrustal level (4.2 kbar or ∼15 km depth), synchronous with contraction across the arc and amphibolite-facies metamorphism of the Skelton Group host metasediments (Allibone et al., 1993a; Cox, 1993). Subsequent emplacement of the DV2 suite occurred at progressively shallower crustal levels during extension and exhumation, with late-stage porphyry dikes recording pressures as low as 1.2 kbar (∼4 km depth; Allibone et al., 1993a). The distinctive compositions and contrasting styles of deformation of various units in the Dry Valleys suggest that the Ross orogen probably developed through several distinct tectono-magmatic phases during the Neoproterozoic through Early Ordovician (Encarnación and Grunow, 1996).

Thirty-one samples ranging from gabbro to granite were selected with an effort to target representative rock types for each suite and to sample the largest-scale intrusions. Here, we offer an abbreviated explanation of the methods employed in this study. A thorough description of the analytical details is provided in the GSA Data Repository.1

Whole-rock major- and trace-element geochemistry was measured by X-ray fluorescence (XRF) and inductively coupled plasma–mass spectrometry (ICP-MS) and used to distinguish samples based on composition and to correlate samples to the specific DV igneous suites of Smillie (1992) and Allibone et al. (1993b).

We employed zircon U-Pb geochronology by laser-ablation–multicollector (MC)–ICP-MS to determine the crystallization ages of individual bodies and constrain the duration of magmatism in the Dry Valleys area during the Ross orogeny. Zircon separates were mounted in epoxy disks, polished to reveal medial sections, and imaged by scanning electron microscope with a cathodoluminescence (CL) detector (Fig. 2). Images were used to guide the placement of spots for laser ablation, with care taken to avoid overlapping multiple growth domains or inherited cores, though in some cases it was impossible to avoid overlapping distinct CL domains (e.g., sample P49944 in Fig. 2). A Photon Machines short-pulse-width (∼4 ns), 193-nm-wavelength excimer laser was used to ablate spots of 24–30 μm diameter, and 238U, 232Th, 208Pb, 207Pb, 206Pb, and 204(Pb + Hg) were measured with a Nu Instruments Nu Plasma high-resolution MC-ICP-MS at the University of California, Santa Barbara. The Lu-Hf-Yb isotope compositions of the same zircons were measured by MC-ICP-MS in later analytical sessions. Laser-ablation spots for Lu-Hf-Yb measurements were 53 μm in diameter and overlapped those used in the U-Pb age determinations (Fig. 2). Age data were coupled to Hf isotopes and whole-rock geochemistry to identify temporal shifts in magma sources and evolutionary trends.

The Hf isotope system has become an important tracer for crustal evolution and magma source studies. Due to the stronger partitioning of Lu relative to Hf in the mantle during differentiation processes and the radioactive decay of 176Lu to 176Hf, over time, the 176Hf/177Hf of the depleted mantle increases relative to the crust (Patchett and Tatsumoto, 1980; Patchett et al., 1981; Patchett, 1983). In a way, this system is analogous to the Sm-Nd system. However, in situ analyses of Hf isotopes in individual growth zones of zircons may record processes such as magma mixing or country rock assimilation that changed the isotopic composition of the reservoir from which the zircons crystallized (e.g., Griffin et al., 2002; Hawkesworth and Kemp, 2006). Additionally, the physical robustness of zircon makes it less susceptible to the open-system behavior sometimes observed in whole-rock Sm-Nd (e.g., Moorbath et al., 1997), and commonly in Rb-Sr (e.g., Field and Råheim, 1979; Ramos et al., 2005).

Zircon readily incorporates Hf into its structure (routinely >1 wt%) and has a low Lu/Hf ratio (Ahrens and Erlank, 1969; Hoskin and Schaltegger, 2003; Hawkesworth and Kemp, 2006); thus (assuming no fractionation of Hf isotopes during crystallization), the Hf composition of a zircon approximates that of the reservoir from which it crystallized. Unlike trace-element patterns, which are affected by fractional crystallization/melting processes, Hf isotopes in zircon may reflect the composition of the source if open-system processes such as contamination and magma mixing are minimal during the evolution of the magma. In a gross sense, large positive εHf(i) values (the 176Hf/177Hf in the sample at the time of crystallization relative to the chondritic uniform reservoir) indicate a depleted-mantle source for the sample, while a negative εHf(i) value suggests a significant recycled crust component (Vervoort and Blichert-Toft, 1999; Belousova et al., 2005, 2010).

Depleted-mantle model ages (TDM) provide an estimated time of separation of a crustal magma source from a depleted-mantle reservoir. Details of the principles behind the calculation of depleted-mantle model ages (TDM) and two-stage depleted-mantle model ages (TDMC) are described thoroughly in DePaolo et al. (1991), and, as applies to Hf isotopes, in Griffin et al. (2000, 2002). It is important to note that model ages involve several untestable assumptions that may significantly affect the calculated “crustal separation age.” Two fundamental assumptions are: (1) that the crustal source of a melt separated from a reservoir that followed the 176Hf/177Hf evolution curve defined by the depleted-mantle model parameters (present-day 176Hf/177Hf = 0.28325; 176Lu/177Lu = 0.0384; Griffin et al., 2000, 2002), and that (2) the crustal source evolved along the 176Hf/177Hf trajectory of “average crust,” with 176Lu/177Hf = 0.015 (Griffin et al., 2002). The latter assumption is specific to TDMC. Model ages are often interpreted to represent a minimum age because any contamination from a more radiogenic (higher 176Hf/177Hf) mantle component or younger crustal component would drive the resulting isotope composition toward a younger model age. Because of these major assumptions, the model ages presented in this study are considered only very general estimates of the age of crust underlying the Ross orogen.

Major- and Trace-Element Geochemistry

The XRF geochemical data for each dated sample (except for two, for which there was no additional material to analyze) are given in the GSA Data Repository (see footnote 1) and are displayed on discriminant diagrams in Figure 3. The whole-rock composition data presented here are a combination of those from Allibone et al. (1993b), Forsyth et al. (2002), and from this study. See Allibone et al. (1993b) for a comprehensive geochemical data set for Dry Valleys samples. Sample locations and brief descriptions of the rock types can be found in Table 1. Additional information on the samples from this study is available in the New Zealand Institute of Geological and Nuclear Science (GNS) rock and mineral database (

Considerable scatter on the Harker diagrams in Figure 3 precludes a simple cogenetic relationship of samples assigned to each suite, while there is also considerable overlap between the different suites. This demonstrates the need for caution when interpreting the samples in the framework of magmatic suites with distinct sources and evolutionary trends. It should also be noted that the original DV suite subdivision of Allibone et al. (1993b) was primarily for samples with >60 wt% SiO2; therefore, the suite assignment of mafic (<60 wt% SiO2) samples from this study should be considered tentative. However, the major conclusions drawn in this study do not change significantly when the data are interpreted independently of the suite framework.


The DV1a samples from this study range in SiO2 content from ∼50 to 70 wt%. On Harker diagrams, DV1a samples display steep negative slopes for Al2O3, MgO (not shown in Fig. 3), and CaO. The Na2O contents are relatively constant, while K2O increases with increasing SiO2 content. The DV1a suite generally has higher Na2O and lower K2O than the DV1b and DV2 suites. There is a wide range in the concentration of the trace elements Y, Sr, Rb, Zr, and Ba, with no strong correlation to SiO2 content (except in Sr). DV1a samples have Sr/Y ratios (not shown in Fig. 3) similar to DV2 samples and lower than most DV1b samples. The evolved DV1a samples typically plot near the boundary of the metaluminous and peraluminous fields, within the continental arc granitoid field (Fig. 3; Maniar and Piccoli, 1989). On tectonic discrimination diagrams (Fig. 3), the samples overlap the boundary of I&S-type and A-type granitoids on the diagram of Whalen et al. (1987) and plot within the “volcanic arc granitoids” (VAG) field on the diagram of Pearce et al. (1984).


The DV1b samples from this study are all highly evolved, with a restricted range in SiO2 from ∼70 to 75 wt% and major-element oxides that largely overlap with the evolved DV1a and DV2 samples. These samples have greater K2O/Na2O ratios than most DV1a samples and are distinguished from evolved DV2 samples by generally higher Ba and Sr and low Zr and Y. The DV1b samples are weakly peraluminous and primarily plot within the I&S-type granitoid and volcanic arc granitoids fields on tectonic discrimination diagrams (Fig. 3).


The DV2 suite samples range in SiO2 content from ∼48 to 75 wt%, with major oxides that largely overlap with DV1a and DV1b samples at high SiO2 content. Two primitive DV2 samples have lower Al2O3 and CaO and slightly higher K2O and Rb than DV1a samples of comparable SiO2 content. Evolved DV2 samples are enriched in Y and Zr and depleted in Sr and Ba relative to most other samples. They typically have high K2O/Na2O and low Sr/Y (not shown in Fig. 3) ratios relative to DV1a and DV1b samples, respectively. Evolved samples plot near the boundary of the metaluminous and peraluminous fields, within the overlapping continental arc granitoid and postorogenic granitoid fields (Fig. 3). They plot near the boundary between the I&S-type and A-type fields and within the volcanic arc granitoids field on tectonic discrimination diagrams.

Tentative DV2

Three samples from Miers Valley, in the southern end of the study area (Fig. 1), are tentatively assigned to the DV2 suite based on composition and field characteristics (Simpson and Aslund, 1996; Cox et al., 2012). These samples are relatively primitive (49–56 wt% SiO2), Ca-poor, and alkali-rich samples. They have high Ba, Sr, and Zr and low Y concentrations. They have some chemical characteristics similar to the DV1b suite, most notably very high Sr/Y ratios. These samples are all relatively primitive (<57 wt% SiO2) and are not plotted on the tectonic discrimination diagrams that were devised for granitoids. However, the alkali-rich nature of these samples is consistent with the other DV2 samples and the interpretation of Smillie (1992) and Allibone et al. (1993b) that the DV2 suite represents post-tectonic magmatism in the Dry Valleys area.

U-Pb Geochronology

The U-Pb ages and Hf isotope populations for all of the samples from this study are summarized in Table 1. The ages are presented as 206Pb/238U weighted mean ages. They are categorized by suite to emphasize the timing and duration of magmatism. For several samples, the mean square of weighted deviates (MSWD) of the analyses used in the age calculation is greater than that expected for a single population given the sample size. This indicates that either the uncertainties assigned to each individual data point are underestimated, or that the data used in the weighted mean may not be representative of a single population; that is, there is geologic significance to the scatter in the data (Wendt and Carl, 1991). Given that repeated measurements of secondary reference materials returned a mean age within uncertainty of the reference isotope dilution–thermal ionization mass spectrometry (ID-TIMS) age and an MSWD of near 1 (Fig. 4), we are confident that the uncertainty assigned to each data point is appropriate, and therefore samples with MSWD values >1 likely represent minor inheritance and/or Pb loss. We take the weighted mean 206Pb/238U ages reported in Table 1 as the best estimates of the crystallization ages of the samples.

The samples range in age from 515.4 ± 6.6 Ma to 492.2 ± 6.1 Ma. Ages from the DV1a suite range between 515.4 ± 6.6 Ma and 500.1 ± 6.6 Ma (10 samples). Only one reliable age of 508.9 ± 6.4 Ma was obtained for a sample from the regionally extensive Bonney Pluton. Four of six DV1b samples yielded robust ages, ranging from 515.3 ± 6.4 Ma to 500.6 ± 6.6 Ma. Sixteen samples from the DV2 suite range between 505.0 ± 6.5 Ma and 492.2 ± 6.1 Ma. Concordia diagrams and 206Pb/238U age distributions of several samples are shown in Figure 5. These examples demonstrate the issue of Proterozoic and “young” (ca. 550 Ma) inheritance present in some samples, the origin of which is discussed further in a following section.

Hf Isotopes

The samples from this study have a restricted range of εHf(i)values between −0.3 and –7.2 (Table 1; Fig. 6), with values overlapping for all of the suites. The DV1a suite ranges from −0.3 to –3.7. The DV1b suite ranges between –1.4 and –7.2. The DV2 samples range from –1.8 to –5.9. The oldest DV1a and DV1b samples have a restricted range of ∼3 εHf(i) units, and younger samples, including the DV2 suite, display a wider range (Fig. 6B). There is a broadening of the range of εHf(i) and lower εHf(i) values with increasing silica content (Fig. 6A) and decreasing age (Fig. 6B) of the samples. Depleted-mantle model ages range between 1.1 Ga and 1.4 Ga, and two-stage depleted-mantle model ages range between 1.5 and 1.9 Ga (Table 1).

Here, we discuss the interpretation of the U-Pb and Hf isotope data with respect to the timing and duration of magmatism and the potential source(s) for the Dry Valleys granitoids.

Timing and Duration of Magmatism in the Dry Valleys

Previously reported U-Pb zircon ages for intrusive rocks from the Dry Valleys area suggest that magmatism spanned ca. 530–485 Ma (Encarnación and Grunow, 1996; Cox et al., 2000; Allibone and Wysoczanski, 2002). On Mount Morning, ∼70 km southeast of the Ferrar Glacier (Fig. 1), Martin et al. (2014) identified felsic granulite xenoliths (hosted in Cenozoic volcanic rocks) with DV suite geochemical affinities. They reported an age of ca. 545 Ma for one granulite xenolith with DV1b geochemical affinity, suggesting that subduction-related magmatism had commenced by this time in southern Victoria Land (Martin et al., 2014). Our samples do not represent the entire range of magmatism in this area during the Ross orogeny, with an oldest and youngest age of ca. 515 Ma and 492 Ma, respectively (Fig. 7). Allibone and Wysoczanski (2002) reported ages of 531 ± 10 Ma and 516 ± 10 for small orthogneiss bodies with DV1b suite affinity. Encarnación and Grunow (1996) reported an age of 484 ± 7 Ma for a DV2 plagioclase-porphyry dike. These bodies are inferred to be volumetrically insignificant, composing <<1% of the area of exposed basement rocks mapped in southern Victoria Land (Cox et al., 2012). All other U-Pb zircon ages for DV suite granitoids (summarized in Allibone and Wysoczanski, 2002) fall within the 515–492 Ma range represented by the samples from this study. One sample from the large Bonney Pluton yielded a robust age of 508.9 ± 6.4 Ma, which is consistent with ages of samples from other parts of the pluton, including ca. 505 Ma ages reported by Encarnación and Grunow (1996) and Cox et al. (2000) and an age of 499 ± 6 Ma from Allibone and Wysoczanski (2002). The four ages from this pluton agree within uncertainty, but the dispersion in the mean ages may reflect its composite emplacement. Our samples are not distributed evenly across the entire Dry Valleys area (Fig. 1), however; we targeted the largest plutons, including the Bonney and Discovery Plutons, and all the units dated in this study collectively represent over 40% of the exposed Granite Harbour intrusives in this area (Cox et al., 2012). We therefore interpret our data to record the largest pulse of subduction-related magmatism in the Dry Valleys area during the Ross orogeny.

Transient Initiation and Synchronous Cessation of Magmatism

In the northern Victoria Land segment, intrusions spanned ca. 545–485 Ma (Black and Sheraton, 1990; Tonarini and Rocchi, 1994; Rocchi et al., 1998; Fioretti et al., 2005; Bomparola et al., 2006; Tiepolo and Tribuzio, 2008). In the central Transantarctic Mountains, calc-alkaline magmatism may have initiated as early as 590 Ma, and persisted for over 100 m.y. (Goodge et al., 2012). Immediately to the south of the Dry Valleys, alkaline intrusive rocks of the Koettlitz Glacier alkaline suite (Read et al., 2002; Cox et al., 2012) were emplaced ca. 557–517 Ma (Rowell, et al., 1993; Encarnación and Grunow, 1996; Cooper et al., 1997; Mellish et al., 2002; Read et al., 2002; Cottle and Cooper, 2006a; Read, 2010). Although magmatism in the Dry Valleys area may have persisted for ∼45 m.y. during the early Paleozoic (based on the ca. 530 Ma age from Allibone and Wysoczanski [2002] and the ca. 485 Ma age of Encarnación and Grunow [1996]), the majority of granites were likely emplaced between ca. 515 and 492 Ma. This period was significantly abbreviated compared to other segments of the orogen, and it coincides almost exactly with the short period of deformation and magmatism recorded in the Delamerian orogen of southeast Australia (ca. 514−487 Ma; Foden et al., 2006). These results, compared with data from northern Victoria Land and central Transantarctic Mountains, highlight the transient nature of continental arc magmatism and deformation along the paleo–Pacific Ocean margin of Gondwana, with no clear trend of younging in one direction along the margin (Fig. 7).

Overlap in the ages of the subduction-related DV1 suites and the post-tectonic DV2 suite records a transitional shift in tectonic regime. Post-tectonic magmatism occurred within a relatively short period (ca. 495-485 Ma) along at least 1500 km of the arc from southeast Australia through southern Victoria Land, with the emplacement of mafic complexes and A-type granites in southeast Australia (Foden et al., 2006, and references therein), potassic Irizar granites and associated Vegetation lamprophyres in central Victoria Land (Rocchi et al., 2009), and late, alkalic DV2 granites and the Vanda dike swarm in southern Victoria Land (this study; Allibone et al., 1993a, 1993b; Encarnación and Grunow, 1996). In the central Transantarctic Mountains, the magmatic arc broadened from ∼80 km wide to over 160 km wide by ca. 510 Ma, and postkinematic intrusions persisted until ca. 485 Ma (Goodge et al., 2012). Despite its transient initiation, magmatism seems to have ceased nearly synchronously along the entire arc (Fig. 7). Various mechanisms have been proposed to explain the synchronous late-stage magmatism and extension across the orogen. Foden et al. (2006) suggested that the subducting slab beneath southeast Australia may have reached the base of the mantle transition zone (∼660 km depth), resulting in hinge rollback, decoupling of the subducting and overriding plates, and exhumation of the upper plate. Hinge rollback, coupled with lithospheric delamination, was proposed to explain the late-stage lamprophyres and extensional collapse in Victoria Land (Di Vincenzo and Rocchi, 1999; Rocchi et al., 2009, 2011). There is also evidence for the suturing of discrete continental fragments in northern Victoria Land, though there is some debate over whether the fragments are exotic terranes (e.g., Borg et al., 1987; Kleinschmidt and Tessensohn, 1987; Borg and DePaolo, 1991) or autochthonous mobile belts (e.g., Rocchi et al., 2011). Allibone et al. (1993a) postulated that there could have been a similar collisional event in southern Victoria Land at approximately the same time, with remnants of the colliding terrane composing part of the Ross seafloor east of the southern Victoria Land coast.

The short period of subduction-related magmatism also contrasts with other well-studied continental arcs, for example, the Sierra Nevada batholith, in which Ca-alkaline magmatism persisted for over 160 m.y. (Saleeby et al., 2008, and references therein), albeit through high-flux episodes separated by relative magmatic quiescence (Barton et al., 1988; Barton, 1996; Ducea, 2001; Ducea and Barton, 2007). However, a picture of periodic magmatism in the Ross orogen emerges if the entire arc is considered (Fig. 7); the abbreviated magmatism in the Dry Valleys records only one major high-flux episode.

Inheritance of ca. 550 Ma components (Fig. 5) records Ross-stage (magmatic?) zircon growth and/or recrystallization in the Dry Valleys area several tens of millions of years before the emplacement of the voluminous DV igneous suites. Several samples contain a range in concordant 206Pb/238U individual spot ages (>50 m.y. for some samples; see GSA Data Repository [see footnote 1]). Some dispersion of ages may be attributable to protracted zircon crystallization from a melt, but 50 m.y. is an unreasonably long period of time to support such a hypothesis. We interpret the spectra to represent inheritance of zircon grains from slightly older (tens of millions of years) sources. This indicates a complex intrusive history, with partial assimilation of the earliest emplaced plutons and/or juvenile lower crust into later magmas.

Potential Magma Sources

Several prominent observations arise from the zircon Hf data in Figure 6: (1) εHf(i) values of all samples from this study are negative (−0.3 to –7.2); (2) more felsic samples have a wider range and lower εHf(i) values than mafic samples; and (3) the range of εHf(i) broadens with decreasing age. Several scenarios potentially explain the negative εHf(i) values: (1) Low-εHf(i) crustal reservoirs, which separated from the mantle at times indicated by the model ages of the samples, were the direct magma sources for the DV suites; (2) the DV suites were sourced from depleted mantle with a relatively radiogenic Hf isotopic composition and subsequently assimilated a significant proportion of low-εHf(i) crust; (3) melts sourced from an enriched (or primitive) subcontinental lithospheric mantle, with a subchondritic Hf isotopic composition, were contaminated by a subordinate amount of crust during magma differentiation. There are essentially three end-member components involved in these hypotheses: depleted mantle, subcontinental lithospheric mantle, and recycled crust (Proterozoic or older in this case), all of which are feasible constituents in the source of arc magmas. The contrasting hypotheses have bearing on the composition of the mantle wedge beneath the arc and on the relative roles of crustal growth versus recycling in the Dry Valleys area during the Ross orogeny. However, the Hf isotope data alone do not uniquely fit any of these potential scenarios.

Presence of Enriched Subcontinental Lithospheric Mantle Beneath the Ross Orogen

The most mafic samples (∼49–50 wt% SiO2) have negative εHf(i) values, from –0.6 to –2.7. A viable interpretation is that the gabbroic (sensu lato) rocks were derived from partial melting in the mantle wedge beneath the arc. Alternatively, they could have been sourced from mafic lower crust with 1.5–1.6 Ga depleted-mantle model ages; however, this would require total fusion of a gabbroic source. It is also possible that the gabbros are cumulates or the residuum of partial melting of an intermediate crustal source, and that the intermediate and felsic samples represent the extracted differentiates/partial melts. The most mafic DV1a sample (50 wt% SiO2) has a high Mg# (74; not shown in Fig. 3) and very high Ni (120 ppm) and Cr (825 ppm) contents (not shown in Fig. 3; see GSA Data Repository [see footnote 1]), suggesting a cumulate origin. However, the two DV2 suite samples with SiO2 <50 wt% have relatively low Mg# (∼44; not shown in Fig. 3) and are relatively enriched in incompatible elements, such as Rb, Zr, and Ba (Fig. 3), and it is unlikely that they are cumulates. All of the mafic to intermediate samples (<60 wt% SiO2) range in εHf(i) from −0.3 to −4.0.

The composition of the mafic samples is consistent with derivation from an isotopically enriched subcontinental lithospheric mantle beneath the arc. A depleted-mantle component would have a radiogenic Hf composition (εHf[i] of +14.9 at 500 Ma, according to the model parameters of Griffin et al., 2000) that is not observed in any of the DV samples. The introduction of old crustal material (with low εHf[i]) into the source region, perhaps through underthrusting of pelagic sediments or subduction erosion of the overriding plate, could have “contaminated” the Hf isotope composition of the resulting melts, potentially driving a depleted mantle–sourced melt toward a negative εHf(i).

To evaluate this effect, Figure 8 shows a binary mixing model between a basalt sourced from partial melting of depleted mantle and Proterozoic crustal material with low εHf(i). A binary mixing is perhaps oversimplified, as the heat budget in an isolated system would require fractional crystallization to accompany melting of assimilated crustal material (i.e., AFC mixing; DePaolo, 1981). However, the purpose of the model is to derive a first-order estimate of the proportion of old crust in the source necessary to drive a depleted mantle–derived melt to the Hf isotope composition observed in the most mafic DV samples. The εHf(i) of depleted mantle in the model was fixed at +14.9 (the value at 500 Ma based on the parameters of Griffin et al., 2000), and the Hf concentration of a melt sourced from the depleted mantle is fixed at 2 ppm, based on the average of 12 samples of island-arc basalts from White and Patchett (1984). The conditions, and therefore derivatives, of partial melting beneath island arcs may be different from those produced in continental arcs, but these island-arc basalts, with highly radiogenic Hf (εHf ranging from +9.1 to +15.1; White and Patchett, 1984), are demonstrably sourced from the depleted mantle and are therefore an appropriate analog for this end member. The Hf concentration and isotope composition of the “old crust” end member are difficult to estimate because Precambrian cratonic rocks are not exposed in southern Victoria Land. The Hf isotope composition of the crustal component was estimated from the Nd isotope composition of Skelton Group metasedimentary rocks reported by Cox et al. (2000). The least radiogenic sample from that study has εNd(i) = −10.2, which corresponds to an εHf(i) = −10.9 (TDMC = 2.1 Ga) when converted using the regression of the Hf-Nd “terrestrial array” of Vervoort and Blichert-Toft (1999; εHf = 1.36εNd + 3). We chose the lowest-εHf(i) value in order to estimate a minimum amount of Proterozoic crust necessary to drive a depleted mantle–sourced melt to the negative εHf(i) values of the mafic DV samples. The Hf concentration of the crustal end member was set at values of 1.9, 3.7, and 5.3 ppm, estimates for mafic lower crust, intermediate “bulk” continental crust, and upper crust, respectively (Rudnick and Gao, 2003).

The model demonstrates that ∼35% by mass of Proterozoic crust in the source would be required to depress the Hf composition of a depleted mantle–derived melt to the negative εHf(i) values observed in the DV samples (Fig. 8). An even greater proportion of crust would be required if mixing involved crust with a lower Hf concentration (∼60% of crust with 1.9 ppm Hf; Fig. 8). Such a large proportion of crustal material in the source would have significantly affected the bulk composition of the resulting melt and, accompanied by fractional crystallization, would have shifted the melts to higher SiO2 concentrations.

A qualitative assessment of Nd-Sr isotope data from along the Ross orogen supports the presence of an enriched subcontinental lithospheric mantle beneath the arc (Fig. 9). Parabolic trends of the Granite Harbour intrusive samples suggest mixing between old crust with low εNd(i) and high 87Sr/86Sr and a juvenile component with higher εNd(i) and low 87Sr/86Sr. However, the juvenile component in each segment of the orogen only extends to moderately high εNd(i) and low 87Sr/86Sr, far from the composition of depleted-mantle sources. This is consistent with the interpretation of Rocchi et al. (2009), that late-stage lamprophyres with low εNd(i) in central Victoria Land were sourced from enriched subcontinental lithospheric mantle. The Vanda dike swarm (including lamprophyres and felsic porphyries) and DV2 suite in the Dry Valleys may be the equivalent of the Vegetation lamprophyres and Irizar granites of Rocchi et al. (2009).

Figure 10 shows Hf isotope evolution trajectories for several hypothetical mechanisms for the enrichment of the subcontinental lithospheric mantle beneath the Ross orogen. The metasomatism may have occurred at some time (or progressively) prior to the Ross orogeny, decreasing the Lu/Hf ratio of the subcontinental lithospheric mantle beneath the edge of the East Antarctic craton. Over time, this reservoir, with a low Lu/Hf ratio, could develop a low-εHf composition that is indistinguishable from old crust in terms of radiogenic isotopes. Alternatively, “young” metasomatism may have occurred during the Ross orogeny, decreasing the εHf of the subcontinental lithospheric mantle through the introduction of crust-derived Hf through fluid metasomatism, subduction of pelagic sediments, or subduction erosion of the overriding continental plate. Zircon oxygen isotope data could provide useful information about the nature of the mantle source but are currently unavailable.

Crustal Recycling and Temporal Trends

If the felsic DV samples were direct differentiates of the mafic samples, the data should fall on horizontal trajectories on the εHf(i) versus SiO2 diagram in Figure 6A (e.g., Fig. 9 of Rocchi et al., 2009), as fractionation of heavy Hf isotopes would be negligible during partial melting and fractional crystallization processes. The wider range and lower εHf(i) in the more evolved samples (Fig. 6A) suggest some involvement of old (low-εHf[i]) crust in the evolution of the DV magmas—crustal assimilation increased the silica-content and “contaminated” the Hf isotope composition of the melts.

Archean and Proterozoic cratonic rocks are exposed only in the Miller and Geologist Ranges in the central Transantarctic Mountains (Goodge and Fanning, 1999). The nature of the crust underlying the metasedimentary rocks and granitoids of the Ross orogen in southern Victoria Land is known only indirectly through detrital zircon and isotopic studies. Skelton Group metasedimentary rocks have detrital zircon age spectra with strong Grenville (ca. 1050 Ma) and subordinate Mesoproterozoic and Archean peaks (Wysoczanski and Allibone, 2004). However, the provenance of the Skelton Group sediments is not well constrained, and the detrital zircons may represent distal sources. Neodymium depleted-mantle model ages for the Granite Harbour intrusives in the Dry Valleys range from 1.1 to 2 Ga (Cox et al., 2000), but these should only be loosely interpreted as the age range of underlying crust. The presence of Proterozoic zircon xenocrysts in some samples (Figs. 2 and 5; Fig. DR1 [see footnote 1]), particularly those of granitic composition, is direct evidence of the assimilation of old crust during the evolution of the DV suites. The assimilated material may have been some combination of cratonic rocks in the lower crust and Skelton Group rocks near the level of pluton emplacement.

The broadening of the array of εHf(i) in younger samples (Fig. 6B) may reflect a thickening of the crustal column through which the magmas ascended or underplating of supracrustal (low-εHf[i]) material into the source region. Refertilization of the source region beneath continental arcs by periodic underthrusting of evolved continental material has been demonstrated in the North American Cordillera (e.g., DeCelles et al., 2009) and in the Australian Tasmanides (Kemp et al., 2009). Those periods were associated with negative excursions in εNd and εHf in concomitant intrusions. The pattern in Figure 6B seems to reflect a portion of one cycle of thickening and source refertilization. The conundrum here is that the earlier gneissic plutons in the Dry Valleys (including the Bonney Pluton), which have a more restricted range and higher εHf(i) (Fig. 6B), were likely emplaced during contraction across the arc (Cox, 1993). The younger, nonfoliated DV2 rocks were interpreted to have been emplaced in a postsubduction extensional regime (Smillie, 1992; Allibone et al., 1993b). However, Allibone et al. (1993b) suggested that the DV2 suite plutons had various sources, and this suite has a wide range of εHf(i). The mafic samples with higher εHf(i) (Fig. 6A) may have been sourced from the lithospheric mantle, while the evolved samples with lower εHf(i) may have been sourced primarily from fertile lower crust.

Whole-rock geochemistry, U-Pb zircon geochronology, and Hf isotopic data for 31 samples from across the Dry Valleys area yield new insights into the magmatic processes that operated in the southern Victoria Land segment of the Ross orogen during the early Paleozoic. Geochronology reveals that the Dry Valleys segment of the arc was predominantly constructed in a relatively short pulse (∼23 m.y.) of magmatism from 515 to 492 Ma. The style of magmatism transitioned from subduction-related to post-tectonic at ca. 505 Ma, which we interpret as a response to the cessation of subduction. Age spectra for individual samples suggest complex intrusion dynamics and inheritance within individual plutons. Mafic samples show geochemical and isotopic evidence for derivation from an enriched subcontinental lithospheric mantle source. Large granitic bodies may have had a similar source but involved crustal assimilation during their evolution. The crustal component in the DV suites increased over time, perhaps reflecting crustal thickening or the underthrusting of fertile continental crust into the source region. The isotopic composition of the Granite Harbour intrusives along the Ross orogen suggests the presence of enriched subcontinental lithospheric mantle beneath the entire arc. Melts derived from this source may be incorrectly interpreted as derivatives of crustal melting based on isotopic composition alone.

Funding for this work was provided by National Science Foundation grant ANT-1043152 to J.M. Cottle. We thank New Zealand Institute of Geological and Nuclear Science (GNS), New Zealand, for providing samples for this study. We also thank Michael Flowerdew, Alan Cooper, two anonymous reviewers, and the editor, John Goodge, for providing many useful suggestions that significantly improved this article.

1GSA Data Repository Item 2015047, detailed description of analytical methods, data tables, cathodoluminescence images, concordia diagrams, and U-Pb age distributions, is available at, or on request from, Documents Secretary, GSA, P.O. Box 9140, Boulder, CO 80301-9140, USA.