1. Introduction
The Antarctic Peninsula of West Antarctica contains some of the most extensive outcrops in Antarctica and is key to understanding the SW Pacific margin of the Gondwana supercontinent. During the break-up of Gondwana, which started in the Jurassic (König & Jökat, Reference König and Jokat2006), the Antarctic Peninsula separated from southernmost South America (Patagonia) forming the Weddell Sea. Although there are geophysical constraints on the evolution of the Weddell Sea crust (Ghidella, Yáñez & LaBrecque, Reference Ghidella, Yáñez and LaBrecque2002; König & Jökat, Reference König and Jokat2006), the relative pre-break-up position of the Antarctic Peninsula with respect to other components of Gondwana is a still matter of debate. Plate tectonic reconstructions have variably placed the Antarctic Peninsula as a straight prolongation of the Patagonian Andes or in a position parallel to Patagonia along the Pacific coast (see Miller, Reference Miller, Cooper and Raymond2007). This debate is partly due to the poorly exposed pre-Jurassic rocks throughout the region, the difficulty of defining accurate sea floor isochrons older than 83.5 Ma (Ghidella, Yáñez & LaBrecque, Reference Ghidella, Yáñez and LaBrecque2002; Ghidella et al. Reference Ghidella, Lawver, Marenssi and Gahagan2007), and the remagnetization of Permo-Triassic and Jurassic rocks in the Antarctic Peninsula (Poblete et al. Reference Poblete, Arriagada, Roperch, Astudillo, Hervé, Kraus and Roux2011). In the Antarctic Peninsula, the spatial distribution and stratigraphic relationships of different units are not well known, and large areas still do not have a stratigraphic allocation.
Upper Palaeozoic to lower Mesozoic sedimentary rocks represent some of the oldest vestiges of Gondwana in the Antarctic Peninsula. In particular, rocks from the Carboniferous-Triassic Trinity Peninsula Group (TPG) occur widely in the northern Antarctic Peninsula (Fig. 1). They have been interpreted as turbidites or debris flow deposits along an active continental margin (Hyden & Tanner, Reference Hyden and Tanner1981; Bradshaw et al. Reference Bradshaw, Vaughan, Millar, Flowerdew, Trouw, Fanning and Whitehouse2012), which was possibly glaciated (Willan, Reference Willan2003). The geological setting of the TPG is not clear. Views are conflicted between an accretionary complex (e.g. Dalziel, Reference Dalziel1984; Storey & Garrett, Reference Storey and Garrett1985) and an upper slope basinal setting (Smellie, Reference Smellie and McKenzie1987, Reference Smellie, Thomson, Crame and Thomson1991; Bradshaw et al. Reference Bradshaw, Vaughan, Millar, Flowerdew, Trouw, Fanning and Whitehouse2012). The source of the TPG was probably regional and located on Western Gondwana (Barbeau et al. Reference Barbeau, Davis, Murray, Valencia, Gehrels, Zahid and Gombosi2010), but the Northern Patagonia Massif and the Choiyoi siliceous volcanic province in South America are the most commonly suggested candidates (e.g. Willan, Reference Willan2003; Fanning et al. Reference Fanning, Hervé, Pankhurst, Rapela, Kleiman, Yaxley and Castillo2011). Clarifying the tectonic setting and the provenance of the TPG will help us to evaluate between different models proposed for the pre-break-up position of the Antarctic Peninsula.
Map of localities mentioned in the text, sample location, tectonic blocks and domains. AP – Antarctic Peninsula, with domains modified after Vaughan & Storey (Reference Vaughan and Storey2000); EWM – Ellsworth-Whitmore Mountain; MBL – Marie Byrd Land; TI – Thurston Island. The insert shows a simplified geological map prepared by the British Antarctic Survey (1985).
We present a provenance study of 24 sedimentary samples from the TPG from seven localities in the northern Antarctic Peninsula (Fig. 1). This study grants insight into the palaeogeological setting of the SW Pacific margin of Gondwana and uses different petrofacies, which are derived from the same source, to suggest the stratigraphic order of this sequence. Furthermore, we combine petrographic and geochemical whole-rock data as well as the cathodoluminescence characteristics of detrital quartz for this purpose.
2. Geological setting
The TPG is a metasedimentary sequence of Carboniferous-Triassic age which crops out in the northern Antarctic Peninsula for about 500 km (Hyden & Tanner, Reference Hyden and Tanner1981). It probably belongs to the Eastern Domain of the Peninsula (Fig. 1), which represents the continental Gondwana margin (Vaughan & Storey, Reference Vaughan and Storey2000). The TPG is divided into the Hope Bay, Legoupil, View Point, Bahía Charlotte and Paradise Harbour formations (Alarcón et al. Reference Alarcón, Ambrus, Olcay and Vieira1976; Hyden & Tanner, Reference Hyden and Tanner1981; Birkenmajer, Reference Birkenmajer1992), which presumably overlie Palaeozoic basement orthogneiss and paragneiss (Millar, Pankhurst & Fanning, Reference Millar, Pankhurst and Fanning2002; Flowerdew, Reference Flowerdew2008), but no contact relationships have been observed and the stratigraphic order between the formations is unclear. A Permian–Triassic depositional age has been assumed for most of the TPG based on detrital zircon U–Pb ages (Hervé, Miller & Pimpirev, Reference Hervé, Miller, Pimpirev, Fütterer, Damaske, Kleinschmidt, Miller and Tessensohn2005; Barbeau et al. Reference Barbeau, Davis, Murray, Valencia, Gehrels, Zahid and Gombosi2010). Fossil evidence suggests that at least part of the TPG at Cape Legoupil is Triassic (Thomson, Reference Thomson1975). The View Point Formation is the only part of the TPG that may be Late Carboniferous – Early Permian in age (Bradshaw et al. Reference Bradshaw, Vaughan, Millar, Flowerdew, Trouw, Fanning and Whitehouse2012).
Metamorphism of the TPG shows an uninterrupted progression of prehnite–pumpellyite-, pumpellyite–actinolite- and greenschist-facies conditions with maximum temperatures of 300°C to 380°C at low to intermediate pressures (Smellie, Reference Smellie, Thomson, Crame and Thomson1991; Smellie & Millar, Reference Smellie and Millar1995; Smellie, Roberts & Hirons, Reference Smellie, Roberts and Hirons1996). The metamorphic grade and intensity of deformation generally increases from Hope Bay (almost unaltered mudstone and sandstone rocks) to the Nordenskjold Coast (fully recrystallized schists) as indicated by Smellie (Reference Smellie, Thomson, Crame and Thomson1991) and Smellie, Roberts & Hirons (Reference Smellie, Roberts and Hirons1996).
The provenance of the TPG has been studied using traditional petrographic tools (Smellie, Reference Smellie and McKenzie1987, Reference Smellie, Thomson, Crame and Thomson1991; Birkenmajer, Reference Birkenmajer1992), whole-rock chemistry (Willan, Reference Willan2003) and detrital zircon ages (Barbeau et al. Reference Barbeau, Davis, Murray, Valencia, Gehrels, Zahid and Gombosi2010; Bradshaw et al. Reference Bradshaw, Vaughan, Millar, Flowerdew, Trouw, Fanning and Whitehouse2012). Based on sandstone petrofacies characteristics, Smellie (Reference Smellie and McKenzie1987, Reference Smellie, Thomson, Crame and Thomson1991) suggested a dissected arc provenance for the Hope Bay Formation, and a mix between recycled-orogen and dissected arc types for the View Point and Legoupil formations. Low CIA values (chemical index of alteration; Nesbitt & Young, Reference Nesbitt and Young1982) for samples from the Hope Bay, Legoupil and View Point formations suggest derivation of the detritus from a glaciated continental margin (Willan, Reference Willan2003). Moreover, detrital zircon U–Pb age patterns have revealed a large proportion of Permian igneous detrital zircon grains (Hervé, Miller & Pimpirev, Reference Hervé, Miller, Pimpirev, Fütterer, Damaske, Kleinschmidt, Miller and Tessensohn2005; Barbeau et al. Reference Barbeau, Davis, Murray, Valencia, Gehrels, Zahid and Gombosi2010; Fanning et al. Reference Fanning, Hervé, Pankhurst, Rapela, Kleiman, Yaxley and Castillo2011).
Based on detrital zircon U–Pb data, some authors (e.g. Barbeau et al. Reference Barbeau, Davis, Murray, Valencia, Gehrels, Zahid and Gombosi2010; Fanning et al. Reference Fanning, Hervé, Pankhurst, Rapela, Kleiman, Yaxley and Castillo2011) assumed that the source of the TPG is the Permian igneous rocks from the North Patagonian Massif and the Permo-Triassic Choiyoi siliceous volcanic province in Patagonia. This interpretation is based mostly on the absence of any suitable Permian proximal source for the TPG. However, Permian magmatism and metamorphism was recently reported in the Antarctic Peninsula, providing a proximal source for the TPG (Riley, Flowerdew & Whitehouse, Reference Riley, Flowerdew and Whitehouse2012). The Permo-Triassic Duque de York Complex (Forsythe & Mpodozis, Reference Forsythe and Mpodozis1983) and the Devonian–Permian Eastern Andes Metamorphic Complex (Hervé, Reference Hervé, Ortega-Gutiérrez, Coney, Centeno-García and Gomez-Caballero1993) in Patagonia have been correlated with the TPG owing to similarities in detrital zircon age patterns (Hervé, Miller & Pimpirev, Reference Hervé, Miller, Pimpirev, Fütterer, Damaske, Kleinschmidt, Miller and Tessensohn2005; Barbeau et al. Reference Barbeau, Davis, Murray, Valencia, Gehrels, Zahid and Gombosi2010; Bradshaw et al. Reference Bradshaw, Vaughan, Millar, Flowerdew, Trouw, Fanning and Whitehouse2012).
3. Sampling and methods
Sandstone and mudstone samples from the TPG were collected from seven localities in the Antarctic Peninsula (Fig. 1). Twenty-four sandstone samples were selected for point-count analysis on the basis of their low deformation and recrystallization, ascertained from petrographic examinations. The ≥ 63 μm grain-size population of quartz, feldspar and lithic fragments were point-counted (400 in each thin-section) following the Gazzi–Dickinson method (Ingersoll et al. Reference Ingersoll, Fullard, Ford, Grimm, Pickle and Sares1984). Whole-rock major and trace element concentrations of 17 sandstone and 16 mudstone samples were determined by inductively coupled plasma mass spectrometry (ICP-MS) and inductively coupled plasma atomic emission spectroscopy (ICP-AES) at ACME Labs, Canada and the Geology Department, Universidad de Chile. For all diagrams the data were normalized to 100% on a loss-on-ignition (LOI) free basis.
Sandstone samples with cathodoluminescence (CL) colours in quartz unaffected by postdepositional metamorphism were chosen for CL investigations of detrital quartz at the Institut für Geologie und Paläontologie, Münster, Germany. We analysed a total of 358 quartz grains in 13 sandstone samples (32–68 grains/sample) in polished thin-sections coated with carbon. We used a hot-cathode luminescence microscope (HC-1LM) coupled with a water-cooled ANDOR OE-CCD detector, following procedures described in Augustsson & Reker (Reference Augustsson and Reker2012). Wavelength spectra at 375–890 nm with a resolution of ≤ 1.04 nm were measured for randomly selected quartz grains of ≥ 130 μm size, using a spot diameter of 40 μm. The system was operated at 10 kV with a sample current of c. 0.45 μA mm−2. The spectra were measured at temperatures between −70° and −64°C. Wavelength calibrations were made with Hg and Ar lamps. All spectra were background corrected and had measuring times of 50 s. Interval measurements of 5 × 10 s were carried out for some grains to rule out time-dependent spectral changes. We only recorded minor changes which did not exceed the range presented by Augustsson & Reker (Reference Augustsson and Reker2012). For data evaluation, we used the relative intensity of the two dominant peaks of quartz grains with respect to the position of the trough between both peaks, following the classification scheme proposed by Augustsson & Reker (Reference Augustsson and Reker2012).
4. Results
4.a. Sandstone petrography
The analysed sandstone samples are fine- to medium-grained and moderately to poorly sorted arkose (according to the classification scheme of Folk, Reference Folk1980), while lithic arkose dominates at Hope Bay. Monocrystalline quartz and feldspar are the dominant components, ranging from 29% of the total modal composition to 59%, and 30% to 51%, respectively (Table 1). Plagioclase is more abundant than K-feldspar and both show varying degrees of alteration, mainly being replaced by sericite. Fresh feldspar grains usually have albite and Carlsbad twinning. Lithic fragments are less abundant, ranging from 13% to 23% of the total modal composition at Hope Bay and Joinville Island, and less than 13% at other localities. Sericitized volcanic clasts are the most abundant and two types of these have been recognized: intermediate volcanic rock fragments with plagioclase and opaque mineral phenocrysts, and more commonly, very fine-grained, less altered felsic volcanic rock. Metamorphic quartzite clasts have also been recognized, especially at Charlotte and Wilhelmina bays, but they represent less than 3% of the total modal composition of those areas.
Sandstone samples framework composition
Q:F:L – framework quartz, feldspar and lithic fragment content recalculated to 100%
Q – total quartz (Qm+Qp); Lt – total lithic fragments (L+Qp); Pf – petrofacies.
Accessory framework minerals make up less than 2% and include muscovite, biotite, chlorite, other heavy minerals (apatite, zircon, garnet, titanite and epidote) and opaque minerals. There is only a small variation between different samples. The matrix content is usually 5–12% and consists of phyllosilicates (mainly smectite), which give it a black appearance. At Cape Legoupil, the matrix consists of chlorite and other phyllosilicates. Most sandstone samples contain veinlets of sericite, quartz and epidote.
We have distinguished three distinct groups with different detrital modes and geographic allocations (Fig. 2). Group A correspond to samples from Hope Bay and Joinville Island with a higher content of lithic volcanic fragments. Group B have low amounts of lithic fragments and include samples from Paradise Harbour, Wilhelmina Bay, Charlotte Bay and Bone Bay. Finally Group C, samples from Cape Legoupil, have a content of lithic fragments between groups A and B, and a higher content of quartz. The modal composition in and between the groups is distributed continuously.
Modal composition of TPG sandstones (discrimination fields after Dickinson et al. Reference Dickinson, Beard, Brakenridge, Erjavec, Ferguson, Inman, Knepp, Lindberg and Ryberg1983). Q – total quartz; F – total feldspar; L – lithic fragments. Right triangle shows data from Smellie (Reference Smellie and McKenzie1987, Reference Smellie, Thomson, Crame and Thomson1991), Birkenmajer, Doktor & Swierczewska (Reference Birkenmajer, Doktor and Swierczewska1997), Willan (Reference Willan2003) and the compositional fields of the petrofacies I, II and III (dashed lines) proposed by Smellie (Reference Smellie, Thomson, Crame and Thomson1991). Hexagonal dashed fields are petrofacies from Smellie (Reference Smellie and McKenzie1987), but most of the samples from Smellie (Reference Smellie, Thomson, Crame and Thomson1991) were included in Smellie (Reference Smellie and McKenzie1987). Red areas are groups proposed in this study.
4.b. Whole-rock geochemistry
The major element composition suggests a felsic igneous origin for the detritus. The major elements show a systematic geochemical contrast between the sandstone and mudstone samples (Table S1 in the online Supplementary Material available at http://journals.cambridge.org/geo). The SiO2 and Na2O contents are 70±5 wt% and 3.6±0.7 wt%, respectively, for the sandstone but only 65±7 wt% and 2.6±0.6 wt%, respectively, for the mudstone samples. Ti2O, Fe2O3(T), MgO and K2O are positively correlated with Al2O3, and thus have higher average contents in the mudstone than in the sandstone samples (Table S1 in the online Supplementary Material available at http://journals.cambridge.org/geo). The remaining major oxides, CaO, MnO and P2O5, show no correlation with Al2O3. CaO contents are 1.2–1.8 wt% in both the sandstone and mudstone samples. MnO and P2O5 contents are lower in the sandstone than in the mudstone, except for the MnO content at Hope Bay, and P2O5 at Paradise Bay. Both the sandstone and mudstone samples have been affected by low to moderate degrees of chemical alteration, as indicated by the chemical index of alteration (CIA = 53–67; Nesbitt & Young, Reference Nesbitt and Young1982), with the sandstone having lower values (CIA = 53–61) than the mudstone samples (CIA = 56–67). Samples in the A–CN–K plot (Fig. 3) present a linear trend, not parallel to the ideal weathering line (WL), which intersects the primary source composition of granodiorite to tonalite and trends towards muscovite.
A–CN–K diagram (Nesbitt & Young, Reference Nesbitt and Young1984), with indication of the weathering index CIA (chemical index of alteration; Nesbitt & Young, Reference Nesbitt and Young1982). A – Al2O3; CN – CaO*+Na2O; K – K2O. Ideal mineral compositions plotted are Ka – kaolinite; Gb – gibbsite; Chl – chlorite; Mu – muscovite; Pl – plagioclase; Ksp – K-feldspar; and illite. Typical igneous rock averages from Le Maitre (Reference Le Maitre1976): Ga – gabbro; To – tonalite; Grd – granodiorite; G – granite. UCC – upper continental crust; PAAS – Post-Archaean Australian Shale. The solid horizontal line is the plagioclase–K-feldspar join. The dashed line parallel to the A–CN join is the predicted ideal feldspar weathering trend (WL) from granodiorite. The dashed arrowed line is the trend for the TPG.
The rare earth element (REE) composition is very similar to the average upper continental crust (McLennan, Reference McLennan2001) and the Post-Archaean Australian Shale (Taylor & McLennan, Reference Taylor and McLennan1985). The mudstone samples have higher total REE abundances than the associated sandstone samples (Fig. 4). Compared to chondritic values, light REEs are enriched relative to heavy REEs as reflected by a LaN/YbN of 12.5 and 9.8 and GdN/YbN of 1.7 and 1.6 for the sandstone and mudstone samples, respectively. The Eu anomaly is negative with an average Eu/Eu* of 0.78 and 0.67 in the sandstone and mudstone samples, respectively (Fig. 4).
Chondrite-normalized REE compositions for the TPG sandstone and mudstone samples. UCC – average upper continental crust (McLennan, Reference McLennan2001); PAAS – Post-Archaean Australian Shale (Taylor & McLennan, Reference Taylor and McLennan1985).
In general, trace element contents of the mudstone samples are higher than for the sandstone samples (Table S1 in the online Supplementary Material available at http://journals.cambridge.org/geo). A positive correlation exists between trace elements and Al2O3 contents, reflecting the association of most trace elements with the clay fraction. This is particularly marked for Sc with 9.4±2.7 and 14.5±5.5 ppm, V with 86±36 and 120±50 ppm, Co with 10±5 and 18±7 ppm, Y with 23±8 and 32±9 ppm, Nb with 10±2 and 17±4 ppm, and Zn with 55±26 and 95±45 ppm for the sandstone and mudstone samples, respectively. Conversely, the mudstone samples are relatively depleted in Zr and Sr with 307±211 and 165±61 ppm, respectively, compared to the sandstone samples with 378±134 and 244±100 ppm, especially in the Paradise and Hope Bay rocks.
Values of provenance-indicative elements in the sandstone samples, such as Th/Sc and La/Sc, range from 0.9 to 1.6 and 3.0 to 5.8, respectively, indicating an upper continental composition of the source (Fig. 5). The zircon and recycling indicator Zr/Sc is 16.3 to 39.9 in the sandstone and 5.2 to 23.4 in the mudstone samples. Differentiation between the sandstone and mudstone is marked in terms of the Zr/Sc ratio, with the sandstones having clearly higher ratios. Ti/Zr ranges from 8.9 to 20.9 in the sandstone and La/Th from 2.4 to 4.1 in the sandstone and mudstone samples. Sandstone samples from Group A have lower La/Sc and higher Ti/Zr than groups B and C, the latter plotting close to the boundary of the passive margin field (Fig. 5c). Sandstone samples from Group C have lower contents of Fe2O3(T)+MgO wt%, TiO2 wt% and Al2O3/SiO2 than groups A and B (Table S1 in the online Supplementary Material available at http://journals.cambridge.org/geo), reflecting more maturity in those sediments, but Group B has a higher concentration of Zr.
(a) La/Th v. Hf (discrimination fields from Floyd & Leveridge, Reference Floyd and Leveridge1987). LCC – lower continental crust; UCC – upper continental crust. (b) Th/Sc v. Zr/Sc (McLennan et al. Reference McLennan, Hemming, McDaniel, Hanson, Johnnson and Basu1993). UCC – upper continental crust; HMC – trend expected for heavy mineral sorting and concentration. Rock averages: An – andesite; Da – dacite; Rh – rhyolite. (c) Ti/Zr v. La/Sc sandstone diagram (Bhatia & Crook, Reference Bhatia and Crook1986). The discrimination diagram is for sandstone.
4.c. Cathodoluminescence spectra of detrital quartz
According to the Augustsson & Reker (Reference Augustsson and Reker2012) discriminant colour diagram (Fig. 6), most quartz grains show typical blue and brown spectra. Bright and medium blue spectra have a higher intensity peak in the blue wavelength interval than in the red. On the other hand, dark blue and brown spectra have a slightly higher intensity in the red wavelength interval (Augustsson & Bahlburg, Reference Augustsson and Bahlburg2003; Augustsson & Reker, Reference Augustsson and Reker2012). This is typical for a mixture of metamorphic and plutonic quartz. Felsic plutonic quartz grains make up 62% of the total quartz grains measured, low-grade metamorphic or mafic plutonic quartz makes up 37% and only 1% have spectra typical for volcanic quartz (Fig. 7). Group A is dominated by felsic plutonic quartz grains with 86% of the total quartz in this group, showing a trend ranging over the whole felsic plutonic region even extending a little into the mafic plutonic (or low-T metamorphic) region (Fig. 6a). Low-grade metamorphic or mafic plutonic quartz grains are more common in Group B with 71% (Fig. 6b). Group C also displays the broad trend for plutonic quartz grains (Fig. 6c), medium blue being the most common CL colour, but with a larger proportion of the dark blue CL colour than in Group A.
Relative peak intensities for measured quartz CL spectra (discrimination fields from Augustsson & Reker, Reference Augustsson and Reker2012). (a) Petrofacies A. (b) Petrofacies B. (c) Petrofacies C.
Petrofacies interpretation. Relative peak intensities for measured quartz CL spectra are plotted in mean values and first standard deviation fields for the three petrofacies. Arrows show the possible evolution from oldest to youngest petrofacies. The extension of the boundary for petrofacies B is speculative.
Non-quartz-related additional peaks at c. 560 and 705 nm have been observed, especially in Group B. They reflect interference with neighbouring plagioclase or K-feldspar (see fig. 5 in Augustsson & Reker, Reference Augustsson and Reker2012). This peak could affect the displayed spectra, especially for those grains that do not show strong luminescence (e.g. dark blue and brown). The proportion of quartz grains with dark CL colours could be underestimated in Group B. No grains with authigenic overgrowths have been observed.
5. Discussion
5.a. TPG petrofacies
Our petrographic results are similar to those of previous results reported by Smellie (Reference Smellie and McKenzie1987, Reference Smellie, Thomson, Crame and Thomson1991), Birkenmajer (Reference Birkenmajer1992) and Willan (Reference Willan2003). Using this large amount of data from earlier studies we can define four petrofacies of which three were grouped in Section 4 above (Fig. 2). Petrofacies A, B and C are defined from samples from groups A, B and C, respectively (Figs 1, 2). Petrofacies X, with high lithic and quartz contents, is not identified in our data, but it was recognized in the View Point area (Smellie, Reference Smellie and McKenzie1987, Reference Smellie, Thomson, Crame and Thomson1991; Willan, Reference Willan2003).
The sandstone framework composition from petrofacies A is consistent with a dissected arc source (Fig. 2). This implies derivation from an eroded arc in which the plutonic roots had been exposed by erosion (Dickinson et al. Reference Dickinson, Beard, Brakenridge, Erjavec, Ferguson, Inman, Knepp, Lindberg and Ryberg1983). The blue CL colour trend shown in Figure 6a suggests a plutonic origin for this petrofacies, but a high-T metamorphic origin should not be ruled out (Augustsson & Reker, Reference Augustsson and Reker2012). Nevertheless, high contents of fresh feldspar and volcanic lithic fragments support the case for an igneous source. Volcanic quartz grains are practically absent in petrofacies A, even though the occurrence of volcanic fragments is unquestionable. For lithic felsic volcanic grains, the crystal size may be an important issue. The quartz phenocrysts observed in the volcanic fragments are too small to be measured by CL (< 130 μm) and for the more mafic volcanic fragments, quartz phenocrysts have not been observed.
Sediments from petrofacies B were derived from a continental block (Fig. 2), where erosion had exposed deeper levels of the continental crust (Dickinson et al. Reference Dickinson, Beard, Brakenridge, Erjavec, Ferguson, Inman, Knepp, Lindberg and Ryberg1983). This petrofacies is dominated by low-grade metamorphic or dioritic quartz grains (Fig. 6b). However, it is clear that diorite can supply only a small amount of quartz, if any. The occurrence of more low-temperature metamorphic quartz, but the lack of an important metamorphic lithic content in this petrofacies, may reflect a grain-size issue. If the metamorphic rocks contain crystals in the range of sand sizes, as the quartz measured in this study does, they do not necessarily contain crystals of smaller sizes from which lithic fragments could be formed.
The combination of these data shows a contribution of magmatic and metamorphic sources. Hence, it is indeed possible that different parts of the same complex were eroded: a shallower plutonic-volcanic (petrofacies A) and a deeper plutonic-metamorphic (petrofacies B) source. In petrofacies C, the amount of quartz increases with respect to petrofacies A owing to its maturity (Fig. 2), showing recycling or a greater selection of grains and a longer transport pattern. Sediments from petrofacies C could have been deposited contemporaneously with deposition of petrofacies A, but in a distal basin, or it could represent the recycling of older sedimentary rocks.
Based on the quartz and feldspar ratio of previous TPG samples, Smellie (Reference Smellie and McKenzie1987, Reference Smellie, Thomson, Crame and Thomson1991) distinguished three sandstone petrofacies (Fig. 2) and suggested a conceptual stratigraphic order: the Cape Legoupil and View Point formations followed by the Bahía Charlotte Formation, followed by the Hope Bay Formation. This model suggests that a volcanic-plutonic arc was being established on a pre-existing passive continental margin. In this case, the Cape Legoupil and View Point formations would represent the erosion of this passive margin, and the Hope Bay Formation, with more volcanic components, would represent the establishment of the arc. However, detrital U–Pb zircon patterns indicate that the Cape Legoupil Formation may be younger than the View Point Formation (Bradshaw et al. Reference Bradshaw, Vaughan, Millar, Flowerdew, Trouw, Fanning and Whitehouse2012), and has the same Permian principal component as the Hope Bay Formation (Barbeau et al. Reference Barbeau, Davis, Murray, Valencia, Gehrels, Zahid and Gombosi2010). This indicates that the Permian igneous principal source was established previous to deposition of the Cape Legoupil Formation, or petrofacies C. Moreover, the lack of authigenic overgrowths around detrital quartz grains in petrofacies C suggests a first-cycle origin (e.g. Sanderson, Reference Sanderson1984) as well as the fresh and angular feldspar grains, supporting a hypothesis of sorting and more distal deposition over reworking of an already existing older passive margin.
MacKinnon (Reference MacKinnon1983) showed that during gradual erosion of a volcano-plutonic arc from an active continental margin exhumation, the amount of lithic grains decreases progressively, similar to the trend between petrofacies A to B of the TPG (Fig. 2). MacKinnon's (Reference MacKinnon1983) study is based on modal petrographic data from the Rakaia Group, New Zealand (which has similar characteristics to the TPG) using a well-determined fossil stratigraphy. In this particular case, after the decrease in lithic grains, there is an increase in sedimentary lithic fragments. He explains this feature as being due to the cannibalism of the same rising basin. However, in the TPG, petrofacies C does not show evidence of an increase in sedimentary lithic fragments, as would be expected for the cannibalism proposed by MacKinnon (Reference MacKinnon1983).
Petrofacies X, with a recycled orogeny provenance (Fig. 2), was not sampled in this study. It does not follow the modal content trend of TPG petrofacies A, B and C, and has different U–Pb detrital zircon patterns. The matrix of conglomerates and clasts from View Point lack the typical Permian component of the TPG, implying that the deposition occurred before or shortly after the Carboniferous–Permian boundary (Bradshaw et al. Reference Bradshaw, Vaughan, Millar, Flowerdew, Trouw, Fanning and Whitehouse2012). Thus, petrofacies X was deposited before the establishment and erosion of the Permian magmatic source, probably in an older passive margin or in the early stage of development of an active setting. Unlike a passive margin, an active tectonic regime implies rapid exposure of potential source rocks; therefore, sediments are less affected by recycling and have a large proportion of zircons with ages close to the depositional age of sedimentation (Cawood, Hawkesworth & Dhuime, Reference Cawood, Hawkesworth and Dhuime2012). This is the case for petrofacies A, B and C. Furthermore, sandstones from View Point have detrital U–Pb zircon patterns with different age components, principally early Palaeozoic and Neoproterozoic (Bradshaw et al. Reference Bradshaw, Vaughan, Millar, Flowerdew, Trouw, Fanning and Whitehouse2012).
Here we suggest an evolution in the source area (a volcano-plutonic arc), reflected in the sedimentation of different petrofacies. Sediments from petrofacies A were deposited first, derived from the dissected arc. Petrofacies C represents a more distal deposition, probably contemporaneous with the older part of petrofacies A. Petrofacies B is the youngest deposit and is derived from the plutonic roots and deeper levels of the continental crust. Petrofacies X was likely derived from the erosion of an older passive margin and deposited before the other petrofacies. This evolution seems to be oriented from SE to NW in the Antarctic Peninsula (Fig. 7). Although there is no age control within the TPG, a westward subduction zone retreat has been suggested in the Permian period and Mesozoic era (Bradshaw et al. Reference Bradshaw, Vaughan, Millar, Flowerdew, Trouw, Fanning and Whitehouse2012).
5.b. Geochemical provenance signatures
Petrofacies A, B and C sediments were mainly fed by a felsic igneous source, typical of rocks derived from a silicic crystalline (plutonic-metamorphic) terrain with a lesser intermediate-acid volcanic component. The major element content, as shown in Figure 8 (Roser & Korsch, Reference Roser and Korsch1988), is indicative of a principal input from a primary felsic source, with an average composition ranging from tonalitic to granodioritic (Fig. 3). With increasing maturity, sediments plot progressively deeper in the P4 field of Roser & Korsch (Reference Roser and Korsch1999), interpreted as recycling. In this study, samples from petrofacies C do not plot in the P4 field (Fig. 8), as expected for more-recycled sediments. On the other hand, the sandstones from Hope Bay that plot in P4 could indicate a local recycling and explain the higher quartz contained in this sample (Fig. 2)
Provenance discriminant diagram (Roser & Korsch, Reference Roser and Korsch1988). P1 – mafic provenance; P2 – intermediate provenance; P3 – felsic provenance; and P4 – quartzose recycled provenance. Discrimination functions are F1 = −1.773*TiO2+0.607*Al2O3+0.76*Fe2O3T−1.5*MgO+0.616*CaO+0.509*Na2O−1.224*K2O−9.09; F2 = 0.445*TiO2+0.07*Al2O3−0.25*Fe2O3T−1.142*MgO+0.438*CaO+1.475*Na2O+1.426*K2O−6.861. Typical igneous rock averages from Le Maitre (Reference Le Maitre1976): Rh – rhyolite; Rd – rhyodacite; Da – dacite; An – andesite; Ba – basalt.
Incompatible to compatible element ratios also indicate a felsic provenance (Fig. 5a) similar to the upper continental crust composition (Fig. 5b). La/Th reflects the influence of a magmatic arc in the hinterland (Bhatia, Reference Bhatia1985; McLennan et al. Reference McLennan, Hemming, McDaniel, Hanson, Johnnson and Basu1993) and ratios such as La/Sc and Ti/Zr support the active continental margin provenance hypothesis (Fig. 5c). Differences within the petrofacies, such as a lower La/Sc in petrofacies A and lower Fe2O3(T)+MgO%, TiO2% and Al2O3/SiO2 in petrofacies C, suggest a lesser contribution of volcanic rocks in petrofacies B and C and more maturity in petrofacies C.
The weathering pattern shown by TPG samples in the A–CN–K plot (Fig. 3) and relatively low CIA values attest to a low to moderate degree of source rock weathering (Fedo, Nesbitt & Young, Reference Fedo, Nesbitt and Young1995). This can indicate a cold and arid climate (cf. Nesbitt & Young, Reference Nesbitt and Young1982). In a similar scenario, Willan (Reference Willan2003) suggested that the TPG was possibly derived from a glaciated continental margin. However, no textural evidence, such as striated clasts or dropstones, supports this interpretation (see discussion by Bradshaw et al. Reference Bradshaw, Vaughan, Millar, Flowerdew, Trouw, Fanning and Whitehouse2012). A glaciated continental margin would transport a wide range of continental materials from the Gondwana hinterland, but detrital zircons from the TPG are mostly Permian (Hervé, Miller & Pimpirev, Reference Hervé, Miller, Pimpirev, Fütterer, Damaske, Kleinschmidt, Miller and Tessensohn2005; Barbeau et al. Reference Barbeau, Davis, Murray, Valencia, Gehrels, Zahid and Gombosi2010; Fanning et al. Reference Fanning, Hervé, Pankhurst, Rapela, Kleiman, Yaxley and Castillo2011). In addition to a cold, but not necessary glaciated, arid climate, rapid exhumation of the Permian source, erosion and proximal deposition could explain the low chemical weathering of TPG samples. Furthermore, the linear trend in the A–CN–K diagram and the separation of sandstone and mudstone samples suggests a tectonically active region, with dissection under non-steady-state weathering conditions (Nesbitt, Fedo & Young, Reference Nesbitt, Fedo and Young1997). The inclination of the TPG trend in Figure 3 relative to the ideal weathering line (WL) indicates that the variation in CIA cannot simply be accounted for by varying grain sizes and sorting. This feature suggests that the TPG samples have been affected by post-depositional K-metasomatism (Fedo, Nesbitt & Young, Reference Fedo, Nesbitt and Young1995).
Hydrodynamic separation of clay from sand-sized quartz and feldspar controls the REE distribution (Fig. 4). Also the Zr/Sc, Th/Sc, Ti/Zr and La/Sc ratios (Fig. 5; Table S1 in the online Supplementary Material at http://journals.cambridge.org/geo) are affected by the grain-size effect rather than recycling, which agrees with the immature textural characteristics. Separation between sandstone and mudstone samples, in terms of the Zr/Sc ratio (Fig. 5b), is indicative of higher zircon and heavy mineral concentrations in the TPG sandstone but it does not necessary reflect recycling. According to Morton (Reference Morton and Sylvester2012), the processes that have the greatest impact on heavy mineral accumulation are hydrodynamic sorting during transportation.
5.c. Palaeogeographic inferences
Contemporaneous sediments from Patagonia and New Zealand share similar characteristics with the TPG. The sediments could also be derived from volcanic-plutonic continental arcs prior to the Gondwana break-up, as part of the Palaeo-Pacific margin of Gondwana. Sedimentary rocks of the Duque de York Complex (DYC; part of the Madre de Dios Complex; Forsythe & Mpodozis, Reference Forsythe and Mpodozis1983) in Patagonia have the same dominant Permian zircon U–Pb peak (Hervé, Fanning & Pankhurst, Reference Hervé, Fanning and Pankhurst2003; Sepúlveda et al. Reference Sepúlveda, Palma-Heldt, Hervé and Fanning2010) and a similar sandstone framework composition (Faúndez, Hervé & Lacassie, Reference Faúndez, Hervé and Lacassie2002; Lacassie, Roser & Hervé, Reference Lacassie, Roser and Hervé2006; A. Quezada, unpub. Graduation thesis, Univ. Chile, 2010; Fig. 9). The Permian to lower Upper Triassic part of the Rakaia sub-terrane, in the New Zealand microcontinent, also has Permian zircon U–Pb peaks (Ireland, Reference Ireland1992; Pickard, Adams & Barley, Reference Pickard, Adams and Barley2000), and petrofacies equivalent to A and part of B (MacKinnon, Reference MacKinnon1983; Fig. 9)
Modal composition of Duque de York Complex sandstone (Dickinson et al. Reference Dickinson, Beard, Brakenridge, Erjavec, Ferguson, Inman, Knepp, Lindberg and Ryberg1983). Dashed arrowed line is the evolutionary trend of the Rakaia sub-terrane (MacKinnon, Reference MacKinnon1983). Hexagonal fields are Permian–Upper Triassic Rakaia petrofacies (MacKinnon, Reference MacKinnon1983). Grey arrowed line is the evolutionary trend of the TPG proposed in this study. Ellipses (red in online version) are the TPG petrofacies A, B and C as in Figure 7.
In terms of geochemistry, the TPG, DYC and Rakaia rocks are also similar. The major element contents also indicate felsic sources (Roser & Korsch, Reference Roser and Korsch1999; Lacassie, Roser & Hervé, Reference Lacassie, Roser and Hervé2006). Similar matches are evident for the trace elements, although DYC samples have higher Zr/Sc (Lacassie, Roser & Hervé, Reference Lacassie, Roser and Hervé2006), indicative of greater zircon concentration, possibly due to longer transport or more intense weathering of the sand. CIA values of DYC sandstone and mudstone samples are higher than those of the TPG (Lacassie, Roser & Hervé, Reference Lacassie, Roser and Hervé2006) but they overlap the TPG CIA values. This suggests that the DYC was derived from a source with a comparable, although somewhat more severe, weathering regime. On the basis of the palynological record of the DYC, Sepúlveda et al. (Reference Sepúlveda, Palma-Heldt, Hervé and Fanning2010) suggested a humid forest environment with an undergrowth of ferns for the DYC, contrasting with the cold and arid climate of the TPG based on CIA values. Although some palynological data from the TPG were recorded in the sixties, these data are contested and discounted now. CIA values from Rakaia sediments overlap those from the TPG and DYC (Roser & Korsch, Reference Roser and Korsch1999) and the slope of the respective trends are comparable, suggesting similar degrees of K-metasomatism and derivation from a primary source whose composition was in the tonalitic–granodioritic range.
Limestones and metabasalts are not common in the TPG. However, accreted Upper Carboniferous – Lower Permian limestone, metabasalt and chert are an important part in the Madre de Dios Accretionary Complex (Forsythe & Mpodozis, Reference Forsythe and Mpodozis1983). This assemblage suggests a subduction-related accretionary setting for the DYC (e.g. Hervé et al. Reference Hervé, Davidson, Godoy, Mpodozis and Covacevich1981; Forsythe, Reference Forsythe1982). Limestones associated with volcanic rocks have also been recognized in the Rakaia sub-terrane (MacKinnon, Reference MacKinnon1983), and studies of the depositional and deformational settings have favoured a fan depositional and accretionary wedge deformational model (Wandres et al. Reference Wandres, Bradshaw, Weaver, Maas, Ireland and Eby2004 and references therein).
Major elements together with petrographic data (especially the presence of first-cycle quartz-feldspathic sand) support the idea that the petrofacies A, B and C of the TPG were deposited in a tectonic setting corresponding to an active continental margin. Active tectonics allow rapid discharge of sediments to the depocentre, and hence less weathering and chemical modification during transport, resulting in low CIA values. However, there is no clear evidence to suggest an accretionary complex origin for the TPG (Bradshaw et al. Reference Bradshaw, Vaughan, Millar, Flowerdew, Trouw, Fanning and Whitehouse2012; Smellie, Reference Smellie and McKenzie1987, Reference Smellie, Thomson, Crame and Thomson1991), unlike for the DYC (Hervé et al. Reference Hervé, Davidson, Godoy, Mpodozis and Covacevich1981; Forsythe, Reference Forsythe1982; Forsythe & Mpodozis, Reference Forsythe and Mpodozis1983). In this context, the TPG petrofacies A, B and C could have been deposited in a fore-arc basin, autochthonous to the Antarctic Peninsula (Bradshaw et al. Reference Bradshaw, Vaughan, Millar, Flowerdew, Trouw, Fanning and Whitehouse2012; Smellie, Reference Smellie and McKenzie1987, Reference Smellie, Thomson, Crame and Thomson1991), while the DYC was deposited on the subducted plate. In this configuration, the Rakaia sub-terrane could have had the same setting as the DYC, but only sharing the source in Permian to early Late Triassic times. In the Late Triassic, Rakaia sediments seem to have gone through increased recycling, with more sedimentary lithic components (MacKinnon, Reference MacKinnon1983; Roser & Korsch, Reference Roser and Korsch1999), not recorded in the DYC or TPG.
The DYC samples lack the corresponding petrofacies X of the TPG, but that part probably correlates with the Eastern Andes Metamorphic Complex (e.g. Bradshaw et al. Reference Bradshaw, Vaughan, Millar, Flowerdew, Trouw, Fanning and Whitehouse2012), northeast of the DYC in Patagonia. This complex is mostly Devonian to Carboniferous in age, but also has some younger parts with detrital zircon peaks of Carboniferous–Permian age (Hervé, Fanning & Pankhurst, Reference Hervé, Fanning and Pankhurst2003; Augustsson et al. Reference Augustsson, Münker, Bahlburg and Fanning2006). Based on provenance considerations from petrographic and geochemical data, it has been suggested that the Eastern Andes Metamorphic Complex records deposition in a passive continental margin and was derived from a cratonic source (Faúndez, Hervé & Lacassie, Reference Faúndez, Hervé and Lacassie2002; Augustsson & Bahlburg, Reference Augustsson and Bahlburg2003). However, the Permian part of this complex (which crops out at the same latitude as the DYC but east of the Patagonian Batholith) shares similar U–Pb zircon patterns with petrofacies A, B and C of the TPG and the DYC. Augustsson et al. (Reference Augustsson, Münker, Bahlburg and Fanning2006) interpreted the differences in zircon populations from rocks of this complex as marking the change from a passive margin regime to the onset of eastward subduction and the development of a magmatic arc in the Late Carboniferous period in southern Patagonia. This is consistent with changes in the geochemical and petrological characteristics of sandstone samples from the same complex (Augustsson & Bahlburg, Reference Augustsson and Bahlburg2008). We interpret the differences in U–Pb detrital zircon patterns and petrography of petrofacies X and the rest of the TPG in the same way.
5.d. Location of the source: Antarctica or South America?
The combined petrographic, geochemical and the published geochronological data of the TPG (petrofacies A, B and C) and the DYC indicate that the sediments were mainly derived from a similar Permian volcanic-plutonic continental arc source which was relatively proximal to the depositional basin. Although the North Patagonian Massif and the Choiyoi volcanic province in north Patagonia are the most named sources for both complexes (e.g. Willan, Reference Willan2003; Pankhurst et al. Reference Pankhurst, Rapela, Fanning and Márquez2006; Fanning et al. Reference Fanning, Hervé, Pankhurst, Rapela, Kleiman, Yaxley and Castillo2011), this would imply a long transportation path, involving more sorting and sedimentary recycling. Riley, Flowerdew & Whitehouse (Reference Riley, Flowerdew and Whitehouse2012) and Millar, Pankhurst & Fanning (Reference Millar, Pankhurst and Fanning2002) have identified Permian magmatism coupled with metamorphic events in the basement of the Antarctic Peninsula. This implies a new proximal potential source for the TPG and suggests an extension of the early Palaeozoic continental margin magmatism from Patagonia (Riley, Flowerdew & Whitehouse, Reference Riley, Flowerdew and Whitehouse2012).
The U–Pb detrital zircon patterns of the TPG and DYC (petrofacies A, B and C) samples indicate a strong input of Permian magmatic zircons, with age peaks ranging from c. 260 to 295 Ma for the TPG (Barbeau et al. Reference Barbeau, Davis, Murray, Valencia, Gehrels, Zahid and Gombosi2010; Fanning et al. Reference Fanning, Hervé, Pankhurst, Rapela, Kleiman, Yaxley and Castillo2011), and 270 to 290 Ma for the DYC (Hervé, Fanning & Pankhurst, Reference Hervé, Fanning and Pankhurst2003; Sepúlveda et al. Reference Sepúlveda, Palma-Heldt, Hervé and Fanning2010). The Choiyoi magmatic province ranges from the arc-related magmatism (c. 281 Ma) of the Lower Choiyoi (Kleiman & Japas, Reference Kleiman and Japas2009) to the intraplate post-orogenic suites of the Upper Choiyoi at c. 264 and c. 251 Ma (Rocha-Campos et al. Reference Rocha-Campos, Basei, Nutman, Kleiman, Varela, Llambias, Canile and da Rosa2011). Although the Choiyoi lacks the early Permian component of the DYC and TPG, Fanning et al. (Reference Fanning, Hervé, Pankhurst, Rapela, Kleiman, Yaxley and Castillo2011) suggested, based on Lu–Hf isotope data in zircons, that the widespread Permian magmatism in the North Patagonian Massif (ages ranging from 295 to 257 Ma and extending to the Triassic; Pankhurst et al. Reference Pankhurst, Rapela, Fanning and Márquez2006) could be regarded as a deeper extension of the Choiyoi, and that both areas are suitable sources for the TPG and DYC.
One of the most widely accepted palaeogeographic reconstructions locates the Antarctic Peninsula west of Patagonia in Middle Jurassic time (König & Jokat, Reference König and Jokat2006). However, this reconstruction implies a long overlap between Patagonia and the Antarctic Peninsula, and the DYC would have been accreted to the Antarctic Peninsula. These may indicate that there was a smaller degree of overlap, or a ‘short-legged tango’ configuration (Miller, Reference Miller, Cooper and Raymond2007) during Permian and Triassic times or at least during Late Triassic time when the Madre de Dios Accretionary Complex accreted to the fore-arc (Willner et al. Reference Willner, Sepúlveda, Hervé, Massonne and Sudo2009). If northern Patagonia was the principal source of the TPG and DYC, this would imply long transport patterns of the sediments, which is not found in the textural characteristics of the detritus. Ramos (Reference Ramos2008) proposed an extension of the Permian magmatism further to the south (present coordinates) in his ‘western magmatic belt’, even though no upper Palaeozoic crystalline rocks are exposed in southernmost Patagonia. According to Pankhurst et al. (Reference Pankhurst, Rapela, Fanning and Márquez2006), Permian magmatism is the result of the collision of the Deseado block and a following break-off of the slab under the North Patagonian Massif. In this model, no prolongation of this Permian magmatism into southern Patagonia or the Antarctic Peninsula is allowed. However, Hervé et al. (Reference Hervé, Calderón, Fanning, Kraus and Pankhurst2010) reported low Th–U metamorphic Permian zircons in migmatitic gneiss within drill cores in the Magallanes Basin. In the Antarctic Peninsula, Permian magmatism, with age peaks at c. 255, 260 and 275 Ma, is coupled with similar high-grade metamorphism and migmatization (Millar, Pankhurst & Fanning, Reference Millar, Pankhurst and Fanning2002; Riley, Flowerdew & Whitehouse, Reference Riley, Flowerdew and Whitehouse2012). In Tierra del Fuego, Hervé et al. (Reference Hervé, Calderón, Fanning, Kraus and Pankhurst2010) suggested at least 8 to 12 km of erosion before deposition of Middle–Late Jurassic volcanic rocks, which would be an important source of detritus. However, no Permian metamorphic zircons have been reported yet for the TPG and DYC.
With the available data it is difficult to discern whether the magmatism of the Choiyoi–North Patagonian Massif extended into Tierra del Fuego and the Antarctic Peninsula or whether they were different but contemporaneous magmatic events. The North Patagonian Massif and the Choiyoi lack the high-grade Permian metamorphic event (no development of metamorphic zircons) which occurred in Tierra del Fuego and the Antarctic Peninsula, and have an early Permian component not yet reported in the Antarctic Peninsula crystalline basement. A key element is the Deseado Massif, which is located between the North Patagonian Massif and Tierra del Fuego. This block is characterized by weathered and altered granitoids of Silurian and Devonian age intruding into Neoproterozoic metasedimentary rocks (Pankhurst et al. Reference Pankhurst, Rapela, Loske, Márquez and Fanning2003), which differs from the Cambrian basement of Tierra del Fuego (Hervé et al. Reference Hervé, Calderón, Fanning, Kraus and Pankhurst2010; Pankhurst et al. Reference Pankhurst, Rapela, Loske, Márquez and Fanning2003; Sölner, Miller & Hervé, Reference Sölner, Miller and Hervé2000). However, basement rocks are poorly exposed in this area and it is possible that Permian igneous rocks lie hidden beneath younger sedimentary or volcanic rocks.
6. Conclusions
Our results, in combination with published data, permit the identification of different petrofacies, with compositional changes that may record the uplift and exposure of a volcano-plutonic continental arc. This source was located relatively close to the depositional basin and has a tonalitic to granodioritic composition. Deep erosion allowed the exposure of the plutonic-metamorphic basement. Despite this, the chemical weathering in the source area was low, suggesting a rapid exhumation in a dry and cold climate.
The volcano-plutonic arc was likely located along the active Gondwana continental margin, probably in the south Patagonia – Antarctic Peninsula sector. This source could contribute detritus to other sedimentary sequences such as the DYC in southern Patagonia, as indicated by the strong chemical, petrographic and chronological similarities between those sediments.
In this case study, a combination of petrographic, geochemical and cathodoluminescence of quartz provenance analyses were used to identify petrofacies and reconstruct the depositional setting of the TPG. Given the lack of stratigraphic information, this interpretation would not have been possible without the combination of different methods, highlighting the importance of using multiple tools for provenance analysis.
Acknowledgements
This work was supported by the Anillo Antártico ART-105 and Inach B-01–08 projects. Special thanks to the crew of the Almirante Oscar Viel, Leucoton, Lautaro and Aquiles from the Chilean Navy and Thomas Haber, Fernando Poblete, Manfred Brix, Alain Demant, Mark Fanning, Carolina Guzmán and Millarca Valenzuela for participating in the field work and discussion. Stefan Kraus is acknowledged for rock samples and Matthew Callaghan for checking the English grammar. Two anonymous reviewers are acknowledged for comments that improved the paper.
Supplementary material
To view supplementary material for this article, please visit http://dx.doi.org/10.1017/S0016756814000454.
