Tectonophysics 683 (2016) 182–199

Contents lists available at ScienceDirect

Tectonophysics

j ourna l homepage: www.e lsev ie r .com/ locate / tecto

Flow dynamics in mid-Jurassic dikes and sills of the Ferrar large igneousprovince and implications for long-distance magma transport

Giulia M. Airoldi a,b,⁎, James D. Muirhead c,1, Sylvan M. Long d,2, Elena Zanella b,e, James D.L. White a

a Geology Department, University of Otago, Leith street, PO Box 56, Dunedin 9054, New Zealandb ALP, Alpine Laboratory of Paleomagnetism, Via Luigi Massa 6, 12016, Peveragno, Italyc School of Environment, University of Auckland, Private Bag 92019, Auckland, New Zealandd Pomona College, Geology Department, 185 E. 6th St., Rm 232, Claremont, CA 91711, USAe D.S.T., Università di Torino, via Valperga Caluso 35, 10125 Torino, Italy

⁎ Corresponding author at: Via Saluzzo 53, 10125 TorinE-mail addresses: g.airoldi.a@gmail.com (G.M. Airoldi)

james.muirhead@fulbrightmail.org (J.D. Muirhead), sylvanelena.zanella@unito.it (E. Zanella), james.white@otago.ac.

1 Present address: Department of Geological Science443022, Moscow, ID 83844-3022, USA.

2 Present address: Leggette, Brashears & Graham, Inc., C

http://dx.doi.org/10.1016/j.tecto.2016.06.0290040-1951/© 2016 Elsevier B.V. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 17 February 2016Received in revised form 20 June 2016Accepted 21 June 2016Available online 23 June 2016

Magma flow paths in sill-fed dikes of the Ferrar large igneous province (LIP), contrast with those predicted byclassic models of dike transport in LIPs andmagmatic rift settings. We examine anisotropy of magnetic suscepti-bility (AMS) flow paths in dike networks at Terra CottaMountain andMt. Gran, which intruded at paleodepths of~2.5 and ~1.5 km. These intrusions (up to 30 m thick) exhibit irregular, interconnected dike-sill geometries andadjoin larger sills (~200–300 m thick) at different stratigraphic levels. Both shallowly dipping and sub-verticalmagma flow components are interpreted from AMS measurements across individual intrusions, and oftenmatch macroscopic flow indicators and variations in dike attitudes. Flow paths suggest that intrusive patternsand magma flow directions depended on varying stress concentrations and rotations during dike and sill propa-gation, whereas a regional extensional tectonic control was negligible or absent. Unlike giant dike swarms in LIPselsewhere (e.g., 1270 Ma MacKenzie LIP), dikes of the Ferrar LIP show no regionally consistent vertical or lateralflow patterns, suggesting these intrusion were not responsible for long-distance transport in the province. In theabsence of regionally significant, colinear dike swarms, or observed intrusions at crustal depths ≥4 km, we sug-gest that long distance magma transport occurred in sills within Beacon Supergroup sedimentary rocks. This in-terpretation is consistent with existing geochemical data and thermal constraints, which support lateral magmaflow for ~3,500 km across the Gondwana supercontinent before freezing.

© 2016 Elsevier B.V. All rights reserved.

Keywords:AntarcticaAnisotropy of magnetic susceptibilityThermo-mechanical modelSill-fed dikesTerra Cotta MountainMount Gran

1. Introduction

Investigating how magma is transported and accommodated in thecrust can yield key insights into the processes governing the growthand breakup of continental lithosphere (Buck, 2004; Ebinger et al.,2013), and the dynamics of magmatic systems that feed volcanic erup-tions (Tibaldi, 2015). In large igneous provinces (LIPs), the intrusivecomponents controlling both the lateral and vertical migration ofmagma transport are often depicted as colinear swarms of giant dikes(Ernst et al., 1995; Ernst et al., 2001). The primary direction of magmaflow documented for these dike systems changes from vertical nearthe plume head (300–500 km from plume center) to lateral away

o, Italy.,.long@gmail.com (S.M. Long),nz (J.D.L. White).s, University of Idaho, P.O. Box

olumbus, OH 43230, USA.

from the source. Examples include the 1270 Ma MacKenzie and~180 Ma Okavango dike swarms (Ernst and Baragar, 1992; Aubourget al., 2008). However, the shallow plumbing systems (b10 km depth)of a few LIPs, such as the 250 Ma Siberian LIP, form interconnected sillnetworks capable of feeding voluminous outpourings of lavas(Naldrett et al., 1995; Cartwright and Hansen, 2006; Muirhead et al.,2014). The geometries of dikes within these sill-dominated provincesdiffer from classic depictions of LIP dike systems. These intrusions,termed by Muirhead et al. (2014) as sill-fed dikes (but also referred topreviously as inclined “sheets”; Airoldi et al., 2011), exhibit shortlengths (b5 km), variable dips (10–90°), form at sill peripheries, andlink sills at different stratigraphic levels (Naldrett et al., 1995;Cartwright and Hansen, 2006; Muirhead et al., 2014).

Although sill-fed dikes form a key component of the shallow plumb-ing systems of sill-dominated LIPs, magma flow dynamics within theseintrusions remain largely unknown. Many studies focus on magmatransport through the outer sheets and internal sills of saucer-shapedintrusions (Ferré et al., 2002; Thomson and Hutton, 2004; Hansen andCartwright, 2006a; Maes et al., 2008; Polteau et al., 2008b; Galland

183G.M. Airoldi et al. / Tectonophysics 683 (2016) 182–199

et al., 2009). In the Karoo LIP arrangement of intrusive segment ‘lobes’and flow kinematics from anisotropy of magnetic susceptibility (AMS)suggest an up-dipflow component in the outer inclined sheets that con-nect to sill peripheries (Polteau et al., 2008a; Schofield et al., 2010;Galerne et al., 2011). Airoldi et al. (2012), however, revealed complexlateral and vertical magma flow patterns in shallowly dipping, sill-feddikes in the Allan Hills region of Ferrar LIP, Antarctica. These datawere interpreted to record intermittent phases of ‘passive’ magma in-jection into fracture networks forming in response to the forceful injec-tion of underlying sills. However, it is currently unknown whether thismodel of dike growth is regionally consistent throughout the FerrarLIP. The role that dikes played in controlling the regional distributionof Ferrar magmas is therefore poorly constrained.

Fig. 1. Left: DEMmap of south Victoria Land, modified from Muirhead et al. (2014). Areas of peRight: simplified stratigraphy of south Victoria Land. Beacon Supergroup rocks are undifferentiatand Terra Cotta Mountain. TCM: Terra Cotta Mountain-Mt. Kuipers; TAM: TransantarcticOrthoquartzite; ASst: Arena Sandstone; AMFm: Altar Mountain Formation; NMS: NewMountascription of Beacon Supergroup stratigraphy, refer to Barrett (1991) and Stump (1995).

We analyze magma transport dynamics at various depths in themagmaticplumbingsystemof theFerrarLIP. Emplaced~10millionyearsprior to the breakup of East fromWest Gondwana, this widespread (c.a.4,100 km long) magmatic province forms part of the ~183 Ma Karoo-Ferrar LIP (Encarnaciòn et al., 1996), and provides important insightsinto the tectono-magmatic conditions across Antarctica during this con-tinental breakup event. AMS is applied to Ferrar intrusions at Terra CottaMountain and at Mt. Gran, south Victoria Land (Fig. 1), to constrain amodel for magma transport dynamics throughout the province. Theseanalyses are used to infer (1) controls on intrusion propagation at differ-ent levels of the plumbing system, (2) characteristic flowmodes withindikes and sills, and (3) the intrusive structures responsible for broad-scale magma transport throughout the Ferrar LIP.

rsistent ice and snow are white. Inset map (top): location of the study area in Antarctica.ed. Also indicated are the observed stratigraphy and corresponding paleodepth ofMt. GranMountains; MFm: Metschel Formation; ASltst: Aztec Siltstone; BHO: Beacon Heightsin Sandstone; TCS: Terra Cotta Sandstone; WGS: Windy Gully Sandstone. For detailed de-

Fig. 2. Examples of intrusions exposed at Terra Cotta Mountain. [a] A network ofinterconnected segmented dikes. View to the north. [b] Transgressive intrusion exposedon the south-western face of Terra Cotta Mountain. The intrusion exhibits a segmentstep and a curved tip that also intrudes bedding in places. View to the SSE.

184 G.M. Airoldi et al. / Tectonophysics 683 (2016) 182–199

2. The Ferrar large igneous province

Ferrar LIP rocks are exposed for 3,500 km along the TransantarcticMountains of East Antarctica. These intrusive and extrusive rocks,with a total estimated volume of around 300,000 km3 (Ross et al.,2005), represent themost laterally extensive LIP system on Earth. Stud-ies addressing the broad-scale emplacement of the province support alateral transport model, with Ferrar magmas travelling N3,000 kmfrom the Weddell Sea across the east Antarctic margin into south-eastern Australasia (Elliot et al., 1999; Elliot and Fleming, 2000; Leat,2008). Emplacement of the province occurred over 349 ± 49 kyr(Burgess et al., 2015), during (or just prior to) the earliest stages ofGondwanaland breakup, and coincided with the emplacement of theearliest Karoo lavas and sills (U-Pb ages on zircon and baddeleyite be-tween 183.6 ± 1.0 and 182.8 Ma, cf. Encarnaciòn et al., 1996; Svensenet al., 2012; Burgess et al., 2015). The cross-continental distribution ofFerrar LIP rocks has led authors to suggest that Ferrar magmas intrudedand erupted in a continental rift system driven by regional extension, ortrans-tension, in a back-arc setting (Wilson, 1993; Storey, 1995; Elliot,2013). However, intrusion and fracture systems trends consistent withregional extension in the Jurassic are lacking (Muirhead et al., 2012). In-stead, regional dike patterns are consistent with magma emplacementunder a far-field neutral stress regime (Muirhead et al., 2014).

Ferrar intrusions are observed dissecting the flat-lying, ~2.5 km-thick Beacon Supergroup sedimentary sequence and the upper~0.5 km of underlying basement granitoid, amphibolite andmetasedimentary rocks (Elliot and Fleming, 2008). Sills are significantlymore voluminous than dikes (Muirhead et al., 2014). In south VictoriaLand, sills reach ~5,000 km2 in area and up to 450 m in thickness(Gunn and Warren, 1962). Some sills are observed ascending the stra-tigraphy in a ‘step-wise’ fashion (Elliot and Fleming, 2004; Airoldiet al., 2011) and, within the upper Permian and lower Triassic membersof the Beacon sequence, sills become progressively thinner (0–100 m)in places and laterally less continuous (Elliot and Fleming, 2004, 2008).

Swarmsof shallow tomoderately dipping dike intrusions are report-ed from various localities in the central Transantarctic Mountains (e.g.Hornig, 1993; Leat, 2008 and references therein) and south VictoriaLand (Skinner and Ricker, 1968; Wilson, 1993; Morrison and Reay,1995; Muirhead et al., 2014). Regional field and remote sensing studiesreveal that these intrusions connect sills at different stratigraphic levels,and are inferred to assist in the vertical transport ofmagma in the upper4 km of the plumbing system to the surface (Muirhead et al., 2012;Muirhead et al., 2014). Magma flow dynamics within these sill-feddikes are, however, poorly constrained.

3. Field sites

3.1. Terra Cotta Mountain

Dike intrusions in the Terra Cotta Mountain area are exposed alongNE- and SW-facing cliffs (Fig. 2 inMorrison and Reay, 1995). These cliffsreveal a swarm of moderately dipping (mean dip 51°: Muirhead et al.,2012) intrusions dissecting Beacon Supergroup rocks and connectingto the lower contact of a sill capping the mountain (Muirhead et al.,2012). Mt. Kuipers lies immediately east of Terra Cotta Mountain,where a ~ 200 m-thick sill and two dike intrusions can be seen on thewestern flanks. Basement granitoids exposed at the northern foothillof the nunatak underlie a sequence of quartz-rich sandstones, siltstonesand minor mudstones and conglomerates. These sedimentary se-quences belong to units from theWindy Gully Sandstone to the BeaconHeights Orthoquartzite of the ~1.5 km thick Taylor Group rocks(Harrington, 1958; Gunn and Warren, 1962; Ross et al., 2008).A ~ 1.0 km-thick sequence of relatively undeformed, flat-lying, sedi-mentary rocks of the Victoria Group lie unconformably on TaylorGroup rocks. At Terra Cotta Mountain, Ferrar intrusions are observeddissecting rocks of the Windy Gully Sandstone up to the Arena

Sandstone, which suggests emplacement at paleodepths of ~1.5–2.5 km (cf. Fig. 1) (Morrison, 1989; Muirhead et al., 2012).

Dike attitudes at Terra Cotta Mountain are irregular, with orienta-tions varying along strike to form zig-zag patterns. Intrusion dips alsovary up-section, where some intrusions change into sills that connectoffset dike segments, resulting in a transgressive dike geometry(Airoldi et al., 2011). Some intrusions bifurcate locally into smallerdikes. For example, dikes up to 10 m thick can be seen connecting tosmaller (2–6 m thick), ‘offshoot’ dikes, some of which in turn feed intothinner (b2 m thick) ones (Fig. 2a). Offshoot dikes are characterizedby irregular geometries, with different segments exhibiting left- andright-steps, both along-strike and up-section, and curved tips (Fig. 2b).

The relative timing of diking events is ambiguous on the SE slopes ofTerra CottaMountain. Although dikes do cross one another in places, nochilled margins are observed along intrusive contacts that would allowinterpretation of the relative timing of intrusion events. However, with-in individual dikes, chilled contacts are observed trending sub-parallelto the plane of the intrusion (Fig. 3a–c). Chilled zones within intrusionsexhibit either sharp or diffuse contacts. Thin zones comprising amixtureof un-melted, baked and thermo-mechanically deformed host rockma-terial, host rock fragments, calcite veins, and chilled dolerite fragmentsare observed within some dikes, and trend sub-parallel to the nearestintrusion margin. Similar contact relationships also appear along dikeselvages (Fig. 3b–c).

‘Baked’ zones are common in the host rock alongside intrusion mar-gins and are typically 1–2 cmwide. Evidence of thermo-mechanical de-formation affecting both country rock and dolerite is observed at severallocations. For example, where sharp selvages are present, country rockat the margins of intrusions exhibits deformed surfaces (Fig. 4). Stria-tions and, locally, discontinuous veins 1–10 cmwide, are also observedon country rockwalls, along dikemargins. These locally exhibit mineral

Fig. 3. Examples of chilled contacts with annotations showing the main geological features. [a] Evidence of multiple intrusion events marked by an internal chilled margin and smallfragments of host rock preserved within the dolerite. [b] View to the NNW of an intrusion with an internal chilled contact. Person for scale. [c] Close up view of the internal contact in‘b’ showing a mixture of calcite and chilled dolerite. White lines represent mineralized slickensides. Permanent marker for scale (~18 cm long). [d] Peperite situated along a dikemargin. Roman numbers indicate (i) Beacon Supergroup, (ii) dolerite, (iii) chilled dolerite, (iv) relative displacement of host rock roof wall and dolerite, (v) flow lineations constrainedfrom linear crests of ‘cusp’ features (black lines), (vi) calcite vein.

185G.M. Airoldi et al. / Tectonophysics 683 (2016) 182–199

striations with near-vertical lineations (Fig. 4a). No fault planes dissect-ing dikes, sills and/or country rock were observed. Variably shapedcusps-and-grooves and drag folds are preserved along country rockmargins (Fig. 4b and c). There are peperite zones in sedimentary rocksnear some intrusion margins. These zones include chaotic arrays of an-gular fragments and/or rounded pods of chilled dolerite, intermingledwith lithified medium-to-fine sand material (Fig. 3d).

3.1.1. Target intrusionsThe principal intrusions investigated in this study are shown in

Fig. 5. At the base of Terra Cotta Mountain is a N20 m-thick intrusion

(d#3). The lower contact of the intrusion is visible, striking 340° anddipping east at 75°. However, the upper contact cannot be seen any-where in the field area, and the intrusion does not appear to have signif-icant lateral continuity. Near the summit of the mountain, multipleintrusions are observed branching out from d#3 upwards, into thethick (N100 m) sill that caps the mountain (s#4). On the northern andeastern slopes of the mountain variably dipping (2–76°) intrusions ex-hibit alternating dike-sill geometries (e.g., t#1a-b, and t#2). On thewestern slopes of Mt. Kuipers is a ~200 m-thick sill, a ~20 m-thick,~090° striking dike (d#5), and a 10 m-thick, ~160° trending dike(d#6). The latter dike (d#6) tapers down and ends to the north-west

Fig. 4. Examples of kinematic flow indicators observed in the field and used to corroborate AMS data. [a] Exposed country rockwall adjacent to a dike margin withmineralized striations.[b] Country rock grooves adjacent to a dike selvagewith the chilledmargin of dolerite outlined by dashed red line. [c] Ropy flowstructure at the lowermargin of a sill. (For interpretation ofthe references to color in this figure legend, the reader is referred to the web version of this article.)

186 G.M. Airoldi et al. / Tectonophysics 683 (2016) 182–199

before reaching d#5, and it was not observed on Terra CottaMountain'ssoutheastern cliff. Further south-east on Mt. Kuipers, d#6 truncates thesill.

3.2. Mount Gran

Mt. Gran is located ~30 km south-east of Allan-Coombs Hills. Here, asteep, ~750mhigh, southeast-facing cliff exposes a complex network ofintrusions (White et al., 2009) (Fig. 5e). These Ferrar dikes and sills in-trude upper Taylor Group and lower Victoria Group rocks. Based onthickness estimates for the Taylor Group (Harrington, 1958; Gunn and

Fig. 5. Target intrusions and structural data. [a, b] Panoramic view and sketch of targeted intrusare numbered as in Section 3.1.1. [c] Rose diagram of intrusion strikes at Terra CottaMountain arepresented by both poles to dike planes and contours of poles to dike planes, and by contourssouthern exposure (photo: M. McClintock). Intrusions numbered as in Section 3.2.

Warren, 1962), we infer that Ferrar intrusions at Mt. Gran wereemplaced at a paleodepth of 1–1.5 km (White et al., 2009; Fig. 1). Themost prominent intrusion at Mt. Gran is a ~ 30 m-thick, sub-verticaldike (d#7), which truncates a N40 m-thick sill (s#8). A network of in-terconnected sills and transgressive dikes, all b20m thick (s#9), are ex-posed on the northeastern side of the cliff. These shallowly dipping(typically b30°) intrusions transgress the stratigraphy up-section tothe southwest, before merging into d#7 (Fig. 5e). Other inclined intru-sions also extend outward from the western margin of d#7. Two dikes(d#10 and d#11) are exposed in a valley a few hundred meters north-west of the cliff, and exhibit 086° and 056° strikes.

ions at Terra Cotta Mountain. View to the west. NFDS= north face dike swarm. IntrusionsndMt. Kuipers. [d] Fieldmeasurements of Terra CottaMountain-Mt. Kuipers intrusions areto sills planes (n = 23), and their mean pole and great circle. [e] Aerial view of Mt. Gran,

187G.M. Airoldi et al. / Tectonophysics 683 (2016) 182–199

4. Methods: anisotropy of magnetic susceptibility

Paleo-magma flow directions can be determined by analyzing thepreferred alignments and orientations of Fe-bearing minerals using amethod known as anisotropy of susceptibility (AMS) (Tarling andHrouda, 1993). This non-destructive approach has become common inthe last few decades due to its time- and cost-effectiveness, and is com-monly used to constrain interpretations of magma flow directions dur-ing dike and sill emplacement within volcanic plumbing systems (seeHrouda, 1982; Baer and Reches, 1987; Dragoni et al., 1997; Walkeret al., 1999; Ferré et al., 2002; Liss et al., 2002; Poland et al., 2004; Ste-venson et al., 2007;Magee et al., 2012b; Delcamp et al., 2014). To distin-guish ‘primary’ flow fabrics from minor, late-stage flow features(e.g., porous flow during final cooling and contraction; Aarnes et al.,2008; Galerne et al., 2010), AMS data is often compared with macro-scopic flow indicators, such as broken bridge structures, country rockgrooves or ropy flow features (Correa-Gomes et al., 2001; Liss et al.,2002; Polteau et al., 2008b; Airoldi et al., 2012; Magee et al., 2012a;see also Section 4.3). The AMS technique has thus been applied to un-derstand the broad scale emplacement of LIPs, such as the MacKenzieLIP (Ernst and Baragar, 1992), the volcanic margin of east Greenland(Callot and Geoffroy, 2004), the Karoo LIP (Aubourg et al., 2008;Polteau et al., 2008b), the British and Irish Paleogene igneous province(Magee et al., 2012a), and the Siberian LIP (Callot et al., 2004).

4.1. Sampling and analyses

Data presented in this study come from 97 oriented block samplescollected close to thewalls of Ferrar dikes and sills at Terra Cotta Moun-tain (81 samples) andMt. Gran (16 samples). The orientations of doler-ite block samples were determined in the field with both solar andmagnetic compasses and a clinometer. Samples sizes were approxi-mately 10 × 10 × 15 cm, to provide sufficientmaterial to perform petro-graphic and magnetic analyses. Depending on intrusion size andoutcrop accessibility, sampling was also performed across intrusion in-teriors in order to detect any significant compositional and/or texturalvariations.Where possible, both walls of intrusionswere sampled to in-vestigate imbricated magnetic foliations (Knight and Walker, 1988).

Samples were prepared for petrographic and magnetic analysis atthe University of Otago Geology Department and Otago PaleomagneticResearch Facility (OPRF), New Zealand. At least one sample per intru-sion was petrographically analyzed. Between 3 and 15 core specimens(diameter = 25 mm, length = 22 mm) were taken from every blocksample for magnetic analyses. Magnetic susceptibility and AMS mea-surements on over 500 core specimens from Terra Cotta Mountainwere made at the inter-university research center Alpine Laboratoryof Paleomagnetism (ALP - Peveragno, Italy), using an AGICO KLY-3Kappabridge. Susceptibility versus temperature analyses were run forone selected specimen per intrusion using a CS-3 furnace at OPRF. Iso-thermal remanent magnetization (IRM) acquisition, thermal demagne-tization and backfield curves were obtained either using a JR-6 spinnermagnetometer (Lowrie, 1990) at ALP, or with a Princeton InstrumentsVibrating Sample Magnetometer at OPRF. Magnetic carriers in igneousrocks from Terra CottaMountain were determined through the analysisof rock magnetic properties. This included magnetic susceptibility, de-fined by the ratio between the induced magnetization of the materialand the inducing magnetic field, IRM, remanence coercivity (BCR),temperature-dependant susceptibility (KT vs T), blocking (TB), andCurie temperatures (TC) atwhichmagnetic minerals lost their magneticproperties.

AMS of Mt. Gran intrusions was measured using a KLY-4sKappabridge apparatus at the University of Southern California. Naturalremanentmagnetization (NRM) and stepwise alternating field (AF) de-magnetization measurements were performed with a 2-G cryogenicmagnetometer with inline AF demagnetizer (up to 200 mT). TB spectraand an estimate of Curie temperatures were determined on an ASC

thermal demagnetizer. The magnetic mineralogy of Mt. Gran sampleswas determined by combining information such as AMS andNRM, rem-anence coercivity and Curie and blocking temperatures.

4.2. Interpretation of magnetic fabrics

Flow textures in intrusive rocks are the result of the hydrodynamicalignment of elongate crystals during magma flow. Fe-Ti oxides suchas (titano-)magnetite mimic this alignment because they form withinand/or along the edges of earlier crystallized, non-ferromagnetic crys-tals (e.g. feldspar) after magma flow has ceased. As a consequence,flow directions are commonly inferred from the arrangement and ori-entations of all magnetic components within the rock fabric and overallintrusive body.

The anisotropy of magnetic susceptibility, or AMS, is modelled as anellipsoid with mutually orthogonal axes k1 ≥ k2 ≥ k3 (respectively, max-imum, intermediate and minimum susceptibility axes). These axes canbe graphically plotted as lineations on equal area stereographic projec-tions (Fig. 6). The anisotropy parameters defined for anymagnetic fabricellipsoid are the mean magnetic susceptibility (Km) and anisotropy de-gree (P or PJ, corrected anisotropy degree) defining the absolute anisot-ropy of a rock specimen, and magnetic lineation (L), foliation (F), andshape parameter (T) (see Tarling and Hrouda, 1993, Table 1.1, p. 18,for their mathematical expression). Together, L, F and T define the ge-ometry of the AMS ellipsoid. Prolate fabric ellipsoids are elongate(L N F) and characterized by −1 ≤ T b 0, whereas oblate ellipsoids areflattened (FN L) and characterized by0 ≤ T ≤ 1. In the directional analysisof AMS fabrics, magnetic lineation and foliation correspond respectivelyto the maximum susceptibility axis direction k1, and to the plane per-pendicular to k3 and defined by k1 and k2 i.e. the magnetic foliationplane (FPL).

Susceptibility and its parameters also depend upon themagnetocrystalline properties and/or distribution of each magneticmineral species. The most common magnetic carrier in mafic igne-ous rocks is magnetite with variable Ti-content. In ferromagnetic(titano-)magnetite, the maximum susceptibility axis (k1) corre-sponds to the easy magnetization (long) axis of the crystal, andthe minimum susceptibility axis (k3) lies perpendicular to theformer (short magnetocrystalline axis) in multi-domain grains(N100 μm grain-size), whereas the magnetocrystalline arrangementis switched (with the k1 axis aligned to the short magnetocrystallineaxis) in single-domain grains (b1 μm grain-size). Ideally, in intru-sive rocks AMS fabrics carried dominantly by multi-domain magne-tite, and the distribution and hydrodynamic alignment of the rocksmagnetic contributions, are typically prolate. The magnetic linea-tion and foliation alignwith the plane of the intrusion (or imbricatedup to 30°, cf. Dragoni et al., 1997) and indicate the flow directionduring magma emplacement (Tarling and Hrouda, 1993). For thistype of fabric, also termed a normal fabric (N, in Fig. 6), the minimumsusceptibility axis is sub-perpendicular to the intrusion plane (IPL).

Magnetic fabrics in intrusive rocks are, however, also known to ex-hibit deviations from thenormal fabric described above. Imbrication an-gles of 30° to 45°between the FPL and intrusion plane, as well as theexchange of the intermediate and minimum axes of the fabric ellipsoid,are commonly related to composite magnetic mineralogy of AMSsources with different properties (e.g. Ferré, 2002; Aubourg et al.,2008). These are known as intermediate fabrics (I1–I3 in Fig. 6), and areclassified after Airoldi et al. (2012, and references therein) as 3 types:

• I1 AMS fabrics are prolate, with the magnetic lineation lying within45° from the intrusion plane and k2 and k3 dispersed on a girdle.

• I2 fabrics are either prolate or oblate, have both k1 and k3 alignedwiththe intrusion plane and FPL orthogonal to it.

• I3 is a ‘nearly normal’ planar fabric, with intrusion and magnetic folia-tion planes sub-parallel to one another, and imbrication or intersec-tion angle N30°; the intermediate susceptibility k2, rather than the

Fig. 6. Examples of normal (N), intermediate (I(1–3)) and reverse (R)magnetic fabric types fromTerraCottaMountain dolerites. FPL:magnetic foliation plane. See later text (Section 6.2) forintrusion reference codes assigned to individual AMS plots.

188 G.M. Airoldi et al. / Tectonophysics 683 (2016) 182–199

magnetic lineation k1, lies closest to the intrusion plane. For fabrics ofthis type, either k2, or the intersection between intrusion andmagnet-ic foliation planes, can be used as proxy of the magma flow direction(e.g. Geoffroy et al., 2002).

The last fabric type presented in this study, termed inverse (R, inFig. 6), is related to the presence of single-domainmagnetic grainswith-in the rock (Rochette et al., 1999; Ferré, 2002; Airoldi et al., 2012). Thisfabric type is characterized by minimum susceptibility axes alignedwithin the intrusion plane, and the magnetic foliation perpendicularto the intrusion.

4.3. Corroboration of AMS data

AMS fabrics were also compared with macroscopic indicators ofmagma flowobserved in thefield to test if these data reflect the primarymagmaflowdirection. In these instances, the shape, trend andplunge ofpreserved macroscopic features both within dikes and along intrusionselvages were used to corroborate AMS data. Some studies use the ori-entation of the long axis of a broken bridge or step structure betweendike and sill segments (Airoldi et al., 2012 and references therein) ormi-croscopic alignments of minerals and vesicles as direct indicators ofmagma flow (e.g. Geshi, 2008; Soriano et al., 2008). Cusps-and-grooves and plumose structures along dike selvages may give informa-tion on both the local magma flow lineation and sense of shear along anintrusion (Varga et al., 1998; Correa-Gomes et al., 2001; Baer et al.,2006; Urbani et al., 2015). Striations on dike walls could representboth magma flow, and shear betweenmagma and encasing rocks relat-ed to dike opening (e.g. Correa-Gomes et al., 2001; Baer et al., 2006)and/or early-stage shear fracturing ahead of a propagating dike tip(Wilson et al., 2016). In the current study, AMS results were comparedwith flow directions indicated by the presence of cusps-and-groovesand drag folds observed along 5 intrusions out of 7 at Terra CottaMoun-tain (all but d#3 and t#1b). No flow indicators were recorded at Mt.Gran.

5. Source of Ferrar dolerite magnetism

The interpretation ofmagnetic fabric properties in rocks requires theidentification of magnetic carriers. The general petrographic character-istics observed in Terra Cotta Mountain and Mt. Gran dolerites are de-scribed below, followed by a description of the results of magneticmineralogy tests and their significance.

5.1. Petrographic characteristics

Ferrar dolerites from the two study locations are compositionallyand texturally similar. They exhibit a narrow range of crystal sizes (com-monly 100 μmto 500 μm) and compositions. All samples contain a com-bination of orthopyroxenes, clinopyroxenes and plagioclase, withvariable amounts of opaques and secondary/alteration minerals. Largerpyroxene crystals occasionally enclose tabular plagioclase. Rutile andmagnetite either included within or between grains are the mainopaque phases observed at Terra Cotta Mountain, whereas smallamounts of magnetite and hematite with variable Fe-Ti content weredetermined from reflected light microscopy in Mt. Gran intrusions,near or within pyroxene crystals.

At Terra Cotta Mountain, d#3 and d#5 are characterized by theabove mineral assemblage, with orthopyroxene enstatite andclinopyroxene augite and pigeonite crystals around 50–60%, and com-monly ≤40% plagioclase crystals.Within t#1a-b and t#2, orthopyroxenebecomes less common, and clinopyroxene and plagioclase increase inabundance up-section. S#4 and the intrusions at Mt. Gran are petro-graphically similar. In these intrusions, plagioclase is themost abundantmineral phase (40–60%), with 25–40% clinopyroxene (Aug ± Pig), andb20% orthopyroxene.

Terra CottaMountain dolerite textures are commonlymicrolitic por-phyritic to glomeroporphyritic, with no visible microscopic or macro-scopic flow textures (Fig. 7a and b). Glass is uncommon, with theexception of a few chilled margins. Iron oxides and alteration productsregularly replace themicrolitic groundmass, and are especially commonin s#4 and d#6 (Fig. 7b).

Fig. 7. Representative textures in Terra Cotta Mountain dolerites. [a] Clusters of pyroxene(grey) and plagioclase (tabular, white) glomerocrysts immersed in partly alteredgroundmass of glass and crystals; [b] microlitic porphyritic texture. A crystal of feldsparis immersed in a finely crystalline groundmass.

189G.M. Airoldi et al. / Tectonophysics 683 (2016) 182–199

At Mt. Gran, variations in crystal size, shape, and texture occur asfunction of proximity to chilled margins. Fine-grained textures withintersertal regions of glassy mesostasis, pervasive opaques and alter-ation products are commonly observed near intrusion margins(e.g., samples Mg 18-1a and 1-1b). Glomeroporphyritic and/ormicrolitic porphyritic textures characterize the internal portions of dol-erite intrusions (e.g., the center of d#7), where opaques and alterationproducts also become less abundant (e.g., Mg 14-1a to Mg 17-1a, seeSupplementary Table S1).

5.2. Magnetic mineralogy properties

Magnetic properties of Terra Cotta Mountain and Mt. Gran intru-sions are rather uniform. Km values range from 724 to 62183 μSI and423 to 40800 μSI, respectively, with ~90% of the data on the orders of10−3 and 10−2 SI. The degree of anisotropy is normally 1–5%,withmax-imum values of 1.047. The least anisotropic samples were collected ond#5.

Thermomagnetic curves obtained from KT vs T tests on Terra CottaMountain dolerites present either a stable (two specimens) or, com-monly, an irregular behavior, where the curves display upward inflex-ions of the bulk susceptibility and decay around 400 °C (Fig. 8a).Significant final alteration of the specimens at high temperature is un-common. For example, there is no sharp variation in susceptibility atthe end of the progressive thermal demagnetization (PTD); instead, agradual removal of the total rock magnetization occurs between 550

and 600 °C. Plots ofmagnetic intensity upon PTD fromMt. Gran samplesalso show drops in the 550–600 °C thermal interval.

Similarly, steep IRM decay occurs as temperatures approach 400 °Cduring PTD of different BCR fractions (Fig. 8b). This IRM decay is notaccompanied by irregular Km vs T paths (Fig. 8d) which, if present,would indicate mineralogical alteration. Soft remanence coercivity(BCR b 500 mT) magnetic components isolated with the Lowrie testare normally over 58%, and the total contributions from the medium(500 b BCR b 1000 mT) and hard magnetic fractions (BCR ≥ 1000 mT)are below 37% and 7%, respectively (Fig. 8b).

Saturation of Ferrar specimens is reached with applied field values(BS) of 300 mT, indicating a dominant low-coercivity magnetic phase.Additional irregular steps observed in the IRM decay curves are likelydue to demagnetization of soft and medium BCR fractions (BCR rangingbetween 30 and 60–70mT) during application of the back field (Fig. 8c).

5.3. Interpretation of magnetic carriers

Magnetic saturation, remanence coercivity and Curie temperaturevalues determined for Terra Cotta Mountain dolerites, with remanencecoercivity overlap in IRM plots, indicate the presence of both soft andmedium remanence coercivity magnetite and/or maghemite (cf.Borradaile and Jackson, 2004 and references therein). Magnetite is thecommon magnetic carrier in basalts (see Tarling and Hrouda, 1993).However, selective oxidation of magnetite can lead to formation ofmaghemite, particularly in hydrothermal environments (see O'Reilly,1983; de Boer and Dekkers, 1996). It is possible that magnetite presentin Ferrar dolerites altered to maghemite during, for example, a post-Ferrar hydrothermal event in south Victoria Land (e.g. Craw et al.,1992; Ballance and Watters, 2002). In fact, the predominance of Km

values N10−3 SI indicates contributions from both ferromagnetic (e.g.(titano-)magnetite and maghemite) and paramagnetic (e.g. pyroxenesand micas) minerals to the magnetic properties of the samples(Owens, 1974; Rochette, 1987; Hrouda, 2002 and references therein).

Uniformity of magnetic properties, with blocking temperaturesaround 550 °C and low coercivities, suggests magnetite with variableTi-content is the dominant magnetic carrier in Mt. Gran rocks.

We infer amagneticmineralogy reflecting contributions by differentmagnetic carriers (for instance, accessory magnetic minerals such aspyrrhotite and titanohematite) from petrographic observation of a dif-fuse oxidation patina in a few samples, occasionally associated withun-differentiated opaque minerals, and magnetic properties. Asdiscussed in Section 5.2, bulk susceptibility inflexion and decay, andsteepening of IRM curves around 400 °C during progressive thermal de-magnetization of Terra Cotta Mountain samples occur in the absence ofany observablemineralogical alteration in the samples. These variationsin IRMare, however, consistentwith breakdown of pyrrhotite at around300–400°, andmay represent reorganization and/or recrystallization ofheated magnetic grains in both the single and multi-domain state(Thompson and Oldfield, 1986; Hopkinson, 1989).

6. Anisotropy of magnetic susceptibility of Ferrar intrusions

The presence of magnetite and/or maghemite as main magneticcarrier(s) in Ferrar samples is demonstrated by themagnetic properties,and validates the interpretation of ‘normal’ and ‘intermediate’magneticfabrics on the basis of the magnetic lineation direction k1 and magneticfoliation plane (see also Section 4.2).

6.1. Magnetic fabric distribution

Terra Cotta Mountain samples are characterized by both prolate(55%) and oblate (45%) magnetic susceptibility ellipsoids. AMS fabrictypes include normal (23%), I-type (59%) and inverse (4%), and 14% ofsamples exhibit (near-) isotropic magnetic fabrics (P b 1.005, F = L,low values in tests of anisotropy). Sample-by-sample AMS parameters

Fig. 8.Representativemagneticmineralogy plots. [a] Kt/T and [b] J/T diagrams (Lowrie, 1990); [c] IRMplots, both direct (left) and backfield (inset, on the right). [d] Plot of variation of bulksusceptibilities for a batch of specimens analyzed with the method after Lowrie (1990).

190 G.M. Airoldi et al. / Tectonophysics 683 (2016) 182–199

defined for Terra Cotta Mountain dataset are presented in Supplemen-tary Table S2. Similarly at Mt. Gran, samples exhibit both prolate(56%) and oblate (44%) magnetic susceptibility ellipsoids. Normal andintermediate AMS fabrics comprise ~20% and ~45% of the total data, re-spectively. 40% ofMt. Gran dataset (s#8, two d#11 sites and three of theeastern sills/shallowly dipping sheets sites) is characterized by anoma-lous oblate reverse fabrics. Samples producing either isotropic or in-verse fabrics (30% of Terra Cotta Mountain samples and ten sites fromMt. Gran) were discarded from directional AMS analysis.

6.2. AMS flow directions

Samples fromTerra CottaMountain andMt. Granwere grouped into35 and 16 sub-sections, respectively (Tables 1 and 2). Each sub-sectioncontains data analyzed from 1 to 4 sample sites on individual intrusions.Samples within each sub-section produced consistent AMS fabrics anddirections.

Maximum and/or intermediate susceptibility axes commonly liewithin 20° of the intrusion plane (Tables 1 and 2). k1 (or k2, in I3 fabrics)is a reliable flow proxy in 50% of all sub-sections. In instances where theintersection between IPL and FPL was used (e.g., t#2 sites, and see d#5-6b in Fig. 6), the flow direction is b30° from the k1 or k2 axes. Except forminor local misfits, AMS data are in good agreement with intrusion ge-ometries (i.e., flow directions sub-parallel to intrusion walls) andmacro-scale kinematic indicators. 71% of AMS flow directions trendwithin 35° of the macroscopic indicators. A similar fit was observed be-tweenmacroscopic andmagnetic flow fabrics at Allan Hills (70% of AMSfabrics are within 35°: Airoldi et al., 2012) and intrusive swarms else-where (Ardnamurchan, Scotland: Magee et al., 2013), suggesting thatmagnetic lineations presented in this study correlate to magma flowaxes.

6.2.1. General magma flow characteristicsFlow components recorded along the margins of analyzed dike in-

trusions are variable. Magnetic lineation plunges of dikes at TerraCotta Mountain and Mt. Gran range from 7 to 79° (Fig. 9). Of the 35dike sub-sections analyzed, 17% of magnetic flow directions plunge

≤20°, 37% plunge 21–45°, and 46% plunge N45°. Similarly, the trendsof magnetic lineations are variable, and almost any orientation is repre-sented (Fig. 9). These multiple flow directions are also reflected in theorientations of cusps-and-grooves along the walls of intrusions (Figs. 4and 10).

6.2.2. Magma flow at Terra Cotta MountainAMS directions in the Terra Cotta Mountain region follow the geo-

metrical variations of the dike intrusions. For example, within trans-gressive dikes (e.g., t#1), shallowly plunging (b25°) flow paths occuralong shallow dipping segments (t#1a-1, t#1a-3, t#1a-4), whereassteeper flow paths (N25°) are observed only in the steeper dike seg-ments (intrusion dips N50°) (t#1b). Dikes are characterized by variablemagma flow paths (Figs. 9 and 11). This is particularly evident in thethickest dikes (d#3 and d#5), where no specific lateral or verticalflow-modes characterize dike selvages and/or the intrusion interiors.For example, analysis of 12 sub-sections along d#5 reveal magma flowplunges ranging 7–79° (Table 1 and Fig. 11), with both shallowly plung-ing and sub-vertical flow lineations aligningwith the plane of the intru-sion. AMS flow trends for d#3 range from 267 to 336°, with shallow-to-moderate plunges (13–37°) along the south-western margin of the in-trusion (d#3-1 and d#3-4). Steeper flow plunges (46°–68°) correspondto the innermost sampling sites (d#3-2a andd#3-3, Table 1 and Fig. 11).Multiple flow directions can also be inferred for all individual intrusionsfrom cusps-and-grooves observed on the walls of dikes. Evidence ofcomposite flow-modes is not, however, always observed in thinnerdikes (width b 10 m, e.g., d#6), although this may in part be the resultof smaller AMS sample sets across some of these intrusions.

AMS flow lineations from the Terra Cotta Mountain summit sill(s#4) exhibit westward trends (268–310°), with sub-horizontalplunges (b10°). Thesemagnetic lineations are sub-parallel to lineationsof the macroscopic flow indicators.

6.2.3. Magma flow at Mt. GranThe magnetic fabric at Mt. Gran exhibits a general consistency with

the overall geometry of the sampled intrusions (i.e., flow sub-parallelto the dike walls). Magma flow in dikes is generally sub-vertical, with

Table 1AMS flow directions constrained for intrusions at Terra Cotta Mountain and Mt. Kuipers, and information frommacroscopic surface lineations. Samples in italics were collected in 2004.

Intrusion Section k1 k2 k3 Angle k1a

- IPLAngleFPL- IPL

Fabrictype

Flowdirectionb

Macroscopic lineation (trend, plunge)c

Ref.nr.

Strike, dip,dipdirection

D I D I D I

T#1a 141/02°SW #1a-1 61 6 330 10 180 79 14° 37° P, N 241/−06° Grooves and ropies - SW trending (200–235°) lineation on horizontalsurfaces053/58°SE #1a-2 76 57 172 4 264 33 15° P, I1 256/−57°

081/26°S #1a-3 237 7 51 83 147 1 4° P, I1 057/−07°#1a-4 93 27 198 29 327 49 5°(k2) 20°imbr. P, I3 032/−20°

T#1b 083/73°S #1b-1 320 75 73 6 164 14 29° 32° P,N(s)

140/-75°

#1b-2 239 55 355 17 95 30 0° P, I1 059/−55°070/76°S #1b-3 54 31 276 51 158 21 34° O, I3 096/−39°105/68°S #1b-4 303 42 82 39 192 22 28° 44°imbr. P, I3 123/−42°

T#2 016/65°E #2-1 95 39 357 9 256 50 21°(k2) 34° P, I3 213/−32° Grooves - shallow plunges to N045°#2-2 89 45 357 2 265 45 18° 26° P, I1 213/–32°

012/50°E #2-3 94 21 3 2 268 69 27° 30°imbr. P, N/I1 273/–21°D#3 340/75°NE #3-1 224 53 128 4 35 37 29.5°(k2) 28° P, I3 331/–37°

#3-2a 87 68 346 4 254 21 7°imbr. P, N 267/–68°#3-3 98 53 192 3 285 37 20° 18° P, N 324/–46°#3-4 125 14 216 4 321 76 30° P, I2 336/–13°

S#4 000/01°E #4a-1 103 1 9 74 193 16 0° O, I2 283/–01° Grooves, NW,05°#4a-2 105 9 8 37 206 52 8° P, I1 285/–09°

S#4 000/01°E #4b-3 134 23 290 65 40 9 22° O, I2 310/–01°#4b-4 90 24 329 49 195 31 23° P, I1 284/–01° 071, 25° SE; grooves - shallow E - W plunge#4b-5 288 25 45 44 179 36 26° O, I1 268/–01°

D#5 245/74°NW #5-1 177 88 28 2 298 1 18° 40° P, I1 221/–63°#5-2 54 30 267 56 153 15 1° 2° O, N 234/–30°#5-3 104 62 216 12 312 26 31°(k2) 42°imbr. P, I3 237/–27° N055° (grooves)#5-4 198 45 15 45 107 1 19°(k2) 40° O, I3 194/–69°

D#5 245/74°NW #5-5a 94 7 4 0 272 83 30° O, I2 243/–07° N055° (grooves)#5-5b 310 79 59 4 149 11 6° 9°imbr. P, N 130/–79°#5-5c 260 5 16 80 169 9 8°(k2) 15° P, N 095/60°#5-6b 220 3 316 69 129 21 10°(k2) 25° O, I3 110/–68° N090°-N100° (grooves)#5-6c 232 43 34 46 134 9 22°(k2) 22° P, I3 201/–67°#5-6d 65 32 290 48 171 24 8° 17° O, N 245/–32°#5-7 278 20 106 70 9 3 23° P, I1 093/–59°

295/75°N #5-8 308 16 130 74 38 1 22° 20.5° P, N 127/–38°D#6 160/69°W #6-1 48 12 316 7 196 76 20°(k2) O, I3 156/–11° Grooves and ropies - plunge 60–70° with trends from NE to S

#6-3 316 42 68 23 178 40 2° O, I2 136/–42°#6-4 287 36 27 13 134 51 28° O, I2 142/–39°

a k2 is indicated for I3 fabrics and may be used for some N fabrics.b Upper hemisphere, trend and plunge is given for magma flow directions.c Macroscopic lineations are given as lower hemisphere, trend and plunge. “Shallow plunges” are b45° and trends only are indicated when lineations parallel the dip of the intrusion

walls.

Table 2Mt. Gran AMS parameters. AMS parameters as defined for Table 1. Lines highlighted in grey are discarded sites (see Section 6.1).

INTRUSION Sample Km

(μSI)L F Pj T k1 k2 k3 Angle

k1-IPLaAngleMFP-IPL

Fabrictype

Flow directionfrommagnetic fabricb

Name Trend (approx.) D I D I D I

D#7 N045°W Mg13-1 16700 1.029 1.010 1.040 −0.477 168 66 326 22 60 8 10° 17° P, N 348/-66°Mg14-1 19300 1.005 1.006 1.011 0.030 290 34 166 40 44 32 32° imbr. O, I3 110/-34°Mg15-1 3990 1.014 1.014 1.028 −0.032 191 63 320 18 56 20 11°(k2) 23° P, N 011/-63°Mg16-1 17800 1.017 1.004 1.022 −0.652 219 78 109 4 18 11 13° P, I1 039/-78°Mg17-1 32000 1.007 1.001 1.008 −0.675 93 68 246 19 339 9 PMg18-1 40800 1.008 1.004 1.012 −0.383 131 18 227 18 359 64 4° P, I2 315/-19°

S#8 Horizontal Mg10-1 423 1.006 1.010 1.016 0.284 217 73 95 9 3 15 O, RS#9 Horizontal and shallowly dipping

sheetsMg19-2 925 1.008 1.001 1.009 −0.798 301 42 82 41 191 21 P, RMg20-1 5260 1.007 1.002 1.009 −0.556 105 18 195 0 286 72 PMg21-1 635 1.000 1.000 1.000 −0.431 282 29 149 51 27 24 PMg22-3 606 1.001 1.003 1.005 0.430 340 1 249 48 70 42 OMg24-2 1410 1.007 1.019 1.027 0.462 112 78 290 12 20 0 O, R

D#10 N086°E Mg1-1 5610 1.002 1.004 1.006 0.401 200 70 319 10 52 17 OD#11 N056°E Mg2-1 798 1.001 1.002 1.003 0.184 206 58 94 13 357 28 O

Mg2-4 5220 1.011 1.007 1.018 −0.244 266 55 144 20 43 27 17° P, I2 086/-55°Mg2-6 7430 1.014 1.016 1.031 0.077 97 25 334 50 202 29 O

a k2 was used as directional proxy at one site characterized by N fabric.b Upper hemisphere, trend and plunge is given for magma flow directions.

191G.M. Airoldi et al. / Tectonophysics 683 (2016) 182–199

Fig. 9. Distribution of interpreted magma flow trends and plunges (see Tables 1 and 2).

192 G.M. Airoldi et al. / Tectonophysics 683 (2016) 182–199

75% of sampled sub-sections exhibiting flow plunges N55°. AMS flowlineations constrained for the 30 m-thick central dike (d#7) define anoverall north-trending, sub-vertical (63°–78° plunges) flow, with twoshallow (19–34° plunges) anomalous AMS directions at the centerand eastern margin of the intrusion (Fig. 12). Due to either isotropicor reverse magnetic fabric, no directional information could beconstrained for the large sill at the base of the cliff (s#8), the shallowlydipping sills and transgressive dikes (s#9) on the northeast end of thecliff face, and one dike (d#10).

6.3. Summary

Terra Cotta Mountain and Mt. Gran samples are characterized byboth prolate and oblate magnetic susceptibility ellipsoids, with normalAMS fabrics adding up to about 20%, and intermediate ones to ~45% of

Fig. 10. Combination of structural and AMS data from t#2. [a] Flow indicators preserved on theoverlays the rose of dike t#2 trends (field measurements).

the total data respectively. Isotropic or inverse fabrics were discardedfrom directional AMS analysis.

Magma Flow directions were inferred from AMS fabrics by applyinga geometric approach based on the orientation of themagnetic lineationand/or the magnetic foliation plane relative to each intrusion's plane tothe magma flow. Over 70% of the magnetic flow indicators and macro-scopic kinematic indicators trendwithin 35° of each other, and are con-sistent with intrusion geometries. Multiple magma flow paths arecommon along individual intrusions, with flow plunges as low as 7°and as steep as 79° (19°–78° at Mt. Gran) along the dikes, and flowtrends of almost any orientation. Magma flow paths defined for theTerra Cotta Mountain summit sill are consistently sub-horizontal, withwestward trends.

7. Discussion

Long-distance magma transport in LIPs is often depicted to occurthrough the emplacement of giant dikes, 100 s of km long and 10 s ofm thick (Ernst et al., 1995). In some instances, these dikes are shownto have transported magma N1000 km laterally away from an inferredplume source (e.g., MacKenzie dike swarm, Ernst and Baragar, 1992).The development of sill-dominated magmatic systems within LIPs,however, has been increasingly recognized over the past decade(e.g., Thomson and Hutton, 2004; Cartwright and Hansen, 2006;Magee et al., 2014; Magee et al., 2016). These sill complexes comprisea stacked series ofmafic intrusions (e.g., the Golden Valley Sill Complex,Karoo LIP, and sill complexes in the North Atlantic igneous province, seeMagee et al., 2016 for a review), contrasting with magma systems con-ventionally depicted for many extensional rift systems (Wright et al.,2012; e.g., magmatic rift segments of Iceland and East Africa:Muirhead et al., 2015; Urbani et al., 2015). AMS studies addressingmagma flow within the intrusive systems of sill-dominated LIPs arerare compared to studies investigating sub-parallel swarms of dikes(e.g., Delcamp et al., 2014; Eriksson et al., 2014 and references therein).Below we discuss magma transport dynamics within dikes and sills ofthe Ferrar LIP.

host rock wall and a rafted sedimentary block. [b] The AMS stereoplot from section t2#2

Fig. 11.Magnetic flow directions at Terra Cotta Mountain [a] and Mt. Kuipers [b]. Aerial photos are used as a base. Simplified structural stereoplots with AMS flow directions overly bothviews.

193G.M. Airoldi et al. / Tectonophysics 683 (2016) 182–199

7.1. Magma transport at Terra Cotta Mountain

Structural and kinematic data at Terra Cotta Mountain suggest a sillsource underlies the exposed dike network (Muirhead et al., 2012). Al-though many dikes exhibit a lateral flow component, 34% of sampledsub-sections exhibit sub-vertical magma flow paths (N45°), suggestingthat the dike swarmprobably transportedmagmaupward from this un-derlying sill. Many of the dikes of this swarm were locally fed upwardfrom large (N10 m thick) “parent” intrusions. For example, a complex

network of dikes is observed branching outward from the top of d#3.Magma flow paths along a 280 m-wide region of d#3 are sub-vertical,suggesting that magma travelled upward into the overlying dikes ad-joining the upper contact of the intrusion. The replacement of ortho-and clino-pyroxene by plagioclase moving up-dip, determined petro-graphically, supports amodel of vertical flow through the central regionof this intrusion. The intrusions overlying d#3 can be seenmerging intothe large sill (s#4) that caps Terra Cotta Mountain, and probably fedmagma vertically into the base of the intrusion.

Fig. 12. Left: southeastern cliff ofMt. Gran and sampling locations. Right: stereoplotswithmean AMS axes and flowdirections (see Table 2).Mt. Gran dike (d#7) results are consistentwiththe AMS flowmodel of Knight and Walker (1988) (see Section 4.1). IPL: intrusion plane; FPL: magnetic foliation.

194 G.M. Airoldi et al. / Tectonophysics 683 (2016) 182–199

We interpret the 42° range in distribution of AMS flowpaths definedat s#4b sub-sections as a consequence ofmultiple injection points at thebase of the intrusion. Indeed, Muirhead et al. (2012) document at least

Fig. 13. Transgressive dikes and sills, some of which intrude concordant to bedding. [a] View t

forty dikes ascending the stratigraphy, many of which connect to thebase of s#4, and our AMS results suggest these dikes fedmagmaupwardinto this sill intrusion. From these feeder intrusions, we infer that

o the west of intrusion at Terra Cotta Mountain. [b] View to NW of intrusions at Mt. Gran.

195G.M. Airoldi et al. / Tectonophysics 683 (2016) 182–199

magma flowed outward along radial paths to produce the observedcomplex magma flow trajectories.

7.2. Magma transport at Mt. Gran

At Mt. Gran, AMS flow directions in dikes define a dominantly sub-vertical flow (67% of data). In the 30 m-thick d#7, 60% of AMS flow di-rections are N60°, despite coarsely crystalline groundmass away fromdike margins, suggesting extensive late-stage crystal growth occurringas the dike centers slowly cooled. Anomalous, shallowly plungingmagma flow, constrained from intermediate AMS fabrics (Mg 14-1andMg 18-1), reflects amix of oblate and prolate contributions bymag-netic particles to the rock's overall AMS fabric (e.g. Ferré, 2002; Aubourget al., 2008), as well as local variations (vertically and laterally) ofmagma flow, perhaps owing to pulsation in magma supply across theintrusion.

7.3. Controls on magma emplacement and fracture dynamics

The number and geometric complexity of the localized intrusive net-works dispersed throughout south Victoria Land (e.g., Allan Hills,Coombs Hills, Mt. Gran, Terra Cotta Mountain) point to the key role oflocal magmatic stresses in driving dike formation by host-rock fractur-ing during the forceful intrusion of sills (White et al. 2009; Muirheadet al., 2012; Muirhead et al., 2014). Our AMS data suggest that dikesascended from these larger sill intrusions, diverging along several tra-jectories, intruding both along the walls of pre-existing intrusions,newly formed fractures, and bedding horizons (Fig. 13). Magma deflec-tion along bedding planes represents the primary control on intrusionpropagation by pre-existing structures (Fig. 13; see also Airoldi et al.,2011). Up-dip and along-strike variations in dike attitude in otherparts of south Victoria Land (e.g., Allan-Coombs Hills: White et al.,2009; Muirhead et al., 2012) represent the response of intrusions tolocal deviations in the principal stress directions in an otherwise homo-geneous and isotropic stress field (Airoldi et al., 2011; Muirhead et al.,2014). Such stress rotations are shown in previous studies to be provid-ed by rigidity contrasts in the layered propagation medium(Gudmundsson and Brenner, 2004; Kavanagh et al., 2006), stress con-centrations and rotations related to sill inflation (Johnson and Pollard,1973; Malthe-Sørenssen et al., 2004; White et al., 2005), and intermit-tent magma propagation resulting in fluctuating stress concentrationsahead of crack tips, upon both dike and sill inception (Kavanagh et al.,2015), and later cooling (Chanceaux and Menand, 2014).

Variations in magma flow paths along individual intrusions suggestthat magma transport cannot be explained purely though a simple ver-tical flow model. The variety of dike orientations and flow directions,coupled with evidence of multiple injections, suggests that magmapropagated intermittently. Variations between shallowly dipping tovertical flow may represent distinct modes of magma flow occurredthrough time across a single intrusion. For example, as dikes widened(in some instances to N10m), variations inmagma crystallinity and vis-cosity between intrusion margins and interiors may have coincidedwith the development of distinct flow paths and velocities. Alternative-ly, dominant vertical flow might have changed with time to a lateralone, or vice versa. Temporal variations in magma flow directions couldresult, for example, from changes in magma buoyancy from crystalliza-tion and/or magma degassing, or changes in magnitude or direction ofdriving pressure resulting from opening of new interconnected dikes/sills. Variable magma flow in Ferrar dikes at Allan Hills was interpretedby Airoldi et al. (2012) as the result of “passive” injection of magma intozones of intense fracturing above inflating sills. In this model, magmapressures generated at the dike-tip are not the primary force drivingdike-fracture growth and propagation through the host. Instead, open-ing of country rock fractures formed during sill-related deformation(e.g., Johnson and Pollard, 1973) creates pressure gradients that drawmagma into these highly strained zones. Field relationships throughout

the region imply that dike intrusions at Terra CottaMountain are under-lain by a N200m-thick sill (Morrison and Reay, 1995;Marsh, 2004) anddike-emplacement orientationswere probably controlled by local stressconditions related to the inflation of a large underlying sill, rather thanby the far-field tectonic stress state (Muirhead et al., 2012; Muirheadet al., 2014). Consequently, variations in dike attitude and magmaflow direction recorded at Terra Cotta Mountain are consistent withthe sill-driven model of fracture growth and magma propagation ofAiroldi et al. (2012).

7.4. Emplacement of the Ferrar LIP during Gondwana breakup

7.4.1. Magmatic-tectonic environment of the Ferrar LIPFerrar magmas were originally proposed to have been emplaced in

extensional basins in a back-arc rift setting (e.g., Elliot and Larsen,1993; Storey, 1995), but structural evidence consistentwith a rifting en-vironment is absent across the Transantarctic Mountains. For example,no significant Jurassic-age normal faults or long, colinear dike swarms,like those in Iceland and East Africa (Wright et al., 2012; Muirheadet al., 2015), are observed. The thickness (~2,500m) and age (Devonianto Jurassic) of the Beacon Supergroup are consistent with subsidencerates of only 0.011–0.014 mm yr−1, which is 1–2 orders of magnitudelower than in active continental rift settings (10−1–100 mm yr−1),even those exhibiting extension rates of only a few mm yr−1 (e.g., theKenya Rift Valley, Birt et al., 1997). Rare observations (n= 2) of mono-clines by Elliot and Larsen (1993) in Ferrar basalt and tuff layers, origi-nally interpreted as fault-related folds (i.e., Grant and Kattenhorn,2004), are more likely the result of folding at the termination of sills(Hansen and Cartwright, 2006a; Magee et al., 2014), like that demon-strated at Allan Hills, Mt. Fleming and Shapeless Mountain (Grapeset al., 1974; Korsch et al., 1984; Pyne, 1984; Airoldi et al., 2011). Elliot(2013) suggested that the substantial thickness of the Kirkpatrick ba-salts (up to 230 m) necessitates confining topography resulting fromrift-related subsidence. However, sill-driven uplift is also shown to pro-duce significant surface topography, resulting in the formation of basins,100 s of meters deep and 10 s of kilometers long, observed above sillcomplexes in seismic reflection imaging (Trude et al., 2003; Hansenand Cartwright, 2006b). At Shapeless Mountain for instance, the intru-sion of Ferrar sills produced differential uplift that, in places, wouldhave produced N200 m-high topography at the surface (Korsch et al.,1984). A distinct absence of Jurassic-age normal faulting, rift basin sub-sidence, and long sub-parallel dike swarms thus provide compelling ev-idence that Ferrar LIP emplacement was not accompanied byextensional tectonics and the formation of continental rift basins. Thisconclusion is further supported by the orientations of N600 Ferrardikes in south Victoria Land (Muirhead et al., 2014). These dikes showno preferred alignments, consistent with emplacement in an isotropicstress regime. Dike orientations were instead controlled by local mag-matic stresses related to the emplacement of sills.

These observations suggest that the East Antarctic Margin was notsubjected to significant regional tectonic stresses prior to and duringFerrar LIP emplacement, which may explain why continental breakupdid not initiate throughout Antarctica during and after Ferrar-Karoomagmatism. Instead, breakup of the Gondwana supercontinent focusedin what is now the Weddell Sea region.

7.4.2. Broad-scale emplacement of the Ferrar LIPIgneous rocks of the Ferrar LIP crop out in present day Antarctica, SE

Australia, and NewZealand. Although the Ferrarmagmatic province ex-hibits a broad geographic distribution, the remarkably homogenouscompositions exhibited by Ferrar rocks suggest a single source (Elliotet al., 1999). Furthermore, decreasing Mg# and MgO contents awayfrom the inferred source area (Weddell Sea) along the length of theprovince are consistent with fractional crystallization during lateralmagmaflow (Elliot and Fleming, 2000; Leat, 2008).WidespreadOrdovi-cian dikes of the Vanda dike swarm have been mapped in basement

Fig. 14. Synthesis of results and conceptual model. [a] Contour plots of poles (black points) ofmagnetic flow directionsmeasured from dikes throughout south Victoria Land. AMS data forTerra Cotta Mountain and Mt. Gran are from the current study and data for Allan Hills are from Airoldi et al. (2012). Note the variable AMS magma flow directions within each intrusiveswarm and across south Victoria Land generally. [b] Conceptualmodel for the Ferrarmagma plumbing systemmodified fromMuirhead et al. (2012). Ferrar sills and dikes are emplaced inan isotropic stress regime in the Beacon Supergroup. No feeder dikes, or sills, are observed in the basement. Dikes fed magma upward between sills, and thus connect the widespread sillnetwork at different levels across the Beacon sedimentary sequence (example in inset [c]). They ultimately fed the voluminous flood basalts at the surface (Muirhead et al. 2014). Long-distance flow in the sills is inferred. Variable magma flow paths in the dikes are based on AMS data shown in [a].

196 G.M. Airoldi et al. / Tectonophysics 683 (2016) 182–199

197G.M. Airoldi et al. / Tectonophysics 683 (2016) 182–199

granitoids and meta-sedimentary rocks (Allibone et al., 1993), yet noFerrar dikes, or sills, are observed at basement depths below the Base-ment Sill, which occurs ~500 m below the lower contact of the BeaconSupergroup (Marsh, 2004; Leat, 2008). As suggested by Leat (2008),these observations imply that long-distance, lateral magma transportin the Ferrar LIP occurred almost exclusively in the Beacon Supergroup(Fig. 14).

AMS flow directions from dike swarms presented in this study,as well as those at Allan Hills (Airoldi et al., 2012), provide insightsinto regional trends in magma flow in dikes across arguably thebest-exposed portion of the province in south Victoria Land(Fig. 14). These results show no consistent lateral or vertical flowcomponents that typically characterize the dike feeder systems ofmany LIPs (e.g., the 1270 Ma MacKenzie LIP; Ernst and Baragar,1992). It is therefore likely that the observed Ferrar dikes werenot responsible for the regional transport of magma laterallyalong the East Antarctic margin during Ferrar magmatism. Instead,magma flow in dikes was of local importance, creating pathwaysthrough which magma could ascend the stratigraphy through in-terconnected sills and, eventually, erupt at the surface (Muirheadet al., 2014).

As Ferrar dikes were not responsible for long distancemagma trans-port along the province, we highlight the Ferrar sills as the probableconduits through which magma was transported laterally ~3,500 kmwithin Beacon Supergroup rocks across the Gondwana supercontinentfrom its source (see conceptual model in Fig. 14). Lateral magma trans-port in sills for ~3,500 km would require both sufficient magma inputand appropriate thermal conditions to avoid solidification and arrest(Annen and Sparks, 2002; Chanceaux and Menand, 2014). One of themain limiting factors to long-distance transport relates to the originalmagma temperature and the temperature at which magma freezesand stalls (solidification temperature). However, magma intrusionsand lava flows in LIPs are shown to travel N1,000 km simply by main-taining flow rates that avoid brecciation and so maintain an insulatingexternal crust (Keszthelyi and Self, 1998), while heat loss from sills iseven more effectively impeded by the enclosing country rock. In manyways, sills are similar to lava flows, as they form through the progressivelateral propagation, inflation and linkage of magma fingers or lobes(Pollard et al., 1975; Thomson and Hutton, 2004; Schofield et al.,2010; Schofield et al., 2012), some of which reach thicknesses of 10 to100 s of meters (Hansen and Cartwright, 2006a; Schofield et al.,2015). Although magma propagation speeds for upper crustal basalticintrusions (e.g., 10−1 to 10−2 m s−1; Wright et al., 2012) are lowerthan for basalt lavas (101 to 10−1 m s−1; e.g. Keszthelyi and Self,1998; Self et al., 1998; Self et al., 2008), intrusions are comparativelymore insulated because they are surrounded by crustal rocks ratherthan cool atmosphere.

The shallow plumbing system of Antarctica comprises a series ofstacked, interconnected sills. The four major sills in south VictoriaLand average ~300 m in thickness (Marsh, 2004; Elliot and Fleming,2008). By applying a heat balance model similar to that developed byKeszthelyi and Self (1998) for lava flows with an insulating crust, thethermal efficiency of sill flow is determined by viscous heating, conduc-tive heat loss from the upper and lower margins of the intrusion, andthermo-physical parameters. Assuming conservative magma propaga-tion velocities of 0.03, 0.05 and 0.1m s−1, ignoring any effects of viscousheating, and applying typical thermo-physical properties for maficmagmas (injection temperature of 1250 °C, density 2700 kg m−3, ther-mal conductivity of 2.1 J s−1 m−1 °C, specific heat of 1200 J kg−1 °C;Barker et al., 1998;Wohletz et al., 1999;Wang et al., 2010), we estimatethat the maximum heat loss for a 300 m-thick Ferrar sill emplaced at2.5 km depth would be between 0.04 and 0.01 °C km−1 (refer toSection 2 of the supplementary material). These thermal constraintswould have allowed transport of magma forming 300 m-thick Ferrarsills for 3500 to 12,000 km along the East Antarctic margin beforesolidifying.

The thermal constraints on long-distancemagma transport in FerrarLIP sills may be tested further by estimating theminimum sill thicknessrequired for sustained magma flow (cf. Holness and Humphreys, 2003,and refer to Section 2 of the supplementary text). Adopting the thermo-mechanical intrusion parameters of Holness and Humphreys (2003),and assuming a constant and conservative overpressure at the magmasource equal to the tensile strength of rock (3 MPa; Schultz, 1995), weestimate that lateral magma transport for 3,500 km across the East Ant-arctic Margin would require a minimum sill thickness of 110 m. Long-distance magma flow would be further assisted by the channeling ofFerrar sills within the Beacon Supergroup sedimentary basin (Leat,2008),which is laterally continuous along the full length of the East Ant-arctic margin (Barrett, 1981). These estimates are similar to those ob-tained for some of the longest identified Deccan lava flows(Rajahmundry lavas: Self et al., 2008), which advanced N1,000 km in acooler subaerial environment, and are consistent with field studiesand thermo-mechanical models of long-distance transport in giantdikes, which in some instances extend for 1000 s of km laterally fromtheir source (e.g., the MacKenzie dike swarm: Ernst and Baragar,1992; Fialko and Rubin, 1999).

8. Conclusions

Magma flow in dike swarms of the Ferrar LIP investigated in thisstudy contrasts with flow paths predicted by classic models of diketransport in LIPs and magmatic rift settings (Ernst and Baragar, 1992;Wright et al., 2012). Our data suggest that dikes transported magmavertically between sills rather than controlling long-distance lateraltransport in sub-parallel swarms throughout SVL.

AMS data provide new insights into the growth of sill-fed dikeswarms in LIPs. The heterogeneity of magma flow and variability indike attitudes at various depths and scales (from a few meters to kilo-meters) suggest that tectonic stresses had little influence on the growthof the intrusive networks. A complex flow model, with both shallowlydipping and sub-vertical flow components, can be defined for most in-trusions. Variable magma flow within individual intrusions may havedeveloped along strike and up-dip either during a single intrusiveevent, and/or as a result of multiple injections, and locally representearly- vs late-stage magma propagation.

The Ferrar LIP formed during the earliest stages of Gondwanalandbreakup and was originally interpreted to have been emplaced in aback-arc extensional setting. However, fracture systems trends, dikepatterns, and magma flow patterns in SVL are consistent with magmaemplacement in an isotropic stress regime, with bedding anisotropyproviding the dominant structural control on intrusion geometries.Flow patterns observed regionally in dike swarms across southVictoria Land show no consistent lateral or vertical flow directions. AsFerrar dikes were not responsible for the long-distance transport ofmagma laterally across Antarctica, the Ferrar sills remain themost likelycandidate for long-distance transport.

Acknowledgements

We acknowledge the University of Otago and New Zealand Divi-sion of Science for postgraduate funding and an Elsevier PostdocFree Access Passport to G. Airoldi, and the Fulbright NZ-Ministry ofScience and Innovation Award for financial support to J. Muirhead.Field activities were supported by Antarctica New Zealand, Helicop-ters New Zealand, and a University of Auckland research grant (UARF3607851). Dr. C. Ohneiser and Dr. C. Tapia are thanked for assistanceduring sample preparation and measurements at Otago. E. Ferré andS. Planke are thanked for helpful reviews that greatly improved themanuscript. G. Airoldi dedicates this work to A. Cicchino andM. Beltrando.

198 G.M. Airoldi et al. / Tectonophysics 683 (2016) 182–199

Appendix A. Supplementary data

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.tecto.2016.06.029.

References

Aarnes, I., Podladchikov, Y.Y., Neumann, E.-R., 2008. Post-emplacement melt flow inducedby thermal stresses: implications for differentiation in sills. Earth Planet. Sci. Lett. 276,152–166.

Airoldi, G.M., Muirhead, J.D., White, J.D.L., Rowland, J., 2011. Emplacement of magma atshallow depth: insights from field relationships at Allan Hills (South Victoria Land,East Antarctica). Antarct. Sci. 23, 281–296.

Airoldi, G.M., Muirhead, J.D., Zanella, E., White, J.D.L., 2012. Emplacement process of Ferrardolerite sheets at Allan Hills (South Victoria Land, Antarctica) inferred frommagneticfabric. Geophys. J. Int. 188, 1046–1060.

Allibone, A.H., Cox, S.C., Graham, I.J., Smillie, R.W., Johnstone, R.D., Ellery, S.G., Palmer, H.,1993. Granitoids of the Dry Valleys area, southern Victoria Land, Antarctica: plutons,field relationships, and isotopic dating. N. Z. J. Geol. Geophys. 36, 281–297.

Annen, C., Sparks, R.S.J., 2002. Effects of repetitive emplacement of basaltic intrusions onthermal evolution and melt generation in the crust. Earth Planet. Sci. Lett. 203,937–955.

Aubourg, C., Tshoso, G., Le Gall, B., Bertrand, H., Tiercelin, J.J., Kampunzu, A.B., Dyment, J.,Modisi, M., 2008. Magma flow revealed by magnetic fabric in the Okavango giantdyke swarm, Karoo igneous province, northern Botswana. J. Volcanol. Geotherm.Res. 170, 247–261.

Baer, G., Aharon, L., Heimann, A., Shaliv, G., Agnon, A., 2006. The Nahal Tavor vent: inter-play of Miocene tectonics, dikes, and volcanism in the Lower Galilee, Israel. Isr.J. Earth Sci. 55, 1–16.

Baer, G., Reches, Z., 1987. Flow patterns of magma in dikes, Makhtesh Ramon, Israel. Ge-ology 15, 3.

Ballance, P., Watters, W.A., 2002. Hydrothermal alteration, contact metamorphism andauthigenesis in Ferrar Supergroup and Beacon Supergroup rocks, Carapace Nunatak,Allan Hills and Coombs Hills, Victoria Land, Antarctica. N. Z. J. Geol. Geophys. 45,71–84.

Barker, C.E., Bone, Y., Lewan, M.D., 1998. Fluid inclusion and vitrinite reflectancegeothermometry compared to heat-flow models of maximum paleotemperaturenext to dikes, western onshore Gippsland Basin, Australia. Int. J. Coal Geol. 37,73–111.

Barrett, P.J., 1981. History of the Ross Sea region during the deposition of the Beacon Su-pergroup 400–180 million years ago. J. R. Soc. N. Z. 11, 447–458.

Barrett, P.J., 1991. The Devonian to Jurassic Beacon Supergroup of the TransantarcticMountains and correlatives in other parts of Antarctica. Oxf. Monogr. Geol. Geophys.17, 120–152.

Birt, C.S., Maguire, P.K.H., Khan, M.A., Thybo, H., Keller, G.R., Patel, J., 1997. The influence ofpre-existing structures on the evolution of the southern Kenya rift valley - evidencefrom seismic and gravity studies. Tectonophysics 278, 211–242.

Borradaile, G.J., Jackson, M., 2004. Anisotropy of magnetic susceptibility (AMS): magneticpetrofabrics of deformed rocks. In: Martin-Hernandez, F., LÜneburg, C.M., Aubourg,C., Jackson, M. (Eds.), Magnetic Fabric: Methods and Applications. The Geological So-ciety of London, pp. 299–360.

Buck, W.R., 2004. Consequences of asthenospheric variability on continental rifting. In:Karner, G.D., Taylor, B.E., Driscoll, N.W., Kohlstedt, D.L. (Eds.), Rheology and Deforma-tion of the Lithosphere at Continental Margins. Columbia University Press, New York,pp. 1–30.

Burgess, S.D., Bowring, S.A., Fleming, T.H., Elliot, D.H., 2015. High-precision geochronologylinks the Ferrar large igneous provincewith early-Jurassic ocean anoxia and biotic cri-sis. Earth Planet. Sci. Lett. 415, 90–99.

Callot, Geoffroy, 2004. Magma flow in the East Greenland dyke swarm inferred fromstudy of AMS: growth of a volcanic margin. Geophys. J. Int. 159, 816–830.

Callot, Gurevitch, E., Westphal, M., Pozzi, J.P., 2004. Flow patterns in the Siberian traps de-duced from magnetic fabric studies. Geophys. J. Int. 156, 426–430.

Cartwright, J., Hansen, D.M., 2006. Magma transport through the crust via interconnectedsill complexes. Geology 34, 929–932.

Chanceaux, L., Menand, T., 2014. Solidification effects on sill formation: an experimentalapproach. Earth Planet. Sci. Lett. 403, 79–88.

Correa-Gomes, L.C., Souza Filho, C.R., Martins, C.J.F.N., Oliveira, E.P., 2001. Development ofsymmetrical and asymmetrical fabrics in sheet-like igneous bodies: the role ofmagma flow and wall-rock displacements in theoretical and natural cases. J. Struct.Geol. 23, 1415–1428.

Craw, D., Morrison, A.D., Walcott, C.R., 1992. Fluid inclusion evidence for widespread shal-low hydrothermal activity in South Victoria Land, Antarctica. N. Z. J. Geol. Geophys.35, 21–28.

de Boer, C.B., Dekkers, M.J., 1996. Grain-size dependence of the rock magnetic propertiesfor a natural rnaghernite. Geophys. Res. Lett. 23, 2815–2818.

Delcamp, A., Petronis, M.S., Troll, V.R., 2014. Discerning magmatic flow patterns inshallow-level basaltic dykes from the NE rift zone of Tenerife, Spain, using the Anisot-ropy ofMagnetic Susceptibility (AMS) technique. In: Ort, M.H., Porreca, M., Geissman,J.W. (Eds.), The Use of Palaeomagnetism and Rock Magnetism to Understand Volca-nic Processes. The Geological Society of London.

Dragoni, M., Lanza, R., Tallarico, A., 1997. Magnetic anisotropy produced by magma flow;theoretical model and experimental data from Ferrar dolerite sills (Antarctica).Geophys. J. Int. 128, 230–240.

Ebinger, C.J., VanWijk, J., Keir, D., 2013. The time scales of continental rifting: implicationsfor global processes. Special Paper of the Geological Society of America, pp. 371–396.

Elliot, D., Fleming, T., 2000. Weddell triple junction: the principal focus of Ferrar andKaroo magmatism during initial breakup of Gondwana. Geology 28, 539–542.

Elliot, D.H., 2013. The geological and tectonic evolution of the Transantarctic Mountains; areview. Geol. Soc. Lond. Spec. Publ. 381, 7–35.

Elliot, D.H., Fleming, T.H., 2004. Occurrence and dispersal of magmas in the Jurassic FerrarLarge Igneous Province, Antarctica. Gondwana Res. 7, 223–237.

Elliot, D.H., Fleming, T.H., 2008. Physical volcanology and geological relationships of theJurassic Ferrar Large Igneous Province, Antarctica. J. Volcanol. Geotherm. Res. 172,20–37.

Elliot, D.H., Fleming, T.H., Kyle, P.R., Foland, K.A., 1999. Long-distance transport of magmasin the Jurassic Ferrar Large Igneous Province, Antarctica. Earth Planet. Sci. Lett. 167,89–104.

Elliot, D.H., Larsen, D., 1993. Mesozoic volcanism in the central Transantarctic Mountains,Antarctica: depositional environment and tectonic setting. In: Findlay, R.H., Unrug, R.,Banks, M.R., Veevers, J.J. (Eds.), Gondwana Eight: Assembly, Evolution and Dispersal.Balkema, Rotterdam, pp. 397–410.

Encarnaciòn, J., Fleming, T.H., Elliot, D.H., Eales, H.V., 1996. Synchronous emplacement ofFerrar and Karoo dolerites and the early break-up of Gondwana. Geology 24,535–538.

Eriksson, P.I., Riishuus, M.S., Elming, S.A., 2014. Magma flow and palaeo-stress deducedfrom magnetic fabric analysis of the Alftafjordur dyke swarm; implications for shal-low crustal magma transport in Icelandic volcanic systems. Geol. Soc. Lond. Spec.Publ. 396, 27.

Ernst, R.E., Baragar, W.R.A., 1992. Evidence from magnetic fabric for the flow pattern ofmagma in the Mackenzie giant radiating dyke swarm. Nature 356, 511–513.

Ernst, R.E., Grosfils, E.B., Mege, D., 2001. Giant dike swarms: Earth, Venus, andMars. Annu.Rev. Earth Planet. Sci. 29, 489–534.

Ernst, R.E., Head, J.W., Parfitt, E., Grosfils, E., Wilson, L., 1995. Giant radiating dyke swarmson Earth and Venus. Earth Sci. Rev. 39, 1–58.

Ferré, E.C., 2002. Theoretical models of intermediate and inverse AMS fabrics. Geophys.Res. Lett. 29.

Ferré, E.C., Bordarier, C., Marsh, J.S., 2002. Magma flow inferred from AMS fabrics in a lay-ered mafic sill, Insizwa, South Africa. Tectonophysics 354, 1–23.

Fialko, Y.A., Rubin, A.M., 1999. Thermal and mechanical aspects of magma emplacementin giant dike swarms. J. Geophys. Res. Solid Earth 104, 23033–23049.

Galerne, C.Y., Neumann, E.-R., Aarnes, I., Planke, S., 2010. Magmatic differentiation pro-cesses in saucer-shaped sills: evidence from the Golden Valley Sill in the KarooBasin, South Africa. Geosphere 6, 163–188.

Galerne, C.Y., Galland, O., Neumann, E.-R., Planke, S., 2011. 3D relationships between sillsand their feeders: evidence from the Golden Valley Sill Complex (Karoo Basin) andexperimental modelling. J. Volcanol. Geotherm. Res. 202, 189–199.

Galland, O., Planke, S., Neumann, E.-R., Malthe-Sørenssen, A., 2009. Experimental model-ling of shallow magma emplacement: application to saucer-shaped intrusions. EarthPlanet. Sci. Lett. 277, 373–383.

Geoffroy, L., Callot, J.P., Aubourg, C., Moreira, M., 2002. Magnetic and plagioclase linearfabric discrepancy in dykes: a new way to define the flow vector using magnetic fo-liation. Terra Nova 14, 183–190.

Geshi, N., 2008. Vertical and lateral propagation of radial dikes inferred from the flow-direction analysis of the radial dike swarm in Komochi Volcano, Central Japan.J. Volcanol. Geotherm. Res. 173, 122–134.

Grant, J.V., Kattenhorn, S.A., 2004. Evolution of vertical faults at an extensional plateboundary, southwest Iceland. J. Struct. Geol. 26, 537–557.

Grapes, R.H., Reid, D.L., McPherson, J.G., 1974. Shallow dolerite intrusion and phreaticeruption in the Allan Hills region, Antarctica. N. Z. J. Geol. Geophys. 17, 563–577.

Gudmundsson, A., Brenner, S.L., 2004. Howmechanical layering affects local stresses, un-rests, and eruptions of volcanoes. Geophys. Res. Lett. 31.

Gunn, B.M., Warren, G., 1962. Geology of Victoria Land between the Mawson and MulockGlaciers, Antarctica. N. Z. Geol. Surv. Bull. 71, 1–157.

Hansen, D.M., Cartwright, J., 2006a. Saucer-shaped sill with lobate morphology revealedby 3D seismic data: implications for resolving a shallow-level sill emplacementmechanism. J. Geol. Soc. 163, 509–523.

Hansen, D.M., Cartwright, J., 2006b. The three dimensional geometry and growth of forcedfolds above saucer-shaped igneous sills. J. Struct. Geol. 28, 1520–1535.

Harrington, H.J., 1958. Nomenclature of rock units in the Ross Sea region. Nature 182, 290.Holness, M.B., Humphreys, M.C.S., 2003. The Traigh Bhàn na Sgùrra Sill, Isle of Mull: flow

localization in a major magma conduit. J. Petrol. 44, 196–1976.Hopkinson, J., 1989. Magnetic and other physical properties of iron at high temperature.

Philos. Trans. R. Soc. Lond. 180, 443–465.Hornig, I., 1993. High-Ti and low-Ti tholeiites in the Jurassic Ferrar Group, Antarctica.

Geol. Jahrb. Reihe E: Geophysik 47.Hrouda, F., 1982. Magnetic anisotropy of rocks and its application in geology and geo-

physics. Geophys. Surv. 5, 37–82.Hrouda, F., 2002. Low-field variation of magnetic susceptibility and its effect on the an-

isotropy of magnetic susceptibility of rocks. Geophys. J. Int. 150, 715–723.Johnson, A.M., Pollard, D.D., 1973. Mechanics of growth of some laccolithic intrusions in

the Henry mountains, Utah, I: field observations, Gilbert's model, physical propertiesand flow of the magma. Tectonophysics 18, 261–309.

Kavanagh, J.L., Boutelier, D., Cruden, A.R., 2015. The mechanics of sill inception, propaga-tion and growth: experimental evidence for rapid reduction in magmatic overpres-sure. Earth Planet. Sci. Lett. 421, 117–128.

Kavanagh, J.L., Menand, T., Sparks, R.S.J., 2006. An experimental investigation of sill forma-tion and propagation in layered elastic media. Earth Planet. Sci. Lett. 245, 799–813.

Keszthelyi, L., Self, S., 1998. Some physical requirements for the emplacement of long ba-saltic lava flows. J. Geophys. Res. Solid Earth 103, 27447–27464.

199G.M. Airoldi et al. / Tectonophysics 683 (2016) 182–199

Knight, M.D., Walker, G.P.L., 1988. Magma flow directions in dikes of the Koolau complex,Oahu, determined from magnetic fabric studies. J. Geophys. Res. 93, 4301–4319.

Korsch, R.J., Barrett, P.J., Andrews, J.M., 1984. The structure of Shapeless Mountain,Antarctica, and its relation to Jurassic igneous activity. N. Z. J. Geol. Geophys. 27,487–504.

Leat, P.T., 2008. On the Long-Distance Transport of Ferrar Magmas, Structure and Em-placement of High-Level Magmatic Systems. London, Special Publications, GeologicalSociety, pp. 45–61.

Liss, D., Hutton, D.H.W., Owens,W.H., 2002. Ropy flow structures; a neglected indicator ofmagma-flow direction in sills and dikes. Geology (Boulder) 30, 715–718.

Lowrie, W., 1990. Identification of ferromagnetic minerals in a rock by coercivity andunblocking temperature properties. Geophys. Res. Lett. 17, 159–162.

Maes, S.M., Ferré, E.C., Tikoff, B., Brown, P.E., Marsh, J.S., 2008. Rock magnetic stratigraphyof a mafic layered sill: a key to the Karoo volcanics plumbing system. J. Volcanol.Geotherm. Res. 172, 75–92.

Magee, C., Jackson, C.A.-L., Schofield, N., 2014. Diachronous sub-volcanic intrusion alongdeep-water margins: insights from the Irish Rockall Basin. Basin Res. 26, 85–105.

Magee, C., Muirhead, J.D., Karvelas, A., Holford, S.P., Jackson, C.A.L., Bastow, I.D., Schofield,N., Stevenson, C.T.E., McLean, C., McCarthy, W., Shtukert, O., 2016. Lateral magmaflow in mafic sill complexes. Geosphere 12, 1–33.

Magee, C., O'Driscoll, B., Petronis, M.S., Stevenson, C.T.E., Clay, P.L., Gertisser, R., 2013.Magma rheology variations in sheet intrusions of the Ardnamurchan central complex(Scotland) inferred from gabbro inclusion characteristics. J. Petrol. 54, 75–102.

Magee, C., Stevenson, C., O'Driscoll, B., Schofield, N., McDermott, K., 2012a. An alternativeemplacement model for the classic Ardnamurchan cone sheet swarm, NW Scotland,involving lateral magma supply via regional dykes. J. Struct. Geol. 43, 73–91.

Magee, C., Stevenson, C.T.E., O'Driscoll, B., Petronis, M.S., 2012b. Local and regional con-trols on the lateral emplacement of the Ben Hiant dolerite intrusion, Ardnamurchan(NW Scotland). J. Struct. Geol. 39, 66–82.

Malthe-Sørenssen, A., Planke, S., Svensen, H., Jamtveit, B., 2004. Formation of saucer-shaped sills, physical geology of high-level magmatic systems. Geol. Soc. Lond.,Spec. Publ. 215–227.

Marsh, B.D., 2004. A magmatic mush column Rosetta Stone: the McMurdo Dry Valleys ofAntarctica. Eos 85, 497,502.

Morrison, A.D., 1989. Geological observations from Terra Cotta Mountain, upper TaylorGlacier. New Zealand Antarctic Record. 9, pp. 31–43.

Morrison, A.H., Reay, A., 1995. Geochemistry of Ferrar dolerite sills and dykes at TerraCotta Mountain, South Victoria Land, Antarctica. Antarct. Sci. 7, 73085.

Muirhead, J.D., Airoldi, G., Rowland, J.V., White, J.D.L., 2012. Interconnected sills and in-clined sheet intrusions control shallow magma transport in the Ferrar large igneousprovince, Antarctica. Geol. Soc. Am. Bull. 124, 162–180.

Muirhead, J.D., Airoldi, G., White, J.D.L., Rowland, J.V., 2014. Cracking the lid: sill-fed dikesare the likely feeders of flood basalt eruptions. Earth Planet. Sci. Lett. 406, 187–197.

Muirhead, J.D., Kattenhorn, S.A., Le Corvec, N., 2015. Varying styles of magmatic strain ac-commodation across the East African Rift. Geochem. Geophys. Geosyst. 16.

Naldrett, A.J., Fedorenko, V.A., Lightfoot, P.C., Kunilov, V.I., Gorbachev, N.S., Doherty, W.,Johan, Z., 1995. Ni-Cu-PGE deposits of the Noril'sk region, Siberia - their formationin conduits for flood-basalt volcanism. Trans. Inst. Min. Metall., B, Appl. Earth Sci. 104.

O'Reilly, W., 1983. The identification of titanomaghemites: model mechanisms formaghemitization and inversion processes and their magnetic consequences. Phys.Earth Planet. Inter. 31, 65–76.

Owens, W.H., 1974. Mathematical model studies on factors affecting the magnetic anisot-ropy of deformed rocks. Tectonophysics 24, 115–131.

Poland, M., Fink, J., Tauxe, L., 2004. Patterns of magma flow in segmented silicic dikes atSummer Coon volcano, Colorado: AMS and thin section analysis. Earth Planet. Sci.Lett. 219, 155–169.

Pollard, D.D., Muller, O.H., Dockstader, D.R., 1975. The form and growth of fingered sheetintrusions. Geol. Soc. Am. Bull. 86, 351–363.

Polteau, S., Ferré, E.C., Planke, S., N., E.-R., C., L., 2008a. How are saucer-shaped sillsemplaced? Constraints from the Golden Valley Sill, South Africa. J. Geophys. Res. 113.

Polteau, S., Mazzini, A., Galland, O., Planke, S., Malthe-Sørenssen, A., 2008b. Saucer-shapedintrusions: occurrences, emplacement and implications. Earth Planet. Sci. Lett. 266,195–204.

Pyne, A.R., 1984. Geology of the Mt. Fleming Area, South Victoria Land, Antarctica, TenthAntarctic Issue (McMurdo Sound). Department of Scientific and Industrial Research(DSIR), Wellington, New Zealand, pp. 505–512.

Rochette, P., 1987. Magnetic susceptibility of the rock matrix related to magnetic fabricstudies. J. Struct. Geol. 9, 1015–1020.

Rochette, P., Aubourg, C., Perrin, M., 1999. Is this magnetic fabric normal? A review andcase studies in volcanic formations. Tectonophysics 307, 219–234.

Ross, P.-S., Ukstins Peate, I., McClintock, M.K., Xu, Y.G., Skilling, I.P., White, J.D.L.,Houghton, B.F., 2005. Mafic volcaniclastic deposits in flood basalt provinces: a review.J. Volcanol. Geotherm. Res. 145, 281–314.

Ross, P.-S., White, J.D.L., McClintock, M., 2008. Geological evolution of the Coombs-AllanHills area, Ferrar Large Igneous Province, Antarctica: debris avalanches, mafic pyro-clastic density currents, phreatocauldrons. J. Volcanol. Geotherm. Res. 172, 38–60.

Schofield, N., Heaton, L., Holford, S.P., Archer, S.G., Jackson, C.A.L., Jolley, D.W., 2012. Seis-mic imaging of ‘broken bridges’: linking seismic to outcrop-scale investigations of in-trusive magma lobes. J. Geol. Soc. 169, 421–426.

Schofield, N., Holford, S., Millett, J., Brown, D., Jolley, D., Passey, S., Muirhead, D., Grove, C.,Magee, C., Murray, J., Hole, M., Jackson, C., Stevenson, C., 2015. Regional magmaplumbing and emplacement mechanisms of the Faroe-Shetland Sill Complex: impli-cations for magma transport and petroleum systems within sedimentary basins.Basin Res. (Accepted).

Schofield, N., Stevenson, C., Reston, T., 2010. Magma fingers and host rock fluidization inthe emplacement of sills. Geology (Boulder) 38, 63–66.

Schultz, R.A., 1995. Limits on strength and deformation properties of jointed basaltic rockmasses. Rock Mech. Rock. Eng. 28, 1–15.

Self, S., Jay, A.E., Widdowson, M., Keszthelyi, L.P., 2008. Correlation of the Deccan and Ra-jahmundry Trap lavas: are these the longest and largest lava flows on Earth?J. Volcanol. Geotherm. Res. 172, 3–19.

Self, S., Keszthelyi, L., Thordarson, T., 1998. The importance of pahoehoe. Annu. Rev. EarthPlanet. Sci. 26, 81–110.

Skinner, D.N.B., Ricker, J., 1968. The geology of the region between the Mawson andPriestley glaciers, north Victoria Land, Antarctica; part II, upper Paleozoic to quaterna-ry geology. N. Z. J. Geol. Geophys. 11, 1041–1075.

Soriano, C., Beamud, E., Garcés, M., 2008. Magma flow in dikes from rift zones of the ba-saltic shield of Tenerife, Canary Islands: implications for the emplacement of buoyantmagma. J. Volcanol. Geotherm. Res. 173, 55–68.

Stevenson, C.T.E., Owens, W.H., Hutton, D.H.W., Hood, D.N., Meighan, I.G., 2007. Laccolith-ic, as opposed to cauldron subsidence, emplacement of the eastern Mournes pluton,N. Ireland: evidence from anisotropy of magnetic susceptibility. J. Geol. Soc. Lond.164, 99–110.

Storey, B.C., 1995. The role of mantle plumes in continental breakup: case histories fromGondwanaland. Nature 377, 301–308.

Stump, E., 1995. The Ross Orogen of the Transantarctic Mountains. Cambridge UniversityPress, Cambridge.

Svensen, H., Corfu, F., Polteau, S., Hammer, Ø., Planke, S., 2012. Rapid magma emplace-ment in the Karoo Large Igneous Province. Earth Planet. Sci. Lett. 325–326, 1–9.

Tarling, D.H., Hrouda, F., 1993. The Magnetic Anisotropy of Rocks. Chapman and Hall,London, p. 217.

Thompson, R., Oldfield, F., 1986. Environmental Magnetism. Allen and Unwin, London,p. 227.

Thomson, K., Hutton, D., 2004. Geometry and growth of sill complexes: insights using 3Dseismic from the North Rockall Trough. Bull. Volcanol. 66, 364–375.

Tibaldi, A., 2015. Structure of volcano plumbing systems: a review of multi-parametric ef-fects. J. Volcanol. Geotherm. Res. 298, 85–135.

Trude, J., Cartwright, J., Davies, R.J., Smallwood, J., 2003. New technique for dating igneoussills. Geology 31, 813–816.

Urbani, S., Trippanera, D., Porreca, M., Kissel, C., Acocella, V., 2015. Anatomy of an extinctmagmatic system along a divergent plate boundary: Alftafjordur, Iceland. Geophys.Res. Lett. (accepted).

Varga, R.J., Gee, J.S., Staudigel, H., Tauxe, L., 1998. Dike surface lineations as magma flowindicators within the sheeted dike complex of the Troodos Ophiolite, Cyprus.J. Geophys. Res. Solid Earth 103, 5241–5256.

Walker, G.P.L., Canon-Tapia, E., Herrero-Bervera, E., 1999. Origin of vesicle layering anddouble imbrication by endogenous growth in the Birkett basalt flow (Columbiariver plateau). J. Volcanol. Geotherm. Res. 88, 15–28.

Wang, D., Lu, X., Song, Y., 2010. Influence of the temperature dependence of thermal pa-rameters of heat conduction models on the reconstruction of thermal history of igne-ous-intrusion-bearing basins. Comput. Geosci. 36, 1339–1344.

White, J.D.L., Bryan, S.E., Ross, P.-S., Self, S., Thordarson, T., 2009. Physical volcanology ofLarge Igneous Provinces: update and review. The Legacy of George Walker, Geologi-cal Society of London Ed. Special Publications of IAVCEI, London, pp. 291–321.

White, J.D.L., Thordarson, T., McClintock, M.K., Ross, P.S., 2005. Cracking the lid: dike em-placement above large sills of the Ferrar Province, Antarctica. Eos 86 (Abstract V23A-0690).

Wilson, P.I.R., McCaffrey, K.J.W., Wilson, R.W., Jarvis, I., Holdsworth, R.E., 2016. Deforma-tion structures associated with the trachyte mesa intrusion, Henry Mountains,Utah: implications for sill and laccolith emplacement mechanisms. J. Struct. Geol.87, 30–46.

Wilson, T.J., 1993. Jurassic faulting and magmatism in the Transantarctic Mountains: im-plications for Gondwana breakup. In: Findlay, R.H., Unrug, R., Banks, M.R., Veevers, J.J.(Eds.), Gondwana Eight: Assembly, Evolution and Dispersal. Balkema, Rotterdam,pp. 563–572.

Wohletz, K.H., Orsi, G., Civetta, L., 1999. Thermal evolution of the Phlegraean magmaticsystem. J. Volcanol. Geotherm. Res. 91, 381–414.

Wright, T.J., Sigmundsson, F., Pagli, C., Belachew, M., Hamling, I.J., Brandsdóttir, B., Keir, D.,Pedersen, R., Ayele, A., Ebinger, C., Einarsson, P., Lewi, E., Calais, E., 2012. Geophysicalconstraints on the dynamics of spreading centres from rifting episodes on land. Nat.Geosci. 5, 242–250.