Abstract

Sheet intrusions represent important magma conduits and reservoirs in subvolcanic systems. Constraining the emplacement mechanisms of such intrusions is crucial to understanding the physiochemical evolution of magma, volcano deformation patterns, and the location of future eruption sites. However, magma plumbing systems of active volcanoes cannot be directly accessed and we therefore rely on the analysis of ancient systems to inform the interpretation of indirect geophysical and geochemical volcano monitoring techniques. Numerous studies have demonstrated that anisotropy of magnetic susceptibility (AMS) is a powerful tool for constraining magma flow patterns within such ancient solidified sheet intrusions. We conducted a high-resolution AMS study of seven inclined sheets exposed along the Ardnamurchan peninsula in northwest Scotland, and examined how magma flow in sheet intrusions may vary along and perpendicular to the magma flow axis. The sheets form part of the Ardnamurchan Central Complex, which represents the deeply eroded roots of an ∼58-m.y.-old volcano. Our results suggest that the inclined sheets were emplaced via either updip magma flow or along-strike lateral magma transport. It is important that observed variations in magnetic fabric orientation, particularly magnetic foliations, within individual intrusions suggest that some sheets were internally compartmentalized, i.e., different along-strike portions of the inclined sheets exhibit subtle differences in their magma flow dynamics. This may have implications for the flow regime and magma mixing within intrusions.

INTRODUCTION

The transport of magma within a subvolcanic system is commonly facilitated by interconnected sheet intrusions (e.g., dikes and sills). Because magma plumbing systems of active volcanoes cannot be directly observed, analyzing ancient sheet intrusion complexes exposed at the surface is crucial to understanding magma transport within subvolcanic domains (e.g., Anderson, 1937; Walker, 1993; Schirnick et al., 1999; Gudmundsson, 2002; Muirhead et al., 2012; Schofield et al., 2012b; Cashman and Sparks, 2013; Petronis et al., 2013). Analyses of ancient sheet intrusion complexes provide invaluable insights into magma emplacement mechanisms and thereby contribute to volcanic hazard assessment (e.g., Sparks, 2003; Sparks et al., 2012; Cashman and Sparks, 2013), understanding the distribution of eruption locations (e.g., Abebe et al., 2007; Gaffney et al., 2007), and elucidating controls on crystal growth and geochemical variations (e.g., Latypov, 2003). For example, studies of magma flow indicators (e.g., vesicle imbrication, phenocryst alignment, magnetic fabrics) in solidified intrusions have demonstrated that sheet geometries alone cannot be used as proxies for magma transport directions, i.e., flow within dikes or inclined sheets can range from dip parallel to strike parallel (e.g., Abelson et al., 2001; Holness and Humphreys, 2003; Callot and Geoffroy, 2004; Geshi, 2005; Philpotts and Philpotts, 2007; Kissel et al., 2010; Magee et al., 2012a). All studies focused on elucidating the structure and source of subvolcanic intrusion complexes should therefore consider magma flow patterns.

Anisotropy of magnetic susceptibility (AMS) allows the rapid and precise measurement of magnetic fabrics from large sample sets (Tarling and Hrouda, 1993). Numerous studies have successfully demonstrated that magnetic lineations and foliations, measured by AMS, can record information on primary magma flow in sheet intrusions (Fig. 1; Launeau and Cruden, 1998; Archanjo and Launeau, 2004; Cañón-Tapia and Chavez-Alvarez, 2004; Féménias et al., 2004; Philpotts and Philpotts, 2007; Stevenson et al., 2007b; Polteau et al., 2008; Petronis et al., 2013). For example, the imbrication of magnetic fabrics is related to increasing velocity gradients adjacent to the wall rock, and can be used to establish magma flow directions (Fig. 1; e.g., Knight and Walker, 1988; Tauxe et al., 1998; Callot et al., 2001; Féménias et al., 2004). AMS therefore potentially provides a powerful tool for assessing magma flow in solidified sheet intrusions.

Although several studies have identified variations in magma flow–related AMS fabrics, particularly along strike of the principal emplacement direction in individual intrusions, the processes that generate local variations in magma flow dynamics remain poorly constrained (e.g., Ernst and Baragar, 1992; Cañón-Tapia and Chavez-Alvarez, 2004; Aubourg et al., 2008; Cañón-Tapia and Herrero-Bervera, 2009; Magee et al., 2013a). An AMS analysis of numerous intrusions exposed in the Ardnamurchan Central Complex (northwest Scotland) was conducted (Magee et al., 2013b) and it was found that the magnetic fabric orientations measured occasionally varied along sheet strike. Assuming that the magnetic fabrics record lateral variations in the magma flow pattern, it was speculated (Magee et al., 2013b) that individual inclined sheets were locally compartmentalized because rheological differences between adjacent magma pulses promoted the internal segmentation of otherwise continuous sheet intrusions. The potential preservation of internal compartmentalization implies that mixing (e.g., chemical composition, crystal population transfer, or xenolith transport) within continuous sheet intrusions may be laterally restricted and could result in the preferential channelization of magma (Holness and Humphreys, 2003; Magee et al., 2013a). In this study we present a high-resolution AMS analysis combined with structural measurements and field observations of seven sheet intrusions within the Ardnamurchan Central Complex. An important aim of this study is to assess how magnetic fabric variations that correspond to localized, intraintrusion magma flow dynamics can be elucidated and distilled from overall magma flow patterns.

GEOLOGICAL SETTING

The Ardnamurchan Central Complex is located in northwest Scotland and comprises a suite of major intrusions (e.g., laccoliths and lopoliths) and numerous minor sheet intrusions (Fig. 2) (Emeleus and Bell, 2005). This exposed magmatic network represents the deeply eroded roots of an ancient volcanic edifice that formed ca. 58 Ma during the development of the British and Irish Paleogene Igneous Province (Emeleus and Bell, 2005). Intensive igneous activity at that time (ca. 61–55 Ma) was fundamentally related to the incipient opening of the North Atlantic and associated lithospheric impingement of a mantle plume (Saunders et al., 1997).

Sheet intrusions in Ardnamurchan are primarily diabase, typically <1 m thick, and display a variety of orientations (Magee et al., 2012a). They were emplaced into a complex host-rock stratigraphy on Ardnamurchan that consists of Neoproterozoic Moine Supergroup metasedimentary rocks (i.e., upper Morar Group) unconformably overlain by Mesozoic metasedimentary strata (e.g., the calcareous Blue Lias Formation, interbedded limestones and shales of the Pabay Shale Formation, and the Bearreraig Sandstone Formation) and early Paleogene volcaniclastics and olivine-basalt lavas (Fig. 2) (Emeleus and Bell, 2005; Emeleus, 2009). The sheet intrusions predominantly display a concentric or arcuate strike (Fig. 2) and an inward inclination (Richey and Thomas, 1930; Emeleus, 2009). This apparent inverted conical geometry, also exhibited by similar intrusion suites within the Mull and Skye Central Complexes, forms the foundation of the cone sheet emplacement model developed by Bailey (1924) and Anderson (1936). The assumption that cone sheets and their host fractures can be simply projected downdip to a convergence point has led to the notion that they are fed from a central overpressured magma chamber (Bailey, 1924; Richey and Thomas, 1930; Anderson, 1936). For example, Richey and Thomas (1930) used linear projections of the Ardnamurchan cone sheet dips and the location of the major intrusions to originally define three intrusive foci, which were inferred to reflect three spatially and temporally separate centers of magmatic activity (Fig. 2). However, numerous studies have reevaluated the geometry and emplacement mechanisms of major intrusions on Ardnamurchan and have questioned this hypothesis (e.g., Day, 1989; O’Driscoll et al., 2006; O’Driscoll, 2007; Magee, 2012; Magee et al., 2012b). Burchardt et al. (2013) constructed a three-dimensional (3D) downdip projection of the cone sheets and suggested that the principal zone of convergence corresponds to an ∼6 × 5 km (elongated east-west) ellipsoidal source reservoir emplaced at 3.5–5 km depth (Fig. 2).

In Magee et al. (2012a), an alternative interpretation for cone sheet emplacement based on an analysis of magma flow patterns derived from magnetic fabrics was presented. The subhorizontal, strike-parallel flow fabrics identified in the majority of intrusions led to the proposal that the cone sheets represent laterally propagating regional dikes (i.e., externally sourced), which upon entering the vicinity of the Ardnamurchan Central Complex were deflected by the local stress field into preexisting, inwardly inclined, concentric fractures (Magee et al., 2012a). Although the origin of some of the Ardnamurchan sheet intrusions originating from a central source was not precluded (Magee et al., 2012a), i.e., a prerequisite of the cone sheet model, the term “inclined sheet” is utilized for all sheet intrusions we studied in this work in order to avoid genetic connotations (cf. Gautneb et al., 1989; Gautneb and Gudmundsson, 1992; Siler and Karson, 2009).

METHODOLOGY

Magnetic Fabrics as a Record of Magma Flow

Magma flow petrofabrics in sheet intrusions may be attributed to the hydrodynamic alignment of suspended crystal populations by noncoaxial shear or coaxial shear, dependent on variations in magma velocity gradients across the intrusion (e.g., Fig. 1) (Correa-Gomes et al., 2001; Callot and Guichet, 2003; Cañón-Tapia and Chavez-Alvarez, 2004). Although this hydrodynamic alignment is typically considered to be stable during magma flow (i.e., crystal orientations remain fixed once aligned), experimental work suggests that this assumption is only valid if the crystal content is >20%, because collisions prevent crystal rotation (see Cañón-Tapia and Herrero-Bervera, 2009, and references therein). Below this threshold, crystals within a flowing magma display a cyclic behavior, whereby the rotation of their principal axes means that the crystals transition between flow-parallel and nonparallel orientations (Cañón-Tapia and Chavez-Alvarez, 2004; Cañón-Tapia and Herrero-Bervera, 2009). The time each crystal spends in either stage of the cyclic phase (i.e., flow parallel or nonparallel) is controlled by the aspect ratio of the crystal and the amount of shear, e.g., high-aspect-ratio phenocrysts spend the majority of time in a flow-parallel orientation (Cañón-Tapia and Herrero-Bervera, 2009). These theoretical considerations of crystal cyclicity therefore imply that if a significant proportion of crystals are nonparallel to flow in a specific part of an intrusion during solidification, then the average petrofabric of a corresponding sample may not obviously relate to the magma flow conditions (Cañón-Tapia and Chavez-Alvarez, 2004; Cañón-Tapia and Herrero-Bervera, 2009). Magma flow within an intrusion can also vary with time, potentially producing a range of petrofabric orientations preserved in different zones of a sheet intrusion. For example, petrofabrics within chilled margins are likely to relate to the initial magma propagation conditions, whereas fabrics in thick sheet intrusion cores may correlate to a more mature phase of magma flow (e.g., backflow or convection; Philpotts and Philpotts, 2007). Magma flow fabrics can also be overprinted by postemplacement processes such as convection and tectonic compression (e.g., Borradaile and Henry, 1997; Schulmann and Ježek, 2012).

It is clear that petrofabrics preserved in sheet intrusions may have a complex origin and history. AMS provides a quantitative measure of mineral alignments (e.g., of titanomagnetite phenocrysts in mafic rocks) and is particularly useful for fine-grained rocks where petrofabrics may not be optically resolvable (Tarling and Hrouda, 1993; Dunlop and Özdemir, 2001). It is now widely accepted that, even in weakly anisotropic material, magnetic lineations and foliations commonly reflect the magmatic petrofabric, providing information on magma migration, flow geometries, and regional strain (King, 1966; Owens and Bamford, 1976; Hrouda, 1982; Borradaile, 1987, 1988; Rochette, 1987; Tarling and Hrouda, 1993; Borradaile and Henry, 1997; Bouchez, 1997; Sant’Ovaia et al., 2000; Petronis et al., 2004, 2009; Horsman et al., 2005; O’Driscoll, 2006; Stevenson et al., 2007a; Kratinová et al., 2010). In particular, numerous studies have substantiated the relationship between the orientation of magnetic minerals and magma flow through correlation with visible magma flow indicators (e.g., Callot et al., 2001; Aubourg et al., 2002; Liss et al., 2002; Horsman et al., 2005; Morgan et al., 2008). Knight and Walker (1988) presented an empirical study of AMS and suggested that the magnetic lineation could be equated to the primary magma flow axis. Furthermore, high magma velocity gradients at sheet margins and crystal interactions have been shown to create imbricated fabrics, the closure directions of which coincide with the primary magma flow direction during initial emplacement (Fig. 1) (Tauxe et al., 1998; Correa-Gomes et al., 2001; Callot and Guichet, 2003; Féménias et al., 2004; Philpotts and Philpotts, 2007; Morgan et al., 2008). To interpret magma flow patterns from magnetic fabrics it is therefore important to (1) sample different locations of an intrusion by collecting traverses of varying orientation, with respect to the sheet geometry, and analyzing multiple sites along sheet strike and/or dip (Cañón-Tapia and Herrero-Bervera, 2009); (2) independently determine magma flow patterns within sheet intrusions if possible (e.g., measuring visible magma flow indicators); and (3) consider whether primary fabrics have been modified by later magmatic or tectonic processes.

AMS Technique

In this study seven separate sheet intrusions (S1–S7) that intrude a variety of host rocks and display a range of orientations (i.e., sills to dikes) in the southern portion of the Ardnamurchan peninsula were analyzed. Similar to the majority of inclined sheets on Ardnamurchan, the analyzed intrusions are aphyric and predominantly consist of fine- to medium-grained (<0.05–0.5 mm) plagioclase microlites, skeletal clinopyroxene, and titanomagnetite (Magee, 2011; Magee et al., 2012a, 2013a). The relatively fine grain size of the inclined sheets is challenging for petrological (petrographic) analyses of silicate fabrics. Of the seven inclined sheets examined, AMS fabrics were analyzed for three intrusions (i.e., S2, S4, and S7) in Magee et al. (2012a); their analysis involved the collection of one (i.e., S2 and S4) or three (i.e., S7) block samples for each intrusion, a strategy that was not designed to investigate local magma flow pattern variations in individual intrusions. AMS samples used in this study were collected in 2008, typically from two or more sites along sheet strike, as oriented drill cores using a portable gasoline-powered drill with a nonmagnetic diamond bit. All samples were oriented using a magnetic and (when possible) a sun compass. Depending on exposure quality, suites of samples were extracted at each site and binned into profiles characterizing the intrusions margins and core or an entire sheet-orthogonal traverse. This sampling strategy allows any lateral and vertical variations in the magnetic fabrics to be spatially analyzed.

The AMS fabrics of each specimen were measured on either an AGICO KLY-3S kappabridge (an induction bridge that operates at a magnetic field of 300 A/m and a frequency of 875Hz) at the University of Birmingham (UK) (i.e., S1, S6, and S7) or on an AGICO MFK1-A (an induction bridge operating at 976 Hz with a 200 A/m applied field) at New Mexico Highlands University (USA) (i.e., S2–S5). Some S1, S6, and S7 specimens were remeasured on the AGICO MFK1-A and showed no difference in magnetic fabric results between the two induction bridges. Magnetic susceptibility differences were measured in three orthogonal planes and combined with one axial susceptibility measurement to define the susceptibility tensor. This tensor, which may be visualized as an ellipsoid, comprises the three principal susceptibility magnitudes (K1 ≥ K2 ≥ K3) and a corresponding set of three orthogonal principal axis directions.

Where magnetic fabrics are prolate and the shape of the susceptibility ellipsoid is elongated along the K1 axis, it is at times appropriate to interpret the orientation of the K1 lineation in the context of a flow or stretching direction, although there are many caveats when interpreting the linear fabric (e.g., Ellwood, 1982; Knight et al., 1986; Hillhouse and Wells, 1991; Geoffroy et al., 1997; Le Pennec et al., 1998; Tauxe et al., 1998). Conversely, oblate fabrics correspond to a susceptibility ellipsoid that is flattened in the K1-K2 plane (e.g., Tarling and Hrouda, 1993). The orientation of the K1-K2 susceptibility axes commonly varies between specimens from the same sample, with the overall dispersion of the two susceptibility axes defining a great-circle girdle on a stereographic projection. Therefore, if the fabric elements at a site are strongly oblate and the 95% confidence ellipses of the K1 and K2 axes overlap in the K1-K2 plane, it is often not appropriate to interpret the orientation of the K1 lineation as a flow or stretching direction (e.g., Cañón-Tapia, 2004; Cañón-Tapia and Herrero-Bervera, 2009).

The magnitude parameters are reported in terms of size, shape, and strength (or ellipticity) of the ellipsoid. These include the mean (or bulk) susceptibility, Kmean = (K1 + K2 + K3)/3; the degree of anisotropy {Pj = exp√2[(η1 – η)2 + (η2 – η)2 + (η3 – η)2]}, where η = (η1 + η2 + η3)/3, η1 = lnK1, η2 = lnK2, η3 = lnK3 (Jelínek, 1981) and the shape parameter T = [2ln(K2/K3)/(ln(K1/K3)] – 1. The latter parameters (Pj and T) are reported as dimensionless parameters, whereas Kmean is measured in SI units. A value of Pj = 1 describes a perfectly isotropic fabric, while a Pj value of 1.15, for example, corresponds to a sample with 15% anisotropy (P gives a value that translates directly to percent anisotropy, whereas Pj is a close approximation). The quantitative measure of the shape of the susceptibility ellipsoid (T) ranges from perfectly oblate (T = +1) to perfectly prolate (T = –1).

Mineralogical Controls on Magnetic Fabric Orientation

Magnetic fabrics measured in titanomagnetite-bearing rocks are at times difficult to interpret because the relationship between the magnetite fabric and the mineral fabrics of the volumetrically dominant silicate phases is often uncertain, and titanomagnetite is frequently a relatively low-temperature liquidus phase. It is important that quantitative textural analyses have demonstrated that titanomagnetite shape and distribution (i.e., petrofabric) is commonly controlled by the primary silicate framework (e.g., Cruden and Launeau, 1994; Launeau and Cruden, 1998; Archanjo and Launeau, 2004; O’Driscoll et al., 2008). The magnetic response of titanomagnetite is also controlled by grain size and its shape anisotropy (Tarling and Hrouda, 1993). Multidomain (MD) titanomagnetites (>100 µm) have a strong shape-preferred anisotropy and thus their magnetic lineation will parallel the long axis of the grain. In contrast, single-domain (SD) magnetites (<1 µm) are more susceptible to magnetization along the magnetocrystalline easy axis, orthogonal to the shape long axis (Hrouda, 1982; O’Reilly, 1984; Potter and Stephenson, 1988; Dunlop and Özdemir, 2001). From the dependence of principal susceptibility axis orientation on grain size, titanomagnetite populations consisting purely of MD or SD grain sizes are interpreted to produce normal or inverse magnetic fabrics, respectively (Rochette et al., 1999; Ferré, 2002). A normal magnetic fabric implies that the magnetic fabric mimics the mineral shape fabric, regardless of the fabric origin. Inverse magnetic fabrics are characterized by K1 and K3 principal susceptibility axes that parallel the pole to the mineral foliation and the mineral lineation, respectively, somewhat complicating their interpretation. The term inverse magnetic fabric was coined by Rochette and Fillion (1988), who proposed that such fabrics may form in response to either (1) c-axis preferred-orientation of ferroan calcite grains, whose maximum susceptibility is parallel to the c-axis; or (2) the presence of SD elongated ferromagnetic grains. In magnetite or maghemite-bearing rocks, when the fabric is carried by SD grains, this leads to an inverse fabric (e.g., Potter and Stephenson, 1988; Rochette and Fillion, 1988; Borradaile and Puumala, 1989). A mixture of SD and MD titanomagnetites may yield intermediate fabrics, where either one of or neither of the K1 and K3 principal susceptibility axes align with a component of the mineral shape fabric (Rochette et al., 1999; Ferré, 2002).

When it can be demonstrated that the magnetic fabric is carried by paramagnetic ferromagnesian silicates, multidomian ferrimagnetic grains, or a mixture of both, it is commonly observed that the magnetic fabric and petrofabric agree. However, occasionally the petrofabric and magnetic fabric may still not coincide if there are magnetostatic interactions between individual, closely packed ferrimagnetic grains (Hargraves et al., 1991). These magnetostatic interactions can produce a distribution anisotropy, promoted by the generation of an asymmetric magnetic interaction field that may contribute to the bulk magnetic anisotropy (Hargraves et al., 1991). Theoretical models have shown that when grains become closer and magnetostatically interact, the distribution of grains rather than their individual orientations dominate the petrofabric (e.g., Stephenson, 1994; Grégoire et al., 1995, 1998; Cañón-Tapia, 1996, 2001).

To assess the magnetic mineralogy of the sheet intrusions in question in this study, high-temperature, low-field-susceptibility experiments were conducted, using an AGICO MFK1-A (multifunction kappabridge) susceptibility meter and a CS4 furnace attachment, in a stepwise heating-cooling fashion from 25 °C to 700 °C to 40 °C in an Ar atmosphere. Hysteresis measurements were conducted on a Lakeshore Shore Cryotronics MicroMag 2900/3900 vibrating sample magnetometer (VSM) at the University of Texas-Dallas paleomagnetism laboratory. Hysteresis experiments involved vibrating the sample within a 3.0 T applied field at 83 Hz next to a set of pick-up coils. The vibrating sample creates a time-varying magnetic flux in the coils, generating a current that is proportional to the sample magnetization.

RESULTS

Here we present the field observations and magnetic fabric analysis for each of the seven intrusions studied, as well as data pertaining to a suite of rock magnetic experiments. All orientation measurements are recorded as strike and dip unless otherwise stated. Magnetic data are presented in Table 1.

S1

Field Observations

Diabase inclined sheets in the vicinity of S1 (UK National Grid coordinates NM 492 626; 56°41′16″N, 6°05′45″W) display a wide range of orientations and locally complex intrusion morphologies (Figs. 3 and 4; see also Kuenen, 1937; Magee et al., 2012a). The S1 intrusion is aphyric with grain sizes <2 mm; with the exception of a thin <1 cm chilled margin, no grain size variation is observed across the inclined sheet at hand specimen scale. The ∼1-m-thick S1 intrusion (oriented 142/15°SW) is generally concordant to the local Blue Lias Formation bedding (∼140/10°SW), except for an ∼5-m-wide zone where it transgresses stratigraphy at a steeper angle (018/55°SW) (Fig. 4A). This zone of transgression is bound to the south by an ∼35-cm-thick inclined sheet (160/48°NE) that crosscuts S1 (Fig. 4A). A steeply dipping dike (110/72°SW) impinges onto the base of the transgressive S1 portion, where it rotates into a sill (086/10°S) and exploits the contact between S1 and the host rock before terminating against the ∼35-cm-thick inclined sheet (Fig. 4A). Numerous studies have shown that such deflections of magmatic sheet intrusions may occur along boundaries that mark a significant contrast in the mechanical properties of the host rocks (e.g., Gudmundsson, 2002, 2011; Kavanagh et al., 2006; Burchardt, 2008). The development of the inclined sheet into a sill may imply that its impingement locally uplifted S1. However, it is important to note that (1) the sill is not observed on the southern side of an inclined sheet, which crosscuts S1, suggesting that the sill terminated against a preexisting intrusion; and (2) adjacent bedding planes are not tilted (Fig. 4B). These observations indicate that the rotation of S1 is a primary, emplacement-related feature, although the exact origins of such a perturbation in the sheet geometry remain unexplained and require further study.

Magnetic Fabrics and Susceptibilities

Two sites were sampled, separated by ∼20 m, along the strike of S1. At each site, the base, middle, and top of S1 were sampled and a vertical traverse was also collected (Figs. 3 and 4). The Kmean values (3.03 × 10–2 SI to 5.5 × 10–2 SI) of S1a–S1d describe a broad range, while the Pj values range from 1.025 to 1.046 (Table 1). The T (–0.028 to –0.839) data reveal that the fabrics are triaxial to strongly prolate (Table 1). K1 consistently trends northwest-southeast with plunges ranging from 3° to 29° (Fig. 3B; Table 1). Magnetic foliation strikes are within 10°–23° of the inclined sheet strike (i.e., 129/18°SW) but the base and middle sheet fabrics dip northeast at 58°–77° (Fig. 3B). Toward the top of the intrusion, the magnetic foliation dips southwest at 9° and is subparallel to the orientation of the sheet (Fig. 3B).

The S1e–S1h samples are characterized magnetically by little variation in Kmean (6.16 × 10–2 to 6.91 × 10–2), Pj (1.021–1.039), and magnetic fabrics that are triaxial (T = –0.069) to prolate (T = –0.619) (Figs. 4B, 4C; Table 1). Although the magnetic lineations commonly plunge southeast at ∼21° (ranging from 3° to 45°), the orientation of the magnetic foliation varies with sample position (Figs. 4B, 4D). Magnetic foliations from samples S1e and S1g, which correspond to the top and base of the intrusion, respectively, are close to the plane of intrusion (i.e., 142/15°SW) but dip in different directions; S1e strikes subparallel to the intrusion and dips 20°SW, whereas S1g dips southeast 14° (Figs. 4B, 4D). In contrast to the two marginal samples of S1e and S1g, the girdle of K2 subspecimen axes in S1f (i.e., from the middle of S1) relative to the consistently oriented magnetic lineations suggests that magnetic foliations within the sheet core are variable (Fig. 4B). This is supported by examining discrete sections of the vertical traverse, S1h. Toward the top of the intrusion, the magnetic foliations progressively rotate from subparallel to S1g (i.e., S1h_C) to steep, northeast-dipping orientations (i.e., S1h_B and S1h_A are oriented at 165/54°NE and 130/73°NE, respectively) (Figs. 4B, 4D; Table 1). S1h_B and S1h_A dip oppositely to the immediately overlying S1e fabric (Figs. 4B, 4D). This change in orientation is coincident with a subtle increase in Kmean and change from prolate to triaxial fabrics (Figs. 4B, 4D).

S2

Field Observations

Inclined sheet S2 is 50 cm thick and displays a prominent ramp-flat morphology (Fig. 5A); S2 is observed to transgress the interbedded limestones and shales (160/09°SW) of the Blue Lias Formation at 048/44°NW toward its western extent (NM 49271 62680; 56°41′18″N, 6°05′46″W), before abruptly becoming strata concordant (154/10°SW) (Figs. 5A, 5B). Extrapolation to the east of the flat S2 section highlighted in Figure 5B suggests that a second outcrop of strata-concordant (140/16°SW) S2 is preserved at NM 49278 62664 (i.e., 56°41′18″N, 6°05′46″W) (Fig. 5A). Both outcrops are mineralogically identical, consisting of a medium-grained (<1.5 mm) diabase that contains coarse (to 3 mm) pyroxene and sulfide blebs. No chilled margins were observed and there is no apparent grain size variation at hand specimen scale across the inclined sheet.

Magnetic Fabrics and Susceptibilities

Two sites were selected for analysis within S2; four profiles (i.e., S2a–S2d) were collected from the western outcrop and three profiles (i.e., S2e–S2g) from the eastern outcrop (Figs. 5 and 6). The two sites display a distinct difference in Kmean, with S2a–S2d ranging from 4.13 × 10–2 SI to 5.40 × 10–2 SI and S2e–S2g ranging from 1.62 × 10–2 SI to 1.93 × 10–2 SI (Table 1). No intrasite variation is observed within the Pj values (1.11–1.17) and the T data indicate that, with the exception of S2a (T = 0.358), all profiles contain magnetic fabrics that are near triaxial to prolate (T = –0.181 to –0.819). The subhorizontal magnetic lineations, which trend northwest-southeast, also remain remarkably consistent regardless of sheet orientation and are thus considered reliable (Figs. 5 and 6). Typically, the magnetic foliations strike northwest-southeast, parallel to the magnetic lineation trend, apart from S2a, which is oriented 049/17°NW (plunge azimuth and plunge). Only the S2a and S2d magnetic foliations are located close to the plane of the intrusion (Figs. 5 and 6). However, while the majority of the magnetic foliations are thereby oriented out of the intrusion plane, it is important to note that the extension of the K2 and K3 girdles implies that the magnetic foliations corresponding to S2b, S2c, and S2e–S2g may not be reliable (Figs. 5 and 6). The principal susceptibility axes of S2d are subparallel to those measured in Magee et al. (2012a) for a sample (i.e., CS166) from approximately the same position (Fig. 5C).

S3

Field Observations

The only dike analyzed in this study (i.e., S3; 56°41′18″N, 6°05′48″W) has a diabase composition and is oriented 152/90° (Fig. 3A). Sample S3 is planar and crosscuts the Blue Lias Formation and earlier Paleogene inclined sheet intrusions (Fig. 3A). Crosscutting relationships indicate that dike intrusion postdated tilting of the Blue Lias Formation and emplacement of the inclined sheets (Fig. 3A) that occurred in response to the inflation and growth of the Ardnamurchan Central Complex, i.e., the contemporaneous local stress field was characterized by a radially inclined σ1 and a circumferential σ3 (Magee et al., 2012a). The relatively young age of S3 and its vertical nature (i.e., suggestive of a horizontal σ3), imply that dike emplacement occurred after the cessation of magmatic activity on Ardnamurchan. It is likely that S3 represents a so-called regional dike given that its orientation (152/90°) is parallel to that of the regional dike swarm (160°–340°) exposed locally (Speight et al., 1982). Dike thickness varies along strike from ∼1.5–3 m. The dike consists of fine (≤1 mm) plagioclase, clinopyroxene, and titanomagnetite with no phenocryst phases present. Grain size does not appear to vary across the intrusion at hand specimen scale.

Magnetic Fabrics and Susceptibilities

Two separate sites 60 m along strike were analyzed within S3 (Fig. 3A); at each site, western and eastern contact-parallel profiles and a sheet-normal traverse were sampled (Fig. 7). The Kmean values of S3a–S3c (4.91 × 10–2 SI, 5.21 × 10–2 SI, and 4.92 × 10–2 SI, respectively) are slightly lower compared to S3d–S3f (5.21 × 10–2, 5.22 × 10–2 and 5.90 × 10–2, respectively), but within all six profiles there is a degree of internal variability that is independent of Pj (Fig. 7; Table 1). For all six profiles, the magnetic lineation and the magnetic foliation are located within or close to the plane of intrusion (Fig. 7). The magnetic lineation is typically subvertical, with plunges ranging from 74° to 88°, although the S3a K1 is oriented at 144/26° (plunge azimuth and plunge) (Fig. 7; Table 1). Figure 7 highlights that some subtle variations between the magnetic foliation and intrusion plane occur across the dike. The magnetic foliations in S3a–S3c all dip ∼86° toward the northeast but strike rotates from 146° along the western margin to 164° at the eastern margin. The strikes of the S3d–S3f magnetic foliations display a similar rotation from 138° (western margin) to 151° (eastern margin) across the dike (Fig. 7A). However, it is important to note that the magnetic foliations from the margin samples dip in opposite directions; S3d dips 81° to the southwest while S3f dips northeastward at 84° (Fig. 7A). Within both S3b and S5e, the two sheet-normal traverses, the magnetic fabric orientations remain remarkably consistent (Fig. 7A).

For the three S3a–S3c samples, Pj is relatively consistent (1.065, 1.068, and 1.073, respectively) while T values range from 0.49 to 0.79 (oblate). Although the Pj values of S3d–S3f are similarly consistent (1.028, 1.030, and 1.039, respectively), albeit lower, the shape of the magnetic fabric is triaxial (T = 0.05, 0.26, and 0.09, respectively). Within S3e it is apparent that the most oblate fabrics commonly occur along the dike margins, while the triaxial fabrics occur primarily within a thin (∼25 cm wide) zone offset to the southwest of the dike center by ∼25 cm (Fig. 7B). These triaxial fabrics also spatially correspond to a zone of decreased Pj (Fig. 7B). A similar internal variation is not observed in S3b, where Pj (1.068–1.077) and T (0.49–0.62) are both tightly constrained and uniform (Fig. 7A).

S4

Field Observations

Along its ∼100 m length (centered on 56°41′33″N, 6°04′44″W), S4 displays a highly variable dip of 7°–58°, compared to the consistent orientation (∼046/05°W) of the Pabay Shale Formation host rock (Figs. 8A, 8B). Sheet thickness is similarly variable and ranges from 1 to 5 m (at S4e and S4f) (Figs. 8A, 8B). In two locations, S4 is crosscut by dikes trending 151°–331° and 156°–336° (Figs. 8A, 8B). The intrusion is aphyric with grain sizes <1.5 mm; with the exception of a thin <1 cm chilled margin, no grain size variation is observed across the inclined sheet at hand specimen scale.

Magnetic Fabrics and Susceptibilities

Six sample suites were collected from S4 (Fig. 8): (1) the S4a–S4c profiles sample the base, middle, and top of the 2-m-thick inclined sheet (038/07°W) where a small (∼10 cm high) intrusive step, bearing 163°–343°, occurs; (2) S4d samples the moderate to steeply dipping portion (035/58°W) of S4 to the north of the S4a–S4c site and approximately corresponds to the CSJ1 AMS sample position in Magee et al. (2012a); and (3) S4e and S4f were taken from the southern extent of the inclined sheet (020/46°W), at the low tide mark, where sheet thickness increases to ∼5 m. The range of Kmean values for all samples is relatively limited, from 2.57 to 4.15 × 10–2 SI (Table 1). Overall, the magnetic fabrics show a relatively weak anisotropy (Pj = 1.015–1.031) and are near triaxial to prolate (T = –0.109 to –0.736) (Table 1). Although the magnetic fabric orientation is variable, K1 typically plunges (33°–59°) to the northwest and is within or close to the plane of the intrusion (Fig. 8C). These magnetic lineations are either parallel or oblique (by as much as 50°) to the inclined sheet dip direction (Fig. 8C). The one exception to this is S4a, where K1 is orthogonal to the intrusion plane (Fig. 8C). Magnetic foliations range in dip from 49° to 87° and display variable strike orientations (Fig. 8C). Three profiles reveal magnetic foliation strikes that are parallel to the inclined sheet dip direction (i.e., S4a, S4b, and S4e), while two are oblique (i.e., S4d and S4f) and one is parallel (i.e., S4c) to the sheet strike (Fig. 8C). There are few to no systematic variations in the magnetic fabrics across the sheet width or along strike, regardless of sheet orientation (Fig. 8C). For example, S4d yields a similar magnetic fabric to the CSJ1 sample measured (Magee et al., 2012a) from the same locality (Fig. 8C).

S5

Field Observations

The S5 fine-grained (≤1 mm) diabase inclined sheet (NM46527 62255; 56°41′10″N, 6°08′02″W) is oriented at 042/22°NW and intrudes a massive diabase unit, the overall geometry of which cannot be distinguished in the field due to a paucity of exposure (Figs. 2 and 9). Along strike, the thickness of the inclined sheet varies from <1 m to 3 m (e.g., Fig. 9).

Magnetic Fabrics and Susceptibilities

Within S5, three profiles were analyzed that correspond to the top (S5a), middle (S5b), and base (S5c) of the inclined sheet; a vertical traverse was sampled (S5d) (Fig. 9). Kmean values for all samples range from 3.33 to 4.65 × 10–2 SI (Table 1). With the exception of the basal profile (S5c), which has a Pj value of 1.009, the Pj range is relatively restricted to 1.022–1.029 (Table 1). Overall, the T data suggest that the magnetic fabrics are generally triaxial, although there is a range from near prolate (i.e., S5b = –0.460) to near oblate (i.e., S5a = 0.316) (Table 1). Magnetic lineations all trend northwest-southeast, with plunges ranging from 2° to 27°, subparallel to the strikes of the magnetic foliations (Fig. 9; Table 1). This northwest-southeast trend is subparallel to the dip direction of the inclined sheet (Fig. 9). The spread of individual principal susceptibility axes in the vertical traverse (i.e., S5d) is likely due to poorly constrained magnetic fabrics in the base of the intrusion (cf. S5c) (Fig. 9).

S6

Field Observations

The diabase inclined sheet S6 is fine grained (≤1 mm), oriented at 096/30°N, and located along the Ormsaigbeg shore (56°41′09″N, 6°08′03″W) (Fig. 2). It is emplaced into the Bearreraig Sandstone Formation (083/30°S) and thins eastward along strike from 2 to 1 m. A small intrusive step (∼10–20 cm high), with a long axis bearing 158°–338° (Fig. 10), is observed at the basal contact.

Magnetic Fabrics and Susceptibilities

Four sample suites were collected from S6, including transects along the base, middle, and top of the intrusion as well as a vertical traverse (i.e., S6a–S6d, respectively) (Fig. 10). Kmean ranges from 3.68 to 5.28 × 10–2 SI while Pj (1.026–1.043) and T (–0.049–0.038; triaxial) show little variation (Table 1). Similarly, the magnetic fabric orientations remain remarkably consistent regardless of sample location; K1 is, on average, oriented at 350/18° (plunge azimuth and plunge) and the magnetic foliation (106/20°N average) is subparallel to the plane of intrusion but does display a consistently shallower dip (Fig. 10).

S7

Field Observations

S7 is located to the east of Ben Hiant (at 56°42′22″N, 5°59′50″W), has a medium-grained (∼2–3 mm) diabase composition, consisting primarily of plagioclase, clinopyroxene, and titanomagnetite. It is intruded into a series of vertically stacked, subhorizontal olivine-basalt lavas (<1 mm grain size), but no host-rock contacts are exposed. Figure 11 reveals that S7 can be subdivided into four outcrops (∼30–50 m width), bounded by subtle topographic depressions, which individually display slight variations in thickness (∼1.5–2 m) at regular intervals along strike. Each outcrop represents the southern extremity of an elongated lobe-like ridge (∼<5 m high), which extend northward for as much as ∼200 m and have azimuths ranging from 116°–296° in the northeast to 161°–341° in the southwest (Figs. 11A–11C). From northeast to southwest, the four outcrops have approximate strikes and dips of 037/30°WNW, 058/30°NW, 080/30°N, and 074/30°N (Fig. 11A). Toward the margins of each outcrop, grain size decreases to ∼1 mm and contains an increasing proportion of calcite-bearing amygdales (to 8 mm diameter). Superimposed onto each lobe are a series of subparallel troughs (∼<0.5 m deep) that extend northward from the outcrops for ∼10 m (Fig. 11B) and spatially correspond to the zones of observed thinning (Fig. 11C). Beyond the northern limit of the lobes, a small monocline is developed within the lava flows (Figs. 11A, 11B).

Magnetic Fabrics and Susceptibilities

Four sites within S7 were selected for high-resolution AMS analysis (Fig. 11A); S7a was collected from the northeasternmost outcrop (NM 55401 64253), S7b and S7c are from the same outcrop (NM 55421 64327 and 55513 64422, respectively), and S7d is from the most southwestern outcrop sampled (NM 55324 64217). The Kmean values for each site are 3.68 × 10–2 SI, 6.87 × 10–2 SI, 5.79 × 10–2 SI, and 3.64 × 10–2 SI, respectively (Table 1). Values for Pj and T range from 1.019 to 1.141 (weak to strong anisotropy) and –0.274 to 0.566 (prolate-triaxial to oblate), respectively (Table 1). The magnetic fabric for each site is relatively well constrained and reveals that K1 is approximately orthogonal to the plane of intrusion (Fig. 11D); the magnetic lineation values (plunge azimuth and plunge) are 144/44°, 171/58°, 107/67°, and 141/32° for S7a–S7d. These magnetic lineations are subparallel to the elongation direction of their respective lobe-like ridge (Fig. 11D). Similarly, the magnetic foliation is oriented out of the plane of intrusion and either dips moderately to the south (102/55°S, S7a; 094/59°S, S7b) or steeply to the east (157/72°E, S7c; 144/85°E, S7d) (Fig. 11D).

Rock Magnetic Experiments

Rock magnetic experiments provide important insights into the magnetic mineralogy of a rock, particularly for fine-grained rocks (e.g., those analyzed here) where traditional petrography is difficult. Samples S3b, S3f, S4c, and S5b were selected for low-field susceptibility versus high-temperature experiments because (1) the S3 samples represent apparently normal magnetic fabrics and allow internal variations in magnetic mineralogy to be assessed (Fig. 7); (2) the S4c magnetic fabric is oblique to the intrusion and could therefore be interpreted as an imbricated fabric or an intermediate or inverse fabric (Fig. 8C); and (3) S5b appears to be an inverse fabric (i.e., the magnetic lineation and foliation are approximately orthogonal to the intrusion plane; Fig. 9). The four samples generally show an increase in susceptibility on heating until a sharp downward deflection (i.e., a Hopkinson peak) occurs at 559 °C (Figs. 12A–12D). Convex-upward bumps are superimposed onto this heating trend for S4c and S5b (Figs. 12C, 12D). For S4c, the shallow bump spans a temperature range of 131–350 °C and attains a maximum susceptibility of 274 SI at 287 °C (Fig. 12C). The prominent bump observed in the S5b heating curve spans 104–393 °C and attains a maximum susceptibility of 823 SI at 289 °C (Fig. 12D). Samples selected for hysteresis analysis apparently represent either normal magnetic fabrics (S1a, S3a, S3f, and S6b) or possible inverse fabrics (i.e., S4a and S4c; Figs. 4, 7, 8, and 10). Hysteresis loops for all samples show steep acquisition, reaching saturation by 0.300 T and yielding moderately narrow-waisted loops consistent with a pseudo-SD grain size. Figure 12E shows that the samples chosen for hysteresis analysis all plot within the pseudo-SD field of a standard Day plot.

INTERPRETATION

Magnetic Fabric Origin

Estimating primary magma flow patterns within ancient sheet intrusions is integral to understanding the transport and accommodation of magma within active subvolcanic systems. Although numerous studies have successfully demonstrated the correlation between primary magma flow and magnetic fabrics (e.g., Callot et al., 2001; Aubourg et al., 2002; Liss et al., 2002; Horsman et al., 2005; Morgan et al., 2008), the interpretation of AMS measurements as reliable flow indicators remains controversial. Before AMS data can be used to interpret magma flow patterns, it is essential to define the magnetic mineralogy (i.e., the carrier of the magnetic signature of the rocks) and the origin of the magnetic fabric (e.g., has it been modified by postemplacement tectonic activity?).

Magnetic Mineralogy

From petrographic analyses and rock magnetic experiments, it was suggested (Magee et al., 2012a, 2013b) that the magnetic signature of inclined sheets in Ardnamurchan is dominated by a low-Ti titanomagnetite phase. The following observations support the dominance of low-Ti titanomagnetite on the magnetic signature of the sheets studied here: (1) relatively high Kmean values of >1.62 × 10–2 SI (Tarling and Hrouda, 1993); and (2) general increases in susceptibility on heating to 559 °C (i.e., the Curie point of each sample) before a rapid decrease upon further heating (Fig. 12); based on the equations of Akimoto (1962) this is consistent with a Ti content of ∼0.039 (see Dunlop and Özdemir, 2001). The bumps observed along the heating curves for S4c and S5b in the low susceptibility versus temperature experiments are typically interpreted to result from the homogenization of two Fe-Ti oxide phases, and likely suggest that titanomaghemite may contribute to the magnetic signature of some samples, although monoclinic pyrrhotite also has a Curie point (320 °C) in this temperature range (see Dunlop and Özdemir, 2001). It was also highlighted in Magee et al. (2013b) that some inclined sheets may contain populations of single-domain magnetite, which could potentially alter the orientation of the magnetic fabric by switching the principal susceptibility axes to form intermediate or inverse magnetic fabrics (cf. Rochette et al., 1999; Ferré, 2002). Figure 12E indicates that all of the samples analyzed are dominated by pseudo-SD titanomagnetite populations, implying that, at least for S1a, S3a, S3f, S4a, S4c, and S6c, the magnetic fabrics can be classified as normal.

Magnetic Fabric Origin

In Magee et al. (2012a, 2013b) it was demonstrated that the shape and distribution of titanomagnetite populations within the Ardnamurchan inclined sheets was controlled by the primary silicate framework. This implies that the magnetic fabrics correlate with the petrofabric of the silicate grains. If the mineral fabrics were generated by magma flow, it is typically expected that K1 and the magnetic foliation will be located within or close to the plane of intrusion (i.e., the magnetic fabrics are normal) and that the magnetic lineation may correspond to the magma flow axis (cf. Knight and Walker, 1988; Rochette et al., 1999; Ferré, 2002). It was argued (Magee et al., 2012a) that magma flow patterns are discernible in inclined sheets on Ardnamurchan by combining the orientation of identified normal magnetic fabrics, particularly magnetic lineations, with measurements of the long axes of visible flow indicators such as intrusive steps, broken bridges, and magma lobe axes. Where similar intrusive steps are observed in the inclined sheets analyzed here (i.e., S4a–S4c, S6, and S7), the orientation of the magnetic lineation is subparallel to that of the step long axes (Figs. 8 and 10). This suggests that the magnetic fabrics can be correlated with magma flow.

However, several alternative options need to be explored when interpreting magnetic fabrics as related to magma flow. For example, many of the inclined sheet intrusions analyzed here were emplaced at relatively shallow levels, but apparently lack chilled margins (see also Magee et al., 2012a, and references therein), implying that either the temperature of the host rock during the emplacement of the inclined sheet swarm was elevated by the local magmatic activity (Day, 1989), inhibiting chilled margin formation, or magma flow within the individual sheets was protracted and instigated melt-back of any chilled margin originally present (e.g., Huppert and Sparks, 1989). It is therefore difficult to discern whether magnetic fabrics correspond to initial propagation or magma flow within a more mature system (e.g., Liss et al., 2002; Philpotts and Philpotts, 2007). Furthermore, it is important to note that measured magnetic lineations and/or magnetic foliations are not always close to the plane of the intrusion (e.g., Figs. 3–6, 8, 9, and 11). Such disparities between the orientation of the intrusion and the magnetic fabrics are commonly interpreted as intermediate or inverse fabrics produced by the presence of a SD titanomagnetite population within a sample (e.g., Potter and Stephenson, 1988; Rochette and Fillion, 1988; Borradaile and Puumala, 1989). It is important that hysteresis experiments demonstrate that S4a and S4c, which record magnetic fabrics that are strongly oblique to the intrusion plane, do not contain SD titanomagnetite populations (Fig. 12E). This implies that apparently intermediate and inverse fabrics cannot necessarily be attributed to complexities in the magnetic mineralogy. Instead, these anomalous magnetic fabrics may result from cyclic crystal behavior during magma flow and/or postemplacement processes (Cañón-Tapia and Herrero-Bervera, 2009). Because Ardnamurchan remained relatively tectonically inactive after the formation of the central complex (Emeleus and Bell, 2005), any postemplacement superimposition of magnetic fabrics would likely have resulted from (1) convection within individual inclined sheets; (2) inflation or deflation of later major intrusions (e.g., the gabbro lopolith); or (3) roof subsidence and intrusion closure, instigated by the waning of magma pressure, within the inclined sheets during the final stages of emplacement.

We consider it unlikely that convection modified most of the magnetic fabrics measured because the majority of sampled sites occur where the inclined sheets have thicknesses <3 m (e.g., Figs. 4, 5, 8, 9, and 11), i.e., heat loss is expected to be relatively rapid, inhibiting convection. If convection did occur in any of the sheet intrusions studied, it may be expected that the thickest intrusion (i.e., the S4e–S4f sample site where inclined sheet thickness increases to 5 m) would record the strongest evidence of convection within the magnetic fabrics. We suggest that if convection were to have occurred in the thicker portions of S4, the associated magnetic fabrics should differ from those measured in thinner sections of the intrusion. However, AMS results for S4 all display magnetic lineations that trend approximately northwest-southeast, parallel to the long axis of an intrusive step (i.e., a visible magma flow indicator) observed near S4a–S4c (Fig. 8). These observations suggest that convection did not modify the magma flow–related petrofabrics.

Deformation of the inclined sheets, induced by either major intrusion growth or roof subsidence, would likely affect entire inclined sheets. We assume that at any one sample site, application of a postemplacement strain capable of modifying petrofabrics will act to homogenize the magnetic fabric orientation, although irregularities in sheet geometry at different sites may promote variations in postemplacement fabrics. Given the subcircular nature of the exposed major intrusions and the arcuate strike of the inclined sheets (Fig. 2), we would expect that any non–magma flow–, compaction-related fabrics should be (1) oblate, with magnetic foliations that parallel intrusion contacts; and (2) typically consistent along the strike of individual inclined sheets. However, the broad range of magnetic fabric orientations measured here (and in Magee et al., 2012a), some of which are not close to the plane of intrusion, suggest that the magnetic fabrics were not formed by postemplacement tectonomagmatic events. Similarly, quantitative textural analyses of several inclined sheets within the Ardnamurchan Central Complex suggest that they have undergone minimal textural equilibration following emplacement (Magee et al., 2013a). Given the lack of evidence for postemplacement fabric modification, as well as the observed parallelism between magnetic lineations and field flow indicators (e.g., Figs. 8 and 10; Magee et al., 2012a), we suggest that the magnetic fabrics dominantly record primary magma flow. Through the integration of magnetic fabric analyses and structural field observations, the following subsections outline the interpretation of the emplacement of the individual sheet intrusions studied.

S1

Regardless of AMS sample location, the magnetic fabrics within S1 are weakly to strongly prolate (–0.169 to –0.839) and K1 gently plunges (∼21°) northwest-southeast (∼134°–314°), subparallel to sheet strike (Fig. 4B). If it is assumed that the magnetic lineation reflects the axis of primary magma flow, the measured K1 would imply that magma within S1 either flowed toward the northwest or southeast, along the strike of the sheet. However, the magnetic foliations display variable orientations, although the majority strike subparallel to S1, and at S1a–S1d define an imbrication suggestive of a southwest-directed magma flow pattern. In contrast, the magnetic foliations derived from S5e–S5h do not display a clear imbrication pattern, but rather describe a progressive rotation from southeast-inclined magnetic foliations at the sheet base to moderately inclined northeast-dipping foliations near the top.

There are a number of interpretations that may be invoked to explain these observed complexities in the magnetic fabrics. Although S1a does not contain an SD titanomagnetite population, we cannot rule out the possibility that magnetic fabrics recorded for other profiles within S1 are intermediate or inverse (cf. Rochette et al., 1999; Ferré, 2002). Two alternative mechanisms for generating different magnetic foliations via variations in primary magma flow dynamics may also be considered. First, several studies have highlighted that different magnetic fabrics may be recorded at intrusion margins, particularly those that are chilled, compared to within the core of the sheet (e.g., Liss et al., 2002; Philpotts and Philpotts, 2007). This is because chilled margins are likely to record sheet initial propagation fabrics and high simple shear gradients, while intrusion cores could preserve either regional magma flow patterns, different magma pulses, or convection in a relatively mature conduit (Liss et al., 2002). We consider it unlikely that the magnetic fabrics measured relate to differences in the style of fabrics recorded at the margins and the core because the chilled margin at S1 is <1 cm thick and therefore below our resolution of sampling (AMS cores are 2.5 cm in diameter). An alternative explanation concerns the common assumption that magma flow remains uniform along the strike of the magma flow direction (e.g., Callot et al., 2001; Correa-Gomes et al., 2001; Féménias et al., 2004). In Magee et al. (2013b) it was suggested that sheet intrusions may be internally compartmentalized, implying that magma flow patterns could vary laterally within individual inclined sheets. Such compartmentalization could be associated with the observation that sheet intrusions are typically emplaced initially as a series of thin, discrete segments, which only coalesce upon continued magma input (see Schofield et al., 2012b, and references therein). Any minor variations in the rheology and/or flow temperature of these discrete segments could promote subtle differences in their magma flow dynamics, which may be maintained upon coalescence and effectively compartmentalize the sheet intrusion (Magee et al., 2013a). In particular, lateral variations in magma flow dynamics would likely produce zones of relatively high-velocity gradients that are orthogonal to intrusion contacts. Figure 13 is a schematic diagram based on the magnetic fabric data from S1 that illustrates a potential interpretation of the spatial variations in magnetic fabrics, in light of the discussion here.

S2

Many sheet intrusions observed in field (e.g., S2) and seismic reflection data have a ramp-flat morphology, i.e., whereby an inclined sheet transgresses stratigraphy before eventually becoming strata concordant as a bedding plane or weak lithology is exploited (e.g., Thomson and Schofield, 2008; Magee et al., 2012a, 2012, 2014). Commonly, the inclined sheets are fed via sills, although this can be difficult to corroborate in the field. It is important to note that these ramp-flat structures are not related to intrusive steps and that magma flow is expected to be close to parallel or parallel to the dip direction of the inclined sheet portion. For S2, this sheet geometry would imply that magma flowed from the northwest to the southeast (i.e., dip parallel overall), consistent with the trend of the measured magnetic lineations (Figs. 5 and 6).

S3

A magnetic analysis of a diabase dike was conducted to provide a comparison with the inclined sheets examined. Magnetic lineations and foliations are all located within the plane of intrusion, with K1 primarily being subvertical (Fig. 7). The exception to this is S3a, where K1 plunges 144/26° (Fig. 7A), but this value may not be reliable due to the strongly oblate nature of the magnetic fabric (T = 0.79) and the spread of observed specimen data. Subtle variations in the magnetic foliation, including S3a, define an imbrication that opens downdip (Fig. 7). Overall, the magnetic fabrics are consistent with an upward-directed magma flow (i.e., dip parallel), slightly offset from vertical toward the southeast. A magma flow origin of the magnetic fabric could be further supported if the decrease in the oblateness of the magnetic fabrics toward the core of the S3e traverse is assumed to relate to the increased friction between magma and host rock toward intrusion contacts, which generates a high-velocity gradient and oblate fabrics (Féménias et al., 2004). Alternatively, the margins of S3 may preserve fabrics from an initial period of higher flow strength compared to the core, which could host magnetic fabrics related to a later phase of decreasing magma flow. Similar fabric variations may not be observed in S3b because (1) the sample spacing could be too coarse, or (2) the increased width of the intrusion (i.e., 3 m relative to 1.5 m at S3e) may not be conducive to the preservation of the full velocity profile. It is, however, difficult to determine the process driving the recorded magma flow, e.g., whether the magnetic fabric related to emplacement or subsequent convection.

S4

The along-strike variation in the dip of S4 (∼7°–58°) can be considered a primary emplacement feature because there is no associated change in bedding orientation (Fig. 8) that would be indicative of subsequent tilting. Sheet thickness is also observed to range from ∼2 to 5 m. Despite this variation in sheet geometry, encompassed by the three sites targeted for AMS, there is little systematic change in the magnetic fabric (orientation, shape, or strength of anisotropy; Fig. 8). For example, with the exception of S4a, which displays a steep magnetic lineation and a weakly defined magnetic foliation, K1 axes plunge northwest at 33°–59° and parallel the long axis of an intrusive step (Fig. 8C). Magnetic foliations are consistently oriented at a high angle to the sheet dip and also occasionally to the intrusion strike (Fig. 8C). Because K1 remains in the same approximate position throughout the samples, the orientation of the magnetic foliation is controlled by the K2 axis, which appears to switch with K3 (Fig. 8C). These deviations in the magnetic foliation orientation may relate to (1) complex and localized variations in magma flow dynamics within a single intrusion (e.g., Fig. 13); (2) the sampling of different magma pulses with differing magma flow patterns (Liss et al., 2002); or (3) the occurrence of a sufficient proportion of SD magnetite, in samples other than S4a and S4c (Fig. 12E), to produce mixed fabrics, as discussed here (cf. Rochette et al., 1999; Ferré, 2002). Although adequate information to distinguish between these hypotheses is lacking, the parallelism between the magnetic lineations, sheet dip direction, and the orientation of an intrusive step long axis implies that magma flow can still be elucidated (at least locally in the sheet) and was dip parallel. The AMS sample analyzed in Magee et al. (2012a) from the northern exposure limit of S4 (i.e., their CSJ1) is parallel to the fabric described from within S4d (Fig. 5C).

S5

The parallelism between the magnetic lineations and the dip direction of S5 suggest that emplacement may have occurred in a northwestward or southeastward direction (Fig. 9). Unfortunately there is not enough information to determine if the magnetic foliations, which are moderately to steeply dipping and strike parallel to the intrusion dip direction, reflect variations in primary magma flow patterns or the development of intermediate and/or inverse magnetic fabrics.

S6

Throughout S6 the AMS data are remarkably homogeneous (Fig. 10). The triaxial fabric ellipsoids consistently display a K1 axis oriented subparallel to the dip and dip direction of the inclined sheet (083/30°N strike and dip) as well as the orientation of a minor intrusive step (∼158°–338° bearing) (Fig. 10). Although S6 thins to the east of the sample site from 2 to 1 m, a morphological feature often inferred as a proxy for the magma flow direction (i.e., sheet intrusions are expected to thin toward their propagating tip; e.g., Hansen et al., 2011), the magnetic fabrics and intrusive step suggest that the magma flow axis was dip parallel (i.e., oriented north-northwest–south-southeast). Thus, intrusion thinning may here be related to increasing proximity toward the lateral tip of the intrusion.

S7

The four discrete outcrops composing S7 are considered to represent a single intrusion because they are petrologically similar and display a consistent approximately northeast-southwest strike and northward inclination (∼30°) (Fig. 11). Apparent lobe-like elongations developed to the northwest of the individual outcrops, distinguished by subtle topographic changes and the presence of small diabase outcrops, and the intervening topographic troughs may reflect either postemplacement erosion, or are a primary morphological feature (Figs. 11A, 11B). Chilled margins and increasing amygdale abundance toward the upper, lower, and lateral contacts of each lobe-like segment support an emplacement-related origin to the outcrop pattern observed. Similar magma lobe geometries have been described from the transgressive, inclined rims of saucer-shaped sills observed both in the field (e.g., Polteau et al., 2008; Schofield et al., 2010) and in seismic reflection data (Thomson and Hutton, 2004; Schofield et al., 2012a; Magee et al., 2013b). These studies have shown that magma lobes form through the coalescence of magma fingers, i.e., thin, elongated magma conduits with an elliptical cross section that may be emplaced in a nonbrittle fashion in response to intrusion-induced host-rock fluidization (Schofield et al., 2012b). Internal variations in the thickness of the S7 segments are consistent with the growth of magma lobes through the amalgamation of inflating magma fingers. Schofield et al. (2012b) described similar magma fingers in a diabase inclined sheet intrusion located ∼300 m to the west of S7 and emplaced into a succession of Neoproterozoic Moine Supergroup metasedimentary rocks and Paleogene volcaniclastics and olivine-basalt lavas. The magma fingers are only observed within the poorly consolidated lavas and volcaniclastics, where intrusion-induced collapse of the host rock pore space accommodated the magma volume and promoted nonbrittle emplacement (Schofield et al., 2012b). It seems plausible that similar processes may have controlled the intrusion of S7 into the olivine-basalt lavas. Long axes of magma lobes and fingers can be used as a proxy for the primary magma flow axis (Schofield et al., 2012b). The northwestward elongation of the magma lobes and fingers documented here therefore implies a dip-parallel, northwest-southeast–oriented magma flow axis (Fig. 11A). Given the radial disposition of the four S7 outcrops, i.e., their long axes rotate from 166°–296° in the northeast to 161°–341° in the southwest, it is suggested that magma was fed from the northwest (Fig. 11A). Figure 11 highlights that the projected source position corresponds to the location of a northeast-southwest–trending monocline in the olivine-basalt lavas. This monocline might be the manifestation of roof uplift and forced folding above a tabular intrusion from which the S7 magma lobes emanated. This model and the S7 field observations are reminiscent of magma lobe structures described from the inclined limbs of saucer-shaped sills, where transgression was promoted by fracturing or fluidization of the host rock at points of maximum flexure on the fold (Thomson and Schofield, 2008; Schofield et al., 2010).

Considering the possibility that S7 represents the southern inclined limb of a saucer-shaped sill centered to the north, it is apparent that the visible magma flow indicators (i.e., magma lobe and finger long axes) are not corroborated by the AMS results presented here (Fig. 11) or those in Magee et al. (2012a, CS111–CS115 therein). The model proposed requires an upward and outward magma flow pattern, implying K1 should plunge to the northwest and be located within the plane of intrusion. Regardless of the sample position, Figure 11D reveals that K1 is located near the normal to the intrusion plane. Similarly, magnetic foliations strike subparallel to the sheet intrusion dip direction and are nearly orthogonal to the intrusion plane (Fig. 11D). These measurements imply that the magnetic fabrics do not correspond to the primary magma flow pattern and may instead reflect an inverse or unstable magnetic fabric (cf. Rochette et al., 1999; Ferré, 2002; Cañón-Tapia and Herrero-Bervera, 2009).

DISCUSSION

Our results show that integrated analyses combining AMS, rock magnetic experiments, and structural field observations allow inferences about magma flow patterns to be made. An important observation emanating from this study is that localized internal variations in the magnetic fabrics of inclined sheet intrusions may result from perturbations in the primary magma flow and are strongly controlled by sheet geometry. In particular, thinner sheet intrusions appear to display more uniform magnetic fabrics relative to thicker intrusions. This may be because (1) chilled margins, which record the initial sheet propagation (e.g., Liss et al., 2002; Philpotts and Philpotts, 2007), form a greater bulk of thinner intrusions; (2) particle rotation and cyclicity during magma flow may be inhibited (see Cañón-Tapia and Chavez-Alvarez, 2004); or (3) thicker intrusions may be composed of multiple magma pulses, each of which may contain subtly different mineralogies or magma flow patterns, or allow convection. The emplacement of subsequent magma pulses may additionally superimpose inflation-related subfabrics onto earlier, subsolidus intrusive phases.

It is also important to consider how magma flow patterns may vary along strike. Many sheet intrusions are not emplaced as long, continuous bodies but rather form through the coalescence of discrete magmatic segments (e.g., Fig. 13A) (see Schofield et al., 2012b, and references therein). If these individual segments become isolated following coalescence, perhaps due to the presence of internal chills or rheological boundaries, continued magma flow will therefore be influenced by high velocity gradients, not just at the major intrusion margins but also at the lateral contacts (e.g., Fig. 13A) (Magee et al., 2013a). The imbrication of magnetic foliations may be more complex than previously considered. Such an internal compartmentalization of sheet intrusions may compromise lateral mixing of magma or crystal populations (Magee et al., 2013a). Our results show that information pertaining to primary magma flow and inclined sheet emplacement can be elucidated given a thorough consideration of fabric relationships, magnetic mineralogy, and field observations.

Ardnamurchan Inclined Sheet Emplacement

The Ardnamurchan and Mull Central Complexes host the archetypal examples of cone sheet intrusions. Cone sheets have a subconcentric strike and dip inward toward a central source (Bailey, 1924; Richey and Thomas, 1930; Anderson, 1936; Phillips, 1974; Schirnick et al., 1999) from which the initial fracture and infilling magma are expected to propagate upward and outward (i.e., K1 should be dip parallel) (Herrero-Bervera et al., 2001; Geshi, 2005; Palmer et al., 2007; Magee et al., 2012a). With the exception of S3, which likely represents a regional dike, the inclined sheets treated here have all previously been attributed to the cone sheet swarm on Ardnamurchan (Richey and Thomas, 1930; Emeleus, 2009; Burchardt et al., 2013).

Our results indicate that the inclined sheets studied across the southern portion of Ardnamurchan, excluding the S3 regional dike, are predominantly characterized by dip-parallel, northwest-southeast magma flow axes (i.e., S2, S4-S7; Fig. 14). The exception to this trend is S1, in which magma either flowed toward the southwest (i.e., dip parallel) or northwest-southeast (i.e., strike parallel), depending on whether magnetic foliation imbrication or magnetic lineation trends, respectively, are used to define the magma flow pattern. These observations generally support the findings of Magee et al. (2012a), who noted that northwest-southeast–oriented magnetic lineations dominated the Ardnamurchan inclined sheets (Fig. 14). From the 69 inclined sheets that were regarded (Magee et al., 2012a) as hosting reliable AMS fabric measurements, dip-parallel magma flow axes were only interpreted for 12 inclined sheets, with the other 57 displaying strike-parallel magma flow patterns. These latter strike-parallel magnetic lineations were considered to reflect lateral magma flow along the inclined sheets, sourced from a reservoir external to the Ardnamurchan Central Complex; dip-parallel magma flow patterns were inferred to be fed from a central source beneath Ardnamurchan (Magee et al., 2012a). In this study, magma flow directions could only be inferred from S2 and S7, with both suggestive of a source to the northwest of the sampled exposures. Of these two inclined sheets, only the magma flow direction data for S2 is consistent with being fed from a central source within the Ardnamurchan Central Complex (Richey and Thomas, 1930; Burchardt et al., 2013). The S7 intrusion appears to form part of a saucer-shaped sill, the source of which remains unknown.

Although the scope of this high-resolution magnetic fabric study is insufficient to determine whether the majority of inclined sheets were fed from a central source within the Ardnamurchan Central Complex (Burchardt et al., 2013) or an external reservoir (e.g., the Mull Central Complex; Magee et al., 2012a), we note that (1) little, if any, postemplacement modification of the magnetic fabrics has occurred; (2) consistent northwest-southeast–trending magnetic lineations and variable magnetic foliation orientations imply that the AMS fabrics likely correlate to primary magma flow; and (3) inferred magma flow axes may be dip or strike parallel to the inclined sheet, indicative of both updip and lateral magma flow patterns, respectively. Burchardt et al. (2013) argued that lateral magma flow patterns inferred from inclined sheet AMS data (e.g., Magee et al., 2012a; this study) could be produced via the vertical translation of magma from a central source if a helical flow regime dominated inclined sheet emplacement. The only documented occurrence of helical flow concerns a composite, cylindrical pluton and is attributed to magma mixing (Trubač et al., 2009). However, for the Ardnamurchan inclined sheets, such a magma flow pattern requires that the sheets are fully concentric along strike, a geometry that is not consistent with geological maps or first-order field observations of the Ardnamurchan Central Complex that reveal that the vast majority of inclined sheets (i.e., those not crosscut by major intrusions) only extend along strike 1–2 km but typically <100 m (Fig. 1) (Richey and Thomas, 1930; Emeleus, 2009). It is also important to note that the inclined sheets are represented diagrammatically on the geological map of Ardnamurchan (see Richey and Thomas, 1930, p. 173 therein), i.e., the mapped inclined sheet traces and dip values utilized by Burchardt et al. (2013) are local averages that have been extrapolated. Overall, our observations and interpretations here support the conclusion in Magee et al. (2012a) that inclined sheets on Ardnamurchan were sourced from magma reservoirs both central and external to the central complex. Petrological and geochemical (isotopic) analyses are required to further test this hypothesis.

Field observations reveal that the inclined sheets are geometrically complex and typically display significant variations in the strike and dip of individual intrusions (see also Richey and Thomas, 1930; Kuenen, 1937; Magee et al., 2012a, 2013a). These observations and the magnetic fabric analysis imply that the majority of sheet intrusions on Ardnamurchan may have a different downdip extension to that previously envisaged, i.e., they do not converge upon a central source reservoir (e.g., S7), and that magma was sourced externally to the Ardnamurchan Central Complex (Magee et al., 2012a). These ideas highlight the danger in assuming that the dips of inwardly inclined sheets can be projected downward to infer magma chamber source locations (Richey and Thomas, 1930; Burchardt and Gudmundsson, 2009; Burchardt et al., 2013), although there are several examples where additional data (e.g., magma flow indicators) suggest that this approach may be applicable for constraining source characteristics (e.g., Geshi, 2005). However, it is clear from field observations elsewhere (e.g., Burchardt, 2008; Tibaldi and Pasquarè, 2008; Muirhead et al., 2012; Schofield et al., 2012b) and seismic reflection data (e.g., Thomson and Hutton, 2004; Planke et al., 2005; Magee et al., 2014) that the orientation of an intrusion at a specific level of exposure does not necessarily reflect that of the entire sheet, questioning the accuracy of models that are solely reliant on the planar projection of surficial strike and dip averages.

CONCLUSIONS

The analysis of ancient sheet intrusions exposed at the surface provides crucial insights into the emplacement mechanisms and magma flow patterns of active subvolcanic plumbing systems. We employed anisotropy of magnetic susceptibility (AMS) to examine magnetic fabrics within a suite of seven inclined sheet intrusions located on Ardnamurchan, northwest Scotland. Despite a broad variation in the orientation of studied sheet intrusions, magnetic lineations predominantly trend northwest-southeast and have shallow to moderate plunges. Magnetic foliations within individual intrusions display more variation in their orientation and are not necessarily subparallel to the plane of intrusion. Through the integration of AMS, rock magnetic experiments, and structural field observations, we demonstrate (1) that the magnetic signature is dominated by low-Ti titanomagnetite populations, which commonly have a pseudo-SD grain size; (2) the measured magnetic fabrics are complex and variable within individual intrusions; (3) little postemplacement modification of the magnetic fabrics has occurred; and (4) the magnetic fabrics likely reflect primary magma flow. By considering the magnetic fabric orientation and their location within each intrusion, we show that inferred magma flow axes for at least five intrusions are typically dip parallel and oriented northwest-southeast. One intrusion potentially displays evidence for strike-parallel magma flow directed toward the southwest. Our results suggest that magma flow dynamics within individual intrusions can vary laterally, promoting the development of magma lobes that can effectively internally (petrologically) compartmentalize seemingly continuous sheets. This has important implications for understanding the channelization of magma within sheet intrusions, which can affect eruption locations and magma mixing trends.

We thank Trevor Potts for providing accommodation during the field campaign, which was funded by National Geographic Grants in Aid of Research award 8106-06 to Petronis. The cored drill holes produced over the course of this study were subsequently infilled under the guidance of Scottish National Heritage. We are grateful to Nobuo Geshi and two anonymous reviewers for their constructive comments. We thank Tom Stone for his preliminary analysis on some of these samples for his Master of Science degree project.