We ally aeromagnetic interpretation with constrained three-dimensional (3D) gravity inversion over the Musgrave Province in central Australia to produce a 3D architectural and kinematic model of the ca. 550 Ma compressional intraplate Petermann orogeny. Our model is consistent with structural, metamorphic, and geochronological constraints and crustal-scale seismic models. Aeromagnetic interpretation indicates that divergent thrusts at the margins of the province are cut by transpressional shear zones that run along the axis of the orogen. Gravity inversion indicates that the marginal thrusts are crustal-scale and shallow-dipping, but that the transpressional shear zones of the axial zone are more steeply dipping, and penetrate the crust-mantle boundary, accommodating offsets of 10–25 km. This thick wedge of mantle within the lower crust has been in isostatic disequilibrium for more than 500 Ma, and we suggest that this load may be supported by local lithospheric strengthening resulting from the emplacement of relatively strong lithospheric mantle within the relatively weak lower crust. Other orogenic processes inferred from the model include: probable inversion of relict extensional architecture; crustal-scale strain partitioning leading to strain accommodation by the vertical and lateral extrusion of relatively undeformed crustal blocks; and escape tectonics directed toward the relatively free eastern margin of the orogen. These processes are consistent with the concept that mechanical and thermal heterogeneities in the lithosphere, and the resulting feedbacks with deformation, are the dominant controls on intraplate orogenesis. This model also demonstrates that the architecture and kinematics of the Petermann orogeny require modification of leading models of Gondwana assembly.

The architecture and kinematics of orogenic belts have been the topic of many studies, mostly focused on actively deforming or recently deformed plate margins, such as the Himalaya (e.g., Molnar, 1988; Yin, 2006) or the European Alps (e.g., Bruckl et al., 2007; Ebbing et al., 2001; Luschen et al., 2004). In contrast, compressional intraplate orogens have been the subject of comparatively few studies.

To date, studies of the structure and kinematics of intraplate compression have concentrated on the inversion of extensional basins in both backarc-hinterland and forearc-foreland settings (e.g., Dickerson, 2003; Sandiford, 1999; Turner and Williams, 2004; Ziegler et al., 1995), with others studying the dynamic evolution of the currently active intraplate compressional orogens of the Tien Shan (Burov et al., 1990; Tapponnier and Molnar, 1979; Zhao et al., 2003) and Altai (Cunningham, 2005) regions of central Asia. These studies highlight the diverse range of settings for intraplate compressional deformation, and also the variety of lithospheric processes that can occur. However, one finding that is common to all studies is the importance of thermal and mechanical heterogeneities in the continental lithosphere as a control on crustal architecture (e.g., Cunningham, 2005; Dickerson, 2003; Hand and Sandiford, 1999; Sandiford and Hand, 1998; Ziegler et al., 1998).

Late Neoproterozoic to Devonian tectonic reworking of central Australia is interpreted to reflect intraplate compressional orogenesis (e.g., Betts et al., 2002; Camacho and Fanning, 1995; Camacho et al., 2002; Hand and Sandiford, 1999; Sandiford and Hand, 1998). Two discrete orogens are recognized: the ca. 600–500 Ma Petermann orogeny that reworked the Mesoproterozoic Musgrave Province (e.g., Major and Conor, 1993; Wade et al., 2008) and the ca. 450–350 Ma Alice Springs orogeny that reworked the Paleoproterozoic Arunta inlier (e.g., Collins and Shaw, 1995; Sandiford, 2002).

The initiation of these orogens has been the topic of much study, and several quite different models may explain the initiation and some of the main features of these orogens (cf. Braun and Shaw, 2001; Camacho et al., 2002; Neil and Houseman, 1999; Neves et al., 2008; Sandiford et al., 2001). A detailed regional-scale model of the 3D architecture and kinematics of these orogens is lacking. This is important because it may indicate the orientation and intensity of the forces driving the system and characterize the feedback processes that control the interaction between crustal architecture and the dynamics of the orogen.

In addition, these orogens provide a record of intraplate continental lithospheric deformation under the influence of one of Earth's most dramatic periods of tectonism associated with the assembly of Gondwana, and their architecture and kinematics may help to recognize the most (and least) credible of many competing tectonic models (e.g., Boger and Miller, 2004; Cawood, 2005; Collins and Pisarevsky, 2005; Jacobs and Thomas, 2004; Meert and Van Der Voo, 1997; Meert, 2003; Rino et al., 2008; Veevers, 2003).

A combination of aeromagnetic data and gravity data can be used to image architecture from the near surface to crust-mantle boundary geometry (e.g., Williams and Betts, 2007), and therefore these data provide the ideal tool to unify the concepts of previous studies of orogenic architecture at multiple scales. In this paper, we combine interpretation of high-resolution aeromagnetic data with 3D gravity inversions to produce a crustal-scale model of the architecture and kinematics of the intra-plate Petermann orogeny in the eastern Musgrave Province. This model is constrained by geological observations at a number of scales, including pressure-temperature-time (P-T-t) data, structural interpretations, petrophysical sampling, and macro-scale geological observations, and constraint is also derived from crustal-scale seismic reflection lines and passive seismic models. From this architectural and kinematic model, we infer the most influential lithospheric processes that have shaped, and been controlled by, the architecture and kinematics of the Petermann orogeny.

The Musgrave Province preserves a variety of Mesoproterozoic gneissic rocks of dominantly felsic lithology with precursors dated at ca. 1600 Ma (Gray, 1978; Wade et al., 2006) that were metamorphosed at amphibolite to granulite facies during the ca. 1200 Ma Musgravian orogeny (Camacho and Fanning, 1995; Gray, 1978; Maboko et al., 1991; Sun and Sheraton, 1992; White et al., 1999). Although, in outcrop, the structural trend of this event is variable throughout the province (cf. Aitken et al., 2008; Aitken and Betts, 2009b; Clarke et al., 1995; Edgoose et al., 2004; Major and Conor, 1993), linking these observations to a coincident structural grain in aeromagnetic data defines this structural trend at the regional scale and shows that it is dominantly northeast trending (Aitken et al., 2008; Aitken and Betts, 2009b). The emplacement of the granitic plutons of the Pitjantjatjara Supersuite occurred during and shortly after this orogeny (Major and Conor, 1993), and their emplacement pattern dominantly reflects the northeast-trending structural grain of the Musgravian orogeny (Aitken et al., 2008; Aitken and Betts, 2009a, 2009b; Edgoose et al., 2004; Major and Conor, 1993). These chains of magnetic granitoids are interpreted to be continuous beneath the Amadeus and Officer basins, defining the extent of the ca. 1200 Ma Musgravian-Albany-Fraser orogeny (Aitken and Betts, 2008).

Subsequent to the Musgravian orogeny, the voluminous mafic intrusions of the Giles Complex and coeval mafic dikes and granitoids were emplaced within the Musgrave Province during the extensional Giles event at ca. 1080 Ma (Glikson et al., 1995; Sun et al., 1996), along with surficial volcanic rocks now exposed at the margins of the Musgrave Province (Glikson et al., 1995). Although not well defined, the extent and orientation of this event may be defined by east-to-east-southeast–trending shear zones that predate or are synchronous with the dike emplacement events (Aitken et al., 2008; Aitken and Betts, 2009b; Clarke et al., 1995); the alignment of Giles Complex mafic intrusions along the axis of the Musgrave Province with no geophysical evidence for buried Giles Complex plutons outside of this zone (Glikson et al., 1995; Glikson et al., 1996); and the orientation and extent of the Warakurna large igneous province (LIP), of which the Giles Complex is a key constituent, which extends from northern Western Australia to the Musgrave Province (Wingate et al., 2004).

After a hiatus of ca. 200 million years, mafic dikes were emplaced at ca. 800 Ma along east-southeast– and southeast-oriented structures (Zhao et al., 1994). The inception of the Officer and Amadeus basins is broadly contemporaneous with these dikes, and probably formed as part of the once contiguous Centralian Superbasin (Walter et al., 1995). This ca. 800 Ma extensional event may represent a northwest-trending aulacogen, related to a mantle plume centered beneath the Adelaide Rift Complex to the southeast (Betts et al., 2002; Zhao et al., 1994). A second hiatus of ca. 200 million years followed this event, before the Musgrave Province was intensely reworked during the ca. 550 Ma Petermann orogeny.

Although locally derived vertical driving forces may have played a significant role (Neil and Houseman, 1999), the Petermann orogeny is typically interpreted to represent the intraplate response to far-field stresses related to Gondwana assembly. Due to the uncertainties regarding the assembly of Gondwana, defining a specific tectonic driver for this event is not straightforward, and several major orogens may have contributed to the stress field. The closest active plate-margin orogen may be found in the ca. 570–530 Ma collision of India with Australia's western margin, termed the Kuunga orogeny (Collins and Pisarevsky, 2005; Meert et al., 1995; Meert and Van Der Voo, 1997; e.g., Meert, 2003). This orogeny was originally interpreted to represent the ca. 570–530 continent-continent collision of Gondwana, resulting in the suturing of Australia, east Antarctica, and the Kalahari craton, onto the remainder of Gondwana, which was previously assembled during the ca. 750–620 Ma east African orogen (Meert, 2003). However, a lack of accreted arc fragments or continental blocks and the limited extent of its component terranes led Squire et al. (2006) to suggest that the Kuunga orogeny is an intracratonic response to the East African–Antarctic orogen, which is interpreted by several authors to record the major event in the amalgamation of Gondwana (Jacobs and Thomas, 2004; Stern, 1994). A third hypothesis for Gondwana assembly recognizes the dominance of transpressional orogenic belts, and proposes that oblique subduction along the Pacific margin of Gondwana from ca. 560 Ma onwards led to continental blocks becoming a “counter-rotating cog” in Gondwana (Veevers, 2003).

The localization of strain in the Musgrave Province during the Petermann orogeny has been the subject of some discussion. An early model suggested thermal blanketing of an upper crust high in heat-producing elements by the thick sediments of the Centralian Superbasin as a mechanism to create anomalously weak lithosphere beneath the deepest part of the basin, which was interpreted to overlie the Musgrave Province (Hand and Sandiford, 1999; Sandiford and Hand, 1998). This model has since been disputed on the grounds of an emergent Musgrave Province as the source for detrital zircons in ca. 700 Ma to ca. 500 Ma Amadeus Basin sedimentary rocks (Camacho et al., 2002), and the possibility of high heat production in the lithospheric mantle as a mechanism for strain localization has also been raised (Neves et al., 2008). Other alternatives have been suggested to drive this orogenesis, including Rayleigh-Taylor instability of the lithospheric mantle (Neil and Houseman, 1999) and strain localization at the interface between regions of contrasting mechanical strength, often related to the weakening effects of previous deformation events (Braun and Shaw, 2001; Camacho et al., 2002).

In outcrop, Petermann orogeny deformation is characterized by the development of mylonite, ultramylonite, and pseudotachylite zones, varying from a few meters in width to several kilometers (Clarke et al., 1995; Edgoose et al., 2004). The primary structure in outcrop, the Woodroffe thrust, is a shallowly south-dipping mylonite zone with an apparent thickness of up to 3 km, and a strike length greater than 500 km (Fig. 1). The Woodroffe thrust forms the boundary between the Fregon subdomain to the south, which is dominated by granulite-facies metamorphic rocks, and the Mulga Park subdomain to the north, which contains amphibolite-facies metamorphic rocks (Camacho and Fanning, 1995; Maboko et al., 1992). Within the Fregon subdomain, several other major shear zones are recognized in outcrop, including the Mann fault, Marryat fault, Ferdinand fault, and Hinckley fault; however, many more that are not observed at the surface due to extensive cover successions are evident in the aeromagnetic data (Fig. 1). As well as defining major metamorphic grade transitions, Petermann orogeny shear zones also delineate the margins of the Levenger and Moorilyanna grabens, interpreted as syntectonic transtensional grabens that formed during the Petermann orogeny (Gravestock et al., 1993; Major and Conor, 1993).

The Musgrave Province records little tectonic activity subsequent to the Petermann orogeny, with deformation being restricted to infrequently observed low metamorphic grade shear zones, thought to be related to the Alice Springs orogeny (Edgoose et al., 2004; Major and Conor, 1993). In contrast, the Officer and Amadeus basins record extensive deformation and subsidence during the 450–350 Ma Alice Springs orogeny (e.g., Haddad et al., 2001; Hoskins and Lemon, 1995; Lindsay, 2002), including a major thrust complex at the southern margin of the Musgrave Province that has deformed the Ordovician to Devonian strata of the Officer Basin (Lindsay and Leven, 1996).

Geochronological and Metamorphic Studies of the Petermann Orogeny

Geochronological and metamorphic studies relevant to the Petermann orogeny have focused on defining the evolution of Petermann orogeny shear zones and the contrast in crustal blocks across them. These studies have been undertaken in three regions: the Musgrave Ranges, the Tomkinson Ranges, and the Mann Ranges (Fig. 1).

In the Musgrave Ranges (Fig. 1), geochronological studies indicate that the Woodroffe thrust was active during the late Neoproterozoic to Early Cambrian (560–525 Ma) (Camacho and Fanning, 1995; Maboko et al., 1992). The similar geochronological evolution either side of the Woodroffe thrust is interpreted to indicate that the metamorphic grade difference across this shear zone reflects differing crustal levels of the same terrane (Camacho and Fanning, 1995). Geochronologically constrained P-T data defined five metamorphic events for this region. The first four reached up to granulite facies and may represent the ca. 1200 Ma Musgravian orogeny (Maboko et al., 1991). The fifth metamorphic event is characterized by muscovite development in mylonite zones, and occurred under greenschist-facies conditions (~4 kbar, <400 °C) at 540 ± 10 Ma (Maboko et al., 1991). In contrast to this greenschist-facies metamorphic event, P-T estimates from the Davenport shear zone, located ~10 km south of the Woodroffe thrust, contain evidence for a subeclogite-facies event (~12 kbar, ~650 °C) dated at 547 ± 30 Ma (Camacho et al., 1997; Ellis and Maboko, 1992). A geodynamic model based on these P-T data proposes crustal thickening in the early stages of the Petermann orogeny, before exhumation begins, progressing to a crustal-scale flower structure (Camacho and McDougall, 2000).

A similar evolution is observed in the Tomkinson Ranges in the western Musgrave Province (Fig. 1), where geochronologically constrained P-T data indicate several late Mesoproterozoic granulite-facies events followed by the development of ultramylonite and pseudotachylite zones, and a metamorphic overprint at up to eclogite facies (14 ± 1 kbar, 700–750 °C). These rocks were subsequently overprinted by mica-rich retrograde shear zones (Clarke et al., 1992, 1995). Although not radiometrically dated in this locality, these shear zones connect into the regional network of major Petermann orogeny shear zones, and are interpreted to be Petermann orogeny aged (Clarke et al., 1995).

Analysis of Petermann orogeny overprints in the Mann Ranges (Scrimgeour and Close, 1999) showed that metamorphosed granites north of the Woodroffe thrust in the Mulga Park subdomain underwent amphibolite-facies metamorphism (6–7 kbar, 650 °C) during the Petermann orogeny, whereas in mylonites immediately across the Woodroffe thrust granulite-facies conditions were observed (9–10 kbar, 700 °C) and, ~40 km farther south, subeclogite-facies conditions were observed (12–13 kbar, 700–750 °C). These subeclogite-facies mylonites are cut by high metamorphic grade migmatitic shear zones that have been U-Pb sensitive high-resolution ion microprobe (SHRIMP) dated at 560 ± 11 Ma (Scrimgeour et al., 1999). These were then cut by mylonites at amphibolite facies (7 ± 2 kbar, 660 ± 50 °C). These overprinting relationships are interpreted to reflect the exhumation of the province from subeclogite facies to amphibolite facies during the Petermann orogeny (Scrimgeour and Close, 1999).

These sharp transitions in crustal level across shear zones indicate the juxtaposition of crustal blocks, within which P-T estimates can be fairly consistent (Scrimgeour and Close, 1999). This is echoed in the mineralogy of igneous rocks from the Giles event, which shallow southwards across sharp, shear zone–related transitions. Ultramafic plutons in the northern Fregon subdomain were emplaced at ~6 kbar (Clarke et al., 1995). Moving south, coeval rocks show a transition through gabbropyroxenite, troctolite, and ultimately surface volcanics at the margins of the province (Glikson et al., 1995). This indicates that since ca. 1080 Ma, the northern Fregon subdomain has been uplifted by ~20 km relative to the margins of the province.

The Architecture of the Petermann Orogeny

Shear zones are important in defining the kinematics and architecture of the Petermann orogeny; however, very little work has been done to define the architecture and kinematics of these shear zones, either in the near surface or at depth. Structural studies (Clarke et al., 1995; Edgoose et al., 2004; Flottmann et al., 2005) have defined the kinematics of some of these shear zones in localized areas, although the lack of a regional framework makes these results difficult to integrate with the lithospheric-scale architecture. Models of the lithospheric-scale architecture of the Musgrave Province based on passive seismic data are characterized by steep lithospheric-scale shear zones, correlated with the Mann fault, Wintiginna lineament, and Lindsay lineament, that define an upwards Moho offset beneath the central Musgrave Province (Lambeck and Burgess, 1992).

Foreland basins are sensitive indicators of the isostatic and geodynamic processes of orogenic belts, and as a result, provide an important record of orogenic evolution (e.g., Berge and Veal, 2005; Burbank, 1992; Karner and Watts, 1983; Lambeck, 1983). The Officer and Amadeus basins that flank the Musgrave Province should therefore record the evolution of the Petermann orogeny.

Provenance studies of both the Amadeus and Officer basins have detected a large influx of sediments between ca. 600 Ma and 500 Ma, with “Grenville-aged” detrital zircon populations consistent with the erosion of the Musgrave Province during the Petermann orogeny being observed in each basin (Maidment et al., 2007; Wade et al., 2005; Camacho et al., 2002; Zhao et al., 1992). In the Amadeus Basin, the Musgrave Province is considered a source of zircon throughout the evolution of the Amadeus Basin, providing a small to moderate contribution prior to 560 Ma, the dominant contribution during the period 540–500 Ma and progressively less influence in later samples (Maidment et al., 2007). In the eastern Officer Basin, provenance studies have detected a large influx of Musgrave Province–derived sediments ca. 600 Ma, indicating the onset of the Petermann orogeny, but a lack of Musgrave Province–derived sediments during the 580–540 Ma period (Wade et al., 2005). Eastward transport of sediments along east-trending structures was proposed as the most likely reason for this lack of sediment input from the Musgrave Province (Wade et al., 2005). Subsidence analysis in the Officer Basin indicates a period of subsidence during the Petermann orogeny, followed by a brief period of nonsubsidence, and then further subsidence until ca. 500 Ma, attributed to the Delamerian orogeny (Haddad et al., 2001).

The high-resolution (200–400 m line spacing) regional aeromagnetic data covering the Musgrave Province permits the definition of plutons, basins, and shear zones in the near surface by their magnetic character, and also the definition of the principal structural trends and their overprinting relationships (Aitken et al., 2008; Aitken and Betts, 2009a, 2009b). Major Petermann orogeny shear zones are defined by narrow, high-amplitude magnetic lows, and were mapped throughout the eastern Musgrave Province (Fig. 2A). In addition to defining the locations of these shear zones, kinematic information for these shear zones was interpreted from the aeromagnetic data using similar methods to those used in structural geology (Betts et al., 2007).

The aeromagnetic data show a variable but broadly northeast-trending magnetic grain, which is interpreted to be Musgravian orogeny aged (Aitken et al., 2008; Aitken and Betts, 2009a, 2009b). This magnetic grain, along with ca. 1200 Ma Pitjantjatjara Supersuite granitoids and ca. 1080 Ma Giles Complex plutons, act as magnetic marker horizons that have been deformed by Petermann-aged shear zones, permitting a kinematic interpretation of the main Petermann orogeny shear zones.

The shallow south-dipping Woodroffe thrust (Fig. 2A) is defined by an abrupt change in magnetic texture from more magnetic rocks with high-amplitude magnetic fabrics to the south of the thrust, to less magnetic rocks with lower amplitude magnetic fabrics to the north of the thrust, reflecting ca. 550 Ma juxtaposition of granulite-facies and amphibolite-facies rocks (Maboko et al., 1992). The aeromagnetic data do not reveal any kinematic information for the Woodroffe thrust, but kinematic indicators within the shear zone consistently indicate south-over-north movement (Edgoose et al., 2004).

The Mann fault is defined in the aeromagnetic data by a broad (2–3 km) and intense aeromagnetic low, and can be traced from the western edge of the interpretation area, through the Levenger graben, to its connection with the Echo lineament. Deflection of the Musgravian structural trend proximal to the Mann fault indicates right-lateral shear sense, consistent with folding of the Levenger Formation within the Levenger graben (Major and Conor, 1993).

The Ferdinand fault extends northeast from the Levenger graben, and the deflection of the Musgravian structural trend indicates that this shear zone is left-lateral, consistent with surface mapping (Major and Conor, 1993). In the aeromagnetic data, the Ferdinand fault is cut by the southeast-trending Marryat fault, which has caused large apparent right-lateral offsets to magnetic granitoids and also the Woodroffe thrust. These two major shear zones may form a conjugate set, indicating N-S compression.

The Mann fault, Ferdinand fault, and Marryat fault define the northern limit of the axial zone of the orogen, which extends south to the Wintiginna lineament, a shear zone that extends across the whole province (Fig. 1). This axial zone is characterized by an anastomosing network of shear zones, many showing apparent right-lateral offset of magnetic marker horizons (Fig. 2A). Major shear zones within this zone include the Wintiginna-Hinckley lineament and Paroora lineament, neither of which show any strike-slip kinematic indicators in the aeromagnetic data, and the Echo lineament, which shows prominent deflection of the Musgravian structural trend indicating right-lateral shear sense. In the aeromagnetic data, the Wintiginna lineament shows apparent right-lateral offset of several magnetic marker horizons (Fig. 2A).

South of the Wintiginna lineament, a lack of strike-slip kinematic indicators indicates that dip-slip movement is dominant in this region. In particular, the major shear zone in this region, the Lindsay lineament, shows no evidence of strike-slip motion.

Although they were active during the Alice Springs orogeny in places, the margins of the Musgrave Province are also interpreted to have been active during the Petermann orogeny (Edgoose et al., 2004; Flottmann et al., 2005; Scrimgeour et al., 1999). The prominent crustal-scale nappe complexes in the vicinity of the Petermann Ranges (Flottmann et al., 2005; Scrimgeour et al., 1999) are characteristic of basement-cored nappes throughout the Mulga Park subdomain (Edgoose et al., 2004) that are interpreted to represent pervasive Petermann orogeny deformation. The southern margin of the Musgrave Province may also have been active during the Petermann orogeny but was extensively reactivated during the Alice Springs orogeny (Aitken and Betts, 2009a; Lindsay and Leven, 1996; Hoskins and Lemon, 1995).

Well-defined crosscutting relationships between these shear zones (Fig. 2A) indicate that the Petermann orogeny had at least two stages of deformation: the first phase, in which granulite-facies crust was emplaced above amphibolite-facies crust, is characterized by north- and south-directed movement on divergent shallow-dipping, crustal-scale thrust faults, principally the Woodroffe thrust, the Lindsay lineament, the Piltardi detachment zone, and possibly also the margins of the province. The second phase is characterized by dextral transpressional movement on more steeply dipping crustal to lithospheric-scale shear zones in the axial zone of the orogen, principally the Mann-Ferdinand-Marryat fault system and the Wintiginna lineament.

Although the relative timing of these deformation events is clear, geochronological estimates for the Petermann orogeny are not sufficiently precise to detect this multiphase evolution, and the absolute age difference between these events is therefore unconstrained.

The Gravity of the Eastern Musgrave Province

The kinematics indicated in the aeromagnetic interpretation gives an estimate of two-dimensional motion in the near surface, but constraining the depth penetration and vertical component of motion on these shear zones is more difficult. If these shear zones are associated with crust mantle-boundary offsets and the juxtaposition of crustal blocks from different levels, then gravity modeling should indicate their deeper geometry.

The granulite-facies Fregon subdomain is associated with a very high amplitude regional gravity anomaly (~150 mGal) and steep regional gravity gradients (~30 eotvos). This intense high, indicating a large subsurface load, is situated within a broad, subcircular gravity low in central Australia, 1000 km in diameter, that corresponds to a region of thick crust (Clitheroe et al., 2000) and may represent a long-wavelength flexural depression.

Gravity data coverage in the eastern Musgrave Province is relatively good (Fig. 2B), with regional grids at 7.5–10 km spacing supplemented by more recent high-resolution profiles at ~1 km spacing (Gray and Flintoft, 2006; Gray and Aitken, 2007; Gray et al., 2007). The main shear zones interpreted in the aeromagnetic data are broadly correlative with the major steep gradients in the gravity field (Fig. 2B), although there are significant differences: The principal surface boundary, the Woodroffe thrust, is not associated with the principal gravity gradient, which is located farther south, whereas in other areas steep gravity gradients occur with no evidence of major Petermann orogeny shear zones in the near surface. A broad, northeast-trending low crossing the regional high (Fig. 2B) is not associated with any structure defined in the magnetic field, and may relate to a deep crustal or lithospheric boundary.

The Gravity Inversion Method

Petrophysically constrained gravity modeling has been shown to be an effective method for defining the architecture of the near surface (Farquharson et al., 2008; Fullagar et al., 2008; McLean and Betts, 2003). These methods can be extended to model the whole crust because constraint on the upper crustal density distribution greatly reduces the uncertainty in modeling the geometry of the lower crust and upper mantle (e.g., Ebbing et al., 2001).

To produce density models of the eastern Musgrave Province, three-dimensional inversion was conducted using VPmg software (Fullagar et al., 2008), which by iteratively modifying an input geological model containing lithological units and density information, seeks to optimize the fit to gravity data. With a uniform target misfit, the fit to the gravity data is defined by the root mean square of the residual anomaly (the RMS misfit). With VPmg, inversion terminates when the RMS misfit is less than the target misfit (convergence), or when the algorithm fails to reduce this parameter on successive iterations (a stalled inversion). Lithological units in the model can be either homogenous (i.e., density is the same throughout) or heterogeneous (i.e., density can be varied within the unit). The software has two main gravity modeling modes—heterogeneous density optimization and geometry optimization.

For heterogeneous density optimization, the subsurface is discretized into cells of regular x, y, and z extent, each represented by a single density value, and the inversion algorithm seeks to replicate the gravity data by modifying the density distribution represented by these cells. In this inversion mode, the density of homogenous units and the boundaries between units cannot change. For geometry optimization, the subsurface is discretized into vertical prisms, within which the depths to lithological boundary intersections are recorded, and the inversion algorithm seeks to replicate the gravity data by varying the depths to these lithological boundaries. In this inversion mode, the density within each prism cannot change, although any preexisting heterogeneity is maintained.

The much-documented inverse problem (e.g., Parker, 1994; Zhdanov, 2002) means that changes in density during inversion must be controlled to avoid an unrealistic density structure. During heterogeneous density optimization the user can impose upper and lower bounds on the range of densities permissible for each lithology and control the maximum change in density permitted per iteration. During geometry optimization, a user-defined parameter controlling the maximum relative change in interface depth per iteration is applied. This constraint means that it is mathematically easier to change the geometry of units at depth, and acts as depth weighting to counteract the loss of sensitivity with depth. More rigid constraints can also be imposed on the model geometry by defining regions in 3D space within which the boundaries of a lithological unit or units cannot change.

The Initial Model Parameters, Constraints, and Boundary Conditions

The spatial limits of the gravity model are broadly defined by the limits of high-resolution gravity data (Fig. 2B). The base of the model was set at 90 km depth, and precise topography from the gravity data was maintained in modeling as the upper surface bound. Prior to inversion, the free-air gravity data were minimum curvature gridded with 5 km cell size, upward continued by 2500 m to remove short wavelength content irresolvable with 5 km cell size, and detrended along a planar surface, to remove the need for density sources outside the model. This planar trend slopes from the south to the north over a total range of 11 mGal.

To provide constraint on the upper crustal density distribution, 146 measurements of the density of major lithologies were made on samples and core from throughout the Fregon subdomain (Fig. 1). These measurements showed that the density distribution is heterogeneous at small scales (tens of meters), and this means that individual density measurements do not reflect the bulk density of large modeling cells, and are not therefore used to directly constrain the densities of measurement localities. However, the statistical distribution of these measurements (Fig. 3) is important in constraining the probable density distribution in the near surface.

Deep seismic reflection studies (Korsch et al., 1998; Lindsay and Leven, 1996) and passive seismic models (Clitheroe et al., 2000) constrain the Moho depth to ~50 km beneath central Australia. To correspond with this constraint, and the crustal layering observed in the seismic reflection studies (Korsch et al., 1998; Lindsay and Leven, 1996), a four-layer model was constructed with the mantle (3.3 g/cm3), eclogitic crust (3.1 g/cm3), lower crust (2.85 g/cm3), and upper crust separated by boundaries at 50 km, 35 km, and 25 km, respectively.

The upper crust was subdivided in accordance with the major geological boundaries (Fig. 1), with units representing the Amadeus and Officer basins (2.55 g/cm3), amphibolite-facies crust (2.67 g/cm3), granulite or transitional granulite-facies crust (2.77 g/cm3), and also the transitional granulite-amphibolite-facies Wataru gneiss (2.75 g/cm3) in the southwest of the area (Gray, 1978) and the Ammaroodinna inlier (2.85 g/cm3) in the southeast (Krieg, 1993). These density values are constrained by both petrophysical data (Fig. 3) and the density contrasts required to satisfy the short-wavelength gravity gradients revealed in high-resolution data across the major density boundaries (Gray and Flintoft, 2006; Gray and Aitken, 2007; Gray et al., 2007).

A Heterogeneous Upper Crust Model

To investigate the density distribution required to satisfy the gravity anomaly from the upper crust alone, petrophysically constrained density inversion was applied within the upper crust using 5 km × 5 km × 1 km cells. The maximum density change per iteration was set at 0.02 g/cm3, and the target misfit was set to 1 mGal. The densities within lithological units were constrained as follows: amphibolite-facies crust was constrained to densities between 2.62 g/cm3 and 2.72 g/cm3; the Wataru gneiss was constrained to between 2.7 and 2.8 g/cm3; and the Ammaroodinna inlier was constrained to between 2.8 and 2.9 g/cm3. Granulite-facies crust was constrained to densities between 2.67 and 2.87 g/cm3, 0.1 g/cm3 either side of the measured median density (2.77 g/cm3). The densities of the homogenous units—the Amadeus and Officer basins, the lower crust, the eclogite layer, and the mantle—were held invariant.

From an initial RMS misfit of 29.16 mGal, the inversion stalled after 21 iterations at 5.06 mGal. Residual anomalies are mostly observed at the margins of the model (Fig. 4A), although there are significant negative residual anomalies (~15 mGal) over regions of the amphibolite-facies crust where the lower density limit of 2.62 g/cm3 is too high to permit a fit to the deep gravity lows. The fit over the granulite-facies crust is generally good, although the lack of Giles Complex mafic intrusions in the model is reflected in short-wavelength positive residual anomalies over major intrusions. The density distribution within this model (Fig. 4A and Animation 1)1 is generally reasonable and shows that there is no inherent requirement in the gravity data for crust-mantle boundary relief beneath the Musgrave Province.

However, the density distribution in this model has large areas of anomalously dense or light upper crust, for which there is little petrophysical evidence (Fig. 3). A particularly large density contrast is required between the Mulga Park subdomain (2.62 g/cm3 or less) and the Fregon subdomain (2.75–2.87 g/cm3). In addition, the major surface boundary juxtaposing crustal levels, the Woodroffe thrust, is only associated with a small density contrast in this model, with a large near-surface density contrast concentrated farther south. We consider this model to be inconsistent with the P-T and density constraints, and it also bears little resemblance to the architecture imaged in seismic models (Lambeck and Burgess, 1992). Some amount of crust-mantle boundary relief is therefore probable.

A Median Density Model

The geometry of the crust-mantle boundary and the amount of relief required to produce the observed gravity anomaly were investigated using a geometry optimization inversion, with 5 km × 5 km vertical prisms of 90 km depth extent and a maximum depth change per iteration of 2%. The target misfit was 1 mGal, and the boundaries of all units were permitted to change.

From an initial RMS misfit of 29.16 mGal, the inversion stalled after 80 iterations at 4.90 mGal. Short-wavelength residual anomalies are observed throughout the model area (Fig. 4B) as a result of the inability of this model to resolve small near-surface features. The geometry derived from this inversion (Fig. 4B and Animation 3[see footnote 1]) shows that the mantle and eclogite layers are, in general, uplifted beneath the east-trending central gravity high, and depressed beneath the gravity lows. The amount of crust-mantle boundary relief changes along strike, with the greatest relief of ~20–25 km in the western part of the model, and a reduction to ~10–20 km of relief in the eastern part of the model. In this model, steep crust-mantle boundary gradients correlate with major Petermann orogeny shear zones, principally the Mann fault, Ferdinand and Marryat faults, Wintiginna-Hinckley lineament, and Wintiginna lineament (Fig. 4B).

The geometry of the lower crust, eclogite, and mantle layers in this model are very sensitive to the density contrast between the granulite-facies gneiss and the amphibolite-facies gneiss, with small changes in density causing large changes in the offsets required to satisfy the gravity data. A sensitivity analysis was conducted to quantify this sensitivity by running geometry inversions in which the density contrast was perturbed in the initial model. For a variety of contrast values, the statistical variance of the depth to the resulting crust-mantle boundary was calculated as a measure of its flatness (Fig. 5).

The results of the sensitivity analysis (Fig. 5 and Animation 4 [see footnote one]) indicate that low or negative density contrast, between −0.05 and 0.025 g/cm3, results in a broad crust-mantle boundary high beneath the gravity high, and therefore high variance. Moderate density contrast, between 0.05 and 0.125 g/cm3, produces a flat but undulating crust-mantle boundary surface and low variance, and high density contrast, between 0.15 and 0.25 g/cm3, produces a broad crust-mantle boundary depression beneath the gravity high, and high variance.

Figure 5 illustrates that a density contrast of between 0.075 and 0.125 g/cm3 produces a low-variance crust-mantle boundary and also a low RMS misfit. This result verifies the constraints from petrophysical data and high-resolution gravity data that indicate ~0.1 g/cm3 of density contrast between granulite-facies and amphibolite-facies gneiss.

A Combined Heterogeneous Density and Geometry Inversion

Neither the heterogeneous upper crust model (Fig. 4A) nor the median density model (Fig. 4B) are a good representation of the crustal architecture, due to the omission of major intrusive suites and sedimentary basins and the geometric and density assumptions imposed on the models. A more detailed initial model was constructed (Fig. 6A) including Giles Complex and Pitjantjatjara Supersuite plutons, and syn-Petermann orogeny grabens. The geometry of the lower crust and crust-mantle boundaries were remodeled to represent offset of crustal layers along the plane of lithospheric-scale shear zones, and to more closely resemble the seismic architecture of the Musgrave Province (Fig. 6A). The source of the northeast-trending low in the gravity data (Fig. 2B) is not known, but this anomaly is associated with an eastward gravity gradient. Forward modeling prior to inversion indicated that west-up offset of the lower crust and crust-mantle boundaries by 8.5 km fits this gradient well.

The lithological densities in this model are similar to those used in the previous models, and the densities of the Amadeus and Officer basins, lower crust, eclogite layer, and mantle were held invariant at their initial density. Greater heterogeneity was incorporated into the upper crust by introducing homogenous units representing the Levenger and Moorilyanna Formations (each 2.55 g/cm3), and heterogeneous units representing the Giles Complex (3 ± 0.1 g/cm3) and the Pitjantjatjara Supersuite, subdivided into granitic (2.7 ± 0.1 g/cm3) and charnockitic (2.8 ± 0.1 g/cm3) lithologies. The location of these upper crustal units was defined in accordance with aeromagnetic data and outcropping geology.

The amphibolite-facies crust in the Mulga Park subdomain, the southern Fregon subdomain, and beneath the Amadeus and Officer basins was heterogeneous in this model, with density of 2.67 ± 0.05 g/cm3. The Wataru gneiss (2.75 ± 0.05 g/cm3) and Ammaroodinna inlier (2.85 ± 0.05 g/cm3) were also heterogeneous.

The granulite-facies upper crust was subdivided into four east-trending units, bounded by the Woodroffe thrust, the Mann-Ferdinand-Marryat fault system, the Wintiginna-Hinckley and Paroora lineaments, the Wintiginna lineament, and the Lindsay lineament (Fig. 6A). The northernmost (granulite facies 1) and southernmost (granulite facies 4) of these units were heterogeneous, with upper density bounds of 2.83 and 2.87 g/cm3, respectively, and lower bounds of 2.77 g/cm3. The central units were homogenous, with densities of 2.77 g/cm3 (granulite facies 2) and 2.75 g/cm3 (granulite facies 3).

For this inversion, by running consecutive density and geometry inversions, we first optimized the densities in the heterogeneous units, before adjusting the geometry of all units. The heterogeneous density inversion was run with 5 km × 5 km × 1 km cells, with a maximum density change per iteration of 0.02 g/cm3, and a target misfit of 1 mGal. From an initial RMS misfit of 18.43 mGal, the inversion stalled after 17 iterations at a RMS misfit of 6.62 mGal. Residual anomalies are concentrated above the granulite-facies core, which was invariant in this inversion.

Prior to geometry inversion, the lower-crustal stratigraphy beneath the amphibolite-facies crust was constrained so that the inversion algorithm would only modify the lower-crustal stratigraphy beneath the granulite-facies crust (Fig. 6A). This constraint is necessary because the inversion algorithm will preferentially modify boundaries at depth with high-density contrast (i.e., the base of the amphibolite-facies regions) and without constraint, the resulting geometry in these regions is inconsistent with deep seismic reflection studies (Korsch et al., 1998; Lindsay and Leven, 1996).

The geometry inversion was run with 5 km × 5 km vertical prisms of 90 km depth extent, a maximum depth change per iteration of 2%, and a target misfit of 1 mGal. This inversion stalled after 14 iterations, reducing the RMS misfit from 6.62 mGal to 5.58 mGal. This inversion greatly reduced residual anomalies above the granulite-facies crust, although short wavelength fluctuations are still observed (Fig. 6B).

A Best-Fit Model

Of the three inversions attempted, the architecture derived from the combined density and geometry inversion (Fig. 6B; Animations 5 and 6 [see footnote one]) is the most consistent with the magnetic interpretation, geological observations (Camacho et al., 1997; Clarke et al., 1995; Ellis and Maboko, 1992; Maboko et al., 1991; Major and Conor, 1993; Scrimgeour and Close, 1999), and seismic observations (Korsch et al., 1998; Lambeck and Burgess, 1992; Lindsay and Leven, 1996). The fit to the gravity anomaly is close to that achieved in the other inversions.

This 3D density model is well constrained in the vicinity of the Woodroffe thrust and the Mann, Ferdinand, and Marryat faults due to the relatively high gravity petrophysical and teleseismic data resolution, and greater degree of geological constraint. Away from this zone the model becomes less well constrained as the resolution of gravity, petrophysical and tele-seismic data decreases, and outcrop becomes sparser, although deep seismic reflection lines in the Amadeus Basin and southern Musgrave Province provide constraint on the architecture of the province margins.

The architecture at depth is similar to that proposed from seismic models (Korsch et al., 1998; Lambeck and Burgess, 1992), with the exception that neither the Lindsay lineament nor the shear zone at the southern margin of the province penetrates the crust-mantle boundary in our model, as suggested previously (Korsch et al., 1998; Lambeck and Burgess, 1992; Lindsay and Leven, 1996). The median density model (Fig. 4B) and the combined model (Fig. 6B) show that a relatively thin wedge of granulite-facies gneiss above a shallow-dipping Lindsay lineament satisfies the gravity anomaly here, and crust-mantle boundary uplift south of the Wintiginna lineament is not supported.

Probable Inversion of Ca. 1080–800 Ma Extensional Architecture

A defining characteristic of intraplate orogenesis in the upper crust is the reactivation of relict architecture, typically identified on the basis of inverted extensional basins and upthrust basement blocks (e.g., Turner and Williams, 2004; Ziegler et al., 1995; Ziegler et al., 1998). The east to southeast orientation and moderate to steep dip of the shear zones in the axial zone parallel the architecture interpreted to have developed during ca. 1080 Ma and ca. 800 Ma extensional events, which are characterized in the Musgrave Province by east-southeast– and southeast-trending shear zones, often co-located with mafic dikes (Aitken et al., 2008; Aitken and Betts, 2009b; Clarke et al., 1995; Edgoose et al., 2004). In existing geologic maps, these two major dike suites are generally undifferentiated, and therefore it is difficult to make a distinction between the geometries of these tectonic events with confidence.

The high-pressure recrystallization of dikes in regions intensely deformed during the Petermann orogeny (Camacho et al., 1997; Clarke et al., 1995; Ellis and Maboko, 1992; Major, 1967; Scrimgeour and Close, 1999) indicates that they were important in partitioning strain, and probably focused this strain into the coincident structures along which their emplacement may have been controlled (Aitken et al., 2008; Aitken and Betts, 2009a, 2009b; Clarke et al., 1995).

Although these criteria are not completely definitive, we infer these similarities in orientation and inferred geometry and the recrystallization of mafic dikes to represent the reactivation of preexisting weaknesses in the mid-to-lower crust during the Petermann orogeny, resulting in crustal-scale inversion of the relict geometry from previous extensional tectonic events.

Strain Accommodation and Escape Tectonics

Strain during the Petermann orogeny appears to have been accommodated in two ways: (1) pervasive ductile deformation and crustal thickening in the Mulga Park subdomain and in the Mann Ranges and (2) vertical and lateral extrusion of relatively rigid crustal blocks along ductile shear zones.

Pervasive Petermann orogeny ductile deformation is recorded throughout the Mulga Park subdomain (Edgoose et al., 2004). The structures associated with this deformation are too small to be resolved in our model, although geological mapping indicates that this deformation is characterized by north-vergent recumbent iso-clinal folding of both crystalline basement and sedimentary cover successions accompanied by a metamorphic overprint at upper-greenschist to upper-amphibolite facies (Edgoose et al., 2004; Flottmann et al., 2005; Scrimgeour et al., 1999).

Significant ductile deformation is also present in the Mann Ranges of the Fregon subdomain, although the origins of this ductile deformation are thought to be quite different to that in the Mulga Park subdomain, resulting from relatively rapid uplift of the crust, causing upward advection of heat and the resulting migmatization (Scrimgeour and Close, 1999).

In contrast to the Mann Ranges, much of the axial zone of the orogen lacks a pervasive metamorphic overprint in granulite-facies gneiss despite intense deformation within shear zones. This indicates that strain during the Petermann orogeny was highly partitioned onto mylonite structures (Camacho et al., 1997; Camacho and McDougall, 2000; Camacho et al., 2001; Clarke et al., 1995) possibly as a result of shear heating (Camacho et al., 2001). Similarly, at the crustal-scale, Petermann orogeny shear zones defined in aeromagnetic data (Fig. 2A) generally bound relatively undeformed crustal blocks, and define numerous structural features normally associated with brittle deformation, including conjugate shear zones and pop-up structures, despite being active at lower-crustal levels. This indicates a deformation regime that at crustal scale was characterized by the motion of competent crustal blocks along ductile shear zones, onto which strain was highly partitioned (Fig. 2A).

The major shear zones that bound the axial zone are transpressional and accommodate strain by both the vertical and lateral extrusion of crustal blocks: In the west of the modeled area, 20–25 km of crust-mantle boundary offset is accommodated on the Mann fault and Wintiginna-Hinckley lineament. In the east of the modeled area, there is significantly less vertical offset (10–20 km) and significant northward and eastward movement of lithospheric blocks accommodated by oblique reverse movement on the Ferdinand fault, Marryat fault, and Wintiginna lineament (Figs. 2A and 6B). At the far-east of the model area, an extensive network of splays (Fig. 2A) indicates the terminal accommodation of intraplate strain by north-directed lateral spreading of the orogen.

There is therefore an along-strike transition from a large amount of vertical extrusion in the west, to the east where vertical extrusion is less but a greater amount of northward and eastward lateral extrusion is observed. This is interpreted to reflect escape tectonics with motion directed toward the less competent lithosphere at the eastern margin of Australia, bounded by an incipient subduction zone (Boger and Miller, 2004; Gray and Foster, 2004) and away from the highly competent lithosphere of the West Australian craton. In the far-eastern Musgrave Province, motion is directed north, away from the Archean to Paleoproterozoic Gawler craton.

Transpressional kinematics and escape tectonics are interpreted to be fundamental in the currently active intraplate orogens of central Asia where they occur as a result of the reactivation of structures oblique to the ~N-S compression driven by the Himalayan collision (Cunningham, 2005; Tapponnier and Molnar, 1979), and a similar situation may have occurred in the Petermann orogeny.

The Gondwana Connection

As noted previously, locally derived vertical forces may have played a significant role in the Petermann orogeny (Neil and Houseman, 1999), but the transmission of far-field stress from the collisional orogens related to Gondwana assembly are likely to be the main source of driving forces. As the most proximal of the potential sources of the intraplate stress field, the Kuunga orogen is a compelling candidate, especially considering that stress need only be transmitted through the West Australian craton, which is underlain by a keel of Archean lithospheric mantle (Simons et al., 1999), and as a result is highly competent.

We therefore consider the N-S compressional stress in central Australia to result from SSE-directed motion of India relative to Australia, as part of the Kuunga orogeny. This resulted in sinistral shear on the N-S–oriented Darling fault (Harris, 1994) and dextral shear on the ESE-SE–oriented shear zones of the Petermann orogeny (Fig. 2A). This also explains sinistral shear on the NE-oriented Ferdinand fault (Fig. 2A).

This relative motion can be achieved within the aforementioned models of Gondwana with some modification: Kinematically speaking, oblique subduction of the paleo–Pacific plate (Veevers, 2003) could generate this relative motion, although to do so would require either a greater degree of obliquity than suggested by Veevers (2003), or intracontinental curvature of the stress field (e.g., Hillis and Reynolds, 2000). The initiation of paleo-Pacific subduction, at its earliest ca. 560 Ma (Goodge, 1997) and probably younger near Australia (Boger and Miller, 2004), may preclude this model on the basis of timing.

The original model for the Kuunga orogeny (Meert et al., 1995; Meert, 2003) and subsequent references to this model (e.g., Boger and Miller, 2004; Collins and Pisarevsky, 2005; Goscombe and Gray, 2008) imply E-W–directed collision of India with Australia and Antarctica, which does not readily explain the N-S shortening within central Australia. An oblique collision of India with Australia during the Kuunga orogeny, at least NW-SE oriented, would be a straightforward way to explain the deformation observed in central Australia.

Although more distant, the East African orogen is a plausible source of ~N-S compressive stress in central Australia, as the collision was essentially parallel, and was of sufficient magnitude to have propagated significant forces into the continents. However, the reactivation of central Australia would require the transmission of stress through the entirety of East Antarctica, which is typically considered as a group of coalesced cratonic blocks (e.g., Fitzsimons, 2000; Jacobs et al., 2008), and is thus prone to reworking. We consider the direct transmission of stress unrealistic, unless achieved by the focusing of strain into a network of transpressional intracratonic mobile belts, including the Kuunga orogen, in response to oblique convergence of the East African orogen (Squire et al., 2006).

Crustal Uplift and Lithospheric Strengthening

In the Musgrave Ranges, amphibolite and greenschist-facies mylonites, dated at 540 ± 10 Ma, crosscut subeclogite-facies shear zones, dated at 547 ± 30 Ma (Camacho et al., 1997; Ellis and Maboko, 1992; Maboko et al., 1991). This close coexistence in space and time of subeclogite-facies metamorphism at ~40 km depth, and greenschist-facies metamorphism at ~20 km depth requires ~20 km of exhumation during the Petermann orogeny. The metamorphic evolution of the Mann Ranges and the Tomkinson Ranges define a similar evolution, with deep-crustal mylonites overprinted by mid-crustal mylonites, although the age of these events is not constrained by radiometric dating (Clarke et al., 1995; Scrimgeour and Close, 1999). In addition, the mineralogy of the ca. 1080 Ma Giles Complex and associated volcanic rocks indicates ~20 km of exhumation in the axial zone relative to the margins of the province. The uplift of the entire crustal pile by between 10 km and 25 km in the axial zone of the orogen (Fig. 6B) is therefore consistent with these geologic data.

The architecture of the Petermann orogeny is unusual, in that with a geologically reasonable density structure in the upper crust, the axial zone of the orogen is underlain by a wedge of lithospheric mantle in the lower crust. Isostatically, this load cannot be supported locally, and studies of central Australia have indicated that at short wavelengths (~200 km), central Australia is not in isostatic equilibrium (Lambeck, 1983; Stephenson and Lambeck, 1985). The pattern of gravity anomalies in central Australia indicates that this load, and a similar one beneath the Arunta inlier (Goleby et al., 1989; Goleby et al., 1990), may be collectively compensated by long-wavelength (~1000 km) lithospheric flexure beneath central Australia.

The short-wavelength isostatic disequilibrium implied by the crust-mantle boundary architecture (Fig. 6B) induces strong forces in the lithosphere. The preservation of this architecture for more than 500 million years, despite a number of tectonic events including both east-west–oriented compressional events and north-south–oriented rifting episodes (e.g., Betts et al., 2002; Bryan et al., 1997; Gray and Foster, 2004; Miller et al., 2002), implies that the isostatic forces are not being counteracted dynamically but that the lithosphere is significantly strengthened, at least locally.

The exhumation of the entire crustal pile during the Petermann orogeny may be the driver for this increase in lithospheric strength, due to a combination of the emplacement of a wedge of relatively strong lithospheric mantle into the relatively weak lower crust, and the erosion of an upper crust containing abundant heat-producing elements (Hand and Sandiford, 1999; McLaren et al., 2005; Sandiford and Hand, 1998). These processes would create a block of cool and strong lithosphere relative to the surrounding regions, causing tectonic stability and enabling the preservation of this crustal architecture in the long term.

In addition, the exhumation of the axial zone of the province by 10–20 km implies that a large amount of material has been eroded from the surface. Although these sediments are distributed across a huge area (Lindsay, 2002), a large proportion of this material was deposited in the foreland Amadeus and Officer basins, which contain up to 3 km of late Neoproterozoic to Early Cambrian sedimentary rocks (Gravestock et al., 1993; Lindsay, 2002; Lindsay and Leven, 1996). This erosion-deposition feedback is likely to have amplified the uplift of the axial zone relative to the foreland, similar to other regions (e.g., Avouac and Burov, 1996; Burov, 2007).

A combination of aeromagnetic interpretation and 3D gravity inversion has enabled the derivation of an architectural and kinematic model of the intraplate compressional Petermann orogeny in central Australia. This model is characterized by the development in the early stages of the orogeny of divergent crustal-scale thrusts at the margins of the province that accommodated crustal thickening, and the juxtaposition of granulite-facies and amphibolite-facies crustal blocks. These thrusts are cut by an axial zone of steep, crustal to lithospheric-scale transpressional shear zones that accommodated the exhumation of the axial zone, causing crust-mantle boundary offsets of up to 25 km.

The architecture of this orogen permits the characterization of several lithospheric processes, including the probable reactivation of relict extensional architecture, strain accommodation by the vertical and lateral extrusion of competent crustal blocks along major ductile shear zones, and escape tectonics directed toward the east.

Perhaps the most interesting aspect of the Petermann orogeny architecture is the emplacement of a wedge of lithospheric mantle within the lower crust, and its preservation for more than 500 million years despite local isostatic disequilibrium and a number of major tectonic events elsewhere in the continent. The preservation of the crust-mantle boundary offset implies that this block of lithosphere is sufficiently strong to resist tectonic and isostatic forces. We propose that this strength is a result of the emplacement of a wedge of lithospheric mantle in the lower crust, accompanied by erosion of the upper crust, which produced an anomalously strong and cool block of lithosphere.

These processes are wholly consistent with the concept that in intraplate orogens the heterogeneity of the continental lithosphere controls the architecture and kinematics of deformation and creates feedbacks with the rheological and thermal structure of the continental lithosphere (e.g., Hand and Sandiford, 1999; Sandiford and Hand, 1998; Sandiford et al., 2001; Ziegler et al., 1998). In this respect, the Petermann orogeny is similar to most other intraplate orogens.

We note, however, that within this paradigm the Petermann orogeny is remarkable for several reasons: (1) deformation is focused to a very discrete region far from the plate margin, within which intense lithospheric-scale deformation has occurred; (2) although none can be excluded on the basis of this study, existing models of this region during Gondwana assembly (Collins and Pisarevsky, 2005; Meert, 2003; Squire et al., 2006; Veevers, 2003) require modification to be consistent with the architecture and kinematics of the orogen; and (3) orogenesis has led to significant lithospheric strengthening, causing the stabilization of a previously much reworked terrane and the preservation of its remarkable crustal architecture to the present.

This work was supported by Primary Industry and Resources South Australia (PIRSA) and Australian Research Council Linkage grant LP0560887. Aitken was also supported by a Monash University Faculty of Science Postgraduate Publication Award. The Musgrave Province aeromagnetic dataset was supplied by PIRSA. Gravity data were obtained under license from Geoscience Australia and PIRSA. We thank Rick Squire, James Evans, and an anonymous reviewer for their constructive comments on the manuscript.

1If you are viewing the PDF of this paper or reading it offline, please visit the full-text article on http://lithosphere.gsapubs.org/to view Animations 1–5. You can also access them at the following respective links: http://dx.doi.org/10.1130/L39.S1, http://dx.doi.org/10.1130/L39.S2, http://dx.doi.org/10.1130/L39.S3, http://dx.doi.org/10.1130/L39.S4, and http://dx.doi.org/10.1130/L39.S5.