Interactions between continent-like ‘drift’, rifting and mantle flow on Venus: gravity interpretations and Earth analogues

Abstract Regional shear zones are interpreted from Bouguer gravity data over northern polar to low southern latitudes of Venus. Offset and deflection of horizontal gravity gradient edges (‘worms’) and lineaments interpreted from displacement of Bouguer anomalies portray crustal structures, the geometry of which resembles both regional transcurrent shear zones bounding or cross-cutting cratons and fracture zones in oceanic crust on Earth. High Bouguer anomalies and thinned crust comparable to the Mid-Continent Rift in North America suggest underplating of denser, mantle-derived mafic material beneath extended crust in Sedna and Guinevere planitia on Venus. These rifts are partitioned by transfer faults and flank a zone of mantle upwelling (Eistla Regio) between colinear hot, upwelling mantle plumes. Data support the northward drift and indentation of Lakshmi Planum in western Ishtar Terra and >1000 km of transcurrent displacement between Ovda and Thetis regiones. Large displacements of areas of continent-like crust on Venus are interpreted to result from mantle tractions and pressure acting against their deep lithospheric mantle ‘keels’ commensurate with extension in adjacent rifts. Displacements of Lakshmi Planum and Ovda and Thetis regiones on Venus, a planet without plate tectonics, cannot be attributed to plate boundary forces (i.e. ridge push and slab pull). Results therefore suggest that a similar, subduction-free geodynamic model may explain deformation features in Archaean greenstone terrains on Earth. Continent-like ‘drift’ on Venus also resembles models for the late Cenozoic–Recent Earth, where westward translation of the Americas and northward displacement of India are interpreted as being driven by mantle flow tractions on the keels of their Precambrian cratons. Supplementary material: Bouguer gravity and topographic images over a segment of the Mid-Atlantic ridge and Ross Island and surrounds in Antarctica, principal horizontal stress trajectories about mantle plumes on Earth, map and interactive 3D representations of cratonic keels beneath North America from seismic tomography, and a centrifuge simulation for comparison with Venus in support of our tectonic model are available at http://www.geolsoc.org.uk/SUP18736.

Venus, although similar to Earth in size, inferred internal composition and surface gravity, does not show any features that characterize plate tectonics on Earth, namely subduction zones, volcanic arcs, obvious seafloor-spreading ridges offset by transform faults, and large translations and rotations of distinct lithospheric plates (Anderson 1981;Phillips et al. 1981;Solomon et al. 1991Solomon et al. , 1992Solomon 1993;Phillips & Hansen 1994;Simons et al. 1994Simons et al. , 1997Grimm 1998;Hansen 2007;Smrekar et al. 2007;Watters & Schultz 2009;McGill et al. 2010;Moores et al. 2013). It has, nevertheless, been suggested by van Thienen et al. (2004) that it is theoretically possible for subduction and plate tectonics to have occurred in Venus' past, but for which no evidence remains following Venus' resurfacing (Strom et al. 1994). The terrestrial planets, with the notable exception of Earth, have traditionally been thought to experience a stagnant lid convection regime where a very viscous -rigid 'lid' that covers the entire planet overlies active convection in the underlying hot mantle (Solomatov & Moresi 1996Moresi & Solomatov 1998;O'Rourke & Korenaga 2012). More recent and detailed two-dimensional (2D) numerical modelling by Armann & Tackley (2012) suggests that stagnant lid convection, alternating with episodic convection with approximately 150 million-year overturn events, best explains Venus's tectonic development. Convective cells on Venus are estimated to be around 600-900 km to (exceptionally) 2000 km wide (Solomatov & Moresi 1996). The present lithospheric thickness of Venus is estimated at ,150 km by Nimmo & McKenzie (1998) and Smrekar & Parmentier (1996); van Thienen et al. (2004) suggested that the lithosphere may be thicker beneath continentlike planae, and to be thus similar to Earth. Orth & Solomatov (2012) proposed an average lithospheric thickness of between 300 and 500 km, and a crustal thickness between 20 and 60 km. James et al. (2013) provided estimates of lithospheric thickness between 100 and 200 km, but with lower estimates of 8-25 km for the average crustal thickness, and crustal thicknesses calculated at between 41 and 60 km beneath Lakshmi Planum and surrounding fold belts (Fig. 1).
The geology and broad stratigraphic framework of Venus are portrayed in a global map by Ivanov & Head (2011. Volcanic zones, broad topographical rises, positive gravity anomalies, radiating rifts and dykes, and annular features such as coronae (Barsukov et al. 1986) are taken as evidence for hotspots or upwelling hot mantle plumes (Barsukov et al. 1986;Phillips et al. 1991;Bindschadler et al. 1992;Smrekar 1994;Phillips & Hansen 1998;Ernst & Desnoyers 2004;Basilevsky & Head 2007). Venus Express VIRTIS (Visible and Infrared Thermal Imaging Spectrometer) thermal emissivity  Ivanov & Head (2010a, b) for the Lakshmi Planum area and Marinangeli & Gilmore (2000) for the Akna Montes-Atropos Tessera region in the west of the interpreted area. The geometry of thrusts and transcurrent/transpressional shear zones on the NW, northern and NE margins is interpreted as Himalayan-style indentation and lateral escape (but without plate tectonics) resulting from the northward displacement of Lakshmi Planum in the first shortening event (D 1 ). Thrusts east of Lakshmi Planum and NE-striking dextral shear zones are attributed to a second event (D 2 ) of bulk WSW-ENE shortening. Images are from NASA/JPL; the interpretation in (b) is modified after Harris & Bédard (2014). measurements were taken by Smrekar et al. (2010) to suggest the presence of young to possibly active hotspot-related volcanism, refuting early ideas that Venus is no longer tectonically or volcanically active (e.g. Kerr 1994). The interpretation of presently active volcanism is, however, questioned by Ivanov & Head (2010c), although they suggest volcanic eruptions have occurred in the last several decades (cf. Bondarenko et al. 2010). The stagnant lid, mantle plume-dominated tectonics of Venus contrast to mobile lid convection on Earth (Moresi & Solomatov 1998) where the lithosphere moves as discrete, rigid plates, and oceanic crust formed at active spreading ridges is consumed at subduction zones. In the tectonic environment presently envisaged for Venus, large, regionally coherent displacements of crustal blocks/terrains have not been considered likely. Harris & Bédard (2014), however, documented horizontal displacements and polyphase folding and shearing on Venus interpreted from Magellan radar images of Venus' surface. The most spectacular of these examples is in western Ishtar Terra where regional folds, thrusts, and transcurrent and transpressional shear zones surrounding the cratonlike Lakshmi Planum define a geometry identical to structures produced during indentation and lateral escape of the Himalayan-Indochina system on Earth (Harris & Bédard 2014; Fig. 1). Our research builds upon radar interpretations made by Kaula et al. (1992; who described detailed radar images of convergent fold belts, thrusts and regional transcurrent shear systems, but without any regional synthesis) and provides structural evidence for comparisons with indentation and lateral escape tectonics on Earth suggested by Crumpler et al. (1986;expanded upon in Moores et al. 2013), Markov (1986) and Head et al. (1990). This paper presents interpretations of enhanced gravity data on Venus to demonstrate that shear zones bounding Lakshmi Planum (interpreted from radar images) are crustalscale structures that form part of a coherent regional system of shear zones, which is identified for the first time. Gravity data are also used to highlight regional crustal-scale rifts, the extent of which appears to have been underestimated in prior maps derived from radar images. The nature of the mechanisms causing large horizontal displacements on Venus without plate tectonics are discussed, and the interaction between rifting, transcurrent faulting, and indentation tectonics as recognized in Ishtar Terra are developed, drawing on comparisons with analogous plume-related features on Earth.
Recent geophysical and GPS data, as well as numerical modelling for Earth highlight the importance of horizontal traction on the base of deep continental roots as a force that helps to drive displacement of continents (e.g. Bokelmann 2002a, b;Liu & Bird 2002;Eaton & Frederiksen 2007;Alvarez 2010;Faccenna et al. 2013;Ghosh et al. 2013), in addition to the action of plate boundary forces such as ridge push, slab pull, and trench suction (Forsyth & Uyeda 1975). This implies that, even on Earth, modern plate tectonics is not required for the displacement ('drift') of continents, as is frequently assumed; if there were no plate boundary forces, then continental drift and resulting orogenesis on Earth would probably still occur, albeit more slowly. There is an active debate about when plate tectonics on Earth started, and whether the formation and deformation of Archaean terrains resulted from plate tectonic processes such as subduction and arc accretion (reviewed in Bédard et al. 2013 andBédard 2014). Studies of Venus thus help us better understand tectonic processes operative in an Archaean Earth (cf. Anderson 1981;McGill 1983;Morgan 1983;Markov 1986;Markov et al. 1989;Glukhovskiy et al. 1995;Sorohtin & Ushakov 2002;Stern 2004;Hansen 2007;Head et al. 2008;Van Kranendonk 2010).

Previous studies of faulting and brittle-ductile shearing on Venus
Despite the absence of plate tectonics, Venus displays diverse volcanic and tectonic features (folds, faults and brittle-ductile shear zones) indicative of regional lithospheric shortening and extension, many of which are similar to those developed on Earth (Crumpler et al. 1986;Head et al. 1990Head et al. , 1992Kaula et al. 1992;Solomon et al. 1992;Watters 1992;Hansen et al. 1997;Hansen 2007;Watters & Schultz 2009;McGill et al. 2010;Harris & Bédard 2014). Whilst detailed structural interpretations of radar images (cited above and in the following sections) have been undertaken, they are mainly concerned with relatively small regions and regional tectonic overviews have focused on the distribution of volcanic features (e.g. Pronin & Stofan 1990;Head et al. 1992;Stofan et al. 1992;Price et al. 1996), impact craters (Price et al. 1996), mafic dykes (e.g. Grosfils & Head 1994;Ernst et al. 1995, wrinkle-ridges and fractures (e.g. Hansen & Olive 2010), and rifts (e.g. Price et al. 1996;Basilevsky & Head 2000). Regional geological mapping coordinated by the USGS (United States Geological Survey) has been based almost exclusively on interpretation of geomorphological features from radar images (Tanaka et al. 2010; e.g. 'ridge crests', 'wrinkle ridges' or 'large ridges' are mapped instead of folds), and guidelines (Tanaka et al. 1994(Tanaka et al. , 2011 advise curtailing the amount of structural elements included in regional maps. The global distribution of principal structural elements is portrayed by Ivanov & Head (2011); a detailed interpretation of much of the planet is also included in Hansen & Olive (2010).

Normal faults and regional rifts
The recognition of rifts on Venus has largely been based on radar-image interpretation of normalfault-bounded graben as linear rifts radiating from volcanic centres termed novae, concentric, annular rifts rimming coronae (e.g. Solomon et al. 1992;Crumpler et al. 1993;Ernst et al. 1995;McGill et al. 2010;Studd et al. 2011), closely spaced graben in tessera terrains (Gilmore et al. 1998) and as deep (Watters & Schultz 2009) canyons/graben termed chasmata that are thousands of kilometres long (e.g. Solomon et al. 1992). Linear rifts and dyke swarms on Venus have been likened to those on Earth (Solomon 1993;Ernst et al. 1995;Foster & Nimmo 1996), but on Venus, rifts are about 3 times greater in width than plume-related rifts on Earth, which is attributed to the absence of thick sediment infill (Foster & Nimmo 1996). Low magnitudes of crustal extension due to normal faulting were calculated by Connors & Suppe (2001) from slope measurements, although they acknowledged that their calculations of crustal extension did not include the likely contribution of ductile to brittleductile extension and necking. Given its high mean surface temperatures (estimated at 462 8C by Seiff et al. 1985 -see NASA http://sse.jpl.nasa. gov/planets/profile.cfm?Object=Venus&Display= Facts&System=Metric (accessed June 2013) -and recently calculated as varying between c. 421 and 441 8C for Venus' southern hemisphere by Mueller et al. 2008), the brittle-ductile transition on Venus is expected to occur at a shallower depth than on Earth (McGill et al. 2010;Violay et al. 2010).
Large volcanic 'flow fields' on Venus are equated to terrestrial flood basalts or large igneous provinces (LIPs) in areas of lithospheric extension and thinning linked to mantle upwelling or mantle plumes (Lancaster et al. 1995;Magee & Head 2001;Ernst & Buchan 2003;Ernst & Desnoyers 2004;Ernst et al. 2007;Hansen 2007). Spreading centres and associated oceanic transform faults similar to those on Earth have not, however, been confirmed on Venus (Phillips et al. 1991;Watters & Schultz 2009), although transform-like faults were proposed by Crumpler et al. (1987; an interpretation we do not agree with, as discussed below). Although most models for rifting on Venus infer mantle upwelling or hot thermal plume origins, normal faulting has also been attributed to diapiric (i.e. Rayleigh -Taylor/density-driven) uplift (Hoogenboom & Houseman 2005) and gravitational collapse (e.g. Solomon 1993;Keep & Hansen 1994).

Strike-slip faulting
Regional transcurrent (strike-slip) faults with up to 450 km of displacement on individual faults, and with cumulative transcurrent displacement of up to 2000 km, were identified in early studies of Venus (e.g. Crumpler et al. 1986Crumpler et al. , 1987Crumpler & Head 1988;Sukhanov & Pronin 1988;Kozak & Schaber 1989;Vorder Bruegge et al. 1990) and are implicit in the interpretations of Head et al. (1990). Pohn & Schaber (1992) interpreted strike-slip fault geometries that they likened to indentation-style tectonics on Earth, and conjugate strike-slip faults were interpreted by Watters (1992). The presence of broad brittle -ductile shear zones, as well as discrete faults, was also recognized early in Venus mapping. For example, Vorder  drew comparisons between en échelon folds and pullapart graben with similar structures formed during wrenching along the San Andreas fault system on Earth (cf. Wilcox et al. 1973;Sylvester 1988). Additionally, Hansen (1992) and Kaula et al. (1992) interpreted sigmoidal deflection of fold-axial traces and their angular relationship to, and truncation by, transcurrent faults as equivalent to regional-scale S/C structures (drawing comparison with structures in mylonitic rocks described by Berthé et al. 1979). Despite these early interpretations, Solomon et al. (1992, p. 13 199) and Solomon (1993, p. 50) stated that 'few large-offset strike slip faults . . . have been observed' and interpreted strike-slip faults as local features developed to accommodate horizontal displacements during crustal shortening or stretching. Bindschadler et al. (1992) considered that development of long strike-slip faults on Venus is inhibited by the absence of water. The ensuing misperception for limited transcurrent displacements on Venus (as noted by Fernández et al. 2010) has been carried forward in subsequent reviews of faulting on planetary bodies (e.g. Schultz 1999;Schultz et al. 2010) and comparisons between the tectonics of Venus and Earth (e.g. Montési 2013). Numerous radar interpretation studies carried out since Solomon (1993) have, nevertheless, continued to illustrate the presence of conjugate strike-slip faults (e.g. Hansen 2007; Yin & Taylor 2011) and regional strike-slip faults and transcurrent wrench zones with displacements of up to hundreds or even thousands of kilometres along them have been mapped (e.g. Raitala 1994Raitala , 1996Brown & Grimm 1995;Ansan et al. 1996;Koenig & Aydin 1998;Tuckwell & Ghail 2003;Kumar 2005;Romeo et al. 2005;Chetty et al. 2010;Fernández et al. 2010). Mechanisms for the formation of regional strike-slip faults and the relationship (if any) to rifting (as postulated by Markov et al. 1989) have, nevertheless, remained uncertain.
Folds and reverse/thrust faults Folds (including, but not limited to, 'ridge belts' and 'wrinkle ridges') deform volcanic plains, and regional folds and reverse/thrust faults have developed along the margins of plateaux, tesserae and coronae (e.g. Head et al. 1990;Suppe & Connors 1992). The formation of these contractional features is variably attributed to compensation of extension in adjacent rifts (Markov et al. 1989), local shortening and crustal thickening (Solomon 1993), flexure and underthrusting of young lithosphere against a more rigid block ), gravitational spreading due to elevation differences , thermal expansion (Solomon et al. 1999) or contraction (McGill et al. 2010), mantle downwelling (Markov et al. 1989;Kiefer & Hager 1991;Bindschadler et al. 1992), or the far-field contraction resulting from mantle flow about upwelling hot plumes (Mège & Ernst 2001). As on Earth, refolded folds on Venus (e.g. Harris & Bédard 2014) are taken as evidence for changes in the orientation of the regional stress and/or displacement field.

Bouguer gravity field and crustal thickness of Venus
A detailed map of the free-air gravity field of Venus was acquired during the 1993 Magellan mission from accelerations and decelerations calculated from line-of-sight doppler shift, where aerobraking was used to establish a low orbit with 36% ellipicity (Solomon 1993;Kaula 1996;Konopliv & Sjogren 1996) (during aerobraking, Magellan's apoapsis was lowered from 8500 to 500 km; its periapsis was 180 km). (Differences between free-air and Bouguer gravity, and their implications relevant to planetary geological interpretations, are explained in simple terms by Lakdawalla 2012.) Gravity data from the previous Venera (Barsukov et al. 1986) and Pioneer Venus Orbiter missions (Solomon 1993) infill many areas lacking Magellan data. As shown by comparison with early satellite gravity for Earth (e.g. Sandwell 1992), the spatial resolution of 110 km for the final, processed Magellan gravity data obtained near the equator and of 180 km at higher latitudes (Kaula 1996) makes these data suitable for regional-scale mapping of crustal-to lithospheric-scale features (especially where the total horizontal gradient is used to delineate structural features). Gravity anomalies have been used in geoid, admittance and flexural rigidity calculations (Solomon 1993 and references therein; Konopliv & Sjogren 1996;Simons et al. 1997;Barnett et al. 2000;Anderson & Smrekar 2006;Wieczorek 2007) to calculate crustal thickness (Simons et al. 1997;Anderson & Smrekar 2006;Wieczorek 2007;James et al. 2013, whose data we present below) and to create models that support the presence of mantle plumes Herrick & Phillips 1992;Solomon 1993;Kiefer & Peterson 2003;Vezolainen et al. 2003). Previous studies of regional free-air anomalies suggest that: † gravity lows are due to less dense crust that extends deeper into the mantle or areas of higher temperature and less dense crust; † there is a high correlation on Venus between the free-air gravitational field and topography, suggesting deep compensation through convective upwelling or downwelling (Solomon 1993;Rummel 2005).
The Bouguer gravitational anomaly field for Venus (Sjogren & Konopliv 2008; developed from the original 1997 gravity data updated with degree 180 harmonic coefficients by Konopliv et al. 1999) was calculated by these authors using a reference density of 2900 kg m 23 based on average basaltic crust (which contrasts to 2650 kg m 23 generally used on Earth, the average density of felsic/continental crust). Despite the excellent gravity coverage, Bouguer anomaly maps have not been (to our knowledge) previously used in interpreting regional, crustal-scale faults on Venus. Spectrally filtered and enhanced gravity data and edges in the horizontal gradient of Bouguer gravity (commonly termed gravity 'worms') are regularly used for mapping regional-scale faults/shear zones and lithological contacts on Earth (e.g. Blakely (Horowitz et al. 2000). The same approach is applied to Venus in the following subsections. The total Bouguer gravity field for northern polar to low southern latitudes of Venus, covering the Ishtar and western Aphrodite terrae and the Sedna, southern Snegurochka and Niobe planitae presented in Figure 2a (place names not shown in this figure are shown in Fig. 1), shows that: † Ishtar Terra (whose geology is described by Kaula et al. 1992;Marinangeli & Gilmore 2000;, 2010a and Aphrodite Terra (including Ovda Regio mapped by Bleamaster & Hansen 2005) are marked by regional Bouguer lows down to 2300 mGal. † Subcircular Bouguer lows in the SW of the map correspond to Beta and Phoebe regionen, which are elevated volcanic centres with thick crust and radiating graben (including Devana Chasma) interpreted to overlie upwelling mantle plumes (Kiefer & Peterson 2003;Vezolainen et al. 2003). Beta Regio is interpreted to overlie a cluster of mantle plumes by Ernst et al. (2007). † An ESE-WNW-trending broad Bouguer high (up to c. 130 mGal) is developed over the 1500-2000 km-wide 'lowland' volcanic plains (Stofan et al. 1987) of Sedna Planitia south of Ishtar Terra. An approximately 5000 km-long Bouguer gravity high bifurcates south of Lakshmi Planum. The northern branch continues for about 14 000 km towards the east and passes approximately 1500 km to the north of Aphrodite Terra, where it is locally punctuated by Bouguer lows over volcanic centres, including Bell and Tellus regionen. The southern branch, which strikes more southeasterly, corresponds to Guinevere Planitia and ends abruptly after about 7000 km to the west of Ovda Regio. † The two linear gravity highs of Sedna and Guinevere planitias are separated by Eistla Regio, a region of lesser Bouguer anomalies dominated by tight groups of volcanos and coronae, attributed to underlying plume clusters by Ernst et al. (2007). † North of Ishtar Terra, Snegurochka Planitia, a region marked by volcanic plains formed in a tensile stress field (Hurwitz & Head 2012), is also marked by a broad, diffuse gravity high. This polar region lies outside the area covered by the present research and has not been examined further.
Figure 2b portrays high-frequency (i.e. 'shallowsource') components of the Bouguer field (cf. data defined by the shallow slope of the MGNP180U power spectrum plot of Wieczorek (2007, fig. 4), equating to a spherical harmonic degree .40).
Thickness variations of Venus' crust calculated from inversion of gravity data by James et al. (2013) were gridded and are presented in 2D and 3D in Figure 3. From the comparison of total and 'shallow-source' Bouguer gravity and crustal thickness images, it can be seen that: † Lakshmi Planum has a uniformly thick crust ( Fig. 3a), yet can be divided into a northern shallow-source gravity high and a southern shallow-source gravity low (Fig. 2b). Similar marked lateral changes occur across all of Ishtar Terra where they correspond to strong gradients highlighted by gravity 'worms' (Fig. 2b). Jull & Arkani-Hamed (1995) and Arkani-Hamed (1996) suggested that the density perturbations over Ishtar Terra that were determined from a similar spectral analysis of gravity data required lateral variations in rock type, and that crust beneath Ishtar Terra and surrounding mountain belts may contain considerable amounts of low-density material. We concur with their conclusions and suggest that gravity lows in Lakshmi Planum may indicate either the presence of low-density, most probably felsic, rocks in the upper crust underlying the surface basaltic flows mapped by Ivanov & Head (2010a, b) or, as the contact between shallow-source gravity highs and lows coincides with their mapped textural boundaries, some flows mapped as basaltic may instead be more felsic. Although Venus has been thought to be dominated by basaltic crust, there is increasing evidence from emissivity data for felsic volcanism, and tessera/highlands material may also be felsic in composition (Nikolaeva et al. 1992;Hashimoto et al. 2008;Helbert et al. 2008;Mueller et al. 2008;Basilevsky et al. 2012), which is supported by petrological modelling (Shellnutt 2013). Given that felsic magma may be generated in oceanic plateaux on Earth above hightemperature mantle plumes and/or spreading ridges such as Iceland (Annen et al. 2006;Willbold et al. 2009), the presence of granites or felsic lavas in the craton-like areas of Ishtar and Aphrodite terrae is consistent with models for initial crustal thickening and plateau formation above an upwelling mantle plume. Felsic crust is not only less dense but also rheologically much weaker (more ductile) than mafic crust; this may have strong consequences for enhancing crustal deformation intensity and, especially, crustal thickening (T. Gerya, pers. comm. 2013) if felsic crust is also present on the margins of these craton-like plateaux. † The linear gravity highs south of Ishtar Terra in the total Bouguer gravity portrayed in Figure 2a correspond to belts of thinned crust  The pattern of structures is explained by two generations of rift-bounding normal faults; however, the lack of any geochronological data on Venus precludes any validation that structures with the same orientation are contemporaneous. The ensuing displacements of areas of continent-like crust in terrae imply changes in regional stress field. Superposed faults are interpreted from gravity (Fig. 2). Volcanic centres: 1, western Eistla; 2, central Eistla; 3, Bell Regio; 4, eastern Eistla; 5, Beta Regio; 6, Laufey; 7, Mnemosyne, the Arachnoid cluster Bereghinya, are interpreted to overlie plume clusters, for which the minimum diameter is estimated (from Ernst et al. 2007). (b) 3D view of the same area as in (a) viewed from underneath, colour-coded for thickness. The two 'continent-like' regions of Lakshmi Planum and Ovda Regio correspond to areas of thicker crust than the surrounding plains. Thick crustal 'roots' underlie fold belts on the north and NW margins of the interpreted Lakshmi Planum D 1 indentor (Fig. 2). The thickest crust occurs beneath Maxwell Montes, where folding and underthrusting (Keep & Hansen 1994) are interpreted to have occurred in the subsequent (D 2 ) event. Although eastern Ovda Regio is offset by a dextral shear zone consistent with the implied D 1 approximately north-south shortening, radar interpretation (Fig. 5) presents evidence for its sinistral reactivation, consistent with the model for an orthogonal rotation of principal stresses between D 1 and D 2 .
in Figure 3. The lack of a similar gravity high in the high-frequency/shallow-source gravity component in Figure 2b indicates a deep, highdensity source for the Bouguer highs in Figure  2a. This can most readily be explained by the presence of denser, underplated mantle-derived mafic rocks beneath the zone of thinned crust. The interpreted rift correlates to the Guinevere, Sedna and Leda (Lednaya) planitias/volcanic lowland plains (Markov 1986). The Guinevere and Sedna planitias are interpreted as being characterized by extensional structures by Sullivan & Head (1984), which is consistent with this rift interpretation. Data from Magee & Head (2001) indicate that the interpreted rift contains over 2 × 10 6 km 2 of volcanic flow fields. However, only several segments, especially those bordering areas of geoid lows, are interpreted as rifts in the global map of the distribution of rift zones on Venus by Ernst et al. (2007) and Krassilnikov et al. (2012).

Interpretation of regional fault and shear zones
Interpretations of gravity and crustal thickness images. Horizontal offsets of Bouguer gravity anomalies, and offset and ductile deflection ('drag') of 'worm' edges in the total horizontal gradient, define a series of linear features (Fig. 2a). The constant horizontal displacement components along their length, the observed offset of both total Bouguer and short wavelength anomalies, and the 100 km-scale resolution of gravity data together indicate that many of these lineaments are crustal-scale transcurrent shear zones. The location and sense of horizontal displacement along shear zones interpreted from gravity images in western Ishtar Terra ( Fig. 2) coincides precisely to those of transcurrent -transpressional shear zones interpreted on the margins of Lakshmi Planum from radar images (Harris & Bédard 2014;Fig. 1b). Sinistral shears on the NW margin of Lakshmi Planum are approximately 1000 km long and the curved, dextral shear zone on its NE margin is around 1800 km long. In these companion studies, Lakshmi Planum is interpreted as having acted as a rigid indentor, forming mountain belts on its northern, SE and NW margins. Gravity data thus greatly strengthen this hypothesis, and attest to the crustal scale of the structures previously interpreted only from radar images. The geometry of these structures resembles the Himalayan-Indochina system on Earth formed due to the indentation of India into Eurasia, and the length of the shear zone on the eastern margin of Lakshmi Planum is similar to the length of the Sagaing fault system on the eastern margin of the Indian indenter (Searle 2006). A similar, indenter-like geometry is defined by shear zones in SE Ishtar Terra (Fig. 2). In western Aphrodite Terra, Ovda and Thetis regiones (which correspond to Bouguer gravity lows similar to Ishtar Terra) are offset by an approximately 2500 km-long, NNW-striking dextral shear zone. Two dextral shears, approximately 1300 and 800 km long, offset the southern margin of Ovda Regio. The linear gravity highs in Guinevere and Sedna planitias, interpreted above as broad rifts, are also cut or bounded by gravity lineaments with apparent horizontal offsets. This is especially evident in Guinevere Planitia, in the central part of the map area, where gravity highs are offset or bounded by NNW-striking linear faults. Gravity 'worms' in the centre of the map area (e.g. at locations A and B in Fig. 2a) are approximately orthogonal to the bounding faults against which they are truncated, and are oblique to the overall trend of the gravity high. This contrasts to the ductile or brittleductile deflection ('drag') of 'worms' along similarly orientated dextral shears that cut the same linear gravity high SW of Lakshmi Planum (e.g. location C in Fig. 2a). Blocks A and B are separated by a fault with dextral horizontal offset, but displacement sense(s) are less clear on their other margins.
Displacement senses along crustal-scale shear zones in western Aphrodite Terra. Dextral strikeslip offsets of the thick crustal blocks/Bouguer gravity lows that correspond to Ovda and Thetis regiones in western Aphrodite Terra, as previously interpreted by Crumpler et al. (1987) from analysis of radar and topographical images (but subsequently interpreted as oceanic fracture zone/transform faults on Earth, instead of strike-slip/transcurrent faults by Crumpler & Head 1988), is apparent on both crustal thickness and gravity images (Figs 2a & 3a). A radar image of the area immediately east of the gravity lineament in Figure 4a, however, portrays transcurrent shear zones where sinistral displacements are evident from horizontal offsets and the deflection of marker layers. Sinistral shears both parallel and step en échelon along the NW-striking feature interpreted from gravity. These sinistral shear zones, NE-striking dextral shears and eastwest-trending graben are similar in geometry to subsidiary structures in transcurrent shear zones mapped on Earth and developed in analogue models (Riedel 1929;Tchalenko 1968Tchalenko , 1970Wilcox et al. 1973;Sylvester 1988;Richard et al. 1995), and define a sinistral wrench regime (Fig. 4b). In this wrench model, en échelon-stepping sinistral shears are interpreted as Riedel shears, and NE-striking dextral shears constitute secondary Riedel (R') shears (Riedel 1929;Tchalenko 1968Tchalenko , 1970). An orthogonal change in the orientations of principal strain axes is thus required, from north-south shortening/east -west extension during the dextral offset of regional blocks observed on gravity images, to bulk east -west shortening and north-south extension during sinistral wrench reactivation. These changes in principal strain axes deduced from the geometry and offset of transcurrent shears is corroborated by the same changes in principal strain axes deduced from refolded fold interference patterns and folding of early formed thrusts within Ovda Regio, documented by Harris & Bédard (2014). Our evidence for shear-zone reactivation and change in displacement sense also reconciles previously apparent contradictions in displacement sense (i.e. sinistral by Tuckwell & Ghail 2003v. dextral by Kumar 2005 established for segments of the nearby 1000 km-long, 50-200 km-wide Thetis Boundary Shear Zone system separating the eastern Ovda and NW Thetis regiones.  Figure 2, generated through combining left-and right-looking images to minimize areas of no data (white lines and rectangles). The shear system interpreted in (b) is parallel to, and eastward of, the gravity lineament (note, however, that the positions of gravity lineaments are approximate given the data resolution). As dextral displacement along this structure is apparent on Bouguer gravity (Fig. 2), sinistral shearing in (b) is interpreted as D 2 reactivation of an existing crustal-scale shear corridor. This interpretation is consistent with the permutations in principal strain axes deduced from overprinting folds and thrusts in Ovda Regio west of this shear zone (c), portrayed by Harris & Bédard (2014).

Tectonics on Earth resulting from mantle plumes and global mantle flow
Before developing a tectonic model to explain our Venus observations, this section briefly reviews salient characteristics of tectonics on Earth related to mantle plumes and global mantle flow. On Earth we have a far better understanding of tectonic processes and 3D geometry through integrated field studies constrained by geochronology, stress measurements and more detailed geophysical data than for Venus. Although there has been much debate as to the existence and tectonic roles of mantle plumes on Earth (e.g. Artyushkov 1973;Forsyth & Uyeda 1975;Anderson 2000Anderson , 2013Foulger 2010;Burke & Cannon 2014), seismic tomography (such as Lithgow-Bertelloni & Silver 1998;Montelli et al. 2006;Boschi et al. 2007; Chang & Van der Lee 2011, whose data is plotted in 2D and 3D in Fig. 5) incontestably shows the presence of deep hot mantle upwellings and cold downwellings. Recent magnetotelluric (e.g. Kelbert et al. 2012) and seismic tomographic data portray far more complex patterns of mantle upwelling and downwellings than early isolated plume models, including vertical walls, horizontal 'fingers' at the base of the oceanic aesthenosphere paralleling overlying plate motion (French et al. 2013), and spatial clusters of smaller plumes instead of single 'superplumes' (Schubert et al. 2014). Three-dimensional numerical modelling suggests that several plumes may rise from deeper upwelling walls/planiform structures (Hanjalić & Kenjereš 2006;Gait et al. 2008) in addition to forming isolated, 'mushroomlike' plumes, and that their form may change with time (Gait et al. 2008). Continental flood basalt/ LIPs on Earth (Pirajno 2000(Pirajno , 2007Ernst & Buchan 2003;Campbell 2005) -for example, the Columbia River Province (Hooper et al. 2007;Camp & Hanan 2008), the Deccan traps (Cande & Stegman 2011;Sen & Chandrasekharam 2011), the Siberian traps (Saunders et al. 2005;Kiselev et al. 2012;Howarth et al. 2014), and the North American Mid-Continent Rift, as well as rifts in Arabia, East Africa, West Antarctica and Iceland (discussed later) -are attributed to mantle plumes. Their structural features and tomographic and gravity expressions help us to interpret the Venus gravity data, and to formulate tectonic models that can explain the interplay between rifting and lateral displacements on Venus.

Regional stress patterns about mantle plumes
Upwelling mantle plumes contribute to, and may be the prime origin of, regional stress patterns for some areas on Earth. For example, Cobbold (2008) noted that maximum horizontal principal stress axes in western Europe and Scandinavia (portrayed in the World Stress Map of Heidbach et al. , 2010 converge on the Iceland mantle plume. Cobbold (2008) and Le Breton et al. (2012) suggested that compressive stresses generated by the Iceland plume explain thrust mechanisms of recent earthquakes in Scandinavia, post-Neogene basin inversion around Iceland, and contribute to basin inversion and onshore crustal shortening on North Atlantic margins. In an Early Tertiary tectonic reconstruction, Mège & Ernst (2001) showed that fold-axial traces in sedimentary rocks of the NW European shelf are concentrically arranged about the same mantle plume that now underlies Iceland when, at 60-50 Ma, Greenland was situated above it. Mège & Ernst (2001) attributed the implied radial shortening to plume-derived horizontal stresses. Similarly, the horizontal compressive stress trajectory in East Africa and Arabia is also controlled by upwelling mantle plumes. Mouthereau et al. (2012) contended that, whilst early shortening and thrusting in the Zagros is subduction and collision-related, the main driving force for Arabian plate motion since approximately 12 Ma is horizontal flow originating from upwelling mantle plumes responsible for the intracrustal extension that created the Red Sea and East African rifts (Fig. 5). Furthermore, 2D numerical modelling by Burov et al. (2007) and Guillou-Frottier et al. (2012) illustrates how plume impingement beneath shallow crust induces compressional stresses in (and, in some models, horizontal displacement of) an adjacent thicker, 'cratonic' crustal block. Zones of mantle downwelling also change the stress field in the overlying crust; for example, in the models of Behn et al. (2004), large compressive stresses developed in the upper crust over a zone of mantle downwelling located ahead of a continent being driven laterally by mantle flow from an upwelling plume. Horizontal projections of the instantaneous flow trajectories in numerical simulations of Hanjalić & Kenjereš (2006) at Rayleigh numbers similar to those used in other Venus simulations (e.g. Robin et al. 2007), although not aimed at modelling geological structures, also suggest that linear zones of focused horizontal displacement may develop between convection cells. The effect of such linear flow discontinuities on an overlying crustal 'lid' was not simulated. If such structures were to develop on Venus, could they produce the network of regional transcurrent shears we interpret from gravity, instead of conventional ideas of bulk regional shortening? Further numerical and physical models are required to test this hypothesis.
On Venus, structures formed during bulk shortening, such as wrinkle ridges and associated conjugate strike-slip faults concentric about the volcanic centres in volcanic plains and folds in crustal plateaux, were also attributed to plumes by Mège & Ernst (2001).

Relationships between rifts and mantle upwelling
Gravity signature, topographical expression and stress patterns. The 1115-1086 Ma (Heaman et al. 2007), approximately 2500 km-long, Mid-Continent Rift in the north-central USA and southernmost Canada in the Lake Superior region cuts Precambrian terrains (Fig. 6), and is marked for the most part by long linear Bouguer gravity highs (Behrendt et al. 1988(Behrendt et al. , 1990Hinze et al. 1997;Miller 2007 and references therein;Stein et al. 2011). The gravity signature of the Mid-Continent Rift closely resembles the Bouguer gravity pattern of the interpreted rift zones on Venus described above.
The Mid-Continent Rift contains an approximately 30 km-thick sequence of volcanic and sedimentary rocks (Behrendt et al. 1988;Hinze et al. 1997), largely covered by younger, Phanerozoic sediments (Miller 2007). Behrendt et al. (1988, p. 81) considered that its northern part 'may contain the greatest thickness of intracratonic rift deposits on Earth'. The rift is underlain by a zone of dense, underplated mafic mantle-derived rocks, which are responsible for its marked gravity high (Behrendt et al. 1988;Chandler et al. 1989;Thomas & Teskey 1994;Allen et al. 2006;Merino et al. 2013). An anomalously deep and generally flat Moho (developed syn-to post-rifting) is apparent on seismic profiles (e.g. Allen et al. 2006). Hinze et al. (1997) and Miller (2007) described successive stages in the tectonic model for rifting as resulting from impingement of an anomalously hot mantle plume at the base of the lithosphere, where the two arms of the rift radiate from the proposed plume head. This plume model, originally proposed by Burke & Dewey (1973), is supported by geological and geochemical data provided by Miller (2007 and references therein). Extension terminated before the rift -drift transition and the development of oceanic crust, presumably due to a change to regional shortening inboard of the Grenville Orogen during which the Mid-Continent Rift underwent minor compressional overprinting. Similarly, positive Bouguer anomalies also mark the 500 km-long and 100 km-wide Bransfield Rift in the Antarctic South Shetland Islands -Bransfield Strait area (Catalán et al. 2012), an actively extending marginal basin where rifting is superposed upon an inactive continental volcanic arc (Lawyer et al. 1995;Fretzdorff et al. 2004). Forward modelling by Catalán et al. (2012) again suggested that thinned continental crust of the Bransfield Rift is underlain by a zone of anomalous, upwelled mantle.
In broad rifts that develop due to mantle flow about an active, upwelling hot plume, the area above the causative plume may correspond to a topographical high, which may or may not be cut by active rifts, and thicker crust than surrounding rifts. For example, Iceland is a topographical high cut by active normal and transform faults and fractures (Fig. 7a, b). Iceland and its surrounding shelf correspond to a region of lower Bouguer anomalies (Fig. 7c) that punctuate the general Bouguer highs along the Reykjanes and Kolbeinsey ridges. The interpreted mantle plume in Iceland is marked by a negative Bouguer gravity anomaly (Darbyshire et al. 2000); the deep source of this anomaly produces a long-wavelength Bouguer low (Fig. 7d). Similarly, Ross Island in West Antarctica, which is interpreted to overlie the Erebus mantle plume (Kyle et al. 1992;Storey et al. 1999;Gupta et al. 2009; Mt Erebus Volcano Observatory: http:// erebus.nmt.edu/index.php/volcanology/51-vol canological-evolution (accessed June 2013)), is a topographical high flanked by intracontinental rifts (see the review by Elliot 2013). It also corresponds to an isolated negative Bouguer anomaly in a zone of rifted crust otherwise marked by Bouguer highs.
We conclude that, from comparison with these rift examples on Earth, the long, linear gravity highs of Sedna and Guinevere planitias on Venus similarly mark broad rifts of basaltic crust that are underplated by higher-density cumulates and mantle that cause their Bouguer highs. As is the case for the Mid-Continent Rift, upper crustal faults are largely obscured by an overlying, late-to postrift sequence (extensive lava flows on Venus), suggesting that the significance of these broad rifts may have been underestimated in previous studies based on radar mapping of surface features. Elevated areas with moderate Bouguer anomalies and slightly thicker crust within Eistla Regio, which are interpreted as overlying plume clusters by Ernst et al. (2007), show distinct similarities to Iceland and Ross Island, where upwelling plumes are postulated, supporting their plume interpretation. This further suggests that Eistla Regio is a topographical high that formed directly above a linear array of upwelling mantle plumes and plume clusters, and that this positive structure formed along the axis of a single rift encompassing both Sedna and Guinevere planitias.
Plume interaction in the Red Sea and East African rifts. Whilst Ernst & Buchan (2003) suggested that plumes in the geological record are characterized by areas of domal uplift, triple-point junction rifting and LIPs, more complex patterns also occur (e.g. Ş engör & Natal'in 2001; Harris et al. 2004), with long linear rifts developing due to linkages between mantle plumes (cf. May 1971). Chang & Van der Lee (2011) illustrated from an S-wave seismic tomographical velocity model that rifts in Arabia and East Africa link well-defined, distinct, upwelling hot mantle plumes in a similar manner to analogue tank experiments summarized by Harris et al. (2004). Three-dimensional isosurfaces  Figure 2a, which also correspond to thin crust (Fig. 3). (b) A simplified geological map, combining the interpretation of (a) and USGS geological map data (http://mrdata.usgs.gov/geology/state/).  (Becker et al. 2009), enhanced to highlight structural features using a combination of vertical and horizontal gradients. Iceland and the surrounding shelf constitute an elliptical elevated area above the Iceland mantle plume. (c) EGM08 (Pavlis et al. 2012) Bouguer gravity anomaly with superposed edges in the total horizontal gradient upwards continued to four levels corresponding to source depths of 15, 20, 25 and 30 km (note that extreme data corrugation of variable orientation precludes worm calculation for shallower levels as too many artefacts are created). Iceland and its surrounding shelf correspond to an elliptical area with anomalies of lower magnitude than surrounding rift areas. Gravity worms highlight linear features subparallel to spreading ridges and transform faults. A similar pattern of worms and Bouguer anomalies is seen over the Sedna and Guinevere planitas on Venus (Fig. 2a). (d) Long-wavelength Bouguer anomaly image over Iceland showing a gravity low reflecting lower-density rocks associated with the underlying mantle plume. Volcanoes are from the Smithsonian Institution Global Volcanism Program database; geology, and faults and fissures are from the GIS version of Jóhannesson & Saemundsson (2009). Late Pliocene-Lower Pleistocene bedrock shows a symmetry of geological units about the central, active spreading centres. (Fig. 5d, e) show that the individual plumes coalesce in the upper mantle to produce a broad linear zone of upwelling. The alignment of interpreted upwelling mantle plumes and plume clusters in Eistla Regio (Fig. 3) is therefore analogously interpreted as a linear zone of mantle upwelling (shown schematically in Figs 7 & 8).
Lateral displacements accompanying upwelling mantle plumes and global mantle flow Plume-related horizontal mantle flow. Upwelling mantle plumes generate a viscous force due to radial flow in the plume head that is approximately proportional to the plume's volume flux (Westaway 1993). Lithgow-Bertelloni & Silver (1998), Behn et al. (2004), Bobrov & Baranov (2011), van Hinsbergen et al. (2011 and Husson (2012) showed that (on Earth) large horizontal components of global mantle flow may arise from deep mantle upwellings, and that convective mantle drag/plume-generated flow is a significant force for driving adjacent plate motions (substantiating early ideas for the role of mantle plumes in driving plate motion mooted by Morgan 1971 andWilson 1973). For example: † Regional mantle flow ensuing from deep mantle upwelling beneath South Africa, documented by Lithgow-Bertelloni & Silver (1998), is a major factor in driving microplate motion as far as the Mediterranean (Faccenna & Becker 2010). Displacement of the Americas due to mantle flow. The westward drift of the Americas in a hotspot reference frame (Gripp & Gordon 2002;Husson et al. 2012) corresponds roughly to the beginning of active subduction tectonics on the west coast, with major orogenic pulses being associated with the accretion of island arc and oceanic plateau terranes (Coney et al. 1980;Dickinson 2006;Nelson & Colpron 2007;Ramos 2009). Can plate tectonic boundary forces account for the westward drift of the American continents? Ridge push against the eastern coast of the Americas would have been a steady westward force since the opening of the Atlantic Ocean. However, the application of a ridge push force sufficient to drive the Americas west and to create the Cordilleras requires that huge compressive stresses be transmitted through the intervening oceanic lithosphere and its junction with the continental lithosphere. Since some of the oldest, coldest, densest oceanic crust on Earth (c. 185 Ma old: Bryan et al. 1977) is located along the east coast of North America, where it is underlain by cold, oceanic mantle lithosphere and is capped Fig. 8. Simplified relationships between upwelling mantle plumes and rifting and indentation of Lakshmi Planum. Horizontal mantle flow, directed outwards from a wall of mantle upwelling linking colinear plumes, produced the northward motion and indentation of the 'craton-like' Lakshmi Planum by pushing against, and tractions at, the base of its deep lithospheric mantle keel. Rifts are segmented and offset by transfer faults that accommodate differential extension resulting from: (i) interaction with outward, oppositely directed flow from adjacent individual plumes or plume clusters; and from (ii) implied differences in flow from where flow is impeded by the presence of Lakshmi Planum's deep keel and thickened crust beneath fold belts on its margins, in comparison with unrestricted flow east of Lakshmi Planum and the fold belts. A colour version of this figure is available in the online paper.
by thick packages of sedimentary rocks (Steckler & Watts 1974), then the transmission of a compressive stress large enough to raise the Rocky Mountains should have triggered the initiation of subduction beneath the east coast of North America. The geological history of the North and South American west coasts are largely devoid of west-subducting episodes (Coney et al. 1980;Dickinson 2006;Nelson & Colpron 2007;Ramos 2009); a westward slab pull contribution therefore seems implausible.
Only the slab rollback configuration of the eastward subduction of oceanic lithosphere beneath the west coast of the Americas could have contributed to westward motion. Paradoxically, terrane accretion or shallow subduction phases, which are commonly associated with phases of compression and orogenesis (Kay & Copeland 2006;Nelson & Colpron 2007;Ramos 2009), should have inhibited or even reversed westward drift, yet westward drift of the Americas was largely oblivious to these shifts in the applied force (compression v. extension). Finally, much of the western coast of North America is defined by strike-slip fault systems and should not contribute to westward motion at all. Given the complete lack of a westward slab pull force, the long strike-slip plate boundaries, and the rough balance between compressional and extensional forces above east-directed subduction zones, it seems implausible to suggest that plate boundary forces are solely responsible for the westward drift of North America. An alternative view is that large-scale convective motions in the mantle push upon the deep, stiff, mantle keels of Precambrian craton-cored continents, and that this mantle traction force plays a major role in the westward drift of the American continents (Bokelmann 2002a, b;Liu & Bird 2002;Eaton & Frederiksen 2007). This model is applied to the indentation of Lakshmi Planum, and to shear displacements between Ovda and Thetis regiones in the Discussion.
Analogue modelling of contemporaneous folding and rifting resulting from underlying flow. Contemporaneous folding, conjugate strike-slip shearing and rifting resulting from underlying flow modelled by Ramberg (1967) using a high-acceleration centrifuge showed that upwelling and downwelling of a ductile layer underlying a semi-brittle upper layer produces shear tractions that result in areas of tensile failure and separation (equivalent to rifts), and areas of localized shortening in which conjugate shear zones and folds were developed. In this model, horizontal flow (induced by a sinking weight) was symmetrical and the resulting geometry was also thus symmetrical, and there was no thickness variation of the upper, brittle-ductile 'crust'. The effects of crustal thickness variations would be expected to enhance and localize areas of folding and shearing.

Tectonic model for linked rifting, indentation and transcurrent faulting
The comparison between observations of Venus and plume-and mantle flow-related structures on Earth outlined above leads to the schematic tectonic interpretation for the western half of the study area in 2D and 3D in Figures 8 and 9, respectively. Rifting in Sedna Planitia (Fig. 8) is attributed to crustal extension on the northern flank of a zone of mantle upwelling, linking plumes and plume clusters. Extension and rifting is produced by tractions of mantle flowing out (horizontally) away from this zone of mantle upwelling. Horizontal mantle flow pushing against the deep keel to Lakshmi Planum drives the rigid planum into the surrounding area of initially thinner crust. A fold and thrust belt is developed ahead of Lakshmi Planum (Harris & Bédard 2014), accompanied by crustal thickening. A sinistral transpressional fold belt developed on the NW planum margin also results in crustal thickening, whilst no thickening occurs in this event on its NE margin, along which dextral transcurrent displacement is interpreted by Harris & Bédard (2014). Where lateral flow away from the zone of mantle upwelling is not impeded by either the deep keel of Lakshmi Planum or the zone of mantle upwelling proposed by Mège & Ernst (2001) beneath the Bereghinya arachnoid cluster (Fig. 3a), the rift in Sedna Planitia is wider and the crust thinned more than where lateral flow is perturbed. Lateral variations in the degree of total extension are partitioned by NW-striking transfer faults. The geometry of structures in Maxwell Montes (Keep & Hansen 1994), polyphase folding in Ovda Regio (Harris & Bédard 2014) and the change in displacement sense along the shear zone separating Ovda and Thetis regiones described above necessitate a change from north -south to approximately NE-SW to east-west shortening in the highland areas. Although the relative timing of events in these different areas remains to be established, the sequence of overprinting is the same. We suggest that this can be explained by the formation of a second set of north-south to NNE-SSW-trending rifts (e.g. Leda Planitia) linking plume centres, as shown in Figure 3a. In this second event, similar forces arising from mantle flow against the thicker crust of East Ishtar Terra formed north-to NW-trending folds during crustal underthrusting and extreme crustal thickening in Maxwell Montes on the eastern margin of Lakshmi Planum, as interpreted by Keep & Hansen (1994). The structures in Maxwell Montes and the other fold and thrust belts surrounding Lakshmi Planum imply crustal shortening, although the structural styles as described by Keep & Hansen (1994) and Harris & Bédard (2014) differ greatly. The difference reflects the initial mobility of Lakshmi Planum, leading to indentation and lateral escape about its NW, north and NE margins, whereas subsequent underthusting against the eastern margin of Lakshmi Planum to create Maxwell Montes (as shown by Keep & Hansen 1994) reflects Lakshmi Planum's subsequent immobility.

Origin of features defined by gravity 'worms'
An intriguing outcome of this study is the identification of subparallel regional features from edges of horizontal gravity gradients in both the Sedna Planitia rift and across the whole northern part of the study area, including Ishtar Terra and for western Aphrodite Terra (Fig. 2). 'Gravity worms' commonly correspond to the margins of shortwavelength Bouguer anomalies (Fig. 2b) and thus act as markers that help identify transcurrent faults. Gravity worms have aided mapping of regional structures on Earth (e.g. Bierlein et al. 2006;Vos et al. 2006;Austin & Blenkinsop 2008;Heath et al. 2009;Harris & Bédard 2014), and the correspondence between transcurrent faults established from the deflection and displacement of 'worms' on the margins of Lakshmi Planum and shear zones interpreted from radar images validates their similar use in regional structural interpretation on Venus. The linear structures defined by 'worms' in Sedna Planitia resemble parallel normal faults and patterns of gravity 'worms' about mid-oceanic spreading ridges on Earth (Fig. 7), whereas the anastomosing, more concentric pattern of 'worms' over Guinevere Planitia, also interpreted as a rift, is the Fig. 9. Schematic, 3D 'cartoon' of indentation and lateral escape about Lakshmi Planum driven by tractions and push-force arising from horizontal mantle flow acting on its deep craton-like keel. A broad zone of mantle upwelling links mantle plumes (cf. Fig. 5 for the present-day Afar-East African rift system). Rifting on the flanks of this zone of upwelling is created through flow away from mantle upwelling. Plains volcanic material is rendered semi-transparent to reveal the underlying mantle interpretation. Bouguer gravity highs imply dense, mafic mantle underplating beneath rifts. same as observed over mantle plumes such as at Beta and Phoebe regiones (see Fig. 3 for the location). Could gravity 'worms' in Venus' rifts act as markers of faulting about a rift axis and be used in a similar fashion to seafloor magnetic anomalies on Earth? Further detailed comparisons between geological maps, radar and emissivity images, and 'worms' are, nevertheless, required to better understand their origin(s) and to determine whether some of Venus' lowlands may have formed during symmetrical outpourings of magma about a rift axis similar to the formation of oceanic crust on Earth.

Implications of fault/shear-zone reactivation
In contrast to studies (discussed earlier) that recognize only limited strike-slip faulting on Venus, and that of McGuire et al. (1996), who suggested that faults are less likely to be reactivated on Venus compared with Earth, we show that large strike-slip displacements are not only widespread but that shear-zone reactivation with reversal of shear sense has occurred. Fault reactivation and inversion on Venus was also proposed by Kumar (2005) and Hansen (2006). Reactivation and reversal of displacement sense along regional transcurrent shear zones is common on Earth along crustal-scale transcurrent structures in Archaean granite -greenstone terrains (e.g. Blewett et al. 2010;Leclerc et al. 2012;Harris & Bédard 2014). Multiple transcurrent (as well as normal and reverse) reactivation was documented along the Archaean-Cretaceous Darling Fault Zone on the western Yilgarn margin (summarized by Harris 1994a, b andWilde et al. 1996) and in younger deformation zones (e.g. Palaeozoic -Tertiary transcurrent displacements on the Great Glen Fault in Scotland: Holgate 1969;Le Breton et al. 2013). Fault reactivation on Earth is attributed to fault-zone weakening (e.g. Handy et al. 2001;Rutter et al. 2001) largely due to the presence of water (e.g. Regenauer-Lieb & Yuen 2003 and references therein). Gurnis et al. (2000, p. 74) concluded that, on Earth, 'old weak structures are reused by the convecting system because it takes less energy to reactivate a preexisting structure than it does to create an entirely new plate margin from pristine, intact lithosphere'. Whilst Gurnis et al. (2000) discussed fault reactivation on Earth in the context of plate tectonics, our results show that transcurrent fault formation and reactivation during changes in principal stresses can occur due solely to changes in the mantle flow field (probably due to changes in plume activity) without plate-tectonic-related stresses. The reactivation of transcurrent faults documented in our study and inversion of normal faults (e.g . Hansen 2006) show that regional faults/shear zones on Venus may be reactivated, similar to faults on Earth, despite the absence of surface water.

Implications for mantle convection and tectonic regime of Venus
The proposed causal link between mantle flow directed horizontally from a zone of mantle upwelling and the development of rifts, indentation tectonics, and strike-slip fault zones, agrees with Kohlstedt & Mackwell (2009) who, from an analysis of a likely rheological profile for Venus, inferred that 'convection and lithospheric deformation will be strongly coupled' (p. 418) and that 'regional and planetary-scale tectonics will likely directly reflect underlying mantle processes such as convection' (p. 419). Our evidence for lateral, 'plate-like' displacement due to mantle flow does not, however, indicate that plate tectonics occurs on Venus. Even in a purely stagnant lid regime (where, by definition, convective stress is less than lithospheric yield stress: Lenardic & Crowley 2012), the lid may still move passively but does not influence convection beneath it (Solomatov & Moresi 1997). Our interpretation for horizontal translation of terrains on Venus is consistent with recent research, based largely on numerical modelling, which suggests that Venus may be better considered as exhibiting intermediate 'transient' (Robin et al. 2007), 'creeping stagnant lid', or transitional (Solomatov & Moresi 1996) or 'episodic' (Turcotte 1993(Turcotte , 1995Loddoch et al. 2006 (O'Neill 2013); that is, the opposite evolutionary trend to that proposed by van Thienen et al. (2004). Our proposed conceptual model shows some similarity with the 'no subduction' regime for the hot early Earth proposed by Sizova et al. (2010) from 2D numerical experiments, as argued for by Bédard et al. (2013) and for which there is recent geochemical support (Debaille et al. 2013). In this regime, horizontal movements of deformable plate fragments driven by mantle flow are accommodated by coexisting zones of shortening and rifting; the latter are also associated with thick mafic crust formation. This similarity supports the proposed analogy of Venus surface dynamics with Precambrian subduction-free geodynamics. The low elevations predicted in the subductionless models of Sizova et al. (2010) are, however, at apparent odds with the high topography present on Venus.
In tectonic models without subduction, lithospheric shortening is accommodated by either delamination (e.g. Turcotte 1989Turcotte , 1995Pysklywec et al. 2010) or 'erosion' (Burov et al. 2007) of mantle lithosphere or through narrow zones of downwelling or 'drips' of basal lithospheric mantle into the aesthenosphere (cf. West et al. 2009;Pysklywec et al. 2010;Sizova et al. 2010). A similar association between mountain belts and zones of symmetrical mantle downwelling was proposed before the advent of plate tectonic theory (Wilson 1961), and analogue models of Pysklywec et al. (2010) show that such a mechanism is plausible to accommodate folding of the upper crust without subduction.
Further detailed integration of the displacement history along interpreted regional faults with interpreted stratigraphic relationships is required to better constrain the sequence of regional tectonic events on Venus. The structures interpreted from gravity data in our study most probably represent deformation corridors and not discrete faults, and more detailed gravity data are required to more precisely map crustal-to lithospheric-scale faults. Mapping of present-day seismicity on Venus (proposed by Balint et al. 2009;Hunter et al. 2012), through wireless seismometers  or remote detection of seismic waves in the atmosphere (Garcia et al. 2005) or ionosphere (Lognonné et al. 2006;Lognonné 2009), is required to provide crucial missing information on whether regional faults interpreted in our study are still active.

Conclusions
Bouguer gravity, enhanced and visualized using modern geophysical software, combined with crustal thickness estimates, constitute important datasets to map regional crustal rift zones and faults on Venus. The high Bouguer gravity signature of the interpreted rifts on Venus is comparable to that of the North American Mid-Continent Rift interpreted to result from mantle underplating. The surface expression of faulting does not provide a true indication of the scale of rifting, and rift-related extension is greater than previously proposed. Aerially extensive lava flows may mask early formed faults, such that visible faults record only the last stages of extension. Criteria for rifts on Venus must be revised to include the presence of long, straight Bouguer gravity highs; the full extent of rifts may thus not be portrayed in current maps that are based solely on structural interpretation of radar imagery.
In additional to local hotspot-like upwellings that may have a direct correlation with discrete volcanic centres and coronae, gravity data suggest the likelihood for an approximately 8000 km-long zone of sheet-like mantle upwelling formed by the coalescence of aligned, upwelling mantle plumes and plume clusters in Eistla Regio, similar to the coalescing and lateral flow of upwelling mantle plume in the Red Sea (but 3.5 times longer). This interpreted active volcanic zone above plumes is flanked by the Sedna and Guinevere rifts, which are characterized by crustal thinning and mafic underplating. Transcurrent shear displacements on Venus are concomitant with, and most probably compensate for, regional extension and rifting. Faults with transcurrent components of horizontal displacement that offset rifts differ from transcurrent faults formed during bulk shortening and indentation in Lakshmi Planum. Their origin compensates for differential extension between blocks where flow is hindered by the presence of either: (i) the deep keel of the adjacent Lakshmi Planum; or (ii) another zone of mantle upwelling (where narrower rifts develop) and blocks where underlying horizontal flow away from the zone of mantle upwelling is unhampered (and wider rifts develop). They show some similarities with transform faults on Earth but do not necessarily imply a central spreading axis. Changes in the locii of active rifting or variations in the amount of extension on orthogonal rifts are interpreted to have produced approximately 908 changes in principal horizontal stress orientations, leading to the reversal of shear sense on transcurrent shear zones separating Ovda and Thetis regiones, refolded folds and underthrusting to produce Maxwell Montes on the eastern margin of the newly immobilized Lakshmi Planum.
On Earth, regional mantle flow and/or plume push acting against deep cratonic keels (along with 'pull' from the related downwelling) contributes to the horizontal displacements of continents in addition to plate boundary forces. The Americas illustrate that continental blocks with the thickest keels are displaced further than terrains with no cratonic keel (e.g. the Caribbean plate). Similarly, the displacement of Ishtar Terra and the driving force for Himalayan -Indochina-like indentation and lateral escape, and lateral displacements between Ovda and Thetis regiones, are attributed to mantle flow acting on their deep crustal keels, whereas terrains with thin crust are not displaced. We concur with previous studies that there is no evidence for subduction and Venusian plate tectonics similar to what is seen on the present-day Earth. The pattern of structures derived from gravity data, especially in the Sedna Planitia rift, however, raises the possibility for features similar to those formed in oceanic crust on Earth. We provide an updated view of Venusian tectonics where large, coherent displacements of its constituent terrains occur without true seafloor spreading or subduction. We surmise that this new perspective of Venus provides an analogue for the tectonics of the Archaean Earth.
T. Gerya, L. Marinangeli, and editors M. Massironi, T. Platz and P. Byrne are thanked for their suggestions and additions to the text. Venus gravity and radar data were provided by NASA, the USGS Astrogeology Science Center, and the Jet Propulsion Laboratory (JPL) at the California Institute of Technology. P. James (MIT) kindly supplied his crustal thickness data, calculated from NASA gravity and presented in James et al. (2013). S.-J. Chang is thanked for providing her seismic velocity data and MEAN velocity model from which seismic tomography figures were generated. The GIS versions of Iceland geological maps were provided by the Icelandic Institute of Natural History. EGM08 and WGM2012 Bouguer gravity data were provided by the Bureau Gravimétrique International. The Holocene Volcanoes of the World database was downloaded from Orr and Associates (Australia). Geosoft's Oasis Montaj TM and software by P. Keating (NRCAN) were used for processing geophysical and crustal thickness data. Radar images were enhanced using the public domain software ImageJ and Adobe Photoshop TM . B. Giroux (INRS-ETE) is thanked for converting NASA gravity data to a readable format. Some features in Figure 9 are modified after symbols courtesy of the Integration and Application Network, University of Maryland Center for Environmental Science (ian.umces.edu/symbols/). This is NRCAN/ESS/GSC contribution number 20130152.