(Received 25 October 2004; revision accepted 13 July 2005)
Abstract–Landsat TM, aerial photograph image analysis, and field mapping of Witwatersrandsupergroup meta-sedimentary strata in the collar of the Vredefort Dome reveals a highlyheterogeneous internal structure involving folds, faults, fractures, and melt breccias that areinterpreted as the product of shock deformation and central uplift formation during the 2.02 GaVredefort impact event. Broadly radially oriented symmetric and asymmetric folds with wavelengthsranging from tens of meters to kilometers and conjugate radial to oblique faults with strike-slipdisplacements of, typically, tens to hundreds of meters accommodated tangential shortening of thecollar of the dome that decreased from ∼17% at a radius from the dome center of 21 km to <5% at aradius of 29 km. Ubiquitous shear fractures containing pseudotachylitic breccia, particularly in themetapelitic units, display local slip senses consistent with either tangential shortening or tangentialextension; however, it is uncertain whether they formed at the same time as the larger faults or earlier,during the shock pulse. In addition to shatter cones, quartzite units show two fracture types—a cm-spaced rhomboidal to orthogonal type that may be the product of shock-induced deformation and laterjoints accomplishing tangential and radial extension. The occurrence of pseudotachylitic brecciawithin some of these later joints, and the presence of radial and tangential dikes of impact melt rock,confirm the impact timing of these features and are suggestive of late-stage collapse of the centraluplift.
INTRODUCTION
Meteorite impacts are catastrophic and extremelycomplex deformational events marked by both extreme andrapidly changing centro-symmetric strain patterns. On the onehand, the creation of first-order structures, such as centraluplifts in complex craters, is currently best reconciled byhydrodynamic models and is consistent with the results ofcontinuum numerical modeling (e.g., Melosh 1989; Meloshand Ivanov 1999). In contrast, however, these central upliftsand the surrounding crater floor rocks contain second-orderstructures that indicate that localized strain heterogeneity andpre-existing or newly formed structural discontinuities play amajor role in the impact-induced structural evolution at scalesfar smaller than can currently be resolved with the models(e.g., Wilshire et al. 1972; Milton et al. 1996; Kriens et al.1999). Several studies of the internal structure of shallowlyeroded complex craters ranging in diameter from a fewkilometers (Upheaval Dome, Kriens et al. 1999; Kenkmannet al. 2005) to several tens of kilometers (Sierra Madera,Wilshire et al. 1972; Gosses Bluff, Milton et al. 1996;
Araguainha, Bischoff and Prinz 1994) have provided someconstraints on the deformation history accompanying impactand central uplift formation. They suggest that, whilelithology exerts some control on the types and geometries ofthe second-order structures observed, the main controllingfactors appear to be the size (duration) of the impact event andthe level of subsequent exhumation of the impact structure(Dence 2004).
The Vredefort impact structure in South Africa is one ofthe three largest impact structures known on Earth (e.g.,Grieve and Therriault 2000), having had an estimated originaldiameter of 250–300 km (Therriault et al. 1997; Henkel andReimold 1998). It is also the most deeply eroded of the three,with independent estimates suggesting removal of 5–10 km ofoverburden following its formation (McCarthy et al. 1990;Gibson et al. 1998; Henkel and Reimold 1998). Consequently,the deep levels of the impact structure that are exposedprovide an unparalleled opportunity to study the subcraterbasement features relating to the impact. This study focuseson the well-exposed Archean supracrustal rocks that occupyan intermediate radial position within the 80 km wide central
1538 F. Wieland et al.
uplift of the structure (the Vredefort Dome). It complementsprevious structural studies of the crystalline basement core ofthe dome (Lana et al. 2003a) and the surrounding marginalsyncline (Simpson 1978; Brink et al. 1997).
GEOLOGICAL SETTING
The Vredefort Dome is located some 120 km southwestof Johannesburg, South Africa, in the center of theeconomically important Witwatersrand Basin (Fig. 1). Itconsists of an ∼40 km wide core of Mesoarchean basementgneisses enclosed by a 20–25 km wide collar of generallyvertical to overturned late Archean to Paleoproterozoicsupracrustal strata, and is surrounded by an ∼50 km widestructural trough known as the Potchefstroom Synclinorium(Figs. 1a and 1c). Evidence of the impact origin of the domeis largely restricted to rocks within a 30–35 km radius of itscenter and includes shatter cones (Hargraves 1961; Manton1962, 1965; Albat 1988; Albat and Mayer 1989; Nicolaysenand Reimold 1999), planar microdeformation features inquartz (e.g., Lilly 1978; Fricke et al. 1990; Grieve et al. 1990;Leroux et al. 1994) and zircon (Kamo et al. 1996; Gibsonet al. 1997), coesite and stishovite (Martini 1978, 1991),microdeformation features, recrystallized diaplectic glass andshock melts in feldspars (Gibson and Reimold 2005), impactmelt breccia dikes (Reimold et al. 1990; Koeberl et al. 1996),and extremely voluminous pseudotachylitic breccia dikes(Dressler and Reimold 2004; Reimold and Gibson 2005).While these features have been studied in some detail, onlylimited investigation has been conducted of the larger-scalestructures (faults and folds) in the collar of the dome by,among others, Manton (1962, 1965) and Lilly (1978).Additional geophysical work by Antoine et al. (1990) andgeological interpretation by Lana et al. (2003a, 2003b)considered the large-scale implications of doming in terms ofdifferential block rotations, but without specifically focusingon any individual structures.
Regional reflection seismic profiles across the dome(e.g., Durrheim 1986; Henkel and Reimold 1998; Therriaultet al. 1996) indicate that the contact between the crystallinebasement and supracrustal sequence has been uplifted by aminimum of 12 km relative to the deepest part of the rimsyncline (Fig. 1c). The impact-related uplift in the centralparts of the dome is estimated at a minimum of 20 km (Henkeland Reimold 1998; Lana et al. 2003a, 2003b). Based on ananalysis of the Archean fabrics in the basement gneisses,Lana et al. (2003a) concluded that little impact-relatedrotation occurred in rocks within a radial distance of ∼12 kmfrom the center, whereas rocks beyond this distance typicallyunderwent rotations of 90° or more. The transition betweenthese two domains appears to be relatively sharp, occurringover a radial distance of only a few kilometers; consequently,the dome displays an overall plug-like geometry (Lana et al.2003a).
The limited outcrop and absence of large-scale layeringin the basement gneisses in the core of the dome hamperedefforts by Lana et al. (2003a) to establish whether it displaysa megabreccia structure similar to that proposed for otherlarge central uplifts (e.g., Ivanov et al. 1996). Assuming thatgneissic fabrics were uniformly oriented prior to impact, Lanaet al. (2003a) did obtain reasonably consistent results whenback-rotating these fabrics in the outer parts of the corearound axes parallel to the local strike of the adjacent collarstrata. In this way, they identified six sectors around theexposed portions of the dome. However, in the absence ofevidence for large radial faults delimiting these sectors in thecore of the dome, they suggested that impact-induceddifferential slip and rotations within the core of the centraluplift could have been accommodated along a pervasivepseudotachylitic breccia vein-fracture network instead.
The quality of exposure of the supracrustal strata in thecollar is better than that of the gneisses in the core of the dome(Fig. 2) and structural interpretation of the collar is aided bythe well-layered nature of the rocks, which providesnumerous marker horizons. These supracrustal rocks range inage from 3.07 Ga to ∼2.1 Ga (Armstrong et al. 1991) andcomprise two volcanic sequences (the 3.07 Ga Dominiongroup and the 2.71 Ga Ventersdorp supergroup) intercalatedwith two major sedimentary sequences (the 2.9–2.71 GaWitwatersrand supergroup and the 2.6–2.1 Ga Transvaalsupergroup). In addition to several small alkali graniteintrusions, mafic and ultramafic sills associated with theextrusive 2.71 Ga Ventersdorp event (Pybus 1995; Reimoldet al. 2000) and the intrusive 2.06 Ga (Walraven et al. 1990)Bushveld magmatic event occur in the collar rocks.
The present study focuses on the well-exposedsiliciclastic strata of the Witwatersrand supergroup that formthe innermost collar, at a radial distance 20–30 km from thecenter of the dome (Figs. 1 and 2). In comparison, the outerparts of the collar, underlain by the Ventersdorp supergroupand lower Transvaal supergroup, are poorly exposed (Fig. 2).The Witwatersrand supergroup strata consist of a lowersequence of pelitic, quartzite, and ironstone units (West Randgroup, ∼4 km thick), and an upper, quartzite-conglomerate-dominated sequence (Central Rand group, ∼3 km thick). Thesedimentary strata and the Ventersdorp-age sills weremetamorphosed before the impact event (Bisschoff 1982;Gibson and Wallmach 1995), with the maximum grade ofmetamorphism increasing with stratigraphic depth from lowergreenschist facies (∼350 °C) in the Central Rand group tomid-amphibolite facies (∼600 °C) in the lower West Randgroup (Gibson and Wallmach 1995). Pressure-temperatureconstraints and geochronological data suggest that thismetamorphism accompanied the 2.06 Ga Bushveldmagmatism that also appears to be linked to the formation ofthe alkali granites (Gibson and Wallmach 1995; Gibson et al.2000; Moser and Hart 1996).
Following the impact, the rocks experienced renewed
Structural analysis of the collar of the Vredefort Dome, South Africa 1539
metamorphism through shock heating and impact-induceduplift. Temperatures attained in the Witwatersrandsupergroup strata during this event decreased radiallyoutwards from ∼525 °C in the lowermost West Rand group to∼300 °C in the Central Rand group, and the deep levels ofburial of the rocks presently exposed at surface facilitatedwidespread recrystallization and new mineral growth,particularly in pelitic rocks (Gibson et al. 1998).
REGIONAL STRUCTURAL FRAMEWORK
Given the large size and great age of the Vredefort impactand the fact that the Witwatersrand rocks were depositedsome 700–1000 Myr prior to the impact, it is necessary toconsider the possibility that the rocks in the Vredefort Domealso record structures related to pre- and/or post-impactdeformation events. In terms of pre-impact tectonics, two
Fig. 1. a) A simplified geological map of the Witwatersrand Basin showing the main structural features (modified from Pretorius et al. 1986),the central position of the Vredefort Dome and its rim syncline, and the approximate limits of a 300 km diameter impact structure. Forsimplicity, the post-Witwatersrand supergroup lithologies are not shown. The dashed line indicates the approximate position of the section in(c). b) A simplified geological map of the Vredefort Dome (modified after Martini 1992) showing the Vredefort Granophyre dikes andtangential fold axial traces in the inner rim syncline. The discordant pre-impact metamorphic isograd (after Nel 1927) and the position of thecore-collar contact in the southeastern sector of the dome (dashed line beneath Karoo cover) are discussed in the text. c) Schematic cross-section based on outcrop, drilling, and seismic data (modified from Friese et al. 1995).
1540 F. Wieland et al.
major faulting events affecting the Witwatersrand basin havebeen identified (e.g., Myers et al. 1990; Roering et al. 1990):NE-SW regional shortening that created a series of largefault-bounded blocks in Central Rand group times (2.89–2.71 Ga) and 2.7 Ga Ventersdorp-age rifting that producedbroadly NE- and E-trending listric normal faults in the basin(Tinker et al. 2002), some of which represent reactivatedCentral Rand group faults.
Evidence exists for at least two major pre-Transvaalsupergroup faults in the northern collar of the Vredefort Dome(Figs. 1b and 2). The larger of the two trends NNE-SSW anddisplays ∼2.5 km offset of Central Rand group andVentersdorp supergroup rocks and thickening of the latter tothe west, whereas the second fault, which lies to the west,trends NW-SE and displays ∼1.2 km of offset (Fig. 1b). If the90° or more of impact-related rotation found in the collar ofthe dome is removed (see below), these faults define aconjugate set bounding a central graben structure.Paleocurrent and clast size data from the Central Rand grouprocks in the dome (Holland et al. 1990) do not support thesefaults being active in Witwatersrand times. The orientation,timing, and magnitude and sense of slip of these faults are,however, consistent with Ventersdorp-age rifting. Althoughoutcrop in the outer collar of the dome is poor, there appearsto be little, if any, offset of the Transvaal supergroup strata,suggesting that impact-related reactivation of these faultsmust have been relatively minor at best. Interpreted seismicprofiles from west and south of the dome (Pretorius et al.1986; Tinker et al. 2002) similarly suggest that most of thelarge-scale fault structures in the central Witwatersrand basin
are related to the Ventersdorp extension at ∼2.7 Ga and showlittle post-Transvaal (<2.1 Ga) reactivation.
Gravity and magnetic geophysical data and limitedborehole information from the southeastern sector of thedome suggests a complicated structure beneath the Karoosupergroup cover rocks (e.g., Pretorius et al. 1986; Antoineet al. 1990; Corner et al. 1990; Martini 1992). The dominantfeature in this sector appears to be a NW-trending horst thatdisplaces the core-collar contact radially outwards by ∼5 kmalong at least two inferred faults (Corner et al. 1990)(Fig. 1b). The pre-impact timing of this feature is indicated bythe apparent lack of offset of impact-related post-shockannealing textures in borehole core samples from the area(Martini 1992). Given the results of Tinker et al.’s (2002)regional study, this is most likely a Ventersdorp age (2.7 Ga)structure, as are the large subsurface faults inferred in thenorthern collar of the dome (Fig. 1c).
In addition to these large faults, Albat (1988) proposed20–30° of pre-impact scissor rotation along a radial fault inthe western collar of the dome to explain anomalous shattercone-bedding relationships. In general, however, thesupracrustal sequence exposed in the northern and westernsectors of the dome shows strong continuity of strata withminimal thickness variations along strike (Holland et al.1990). This points to conformable relationships ofstratigraphic units and thus relatively minor tectonicdisturbance prior to the impact event.
In addition to the inferred large faults in the southeasternsector of the dome, the few small outcrops of Witwatersrandsupergroup strata in this area display moderately steep radialoutward dips, in contrast to the predominantly steepoverturned inward dips in the remainder of the collar (Figs. 1band 3). Various explanations, including post-impact regionaltilting toward the northwest (McCarthy et al. 1990), or NW-directed post-impact thrusting (Friese et al. 1995) have beenproposed. Lana et al. (2003a) noted that the pre-impactmetamorphic isograd in the collar rocks cuts upward throughthe collar stratigraphy toward the northwest (Fig. 1b). Byassuming that this isograd formed horizontally, they suggestedthat the discordance and the slight NW-SE elongation of thebasement-supracrustal contact indicated a pre-metamorphic(and thus pre-impact) northwestward tilt of the strata.However, while a tilt of only a few degrees can explain thediscordant isograd, the tilt required to reconcile the dipvariation in the collar strata would have had to have been ofthe order of 30–60°, which would be implausible on a regionalscale. The most likely explanation is that the anomalous dipsin the poorly exposed southeastern sector reflect pre- or syn-impact fault-block rotations, the former associated with thesignificant zone of Ventersdorp faulting lying southeast of thedome (Pretorius et al. 1986) (Fig. 1a) and the latter related toformation of the central uplift (Fig. 8). As shown in the nextsection, bedding orientation varies considerably in localizedareas throughout the collar of the dome.
Fig. 2. A landsat TM image of the Vredefort Dome showing the well-exposed nature of the Witwatersrand supergroup rocks in the collar.The alternating quartzite-metapelite succession of the West Randgroup can be clearly distinguished from the more massive quartziteunits of the Central Rand group in the northwestern sector. The ovaloutcrop pattern of an alkali granite intrusion in the northwesternsector WNW of Parys is prominent, as are the two large pre-impactfaults in the northern sector (see text for details).
Structural analysis of the collar of the Vredefort Dome, South Africa 1541
Given the metamorphism of the Witwatersrandsupergroup strata to lower greenschist- to mid-amphibolite-facies grades shortly before the impact (Gibson and Wallmach1995), older fault zone rocks are likely to have been partiallyor completely annealed. Some evidence exists for at least twoepisodes of small-scale crenulation folding and foliationdevelopment in the metapelitic units in the collar during thismetamorphism (Gibson 1993; Gibson and Reimold 2001),and similar features have been described associated withbedding-parallel thrusting in the Witwatersrand goldfields(Phillips and Law 1994). However, no map-scale structuresappear to be related to this deformation. The emplacement ofthe alkali granite plutons may account for some of thestructural complexity of the Witwatersrand supergroup stratain their vicinity, with bedding striking radial to the dome inplaces (Fig. 2).
Reconstruction of the Vredefort craterform based ongeophysical data led Henkel and Reimold (1998) to proposethat post-impact SE-directed thrusting, possibly related to thepoorly constrained Mesoproterozoic Kheis orogeny, may
have shortened the northwestern part of the impactstructure by ∼65 km. They also proposed uplift by severalkilometers of the southeastern third of the structure along amajor, NNE-trending lineament or flexure now lying beneaththe Karoo basin, or as a result of NW-directed, 1.1 Ga,Namaqua-Natal thrusting (see also Friese et al. 1995).However, they noted that these events did not substantiallyaffect the gross structure of the Vredefort Dome itself.
RESULTS
The present study involves interpretation of the large-scale structure of the northern and western parts of the collarof the Vredefort Dome using 1:250,000 Landsat TM (Fig. 2),1:25,000 stereoscopic aerial photographs, and 1:10,000orthophotos, together with both regional and detailed localstructural mapping. Most data were collected from quartziteunits of the West Rand and Central Rand groups because oftheir generally excellent exposure (Fig. 2). The principalresults are presented below.
Fig. 3. Orientation of bedding in the Witwatersrand supergroup in the Vredefort Dome. a) Interpretive sketch map of the dome from Antoineet al. (1990) showing their proposed polygonal geometry. b–g) Lower hemisphere equal area Schmidt net projections of contoured poles tobedding, grouped according to Antoine et al.’s (1990) proposed polygonal structure. Apart from the southeast sector and small areas withinthe northern and northeastern sectors, bedding is overturned. Statistical data: b) mean 36°/300°, maximum density 27.8; c) mean 30°/336°,maximum density 25.9; d) mean 24°/349°, maximum density 16.9; e) mean 36°/007°, maximum density 11.8; f) mean 36°/053°, maximumdensity 17.3; g) mean 60°/144°, maximum density 30.7.
1542 F. Wieland et al.
Gross Collar Morphology
The results of our analysis of bedding orientations in theWitwatersrand supergroup strata around the dome (Fig. 3)confirm the mostly overturned nature of the strata in thenortheastern, northern, and western parts of the collar andmoderate, right-way-up dips in the southeastern sector;however, dips in the northern and northeastern quadrants arealso locally right-way-up. Within the limitations of thedataset, which shows significant spread in beddingorientations within comparatively small areas (Figs. 3b–3g),the data do not appear to support Lana et al.’s (2003a)contention that the angle of overturning is discernibly lowerin the southwestern and northeastern quadrants than in thenorthwestern quadrant because of a pre-impact shallownorthwestward dip of the supracrustal strata. This implies thatthe average 120° of overturning of the collar rocks (Fig. 3) isa true reflection of the amount of impact-induced rotation.
Based on interpretation of regional gravity and magneticdatasets, Antoine et al. (1990) suggested that the dome has apolygonal, rather than circular, shape. They divided theexposed section of the dome into six segments, arranged atangles of 40° to 45° to each other, with possibly two moresegments to the south hidden beneath the Karoo Supergroupcover rocks (Fig. 3a). In order to test this polygon model, wesubdivided our dataset to correspond to Antoine et al.’s(1990) segments (Fig. 3a). The resultant stereonets (Figs. 3b–3g) show considerable variation in bedding orientationsstemming from both symmetric and asymmetric folding ofthe bedding on a ten to hundred m-scale (km-scale fold datasuch as shown in Figs. 4 and 5 were excluded from thedatasets) and fault-related block rotations (Fig. 8). The resultsindicate angles of 30° to 45° between the individual maximaof poles to bedding only in the northeastern and westernsectors (Figs. 3b, 3c, 3e, and 3f); however, considerableoverlap exists even for these sectors, indicating that beddingrotation in response to doming was distributed on a smallerscale than that envisaged by Antoine et al. (1990). A furtherargument against the polygonal model is that our study failedto identify any large radial faults extending across the entirecollar of the dome in the positions proposed by Antoine et al.(1990) (compare Figs. 2 and 3a).
A first-order calculation of the amount of tangentialshortening associated with the formation of the central upliftwas made by comparing the cumulative strike length ofthree marker horizons in the Witwatersrand supergroup on the1:50,000 geological map of the Vredefort Dome (Bisschoff2000) with the calculated circumference based on radialdistance from the center of the dome (Table 1). In the HospitalHill Subgroup quartzite horizons of the West Rand group(Figs. 2, 4, and 6), the measured shortening by faulting andfolding is approximately 17%, whereas the amount ofshortening in the upper Government Subgroup and the upperunits of the Central Rand group (Turffontein subgroup) is less
than 5% (Table 1). Given the structural complexity of thecollar, the impersistent outcrop in places, the variation in radialdistance of individual marker units from the center (Fig. 2),and the existence of Ventersdorp-age extensional faults, thesevalues should be regarded as only crude approximations. Forinstance, a 1 km variation in radial distance changes thecalculated shortening by approximately 5%.
Folds
Based on the regional-scale mapping and image analysis,ductile strain in the collar rocks is heterogeneous. Forconvenience, three structural subdomains have beenidentified, although, as argued below, they form a continuum:1) relatively straight segments up to several km long with orwithout internal m- to km-scale gentle to open symmetricundulations; 2) open to tight km-scale symmetric foldsoriented radially to the dome; and 3) asymmetric km-scaleopen homoclinal folds with axial planes trending oblique tothe dome.
Much of the local variation in bedding orientationrecorded in individual sectors around the dome (Fig. 3) relatesto the outcrop-scale gentle folding of bedding, although fault-related block rotations also play a role locally. Smaller-scalefolds (meters to tens of meters) appear to be restricted to thestrongly layered West Rand group. Large, km-scale, broadlysymmetric, gentle to open folds occur in both the West Randand Central Rand group strata, but are more common in theformer. At an even larger scale, the collar rocks show acuspate-lobate pattern indicating heterogeneous strain aroundthe dome. The cusps point both towards and away from thecenter of the dome (Fig. 2) and are invariably faulted.
The most intense km-scale folding occurs in the WestRand group in the northwestern sector of the collar where fivesynformal anticlines are developed in the Hospital HillSubgroup over a strike length of ∼13 km (Fig. 4). Symmetricfolds in the northern and northeastern sectors of the dome aregenerally more open than those in the northwestern sector,with interlimb angles typically between 20° and 30°. All thefolds show synformal anticline geometries, with subverticalaxial planes and steeply inward-plunging hinges, both ofwhich trend radial to the dome. Bedding in the northwesternsegment displays a general southeasterly overturned dip(Figs. 3c and 4). The three central folds (B to D in Fig. 4) arelargely symmetric with radially-striking axial planes and arecharacterized by limbs of roughly equal length that strikeoblique to the general NE-SW bedding trend. In two of thesefolds, the layering on the southwestern limb is partially orcompletely right-way-up (folds B and C, Figs. 4c, 4d, and 5),whereas the more open fold (D, Fig. 4e) displays overturnedlayering on both limbs. Fold D is less disrupted by faults thanfolds B and C, which have been displaced radially outward byconjugate faults by 0.5–2 km (Figs. 4a and 5). Folds B and Dhave steep to vertical radially-trending axial planes and
Structural analysis of the collar of the Vredefort Dome, South Africa 1543
moderate to steeply radially inward-plunging hinges, whereasfold C has a gently southwest-dipping axial plane andtangential gently southwest-plunging hinges, (Fig. 4d). Giventhe general radial trend of the other symmetric folds aroundthe dome, we suggest that post-folding fault-block rotationmay explain this particular anomaly. On a smaller scale, thefold limbs display m- to dm-scale secondary folds and boththe limbs and hinges are disrupted by curviplanar faults thatare typically poorly exposed.
Quartzite units in the hinges of the more intensesymmetric folds (e.g., fold B, Figs. 4 and 5) are highlyfractured and progressive rotation of bedding around thehinge is typically not seen (Fig. 5b). Instead, these hingezones contain irregular networks of pseudotachylitic brecciaup to 20–30 m wide (see also Dressler and Reimold 2004). Incontrast, pseudotachylitic breccias in the quartzites in the foldlimbs or in the straight segments of the collar are considerablyless voluminous and are generally bedding-parallel with onlyminor discordant offshoots (see below). Outcrop of themetapelitic units in the fold hinges is generally poor;however, they do not appear to show similar voluminousnetwork breccias.
Folds A and E in the northwestern sector of the dome areasymmetric and both display sinistral vergence (Fig. 4). Theyare homoclinal, with short southwestern limbs striking
between 45° and 90° to the regional trend of bedding. Whilebedding remains overturned on the short limb of fold E(Fig. 4f), bedding in fold A has a steep right-way-uporientation (Fig. 4b). Fold A is associated with a smaller,open, asymmetric, dextral-verging fold some 500 m to thenortheast that suggests that it is, in fact, part of a largerasymmetric box fold. The short limb of fold E is displacedalong a radial fault with a sinistral slip sense (Fig. 7b),whereas fold A lies within a few km of a major oblique faultwith a sinistral slip sense (compare Figs. 2, 4, and 6). Thehinge of fold E plunges steeply south-southeast and thefoldaxial plane dips steeply to the southeast (Fig. 4f). Incontrast, fold A displays a moderately south-southwestplunging fold hinge and a moderately southwest-dipping foldaxial plane (Fig. 4b). A noteworthy aspect of the asymmetricfolds is that, despite exposure of more than 200° of arc-lengtharound the dome, the major asymmetric folds all show asinistral vergence.
Faults
The quartzites of the Witwatersrand supergroup provideexcellent markers with which to investigate faulting in thecollar of the Vredefort Dome (Figs. 6 and 7). Although thefaults themselves seldom crop out, the exposures afforded by
Fig. 4. Large-scale symmetric and asymmetric folds in the Hospital Hill Subgroup in the northwestern collar of the Vredefort Dome. a) Portionof Landsat TM image of the northwestern collar of the dome (Fig. 2), showing the locations of the folds. b–f) Lower hemisphere equal areaSchmidt net projections showing fold data (dots = poles to bedding, great circles = average bedding in limbs, bold great circle = axial plane).Note that in (c) the limb orientation is more complex, with both right-way-up and overturned portions on the western limb (see Fig. 5b fordetail).
1544 F. Wieland et al.
the significant topographic relief in the collar and analysis offractures in the vicinity of the fault traces suggest that mostfaults are vertical. The orientation of the faults varies fromradial to oblique with respect to the circumference of thedome (Fig. 7) and the largest faults can be traced for more than5 km along strike. Among the larger faults, oblique trendsdominate in the western sector, whereas the northwestern,northern, and northeastern sectors display a preponderance ofradial fault orientations (see also Lilly 1978) (Fig. 6).
Curvature of the faults is common, particularly in the westernsector where the faults are longest. Based on Nel’s (1927)map, Lilly (1978) suggested that most of the curved faults areconcave toward the core of the dome; in contrast, we detectedno preferential pattern. Faults in the Transvaal supergrouprocks in the outer collar show similar oblique-radial strikesand apparent offsets to the inner collar faults (Simpson 1978;Bisschoff 2000), confirming that these structures are youngerthan the ∼2.7 Ga Ventersdorp age faults.
Fig. 5. Structural data from fold B in Fig. 4 and its vicinity showing the relationship between bedding, faults, and joints. The scatter of beddingdata on the western limb of the fold (b and f) is a consequence of dm-scale folding and the change from right-way-up dips near the hinge ofthe fold to overturned dips in the south. a) A sketch map of Hospital Hill quartzite from which data was obtained. b) An annotated aerialphotograph of fold, showing faults and absence of clear hinge curvature. Diagrams (c) to (f) are lower hemisphere equal area Schmidt netprojections of the poles to bedding and joints (contour plots) in the continuations to the (c) SW and (d) NE of the fold, (e) northern limb offold and (f) western limb of fold (great circles refer to average bedding orientations, Roman numerals to joint sets described in text).
Structural analysis of the collar of the Vredefort Dome, South Africa 1545
Horizontal to shallowly plunging grooves on slickensidefracture surfaces in the fault zones indicate that the faultsinvolved principally strike-slip movement. Based on offset ofthe quartzite units, displacement is typically of the order ofhundreds of meters, although a few faults show displacementsbetween 1 and 2 km. Most of the large radial and obliquefaults display sinistral offset, whereas the smaller faultsdisplay either sinistral or dextral offsets and define aconjugate pattern consistent with horizontal shorteninghaving been oriented tangential to the dome. The exception isin the northern sector where one of the large radial faultsdisplays dextral offset; however, this has been interpreted as aVentersdorp age fault (Figs. 1b and 2). The size of the faults interms of both strike length and displacement appears todecrease radially outward, with faults in the Central Randgroup seldom displaying slip magnitudes of more than 100 m.
The faults cut the symmetric and asymmetric folds in theWest Rand group (Figs. 4–6), suggesting that they postdateductile deformation; however, the similar sinistral asymmetryof both the asymmetric folds and the majority of the faults andthe close spatial association between symmetric folds andconjugate faults (e.g., Fig. 5) suggest a consistent strain fieldfor their development.
The fault zones are marked by intense fracturing in thequartzites. They regularly contain pseudotachylitic brecciavein networks up to several meters wide and tens of meterslong, but cataclastic fault breccias have not been found.Together with the pseudotachylitic breccias in the hinges ofthe folds, these are the most voluminous occurrences of thisrock type in the collar of the dome. No surface exposures offaults have been found in the West Rand group metapeliticrocks.
In the wallrocks adjacent to some of the larger faults, thestrike of bedding is progressively rotated in a sense that isconsistent with normal drag folding related to the slip alongthe faults. In addition, however, some radial and oblique-striking faults separate rigid blocks of hundreds of meters toseveral kilometers length that show distinct differences inbedding dip of up to 30–50° (Fig. 8). This most likelyrepresents scissor-type block rotation associated with thefaulting, and may, for example, account for the anomalousorientation of the symmetric fold C shown in Fig. 4.
In addition to the radial and oblique faults, several large,subvertical, tangential faults showing west-side-downdisplacement occur along the core-collar contact in thesouthwestern sector of the dome (Lilly 1978). Such faults are
difficult to locate in the more deeply weathered metapeliticand ironstone units of the Witwatersrand supergroup;however, sharp variations in dip angles of up to 40° betweenstratigraphically-adjacent sedimentary units suggest that suchfaults must exist.
Joints and Shear Fractures
Previous studies of fracture phenomena in the VredefortDome have focused primarily on shatter cones (Hargraves1961; Manton 1962, 1965; Albat 1988; Albat and Mayer1989) and the so-called multiply striated joint sets (MSJS)that Nicolaysen and Reimold (1999) linked to the shattercones. The results of our study of shatter cones are the subjectof another paper and will not be discussed further here, exceptto note that they assist in constraining the formation age ofother fractures (Fig. 9a).
Given the general ease with which joints appear to formin upper crustal environments, the most obvious problem iswhether particular joint sets can be related with any certaintyto the impact event. Apart from the distinctive shatter conesand MSJS, some fractures can be attributed to the impact-related deformation as a consequence of their association withimpact-related melts. The most obvious of these are the ninedikes of impact-melt rock (Vredefort Granophyre) that occurin single or en-echelon radial and tangential fractures of up toseveral km length in the inner collar and outer core of thedome (Therriault et al. 1996) (Fig. 1b). These dikes, whichhave widths ranging from 10 to 60 m, appear to be unaffectedby either the large-scale folds or the radial and oblique faultsin the West Rand group (Bisschoff 2000), consistent withtheir emplacement having occurred after tangentialshortening. This suggests that the latter stages of the evolutionof the dome, prior to final crystallization of the impact melt,were characterized by simultaneous tangential and radialextension. A similar conclusion of radially directed dilationcan also be drawn from the dm-thick bedding-parallelpseudotachylitic breccia veins that extend for tens tohundreds of meters in the West Rand group quartzites, forexample, along the northern limb of fold E in Fig. 4. Like thevoluminous pseudotachylitic network breccias that occurpreferentially within quartzites in the large fault zones and inthe hinges of the large-scale folds, the fillings of these veinsappear to have migrated from their sites of generation (seeDiscussion section).
In the metapelitic rocks of the West Rand group,
Table 1. Calculated shortening of the exposed sections of Witwatersrand supergroup strata in the Vredefort Dome based on the 1:10,000 geological map of Bisschoff (2000).
however, a more complex geometric relationship existsbetween fractures and pseudotachylitic breccia veins. Therocks are pervaded by a network of hairline, irregular tocurviplanar shear fractures that display mm- to cm-scaleoffset of both bedding and pre-impact metamorphicporphyroblasts (Gibson et al. 1997) (Fig. 9b). Gibson (1996)and Gibson et al. (1997) described this phenomenon as a“spaced fracture cleavage” because of the intense, cm- to dm-scale spacing of the fractures. The fractures commonlycontain mm-thick veinlets and pods of pseudotachyliticbreccia that have chemical compositions consistent withlargely in situ derivation, and the veins and fractures areovergrown by the post-impact metamorphic paragenesisfound in the host rock, which commonly obscures thefractures in hand specimen (Gibson et al. 1997). Locally,larger breccia networks are present. Precise slip directionsalong the shear fractures are difficult to determine, and offsetsof bedding locally suggest either horizontal tangentialextension or shortening. The latter may indicate formation ofthese fractures in conjunction with the large-scale folds andfaults, while the former might be compatible with either theshock compression stage or the final collapse of the centraluplift (see Discussion section). Morphologically similar
breccia-filled fractures in the core of the dome (Gibson andReimold 2005) and the Central Rand group quartzites in thenortheastern collar (Martini 1991) have been identified asshock features because of the localization of shockmetamorphic effects along their margins.
Thin psammitic layers within the pelitic rocks locallyshow an orthogonal to rhomboidal fracture pattern on beddingsurfaces involving up to three sets of closely spaced (typicallymm- to cm-scale spacing and tens of centimeters long)fractures, similar to the pattern seen in thinly bedded quartziteunits (Fig. 9c). In both scale and geometry, these fracturesresemble the “shatter cleavage” described by Milton et al.(1996) in the thinly bedded units in the Gosses Bluff structure(Australia) that they linked to the shatter cone phenomenon.Their lack of discernible dilation distinguishes them from thelarger-scale, more irregular joints, which clearly cut them andwhich have a dilational origin (Fig. 9c). These joints form themost striking large-scale fracture phenomenon in these rocks(Fig. 10). Joint spacing and length is generally directlyproportional to layer thickness, with the largest joints in thequartzites being hundreds of meters long and tens of metersapart (Fig. 10a). Although Lilly (1978) suggested that thejoints are parallel to the faults in the collar, his analysis was
Fig. 6. a) Outcrop distribution of Witwatersrand supergroup strata in the Vredefort Dome showing major faults (modified from Bisschoff 2000)and rose diagrams (b–e) of fault orientations compiled after the method of Donath (1962). The arrows indicate boundaries between domainsidentified for rose diagrams. Note the conjugate pattern in all diagrams and the dominance of oblique fault orientations in the western sector(e) and radial orientations in the northwestern (d), northern (c), and northeastern (b) sectors. Several large tangential faults occur in thesouthwestern sector.
Structural analysis of the collar of the Vredefort Dome, South Africa 1547
based purely on strike data. We agree that fracture intensityincreases in the quartzites in the fault zones and that thereforeat least some of the fractures are fault-related. However, a full3-D analysis of numerous sites around the dome indicates thatthe principal geometric control on these joints is beddingorientation, with the main sets typically being oriented eitherperpendicular or parallel to bedding (Figs. 5 and 10).Fractures parallel to bedding have been removed from allstereoplots in Figs. 5 and 10, as they are represented by thebedding orientation.
Apart from the bedding-parallel fractures, the mostprominent set of joints in outcrop is typically subvertical andperpendicular to bedding (labeled I in Figs. 5, 9a, and 10).This set displays a general radial orientation with respect tothe dome. The exceptions are sections where the bedding hasbeen folded or rotated by faulting and, thus, is no longertangential to the dome (Figs. 5e and f). This set bisects theacute angle between two inclined shear fracture/joint sets (IIand III in Figs. 5e, 5f, 9a, and 10), although the two setscommonly show slightly different dips (e.g., Figs. 5 and 10)and may be unequally developed. Locally, these fracturesdisplay mm- to cm-scale normal dip-slip displacement,consistent with tangential extension and vertical shortening(Fig. 9d). A fifth set (IV in Figs. 5 and 10) strikes parallel tobedding and displays a shallow outward radial dip. Joints ofthis set are typically short and may be irregular. Themetapelitic rocks show a similar joint pattern to the quartzites(Figs. 10b and 10c). The lack of prominence of set IV in themetapelitic units might reflect the general lack of verticalrelief in the metapelitic outcrops. The amphibolite-grademetapelitic units in the West Rand group show considerablyless jointing than the intercalated quartzites, whereas the low-grade (or greenschist-facies metamorphic) metapelitic rocksfurther from the center of the dome are typically moreintensely jointed than the adjacent quartzites (Fig. 9a).
Pseudotachylitic breccias are found both within the jointsand shear fractures (Figs. 9b and 9e). In the former case, thebreccias appear to occupy purely dilational structures, aninterpretation supported by the local development of veins inen-echelon tension gash geometries (Dressler and Reimold2004; Reimold and Gibson 2005). The absence of clear-cutgeneration planes for most of the breccias in the dome hasbeen noted by other authors (e.g., Reimold and Colliston1994; Dressler and Reimold 2004). In general, thinner veinsof pseudotachylitic breccia are more likely to be displaced bysmall amounts along cross-cutting fractures (Fig. 9d). Thickerveins show less evidence of displacement by fractures. Whilesome of these joints must be related to post-impact tectonicevents, this evidence suggests that the joint pattern is largelythe product of the final stages of central uplift formation. Theoverall volume of pseudotachylitic breccia and the size ofindividual occurrences decrease radially outwards through theWitwatersrand supergroup rocks, but veins up to 1–2 cm thickare still found locally in the Ventersdorp supergroup rocks.
DISCUSSION
Implications for Central Uplift and Cratering Mechanics
Structural analysis of the Witwatersrand supergrouprocks in the collar of the Vredefort Dome indicates that, ratherthan exhibiting a comparatively simple geometry involvingrigid polygonal segments separated by radial faults assuggested by Antoine et al. (1990), the collar displays ahighly heterogeneous internal structure involving both ductileand brittle strain. The 2.06 Ga metamorphic and alkali graniteintrusive events provide a convenient time marker with whichto distinguish pre-impact brittle structures from thosegenerated during the impact event. The association of thefolds, faults, and fractures in the collar with voluminous
Fig. 7. Aerial photograph interpretations of fault patterns in West Rand group strata in the a) northern (upper West Rand group NNE of Parys)and b) northwestern (lower West Rand group WNW of Parys) sectors of the dome, showing aspects of radial and oblique faulting andpredominance of sinistral offset.
1548 F. Wieland et al.
pseudotachylitic breccias—that are themselves overprintedby the impact-related thermal event (Gibson et al. 1997)—supports their impact origins.
The bulk of the large-scale structures (folds, radial, andoblique faults) display geometries and kinematic indicatorsthat are consistent with formation during tangentialshortening of the collar strata. The amounts of shorteningcalculated in this study agree with those deduced by Manton(1965) using Nel’s (1927) original map of the dome (heestimated 7% shortening in the West Rand group and noshortening in the Central Rand group). They confirm theradial outward decrease in the amount of shortening that wasalso postulated by Simpson (1978), who noted no discernibleshortening in the Transvaal supergroup rocks in the outercollar. The values obtained—17% in the lower West Randgroup and <5% in the Central Rand group—fall well withinthe range of plastic strains predicted by numerical modelingof large central uplifts (Collins et al. 2004).
The similarity in the strain field responsible for theformation of the folds and faults and their spatial coincidence,together with the similar sense-of-shear of faults andasymmetric folds, suggests that formation of the ductile andbrittle features overlapped in time, although ongoing fault-related block rotations may explain the somewhat variedorientation of the folds (Fig. 4) and the generally lesseramounts of overturning of the southwestern limbs of thesefolds. While fold formation indicates ductile strain, the
development of brecciated hinge zones that are commonlypervaded by pseudotachylitic breccias points to brittledeformation as well, which is consistent with the high strainrates typical of impact processes. Ductile behavior in theserocks may have been enhanced by the elevated post-shocktemperatures, estimated at >500 °C in the inner parts of thecollar but decreasing to ∼300 °C in the Central Rand grouprocks (Gibson et al. 1998). This temperature variation—theproduct of differential shock-induced heating and the pre-impact geotherm (Gibson et al. 1998; Gibson and Reimold2005)—may have played a role in the radial outward decreasein the relative importance between folding and faulting seenin the collar. However, this pattern may equally be the resultof the radial outward decrease in the amount of shorteningacross the collar, or the change from a mechanicallyheterogeneous sequence of quartzite and metapelitic units inthe West Rand group, which should have favored folding, tothe more homogeneous and more competent quartzite-dominated succession in the Central Rand group, whichmight have favored faulting.
The predominantly radial to radial-oblique arrangementof structures in the Vredefort Dome and the dominance ofsubhorizontal displacements in their formation contrast withthe tangential to oblique-tangential arrangement ofcentripetally directed thrusts found in the central uplifts ofsmaller craters such as Gosses Bluff (Milton et al. 1996),Upheaval Dome (Kriens et al. 1996; Kenkmann et al. 2005),
Fig. 8. a) Variable dip orientation of West Rand group quartzites disrupted by radial faults in the northwestern part of the collar. The quartzitepackage is 70 m wide. b) Lower hemisphere equal area Schmidt net projection of the poles to bedding in the different fault blocks B, C, andF. Average bedding orientations are shown as great circles. Data indicate more than 90° of rotation about a subhorizontal, broadly tangentialaxis.
Structural analysis of the collar of the Vredefort Dome, South Africa 1549
Fig. 9. Fracture and joint phenomena in the Witwatersrand supergroup rocks in the collar of the Vredefort Dome. a) Shatter cones (center, left)in metapelitic unit cut by four sets of joints, Central Rand group in the northwestern collar. The vertical (II in text) and horizontal (IV) setsare most intense, with set II trending NNE across the image (right side) and set III only poorly developed (bottom, center). b) Pseudotachyliticbreccia-bearing shear fractures showing mm- and cm-scale slip in Hospital Hill subgroup metapelite from the northern limb of fold A (Fig. 4).A weak conjugate pattern is suggested. Although only one set of fractures displays discernible slip, this is consistent with tangential extension(dextral slip sense). However, one fracture (center, left) shows opposing slip. Such complexity is typical of these fractures (length of pen10 cm). c) Intense orthogonal fracture pattern in Hospital Hill quartzite in the northern collar (east of fold E, Fig. 4). The orthogonal fracturesare cut by younger, irregular, conjugate joints (NE and NW orientations in the photo), which are also responsible for the way in which the slabhas broken. d) View of vertical bedding surface showing conjugate shear fractures (sets II and III, see text) cutting thin pseudotachylitic brecciavein (horizontal, east-west), western limb of fold E (Fig. 3) (length of pen 10 cm). e) Shallowly outward-dipping pseudotachylitic breccia(pt 2) filling type IV joint cuts a bedding-parallel vein (pt 1), Hospital Hill subgroup, northeastern collar.
1550 F. Wieland et al.
Haughton (Bischoff and Oskierski 1988), and Sierra Madera(Wilshire et al. 1972). While this might reflect differing levelsof erosion of the central uplifts, with the zone of convergenceof the transient crater wall slumps still preserved in thesmaller central uplifts, it is more likely that the center of theVredefort crater was never marked by such a structure. This isbecause the inward propagation of the faults from thecollapsing walls is not sufficiently fast to reach the center of avery large crater before the floor starts to rebound (Dence2004). Such structural complexity probably exists, instead, inthe peak rings of such craters with the added complexity of astrong component of centrifugal thrusting driven by outwardcollapse of the central uplift, opposing the centripetal motionof the walls. On this basis, it does not appear plausible that thesteeply radially-plunging folds in the collar of the VredefortDome represent rotated radial transpression ridges, aspostulated in Kenkmann and von Dalwigk’s (2000) genericmodel.
At the present levels of exposure of the Vredefort craterstructure, indications of central uplift collapse are providedby the upright to slightly centrifugally verging folds in the rimsyncline (Simpson 1978) (Fig. 1b). Further afield in thegoldfields that lie close to the inferred northern edge of theimpact structure (Fig. 1a), however, mining activity hasuncovered pseudotachylite-bearing listric normal oblique-and dip-slip faults that dip towards the dome (e.g., Killick1993), that could be the product of slumping of the Vredefort
crater walls. In addition, several large vertical faults in thegoldfields that strike radial to the dome also appear to havebeen active, or reactivated, during the impact event (Killick1993; Fletcher and Reimold 1989). These may representtranspressional or transtensional structures akin to the onespostulated by Kenkmann and von Dalwigk (2000) for theouter parts of complex craters.
Collapse of the Vredefort central uplift is also indicatedby the granophyre dikes and joints in the collar rocks thatdeveloped in a strain field involving simultaneous radial andtangential extension. Impact-melt dike intrusion is unlikely tohave been possible until vertical uplift had ceased, a factborne out by the lack of any evidence of displacement of thedikes by the folds and faults. In all likelihood, it indicates thatthe central uplift collapsed to the point that it formed atopographic low, allowing at least some of the impact-melt toaccumulate and be retained above it.
Further evidence for outward collapse of the centraluplift to form a peak ring may be provided by the increasingamounts of rotation measured from the center of the domeoutwards. Based on structural mapping of the crystallinebasement core of the dome, Lana et al. (2003a) concluded thata central region approximately 25 km wide experiencedminimal impact-related rotation, and that this is surroundedby an annulus comprising the outer core of the dome, where∼90° of rotation is needed to explain the present orientation ofthe gneissic pre-impact structures. In contrast, the inner collar
Fig. 10. a) Joint pattern in Hospital Hill subgroup in the northwestern collar (between folds D and E) (Fig. 4). The ridge shows the intense jointspacing typical of the quartzites in the Vredefort Dome, with four joint sets of m- to dm-spacing. b) Lower hemisphere equal area Schmidt netprojections (raw data and contour plot) of poles to joints in the quartzite in (a). Bold great circle shows the average bedding orientation. c)Lower hemisphere equal area Schmidt net projections (raw data and contour plot) of poles to joints in the metapelitic units southeast of theridge (foreground). Contours range from 1–6%.
Structural analysis of the collar of the Vredefort Dome, South Africa 1551
rocks studied here show an average of ∼120° of overturning(Fig. 3). We can find no support for Lana et al.’s (2003a)suggestion that this reflects only 90° of impact-relatedrotation superimposed onto a moderately steeply dipping rocksequence. Instead, we propose that this increase reflects theoutward collapse of the outer parts of the central uplift in thefinal stages of crater modification.
The origin of the uniform sinistral sense-of-shear onfaults and of folds in the collar of the dome is unknown.Pronounced asymmetry in the central uplifts of smallercomplex impact structures such as Gosses Bluff and UpheavalDome has been interpreted as the result of asymmetric massdisplacement of rocks close to the surface caused by obliqueimpact (Milton et al. 1996; Kenkmann et al. 2005), but thelevels exposed in the Vredefort Dome are far deeper thanthose studied in these smaller craters, making such an optionless likely. It is possible that a slight asymmetry in the targetrocks—for instance, pre-impact tilting of the supracrustalsuccession, but by a much smaller amount than proposed byLana et al. (2003a)—might induce asymmetry in the impact-related structures. Unfortunately, nearly half of the VredefortDome is buried beneath younger cover rocks, which hampersa further investigation of this problem.
Numerical modeling suggests that a central uplift the sizeof the Vredefort Dome is likely to have formed within amatter of 2–3 min (e.g., Henkel and Reimold 1998; Meloshand Ivanov 1999). During most of this time, the rocks in thecentral uplift appear to occupy radial positions further fromthe center of the impact structure than their starting positions(e.g., Collins et al. 2004); it is only in the latter stages of upliftthat tangential shortening gains in importance. It thus seemslikely that the radial folds and radial and oblique faultsdeveloped towards the end of central uplift formation.
Within a larger context, the presence of faults in thecollar but apparent lack of them in the core of the dome mayreflect a combination of increased post-shock temperaturestoward the center of the dome (Gibson et al. 1998; Gibson andReimold 2005), different rock types (heterogeneously layeredcollar rocks versus massive crystalline core), and lesseramounts of rotation in the core (plug-like geometry) (Lanaet al. 2003a). One option is that slip could have beendistributed more evenly through the rocks in the core of thedome because of the increased intensity of pseudotachyliticbreccia development toward the center of the dome (Reimoldand Colliston 1994; Gibson and Reimold 2005), therebyobviating the need for widely spaced large-magnitude faults(Lana et al. 2003a). Melosh (2005) rejected this possibility onthe grounds that the melts would have quenched almostimmediately after forming, because of the large temperaturedifference with the host rocks. However, superheated shockmelts such as those described by Gibson and Reimold (2005)may have remained liquid long enough, given the extremehost-rock temperatures found in the center of the dome(Gibson et al. 1998; Gibson 2002). Dence (2004) noted asimilar lack of macroscopic fragmentation (i.e., faults) in the
central parts of the Charlevoix and Manicouagan centraluplifts where shock pressures exceeded 25 GPa, from whichhe suggested that the amount of brittle deformation in centraluplifts may scale inversely with shock pressure.
Pseudotachylitic Breccias
Evidence in support of a syn-impact timing forpseudotachylitic breccia development in the Vredefort Domeis overwhelming (e.g., Dressler and Reimold 2004; Reimoldand Gibson 2005), allowing the breccias to be used toconstrain the impact origin of other structures. However, theexact mechanism(s) by which the breccias formed remainproblematic. Three mechanisms have been proposed (seereview in Reimold and Gibson 2005): a) localized shockmelting caused by extreme fluctuations in shock pressure, b)friction melting along slip surfaces triggered during themodification stage of cratering, or c) a combination of shockand friction melting during the shock stage. Evidence existsfor localized enhancement of shock deformation againstfractures hosting narrow melt breccia veins in both the core(Gibson and Reimold 2005) and collar (Martini 1991) of thedome, but these fractures also invariably show smallamounts of displacement. Given their small volumes, it islikely that these melt breccias crystallized or quenchedvirtually instantaneously, and thus constrain the fractures assyn-shock. However, the same argument cannot be made forthe structures hosting more voluminous breccias, as thesemay have been able to retain a molten matrix for severalminutes, if not hours (Ogilvie, personal communication2004). In contrast to the narrow veinlets, which typicallyshow a close correspondence between the chemicalcomposition of their matrix and that of their wallrocks, thelarger breccias show abundant evidence of mixing of meltfrom a variety of sources, as well as exotic clasts (Reimoldand Colliston 1994), indicating sufficient time for melts andclasts to move distances of at least meters to tens of meters.In the context of the cratering process, it is thus plausiblethat the pseudotachylitic breccias could all have formedsimultaneously during the shock stage by either shockmelting or shock + friction melting, but that crystallizationstraddled the subsequent stages of central uplift formationand collapse, depending on the degree of superheating of themelts, the host rock temperature, and the volume of meltpresent. Equally, however, some or even most of thebreccias could have formed during the modification phase inresponse to high-strain-rate slip events. It is beyond thescope of this study to explore this issue further, except that itis worth noting that thinner breccias in the collar of thedome are more likely to have been displaced by fracturesthan thicker breccias (compare Figs. 9d and 9e) and that amelt capable of surviving for even a few minutes withinsuch a complex structural environment may now reside in avery different structural context from the one in which itoriginated.
1552 F. Wieland et al.
Chronology of Impact-Related Deformation
The oldest mesoscopic impact-related structures seen inthe collar rocks of the Vredefort Dome are the shatter conesthat, together with at least some of the pseudotachyliticbreccia (Martini 1991; Gibson and Reimold 2005), formedduring the shock pulse. The orthogonal to rhomboidalfracture cleavage observed in the quartzite units may formpart of a structural continuum with the shatter cones andMSJS.
The next structures to form were the folds. Recentmodeling by Collins et al. (2004) suggests that rocks in thecentral uplift follow a trajectory in which they are initiallydisplaced downward and radially outward under shockcompression, followed by vertical uplift, only toward the endof which do they move closer to the center of the impactstructure than the position from which they started. Thissuggests that the folds most likely developed close to the endof central uplift formation, although their present overturnedgeometries may reflect subsequent outward collapse of thecentral uplift. Continuing tangential shortening wasaccommodated by asymmetric to conjugate strike-slipfaulting. Pre-existing faults do not appear to have beenparticularly reactivated. It is possible that somepseudotachylitic breccias could have formed at this stagethrough frictional melting along the faults; however, it is alsopossible that sufficiently voluminous melts created during theshock pulse (by shock and/or friction melting) could havesurvived long enough to be driven into extensional sitesopening within the younger structures, where they eitherquenched or crystallized. Given that the central upliftprobably formed within only 2–3 min (e.g., Melosh andIvanov 1999), strain-rates for the tangential shortening wereprobably of the order of at least 10−3 to 100 s−1.
Possibly overlapping the latter stages of the contractionalphase, the collar rocks underwent some tangential faultingand block rotation around horizontal tangential axes. Thesubvertical tangential faults with collar-side-downdisplacement in the western part of the collar (Lilly 1978) andradially inward-dipping faults in the outer collar with center-side-down displacement described by Simpson (1978) maybelong to this phase.
The granophyre dikes and ubiquitous joints indicateradial and tangential extension, which is consistent with late-stage collapse of the central uplift (stress-release). It ispossible that earlier-formed strike-slip faults may haveundergone reactivation or become dilated during this phase,allowing the last vestiges of impact-related melts to infiltratethem.
CONCLUSIONS
In contrast to the massive crystalline core and poorlyexposed outer parts of the Vredefort Dome, the well-layered
siliciclastic rocks of the Witwatersrand supergroup in theinner collar of the dome preserve a variety of fold, fault,fracture, and melt breccia features related to the 2.02 Gaimpact event. The bulk of the fold and fault structures relateto tangential shortening of the strata, which reached ∼17% instrata closer to the center of the dome, decreasing to <5%further out. These figures are compatible with the predictionsof total plastic strain during central uplift formation in largecomplex craters obtained by recent computer modeling(Collins et al. 2004). Tangential shortening was followed byradial and tangential extension related to the collapse of thecentral uplift that produced ubiquitous jointing and rarerfaults, and provided the opportunity for downward intrusionof dikes from the impact melt sheet. The origin and timing ofcrystallization of pseudotachylitic breccias within theVredefort impact event remains problematic and is the subjectof ongoing research.
Acknowledgments–This work was partially funded throughresearch grants from the National Geographic Society and theBarringer Family Fund for Meteorite Impact Research.Without the support of the many landowners in the VredefortDome who granted access to their properties, this work wouldnot have been possible. Comments by U. Riller,F. Schwerdtner and an anonymous reviewer substantiallyimproved earlier versions of this paper. This paper is theUniversity of the Witwatersrand Impact Cratering ResearchGroup Contribution No. 72.
Editorial Handling—Dr. Ulrich Riller
REFERENCES
Albat H. M. 1988. Shatter cone/bedding interrelationship in theVredefort structure: Evidence for meteorite impact? SouthAfrican Journal of Geology 91:106–113.
Albat H. M. and Mayer J. J. 1989. Megascopic planar shock fracturesin the Vredefort structure: A potential time marker?Tectonophysics 162:265–276.
Antoine L. A. G., Nicolaysen L. O., and Nicol S. L. 1990. Processedand enhanced gravity and magnetic images over the Vredefortstructure and their interpretation. Tectonophysics 171:63–74.
Armstrong R. A., Compston W., Retief E. A., Williams L. S., andWelke H. J. 1991. Zircon ion microprobe studies bearing on theage and the evolution of the Witwatersrand Basin. PrecambrianResearch 53:243–266.
Bischoff L. and Prinz T. 1994. Der Araguainha-Krater in Brasilien.Geowissenschaften 12:355–360.
Bisschoff A. A. 1982. Thermal metamorphism in the VredefortDome. Transactions of the Geological Society of South Africa 85:43–57.
Bisschoff A. A. 2000. The Geology of the Vredefort Dome(Explanation of Geological Sheets 2627CA, CB, CC, CD, DA,DC, and 2727AA, AB, BA, 1:50,000 scale). Pretoria, SouthAfrica: Council for Geoscience. 49 p.
Bischoff L. and Oskierski W. 1988. The surface structure of theHaughton impact crater, Devon Island, Canada. Meteoritics 23:209–220.
Structural analysis of the collar of the Vredefort Dome, South Africa 1553
Brink M. C., Waanders F. B., and Bisschoff A. A. 1997. Vredefort: Amodel for the anatomy of an astrobleme. Tectonophysics 270:83–114.
Collins G. S., Melosh H. J., and Ivanov B. A. 2004. Modeling damageand deformation in impact simulations. Meteoritics & PlanetaryScience 39:217–231.
Corner B., Durrheim R. J., and Nicolaysen L. O. 1990. Relationshipsbetween the Vredefort structure and the Witwatersrand Basinwithin the tectonic framework of the Kaapvaal Craton asinterpreted from regional gravity and aeromagnetic data.Tectonophysics 171:49–61.
Dence M. R. 2004. Structural evidence from shock metamorphism insimple and complex impact craters: Linking observations totheory. Meteoritics & Planetary Science 39:267–286.
Dietz R. S. 1961. Vredefort Ring Structure: Meteorite impact scar?Journal of Geology 69:499–516.
Donath F. A. 1962. Analysis of Basin-Range structure, south-centralOregon. Geological Society of America Bulletin 73:1–16.
Dressler B. O. and Reimold W. U. 2004. Order or chaos? Origin andmode of emplacement of breccias in floors of large impactstructures. Earth Science Reviews 67:1–54.
Durrheim R. J. 1986. Recent reflection seismic developments in theWitwatersrand Basin. In Reflection seismology: A globalperspective, edited by Barazangi M. and Brown L. Washington,D. C.: American Geophysical Union. pp. 77–83.
Fricke A., Medenbach O., and Schreyer W. 1990. Fluid inclusions,planar elements, and pseudotachylites in the basement rocks ofthe Vredefort structure, South Africa. Tectonophysics 171:169–183.
Friese A. E. W., Charlesworth E. G., and McCarthy T. S. 1995.Tectonic processes within the Kaapvaal Craton during theKibaran (Grenville) Orogeny: Structural, geophysical, andisotopic constraints from the Witwatersrand Basin and environs.In Information Circular #292. Johannesburg: University of theWitwatersrand. 62 p.
Gibson R. L. 1993. When is a hornfels not a hornfels? Metapeliticrocks from the lower Witwatersrand supergroup, VredefortDome, South Africa. South African Journal of Geology 96:42–48.
Gibson R. L. 1996. The Vredefort hornfelses revisited. South AfricanJournal of Geology 99:93–96.
Gibson R. L. and Reimold W. U. 2001. The Vredefort impactstructure (The scientific evidence and a two-day excursionguide). In Memoir #92. Pretoria, South Africa: Council forGeoscience. 111 p.
Gibson R. L. and Reimold W. U. 2005. Shock pressure distributionin the Vredefort impact structure, South Africa. In LargeMeteorite Impacts III, edited by Kenkmann T., Hˆrz F., andDeutsch A. GSA Special Paper #384. Boulder, Colorado:Geological Society of America. pp. 329–349.
Gibson R. L. and Wallmach T. 1995. Low pressure–high temperaturemetamorphism in the Vredefort Dome, South Africa:Anticlockwise pressure-temperature path followed by rapiddecompression. Geological Journal 30:319–331.
Gibson R. L., Armstrong R. A., and Reimold W. U. 1997. The age andthermal evolution of the Vredefort impact structure: A single-grain U-Pb zircon study. Geochimica et Cosmochimica Acta 61:1531–1540.
Gibson R. L., Reimold W. U., and Stevens G. 1998. Thermal-metamorphic signature of an impact event in the VredefortDome, South Africa. Geology 26:787–790.
Gibson R. L., Reimold W. U., Phillips D., and Layer P. W. 2000. 40Ar/39Ar constraints on the age of metamorphism in theWitwatersrand Supergroup, Vredefort Dome (South Africa).South African Journal of Geology 103:175–190.
Grieve R. A. F. and Therriault A. 2000. Vredefort, Sudbury,Chicxulub: Three of a kind? Annual Review of Earth andPlanetary Sciences 28:305–338.
Grieve R. A. F., Coderre J. M., Robertson P. B., and Alexopoulos J.1990. Microscopic planar deformation features in quartz of theVredefort structure: Anomalous but still suggestive of an impactorigin. Tectonophysics 171:185–200.
Hargraves R. B. 1961. Shatter cones in the rocks of the Vredefortring. Transactions of the Geological Society of South Africa 64:147–161.
Henkel H. and Reimold W. U. 1998. Integrated geophysicalmodelling of a giant, complex structure: Anatomy of theVredefort structure, South Africa. Tectonophysics 287:1–20.
Holland M. J., Stanistreet I. G., and McCarthy T. S. 1990. Tectoniccontrol on the deposition of the Turffontein subgroup around theVredefort Dome. South African Journal of Geology 93:158–168.
Ivanov B. A., Kocharyan G. G., Kostuchenko V. N., Kirjakov A. F.,and Pevzner L. A. 1996. Puchezh-Katunki impact crater:Preliminary data on recovered core block structure (abstract).27th Lunar and Planetary Science Conference. pp. 589–590.
Kamo S. L., Reimold W. U., Krogh T. E., and Colliston W. P. 1996.A 2.023 Ga age for the Vredefort impact event and a first reportof shock metamorphosed zircons in pseudotachylitic brecciasand Granophyre. Earth and Planetary Science Letters 144:369–388.
Kenkmann T. and von Dalwigk I. 2000. Radial transpression ridges:A new structural feature of complex craters. Meteoritics &Planetary Science 35:1189–1201.
Kenkmann T., Jahn A., Scherler D., and Ivanov B. A. 2005. Structureand formation of a central uplift: A case study at the UpheavalDome impact crater, Utah. In Large meteorite impacts III, editedby Kenkmann T., Hˆrz F., and Deutsch A. GSA Special Paper#384. Boulder, Colorado: Geological Society of America. pp.85–116.
Killick A. M. 1993. Pseudotachylites of the West Rand goldfield,Witwatersrand Basin, South Africa. Ph.D. thesis, Rand AfrikaansUniversity, Johannesburg, South Africa. 273 p.
Koeberl C., Reimold W. U., and Shirey S. B. 1996. A Re-Os isotopestudy of the Vredefort granophyre: Clues to the origin of theVredefort structure, South Africa. Geology 24:913–916.
Kriens B. J., Shoemaker E. M., and Herkenhoff K. E. 1999. Geologyof the Upheaval Dome impact structure, southeast Utah. Journalof Geophysical Research 104:18,867–18,887.
Lana C., Gibson R. L., and Reimold W. U. 2003a. Impact tectonicsin the core of the Vredefort Dome: Implication for formation ofcentral uplifts in large impact structures. Meteoritics & PlanetaryScience 38:1093–1107.
Lana C., Gibson R. L., Kisters A., and Reimold W. U. 2003b.Archean crustal structure of the Kaapvaal Craton, South Africa—Evidence from the Vredefort Dome. Earth and Planetary ScienceLetters 206:133–144.
Leroux H., Reimold W. U., and Doukhan J.-C. 1994. A T.E.M.investigation of shock metamorphism in quartz from theVredefort Dome, South Africa. Tectonophysics 230:223–239.
Lilly P. A. 1978. Deformation in the collar rocks of the Vredefort ringstructure. Ph.D. thesis, University of the Witwatersrand,Johannesburg, South Africa. 430 p.
Manton W. I. 1962. The orientation and implication of shatter conesin the Vredefort ring structure. M.Sc. thesis, University of theWitwatersrand, Johannesburg, South Africa. 167 p.
Manton W. I. 1965. The orientation and origin of shatter cones in theVredefort ring. Geological Problems in Lunar Research. 123:1017–1049.
Martini J. E. J. 1978. Coesite and stishovite in the Vredefort Dome,South Africa. Nature 277:495–496.
1554 F. Wieland et al.
Martini J. E. J. 1991. The nature, distribution and genesis of thecoesite and stishovite associated with the pseudotachylite of theVredefort Dome, South Africa. Earth and Planetary ScienceLetters 103:285–300.
Martini J. E. J. 1992. The metamorphic history of the VredefortDome at approximately 2 Ga as revealed by coesite-stishovite-bearing pseudotachylites. Journal of Metamorphic Geology 10:517–527.
McCarthy T. S., Stanistreet I. G., and Robb L. J. 1990. Geologicalstudies related to the origin of the Witwatersrand Basin and itsmineralization: An introduction and a strategy for research andexploration. South African Journal of Geology 93:1–4.
Melosh H. J. 1989. Impact cratering: A geologic process. New York:Oxford University Press. 245 p.
Melosh H. J. 2005. The mechanics of pseudotachylite formation inimpact events. Proceedings of ESF Workshop on ImpactTectonics, Mora, Sweden, vol. 6. Heidelberg-Berlin: Springer.pp. 55–80.
Melosh H. J. and Ivanov B. A. 1999. Impact crater collapse. AnnualReviews of Earth and Planetary Science 27:385–415.
Milton D. J., Glikson A. Y., and Brett R. 1996. Gosses Bluff, a latestJurassic impact structure, central Australia. Part 1: Geologicalstructure, stratigraphy, and origin. AGSO Journal of AustralianGeology and Geophysics 16:453–486.
Moser D. E. and Hart R. J. 1996. Age of impact melting andmetamorphism in the Vredefort structure, South Africa. InProgram and abstracts of the Geological Association of Canadavol. 21. Winnipeg, Canada: University of Manitoba. 67 p.
Myers R. E., McCarthy T. S., and Stanistreet I. G. 1990. A tectono-sedimentary reconstruction of the development and evolution ofthe Witwatersrand Basin, with particular emphasis on the CentralRand group. South African Journal of Geology 93:180–201.
Nel L. T. 1927. The geology of the country around Vredefort.Geological Survey of South Africa Special Publication #6.Pretoria, South Africa: Government Printer. 134 p.
Nicolaysen L. O. and Reimold W. U. 1999. Vredefort shatter conesrevisited. Journal of Geophysical Research 104:4911–4930.
Phillips G. N. and Law J. D. M. 1994. Metamorphism of theWitwatersrand gold fields: A review. Ore Geology Reviews 9:1–31.
Pretorius D. A., Brink W. C. J., and Fouche J. 1986. Geological mapof the Witwatersrand Basin. In Mineral deposits of southernAfrica, edited by Anhaeusser C. R. and Maske S. Johannesburg:Geological Society of South Africa.
Pybus G. Q. J. 1995. Geological and mineralogical analysis of somemafic intrusions in the Vredefort Dome, Central Witwatersrand
Basin. M.Sc. thesis, University of the Witwatersrand,Johannesburg, South Africa. 376 p.
Reimold W. U. and Colliston W. P. 1994. Pseudotachylites of theVredefort Dome and the surrounding Witwatersrand Basin,South Africa. In Large meteorite impacts and planetaryevolution, edited by Dressler B. O., Grieve R. A. F., and SharptonV. L. GSA Special Paper #293. Boulder, Colorado: GeologicalSociety of America. pp. 177–196.
Reimold W. U. and Gibson R. L. 2005. “Pseudotachylites” in largeimpact structures. Proceedings of ESF Workshop on ImpactTectonics, Mora, Sweden, vol. 6. Heidelberg-Berlin: Springer.pp. 1–54.
Reimold W. U., Horsch H., and Durrheim R. J. 1990. The“bronzite”—granophyre from the Vredefort structure—Adetailed analytical study and reflections on the origin of one ofVredefort’s enigmas (abstract). 20th Lunar and Planetary ScienceConference. pp. 433–450.
Reimold W. U., Pybus G. Q. J., Kruger F. J., Layer P. W., andKoeberl C. 2000. The Anna’s Rust Sheet and related gabbroicintrusions in the Vredefort Dome—Part of a widespread Kibaranmagmatic event on the Kaapvaal Craton and beyond? Journal ofAfrican Earth Science 31:499–521.
Roering C., Barton J. M., Jr., and Winter H. de la R. 1990. TheVredefort structure: A perspective with regard to new tectonicdata from adjoining terranes. Tectonophysics 171:7–22.
Simpson C. 1978. The structure of the rim synclinorium of theVredefort Dome. Transactions of the Geological Society of SouthAfrica 81:115–121.
Therriault A. M., Reimold W. U., and Reid A. M. 1996. TheVredefort Granophyre: Field studies. South African Journal ofGeology 99:1–21.
Therriault A. M., Grieve R. A. F., and Reimold W. U. 1997. TheVredefort structure: Original size and significance for geologicalevolution of the Witwatersrand Basin. Meteoritics & PlanetaryScience 32:71–77.
Tinker J., de Wit M., and Grotzinger J. 2002. Seismic stratigraphicconstraints on Neoarchean-Paleoproterozoic evolution of thewestern margin of the Kaapvaal craton, South Africa. SouthAfrican Journal of Geology 105:107–134.
Walraven F., Armstrong R. A., and Kruger F. J. 1990. Achronostratigraphic framework for the north-central KaapvaalCraton, the Bushveld Complex and the Vredefort Structure.Tectonophysics 171:23–48.
Wilshire H. G., Offield T. W., Howard K. A., and Cummings D. 1972.Geology of the Sierra Madera cryptoexplosion structure, PecosCounty, Texas. USGS Professional Paper #599-H. 42 p.
