Tectonophysics 607 (2013) 32–50

Contents lists available at ScienceDirect

Tectonophysics

journal homepage: www.elsevier.com/locate/tecto

Pre-existing oblique transfer zones and transfer/transform relationships in continental margins: New insights from the southeastern Gulf of Aden, Socotra Island, Yemen

N. Bellahsen a,b,⁎, S. Leroy a,b, J. Autin c, P. Razin d, E. d'Acremont a,b, H. Sloan e, R. Pik f, A. Ahmed g, K. Khanbari h a UPMC Univ. Paris 06, UMR 7193, ISTeP, F-75005 Paris, France b CNRS, UMR 7193, ISTeP, F-75005 Paris, France c CGS-EOST, CNRS-Univ. Louis Pasteur, F-67084 Strasbourg, France d ENSEGID, IPB, Univ. Bordeaux 3, 33607 Pessac Cedex, France e Lehman College, CUNY, Environmental, Geographical and Geological Sciences, New York, USA f CRPG-CNRS, BP20, 54501 Vandoeuvre-Lès-Nancy Cedex, France g Seismological and Volcanological Observatory Center, Dhamar, Yemen h Yemen Remote Sensing and GIS Center, Sana'a University, Sana'a, Yemen article info abstract

Article history: Transfer zones are ubiquitous features in continental and margins, as are transform faults in oceanic litho- Received 8 October 2012 sphere. Here, we present a structural study of the Hadibo (HTZ), located in Socotra Island Received in revised form 8 July 2013 (Yemen) in the southeastern Gulf of Aden. There, we interpret this continental transfer zone to represent Accepted 26 July 2013 a reactivated pre-existing structure. Its trend is oblique to the direction of divergence and it has been active Available online 9 August 2013 from the early up to the latest stages of rifting. One of the main oceanic zones (FZ), the Hadibo– Keywords: Sharbithat FZ, is aligned with and appears to be an extension of the HTZ and is probably genetically linked to it. Comparing this setting with observations from other Afro-Arabian rifts as well as with passive margins world- Transfer fault zone wide, it appears that many continental transfer zones are reactivated pre-existing structures, oblique to diver- gence. We therefore establish a classification system for oceanic FZ based upon their relationship with syn- Segmentation structures. Type 1 FZ form at syn-rift structures and are late syn-rift to early syn-OCT. Type 2 FZ form during Structural inheritance the OCT formation and Type 3 FZ form within the oceanic domain, after the oceanic spreading onset. The latter Oblique rifting are controlled by far-field forces, magmatic processes, spreading rates, and . © 2013 Elsevier B.V. All rights reserved.

1. Introduction these two examples, the transfer zones are both pre-existing and oblique to the divergence. This may suggest that transfer zones do Transfer zones and accommodation zones are ubiquitous features in form during early rifting, only if they reactivated pre-existing structures. continental rifts. These zones accommodate along strike structural Geometric consistencies between transfer zones changes within rifts (amount of extension, dip of tilted half ; and mid- ridge transform faults suggest that transform faults orig- e.g. Bellahsen and Daniel, 2005; Colletta et al., 1988; Corti et al., 2007; inate at structures inherited from the rifting stage (Behn and Lin, 2000; Jarrigue et al., 1990; Lezzar et al., 2002; Young et al., 2001) and have Cochran and Martinez, 1988; Brekke, 2000; d'Acremont et al., 2005, been extensively studied (e.g., for conceptual models, Bosworth, 1994; 2006; McClay and Khalil, 1998; Watts and Stewart, 1998; Wilson, Rosendahl, 1987). In the literature, accommodation zones are defined 1965). However, a significant number of studies suggest that transform as areas of diffuse strain transfer between two offset basins, while trans- faults are inherent to spreading processes and thus not related to conti- fer zones are more localized fault systems. Acocella et al. (2005) sug- nental margin segmentation (e.g., Taylor et al., 2009) but to thermal gested that accommodation zones are found in rifts and transfer zones (Sandwell, 1986), and that they may evolve rapidly after their for- in rifted continental margins. However, examples from the Gulf of mation (e.g. d'Acremont et al., 2010). Their origin remains unclear as Suez (e.g. Jarrigue et al., 1990; Moustafa, 1997) and the East African shown in recent reviews (Choi et al., 2008; Gerya, 2012). Analog models Rift System (e.g. Lezzar et al., 2002; Rosendahl, 1987) show that transfer of transform-ridge systems (Freund and Merzer, 1976; O'Bryan et al., zones were active during early stages of rifting. It is noteworthy that, in 1975; Oldenburg and Brune, 1972, 1975) showed that transform faults can nucleate spontaneously in accreting and cooling plates of wax. However, surprisingly, later experiments (e.g. Katz et al., 2005; ⁎ Corresponding author at: UPMC Univ. Paris 06, UMR 7193, ISTeP, F-75005 Paris, France. Tel.: +33 1 44 27 74 64. Ragnarsson et al., 1996) were not as successful in reproducing E-mail address: [email protected] (N. Bellahsen). transform faults (see Gerya, 2012 for more details and the complete

0040-1951/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.tecto.2013.07.036 N. Bellahsen et al. / Tectonophysics 607 (2013) 32–50 33 review). Numerical models confirmed that transform faults may indeed in the proximal margin domain does not necessarily imply that there form spontaneously during oceanic spreading, even without initial is no genetic link between them. Two cases are possible. First, some ridge offset (Choi et al., 2008; Gerya, 2010, 2012 and references transfer zones (parallel to divergence) can be imaged in distal margins therein). and their relations with transform faults can be discussed. Of course, So far, few studies have attempted to determine, from onshore field the imaging of such sub-vertical features is challenging and can be data, whether or not transfer zones influence the location of oceanic questioned, but several datasets show their existence especially in the transform faults. MacDonald et al. (1991) proposed a hierarchy for distal margins (e.g. d'Acremont et al., 2005). Second, the fact that trans- ridge discontinuities. From first order discontinuities (transform faults) fer zones are oblique does not imply that they are unlike precursors of to 2nd, 3rd, and 4th order (overlapping or connecting segments), this transform faults as stated by Taylor et al. (2009). The question is, can classification does not explicitly incorporate considerations about in- we find any structures in the continental margin that determine the lo- heritance from syn-rift times. As noted above, the fact that transfer cation and geometry of the transform faults? And, if so, can we base a zones in rifts are often oblique to the divergence direction may be of im- classification for oceanic FZ on their relationships with continental portance. Indeed, in continental rifted margins, syn-rift faults that are transfer zones? parallel to the future transform faults, i.e. parallel to divergence, are al- In this contribution, we present a study of the structural evolution of most never observed. These results led Taylor et al. (2009) to suggest an onshore oblique transfer zone, based on a new geological map (Leroy that transform faults do not form at transfer zones. It may indeed be et al., 2012; Razin et al., 2010). This feature is located in Socotra Island the case in some passive margins, as demonstrated by the numerous (in Yemen, Fig. 1), which constitutes a part of the southeastern margin above-cited analog and numerical models. However, the observation of the Gulf of Aden. From structural and fault slip data, we detail the that transform faults are rarely or never parallel to any transfer zones structural evolution of transfer zone, that continues offshore as an

Fig. 1. Topographic–bathymetric map of the Atlantic and Afro-Arabian oceanic and rifted domains (from GeoMapApp, Ryan et al., 2009). Two rifts (Gulf of Suez and East African Rift Sys- tem) and one oceanic domain (Eastern Gulf of Aden) are briefly reviewed or studied here. Examples from the Norwegian margin or South Atlantic domain are indicated. 34 N. Bellahsen et al. / Tectonophysics 607 (2013) 32–50 oceanic FZ (Leroy et al., 2012) here named the Hadibo–Sharbithat FZ. 2003, 2006). The rifting pattern is characterized by Tertiary reactivation We show that the transfer zone is composed of reactivated, pre- and formation of E–W to 140°E grabens (Fig. 2) inherited from a Creta- existing faults that have been activated as oblique-slip normal faults ceous intraplate extensional event (see Birse et al., 1997; Brannan et al., since early syn-rift times. We compare these results to Afro-Arabian 1997; Leroy et al., 2012). As a result of the rift obliquity, the reactivated rift systems (Fig. 1) to suggest that most of the transfer zones are grabens as well as newly formed ones are arranged en échelon along the actually pre-existing structures that are oblique to the divergence. Fur- continental margins. Many recent studies have shown that during thermore, we derive a classification for oceanic FZ based on their rela- rifting there were several successive directions of extension ranging tionships with syn-rift structures. from 020°E to 160°E (Bellahsen et al., 2006; Fournier et al., 2004; Huchon and Khanbari, 2003; Lepvrier et al., 2002). These extensional 2. Geological setting of Socotra Island, Yemen, Eastern Gulf of Aden stresses most probably rotated counter-clockwise from 020°E to 160°E. In the eastern Gulf of Aden, the Ocean–Continent Transition (OCT) In the Gulf of Aden (Figs. 1 and 2), rifting commenced at around ridge (Fig. 2) may represent exhumed serpentinized mantle, which 34 Ma (Leroy et al., 2012; Pik et al., 2013-this volume; Robinet et al., may or may not be intruded by post-rift magmatic material that modi- 2013-this volume; Roger et al., 1989; Watchorn et al., 1998). At this fied the OCT after its emplacement (Autin et al., 2010b; d'Acremont time, the of Tethyan slabs beneath the Eurasian plates in- et al., 2006, 2010; Leroy et al., 2004, 2010a, 2010b; Watremez et al., duced extension in the Afro-Arabian plate, mainly because the northern 2011). The continental margin can be divided into several domains Arabian plate entered the collision zone while the remainder of the sub- (Autin et al., 2010b; d'Acremont et al., 2005, 2006) with contrasting duction zone was still active (Bellahsen et al., 2003). The extension was styles of rifting (Leroy et al., 2010a). Three major transform-generated located in the Afro-Arabian rifts (Red , Gulf of Aden and East ), FZ can be identified. They correspond to three major offsets of the coast- which were localized by the activity of the Afar hot spot. The combina- line: from west to east, the Alula Fartak FZ, the Socotra–Hadbeen FZ, and tion of intraplate stresses with the Afar weak zone produced oblique the Hadibo–Sharbithat FZ. The latter is aligned with transverse struc- rifts that did not involve significant reactivation of any major preexisting tures cropping out on Socotra Island, the Hadibo Transfer Zone (HTZ, lithospheric weaknesses (with trend comparable with that of the future Figs. 2, 3, and 4). Between the two former FZ, transfer zones can be Gulf of Aden, Autin et al., 2010a, 2013-this volume; Bellahsen et al., mapped out from seismic data (Fig. 2) in the distal continental margin

Fig. 2. Structure, topography, , and oceanic ages from Leroy et al. (2012) and references therein. The grabens along both margins attest for the rifting that occurred during Oligo– Early Miocene times. The spreading started at A5d (17.6 Ma). Two main transform fault zones (Alula Fartak and Socotra–Hadbeen FZ) offset the Aden–Sheba ridge. Smaller transform fault zones are located between them and migrated through time. The structural map of the Dhofar continental margin is from Bellahsen et al. (2013-this volume). Onshore, the fault network is composed of normal faults trending 070°E to 110°E. The main grabens are disposed en échelon along the margin and witness the rift obliquity (50°E). Offshore, the faults are mainly 070°E near the shore, while they trend 110°E near or within the OCT. Two transfer zones segment the margin within the OCT and trend around 025°E, parallel to the divergence. Two transfer zones were mapped further north, close to the shore within the continental margin; they trend 160°E and are perpendicular to the 070°E-trending rift parallel faults in this area. N. Bellahsen et al. / Tectonophysics 607 (2013) 32–50 35

Fig. 3. Geological map of Socotra Island modified from Leroy et al. (2012). The Island is primarily divided in two domains, separated by the Hadibo Transfer Zone (HTZ). East of the HTZ, the mainly consists of a , with both steep and low-angle normal faults in the northern part, without any significant syn-rift basins. In the western part, three main basin crop out: the Allan Kadarma basin in the Northwestern part, partly controlled by a low-angle segment (see Figs. 5, 6), the Sherubrub–Balan basin in the Southwest, also controlled by low- angle normal fault segments, the central basin that is partly controlled by the transfer zone. and thus probably formed during the final stages of rifting, possibly dur- Central basins) formed during Oligo–Miocene times and few smaller ing formation of the OCT (d'Acremont et al., 2005). These two FZ can be ones formed along the transfer zone (Fig. 3). The two western basins observed to be continuous with transfer zones (e.g. Salalah FZ, Fig. 2). are aligned along E–W to 110°E trending normal faults, while the cen- Oceanic spreading along the central part of the Aden–Sheba ridge tral one is partly aligned along the southern part of the HTZ. The present commenced at 17.6 Ma (Fig. 2, d'Acremont et al., 2006, 2010; Leroy study is focused on the northwestern and the central parts of the island et al., 2012). In the westernmost part it started at 9 Ma (Audin et al., that are described in detail in the following. We describe the structural 2004; Leroy et al., 2012) and in the easternmost part, likely at 20 Ma setting along with fault slip data that provide an idea of the syn-rift (Fournier et al., 2010). From west to east, present-day spreading rates stress states. The tensors have been calculated following range from 13 mm/yr (azimuth 035°E) to 18 mm/yr (azimuth 025°E) the method of Angelier (1984). For the fault slip inversions, only small (Jestin et al., 1994; Vigny et al., 2006). This opening direction demon- faults have been used (mm to dm scale displacements, Fig. 7). Balanced strates oblique divergence as the rift axis trend is about 075°E. cross-sections are presented and based on the assumptions of constant The following description of the is based on previous area and length of sedimentary layers (Dalstrom, 1969). descriptions from South Arabia in Sultanate of Oman (Leroy et al., 2012; Razin et al., 2010; Robinet et al., 2013-this volume; Figs. 6 and 8 3.1. The Allan–Kadarma basin to 10). Pre-rift Formations (Fms.) are Cretaceous to Eocene and include Cenomanian limestones overlain by the Umm Er Radhuma platform In its western part, the Allan–Kadarma basin is controlled by a low- carbonates (Thanetian to Ypresian) and the Rus peritidal platform car- angle normal fault segment that trends WNW–ESE, between the bonates (Ypresian). The middle to late Eocene is represented by the Qalansiya coast and the Jebel Kadarma (Figs. 5 and 8) and was active Dammam (Lutetian to Bartonian) and Aydim (Priabonian) carbonate during deposition of the Mughsayl Fm. (Figs. 3 and 5, upper Oligo- formations. Syn-rift series are composed of the Ashawq (platform, cene–Miocene, Leroy et al., 2012). In the eastern part, south of the Rupelian) and Mughsayl (calciturbidites, Chattian) Fms. Locally, post- Jebel Allan, the basin is controlled by a steeper fault (dip around 60– rift sedimentary rocks can be observed and consist of (late) Burdigalian 70°) (Figs. 5 and 8). Between these two sub-basins, a NE–SW normal to Langhian conglomerates (Ayaft Fm.). fault occurs, which is considered as a transfer fault (see below). In the western part, the main normal fault is a low-angle segment 3. Structural analysis of Socotra Island trending WNW–ESE that was active under an extension around N–S to 020°E (sites 28, 25, 60, 271, Fig. 5). North of the main low-angle Socotra Island is divided in two parts by the large HTZ (Figs. 3 and 4). normal fault, in Jebel Kadarma, two fault trends can be mapped. A The HTZ trends NE–SW and consists of several normal and oblique-slip WNW–ESE south-dipping fault branches at depth from this low-angle faults. In the eastern part of the island, Mount Haggier and surrounding segment (Fig. 8a). A segment trending 070°E was activated subsequent- peaks comprise a large outcrop of basement rocks. The massif is bound- ly, as indicated by its abutment on the WNW–ESE fault segment. There, ed northward by two main north-dipping steep normal faults, one most of the fault slip data indicate a 020°E extension (sites 37, 141, 233, trending ENE–WSW and one trending NW–SE (Fig. 3). Other normal 234, Fig. 5). However, some fault-slip data suggest that there might have faults trending NW–SE, E–W, and 070°E have been mapped in the been a counterclockwise rotation of the extension, from N–S/020°E to field. In the hanging-walls of these steep faults, low-angle normal faults around 160°E: the distribution of some fault slip data is not symmetrical (dip around 10–20°) affect the Mesozoic and Tertiary sedimentary (Fig. 4). Some stereonet analyses have a wide and asymmetrical distri- layers, overlying the basement through tectonic contacts (Figs. 3, 5 bution of fault slip (sites 37 and 234, but also 28, 265, 274, and 275 in and 6). other areas, Fig. 5). They differ from other sites such as sites 25, 326, In the western part of the Island, the structural pattern is more com- 328, or 60B for example (Fig. 5), where the distribution is symmetrical plex. Three syn-rift basins (Allan–Kadarma, Sherubrub–Balan, and around the perpendicular to the fault strike. When the distribution is 36 N. Bellahsen et al. / Tectonophysics 607 (2013) 32–50

Fig. 4. Hadibo Transfer Zone: (a) 3D Google view (earth.google.com) and (b) photo. In the fault's hanging wall, Mesozoic strata are sub-horizontal above the Paleozoic metasediments and basement. In the footwall, basement rocks crop out up to about 1500 m high (Mount Haggier). The Hadibo Transfer Zone presents both a right-lateral and a normal displacement (see Fig. 9).

symmetrical, although wide, the local tectonic history can be considered (on the same fault plane), indicating a 020°E extension followed by a as monophase (if no stria overprinting is observed on fault planes). Such 160°E extension, have been observed in the field (see more examples wide distribution is indeed due to local fault interactions and/or fault in the next sub-section). Moreover, the analysis of some asymmetric plane irregularities. However, when the stria distribution is both wide fault slip distribution (see above) suggests that a counterclockwise rota- and asymmetrical (sites 37 and 234, Fig. 5), it may be interpreted as in- tion of the least compressive stress occurred (also at site 265, Fig. 5). dicative of stress rotations either counterclockwise (sites 37 and 234) or Between the two sub-basins, the NE–SW fault can be considered as a clockwise (site 274A, as well as site 275 in another area, Fig. 5, see cap- transfer fault as it separates two sub-domains with different geometries tion for more details). The counterclockwise rotation (020°E then and amounts of extension (Fig. 8). We have little data from within the 160°E) of the least compressive stress is in accordance with the abutting fault zone, but it might have been active during the entire period of of the 070°E fault against the WSW–ESE one (Jebel Kadarma, Fig. 5). rifting, particularly from early stages, to accommodate the along-strike Finally, above the low-angle fault segment, an E–W extension is variations in geometry and amount of extension. At site 274 (Fig. 5), recorded (Fig. 5, sites 242B and 60B). Such extension is also witnessed two stress tensors are compatible with a N–S extension (one in an ex- by N–S faults that branch on the main low-angle fault segment, East of tensive regime, one in strike-slip regime). Such a stress permutation Qalansiya. may be due to the proximity to the transfer zone (see similar interpre- In the eastern part of the basin, the faults are steeper (Fig. 8b). Two tation in Fournier et al., 2007). main extension directions are observed, 160°E and 020°E, although in- All the faults described above, as well as those described in the termediates are also found (sites 207, 273, 274, 267, 265, 263, 260, next sub-section, are considered to be Oligo–Miocene in age. All Fig. 5). At site 265 (Fig. 5), chronological relationships between striae fault population types affect the formations (see the N. Bellahsen et al. / Tectonophysics 607 (2013) 32–50 37 location of the measurement site on Fig. 5). Faults observed in older 3.2. The Hadibo Transfer Zone formations (Cretaceous, Paleocene, and Eocene) display similar ge- ometry and kinematics. We therefore consider that these are also In the HTZ and surrounding area, normal faults trend NE–SW to Oligo-Miocene in age. 070°E and around 110°E (Figs. 3, 4 and 9). In the transfer zone sensu

Fig. 5. Geological map of the NW part of the Island with fault-slip data. Extension directions mainly range between 020°E and 160°E, with several exceptions around E–W. Cross sections of Fig. 8 are represented by the two thin lines. Few stereonets have a wide and asymmetrical distribution of fault slip: sites 28, 37, 265 and 274, 275. They differ from other sites such as 25, 326, 328 for example, where the distribution is symmetrical around the perpendicular to the fault strike. The asymmetry can be interpreted as witnessing stress rotations that were either clockwise (28, 37, 265) or counterclockwise (274, 275). We have drawn the perpendicular to the fault strike (dashed line). One may note that the fault striae are, in these cases, system- atically lying on a line that was rotated either clockwise or counterclockwise. On these sites, the fault slip has been performed, but manually drawn arrows for extension have been represented to suggest the polyphased history. Numbers 1 and 2 attest for probable first and second phases. 38 N. Bellahsen et al. / Tectonophysics 607 (2013) 32–50

Fig. 6. Field photos of low-angle normal fault segments. a) East of Hadibo (Figs. 3 and 9). Eocene Umm Er Radhuma and Cretaceous limestone over low-angle normal faults, with basement rocks in the footwall. b) East of Qalansya (Figs. 3 and 5). Tertiary sedimentary rocks above low-angle normal fault with basement rocks in the footwall. The cliff is about 500 m high.

stricto, the fault kinematics indicate that it was activated at some point . In the northernmost part, NE–SW and NW–SE steep normal of its history as a normal fault zone trending NE–SW to 070°E, with a faults offset the monocline as well as low-angle faults now observed in sub-perpendicular extension, around 160°E (Fig. 9, sites 18A, 196, the steep fault's hanging wall (Figs 6, 9 and 10). The steep normal faults 259B, 367). There are also evidences of strike-slip events: ENE–WSW are thus considered to post-date the low-angle fault segment. Similar faults were active as right-lateral faults and record strike-slip regimes observation was made in Pik et al. (2013-this volume). with a probable 020°E extension (Fig. 9, sites 18B, 258). These observations (abutting relationships and different domains on Indeed, faults oblique to the HTZ have been observed: there are each side of the transfer zone) suggest that the HTZ was active during faults trending E–W to 110°E with extension directions from N–S to early syn-rift times. Moreover, as it is strongly oblique to the global 020° (Fig. 9, sites 1, 194, 259A for example). Large faults also have the 020°E divergence, it may be a reactivated pre-existing structure. same trends: around site 256, large faults are trending around 110°E (Fig. 9); at about 10 to 20 km west of the HTZ, two E–W normal faults 4. Discussion are mapped (sites 15, 249, and 246) with syn-rift sedimentary rocks in their hanging wall. 4.1. Structural evolution of Socotra Island Relative chronologies between two sets of slickenlines on the same fault plane have been observed at few sites and provide us with a rela- 4.1.1. Pre-existing transfer zones tive chronology between states of stress. At sites 4, 246A, 249B, and Based on our structural analysis, we propose that the HTZ is a 259A (Fig. 9), striations on WNW–ESE normal faults display dip slip reactivated pre-existing structure. The HTZ separates two very different movements (extension around 020°E) followed by more strike-slip domains and thus became active during the earliest stages of extension left lateral movement (extension around NNW–SSE), indicating a coun- prior to other nearby faults that all abut on it. Moreover, the HTZ trends terclockwise rotation of the extension. At sites 194 and 249B (and pos- at angles of about 20°–50° to divergence, strongly suggesting that it was sibly at sites 246A and 259A, although more data is needed), the pre-existing and that it has been reactivated. Without tectonic inheri- asymmetrical fault slip distribution (see above for details) can also be tance, it is most likely that all extensional faults should trend sub- indicative of least compressive stress rotation. Moreover, at sites 138 perpendicular to the divergence or the main directions of extension. (Fig. 9), we found chronologies between striations that suggest both Another NE–SW-striking fault is also observed in the easternmost clockwise and counterclockwise rotations. Finally, at sites 18 and 133 part of the island (Ras Momi, Fig. 3). The fault may be Mesozoic in age (Fig. 9), we found evidences of the opposite chronology on single fault as it offsets the Permo–Triassic sedimentary rocks (600 m of throw). It planes. is similar to faults observed in Huqf (north of Oman) and Sawqrah From a structural point of view, at a larger scale, the overall geome- (southeast of Oman) regions (Leroy et al., 2012). These NE–SW faults try of the area suggests that the 110°E to NW–SE normal faults abut onto thus have polyphased history and are prone to reactivation. Moreover, the NE–SW to ENE–WSW faults of the HTZ (Fig. 5 and 9). Moreover, the Proterozoic acidic and basaltic dykes in the basement at Socotra strike two domains east and west of the HTZ are very distinct and different. NE–SW (Denele et al., 2012). These dykes may very well have been The western one presents many normal faults and associated syn-rift reactivated as faults during the rifting. basins as described above. The eastern domain is almost undeformed, Similar geometries and interpretations have been reported for the except in the northernmost part. It consists of a large, partly eroded Suez Rift (Bosworth, 1994; Jarrigue et al., 1990; Moustafa, 1997) and N. Bellahsen et al. / Tectonophysics 607 (2013) 32–50 39

Fig. 7. Field photos of small normal faults. a) ESE–WNW normal fault at site 141 in Eocene Umm Er Radhuma Fm. (Fig. 5). b) ENE–WSW normal faults in Jebel Allan in Oligo–Miocene Mughsayl Fm. (Fig. 5). c) ESE–WNW normal fault at site 275 in Oligocene Ashawq Fm. (Fig. 5).

East African Rift System (Corti et al., 2007; Rosendahl, 1987). In the Gulf the divergence direction and reactivated pre-existing basement linea- of Suez, the Tertiary rift trends NW–SE to NNW–SSE with an extension ments (e.g. the Rihba zone for the ZAZ, Younes and McClay, direction around 060°E; transverse structures are described as transfer 2002). At a smaller scale, numerous transfer zones trend around NNE– zones that separate asymmetrical rift segments (Bosworth and SSW. These faults are widespread within the rift and also oblique to McClay, 2001; Bosworth et al., 2005; Jarrigue et al., 1990, and references the divergence. therein). Three domains can be defined, the Darag, the October, and the The East African Rifts in Kenya and Tanzania (Fig. 13) formed during Zeit domains (Fig. 12)(Younes and McClay, 2002), separated by the Miocene times (12–10 Ma, see Ebinger, 1989). Within that rift system, Zaafarana (ZAZ, NW–SE) and the Morgan (MAZ, N–S to NNE–SSW) ac- along-strike variability and alternating basin asymmetry are observed, commodation zones, respectively (Fig. 12). These zones are oblique to and transfer zones are well documented (e.g. Corti et al., 2007; Morley et al., 1990; Rosendahl, 1987, and references therein) and trend NW– SE (Fig. 10). This pattern can be mapped at small and large scales. For example, the Rukwa rift (Morley et al., 1992) is a large transfer zone characterized by oblique-slip normal faults. At a smaller scale, within the Lake Tanganyika rift (Fig. 13), several (small) faults trending NW– SE are also considered as transfer faults (Lezzar et al., 2002). Similarly, in the N–S Kenya rift (Fig. 13, Gregory rift), NW–SE to NNW–SSE oblique faults are pre-existing basement shear zones reactivated during rifting (Smith and Mosley, 1993). Similar observations were reported for the Malawi rift (Fig. 13) where pre-existing basement structures controlled the internal rift geometry (Ring, 1994). The direction of divergence along the East African Rifts has long been debated (Fig. 13). Some authors conclude that the main extension is about NW–SE (Chorowicz, 2005; Sander and Rosendhal, 1989; Scott et al., 1992; Specht and Rosendahl, 1989; Versfelt and Rosendahl, 1989) while others conclude that the direction is E–W(Bosworth, 1992; Ebinger, 1989; Morley, 1989). Based upon analog and numerical models, Corti et al. (2007) suggested that it might be E–W. GPS data show that the present-day kinematics consists of E–W to WNW–ESE di- between Nubia and Somalia (Calais et al., 2006; Déprez et al., 2013; Fernandes et al., 2004; Nocquet et al., 2006; Royer et al., 2006; Stamps et al., 2008). Thus, depending on the orientation of the diver- gence, the transfer zones might either be parallel or oblique to diver- gence. However, considering the present-day kinematics and also the probable Miocene kinematics, the transfer zones are oblique to diver- gence. Moreover, most of the transfer zones are strongly influenced by Fig. 8. Balanced cross sections through the Allan–Kadarma basin. Constant lines and thick- nesses of sedimentary layers were respected to balance the sections. The basement area on pre-existing older structures (see Fig. 13 and Rosendahl, 1987 for restored sections is not represented as equal to the basement area on field-based sections. discussion). However, the top basement length is conserved. a) The western cross-section, through the To summarize, the Gulf of Aden, the Gulf of Suez and the East African Jebel Kadarma, displays the low-angle normal fault segment that probably becomes steep- Rifts have common characteristics among which is the geometry of the er at depth. The amount of extension is high (5.6 km). b) The Eastern cross-section, through the Jebel Allan, show steep normal faults that accommodated much less extension transfer zones: they trend oblique to both the divergence direction and than in the western part (0.9 km). the main normal faults (orthogonal to the divergence) and they are 40 N. Bellahsen et al. / Tectonophysics 607 (2013) 32–50 often basement structures reactivated as oblique-slip normal faults. In is accommodated offshore, north of the island. In any case, we show these rifts, it appears that there are no faults (or very few) strictly paral- here that evolve along strike, including the dip angle of the lel to the divergence. main normal faults and associated amounts of extension. During this stage, the transfer zones (especially the HTZ) are pre-existing structures reactivated as oblique-slip faults (right-lateral and normal), most prob- 4.1.2. Early syn-rift depocentres ably with dominant strike-slip motion over dip-slip motion, given the Early transfer zones usually control the location and geometry of angle between the extension (020°E) and the fault zone (~NE–SW). early depocentres (e.g. Bellahsen and Daniel, 2005; Lezzar et al., (2) During a second stage, with increasing extension, 110°E-trending 2002). However, looking at the basin and depocentre distribution on So- normal faults became more numerous, especially in the western part. cotra Island (Fig. 11), it appears that the main ones are located west of During this stage, the extension may have rotated counterclockwise the HTZ. Moreover, the three main basins are located far away from from 020°E to around 160°E resulting in the formation and reactivation this zone and the smaller ones are located at about 10 km westward, ex- of faults trending ENE–WSW to NE–SW. The transfer zone, striking cept the Central Basin (Figs. 3 and 11) that is probably partly controlled NE–SW, was most probably also active as normal faults during this by the HTZ. Why is there no large syn-rift basin in the HTZ hanging wall stage, as they become well-aligned at an angle of 65° to extension. Rift- as it seems active since early syn-rift times? Two answers may be envis- parallel normal faults (trending 070°E) formed (for example in Jebel aged: either the early syn-rift sediments were deposited and subse- Kadarma, Fig. 5). The other normal faults (WSW–ESE) were still active quently eroded or they were not deposited in this area that may have and have an oblique-slip behavior. This stage probably did not last longer been relatively elevated since early rifting times. than few million years but is important as the onset of rift localization oc- There are late (syn-OCT, Burdigalian) detrital sedimentary rocks un- curred during this time (Bellahsen et al., 2013-this volume). This stress conformably overlying the pre-rift, southwest of site 184 (Fig. 9). Thus, rotation has been documented elsewhere in the Gulf (Bellahsen et al., this area was already elevated at the beginning of OCT times. UTh/He 2006; Huchon and Khanbari, 2003; Lepvrier et al., 2002). Although, it thermochronology suggests that the main vertical offset along a 070°E cannot be unambiguously demonstrated by field data at Socotra Island, steep normal faults (star symbol on Fig. 9, east of Hadibo) is rather this is the most probable scenario at the scale of the entire Gulf as strong- late (around 18–20 Ma, syn-OCT, Pik et al., 2013-this volume). Finally, ly suggested by field data and confirmed by analog models (Autin et al., the syn-OCT to post-rift Miocene conglomerates (Burdigalian to 2010a, 2013-this volume) and numerical models (Autin et al., 2013; Langhian) located North of the Mount Haggier, in the hanging wall of Bellahsen et al., 2013-this volume). the normal faults east of Hadibo (star symbol, east of site 138, Fig. 9) (3) During a third stage that led to the present-day fault network, contains many basement pebbles. No other detrital formation is ob- the deformation migrated and localized, probably offshore to the served between these conglomerates and the underlying Cretaceous north, followed by formation of the ocean-continent transition. The and Tertiary carbonates. This suggests that the main uplift of Mount third stage cannot be documented at Socotra Island because of its prox- Haggier can be dated Late Oligocene to middle Miocene. imal position. During this time, the extension was likely 020°E again This does not, however, allow us to determine if the area in the hang- according to Autin et al. (2010a). The field data (clockwise rotation of ing wall of the HTZ was uplifted during early rifting stages or only later. extension from 160°E to 020°E) presented above may support such an Thus, it is difficult to determine if the normal faults connected very early interpretation, however much more data is necessary to confirm it. to the HTZ and controlled significant syn-rift basins or not. The late syn-rift structure and the OCT emplacement may have been In any case, the amount of extension just west of the HTZ is very low controlled by the far-field divergence (Bellahsen et al., 2013-this (Fig. 10a), much lower than in the eastern (Fig. 10b) and western part of volume). Thermochronological data suggest that the HTZ was still active the island (Fig. 8). Thus, the extension is accommodated elsewhere, at this time (Pik et al., 2013-this volume). most likely offshore, further north, where basins may also be found.

4.1.3. Polyphase tectonic history 4.1.4. Hadibo Transfer Zone/Hadibo–Sharbithat FZ continuity The second main result is the structural evolution of the island and The HTZ and the Hadibo–Sharbithat FZ continuity is supported by especially its early structural pattern (Fig. 11). We divide this evolution several observations and interpretations. (1) The FZ is aligned and con- into three stages: (1) a first stage when the main WNW–ESE faults tinuous with the HTZ (Figs. 2 and 11). This alignment can be seen in the formed, along with the oblique reactivation of the NE–SW transfer offset of oceanic magnetic anomalies offshore (Leroy et al., 2012). (2) In zones during extension parallel to 020°E (Fig. 11). The faults with the distal margin between the HTZ and the Hadibo–Sharbithat FZ, a sig- low-angle segments were active in both the western part (Kadarma nificant bathymetric escarpment defines two domains, suggesting the and Sherubrub basins, Figs. 5, 6, and 8) and the eastern part (east of presence of a major structure (Fig. 3). This escarpment roughly links Hadibo, Fig. 6, 9, and 10) of the island. The transfer zones separated the transfer and the fracture zones, though more detailed bathymetry blocks with different geometries and different amounts of extension. coverage and subsurface data are needed for a precise structural map- Thus, the island was already divided into three domains: the ping. (3) U–Th/He thermochronology on apatites from basement rock Kadarma–Sherubrub, the Central, and the Haggier (Fig. 11). The differ- in Socotra Island shows that the tectonic denudation, which started at ent amounts of extension (Figs. 5 and 8) accommodated in these the rifting onset, during late Eocene–early Oligocene time, continues three domains can be explained in two ways. Firstly, part of the exten- during early Miocene (Aquitanian–Burdigalian) time (Pik et al., 2013- sion in the central domain is not expressed at the surface, and some ex- this volume). Thus, the HTZ was still active during the emplacement may, for example, be hidden beneath the coastal plain in the of the OCT, just before the formation of the oceanic transform fault. central-northern part of the island (Fig. 3). Small parts of syn-rift basins We therefore conclude that the HTZ and the Hadibo–Sharbithat FZ are east of this plain can be observed, indicating that there may be normal structurally connected and that the Hadibo transform fault formed at faults hidden below. Secondly, and most likely, some of the extension the Hadibo Transfer Zone.

Fig. 9. Detailed map displaying the HTZ with fault-slip data. Extension directions mainly range between 020°E and 160°E, with several exceptions around E–W. The cross sections in Fig. 10 are represented by the two thin lines. Few stereonets have a wide and asymmetrical distribution of fault slip: sites 15, 246A, 249B. See caption of Fig. 5 for explanation and possible interpretation. N. Bellahsen et al. / Tectonophysics 607 (2013) 32–50 41 42 N. Bellahsen et al. / Tectonophysics 607 (2013) 32–50

Fig. 10. Balanced cross sections through the HTZ. Constant lines and thicknesses of sedimentary layers were respected to balance the sections. The basement area on restored sections is not represented as equal to the basement area on field-based sections. However, the top basement length is conserved. a) The western cross-section, through the HTZ, displays steep normal fault segments; the amount of extension is low (0.8 km). b) The Eastern cross-section, east of Hadibo, show low-angle normal fault segments that accommodated more extension than in the western part. The movements along the low-angle segments was probably along a 020°E direction as attested by fault slip data from this area (Fig. 7, sites 132b, 133a, 138b, 138c) although there are indications of 160°E extension (Fig. 7, site 138a). Thus, we did not estimated the amount of extension as it may occur in the 010° direction along low-angle segments, while it may have occurred before, along a 160°E direction, on the steep faults. In any case, the amount of extension in this cross section is much largerthat on the western one. This justifies that the location of the Hadibo Transfer Zone runs north of the outcrops east of Hadibo.

4.2. Axial segmentation in the Gulf of Aden and classification of oceanic Fracture zones generated by transform faults that formed within the fracture zone OCT during its formation (Figs. 15 and 16) are named Type 2 FZ. In the Gulf of Aden, Type 2 FZ are observed between Shukra El Sheik and Alula In the Gulf of Aden, several transform faults and non-transform discon- Fartak FZ (Fig. 18a, b and in d'Acremont et al., 2005). These FZ offset the tinuities offset the Aden–Sheba oceanic spreading ridge (Figs. 2 and 18). distal OCT boundary but do not offset its proximal boundary. Here, we describe in detail the segmentation of the spreading ridge and Detailed studies of each FZ are required, but they clearly display geom- present a new classification of the oceanic fracture zones. As they repre- etries that differ from Type 1 FZ. They are more numerous in the central sent paleo-transform fault zones, the new classification will apply only part of the Gulf because there the orientation of the rift and OCT are to paleo-1st order transform fault zones (sensu MacDonald et al., 1991). oblique to the direction of divergence (Bellahsen et al., 2013-this First, in the light of our results regarding the structural evolution on volume). Some questions arise: What is their trend relative to the diver- Socotra Island, we propose that a classification of oceanic fracture zones gence? Are they reactivated pre-existing structures? Are they linked to can be made based on the stage of rifting during which they or their pre- any continental transfer zones? More data from distal margin is re- cursors formed, and their relationships to the continental margin geom- quired to answer these questions, in particular high-resolution 3D seis- etry and structures. In particular, we base our classification on whether mic data. the paleo-transform faults/FZ had offset and deformed either thinned We name FZ generated by transform faults that formed after the continental crust and OCT, OCT only, or oceanic crust only. We also em- onset of oceanic spreading as Type 3 FZ or simply oceanic FZ. This phasize their relationship with pre-existing continental structures type of FZ is not linked to any syn-rift structures and formed after reactivated during rifting. rifting and OCT emplacement. Turcotte (1974) and Sandwell We name FZ's generated by transform faults that deformed conti- (1986) showed, on the basis of their spacing, that the oceanic trans- nental crust Type 1 FZ. The Type 1 FZ offsets are usually (but not neces- form faults are thermal contraction cracks, mainly due to the sarily) large and result from a large offset between the two thinned thermomechanical evolution of the oceanic (Detrick continental domains that the proto-transform joined. Such geometry et al., 1995; Gente et al., 1995; Hooft et al., 2000; and see Gerya, implies that part of the margin is a transform margin. Figs. 15 and 16 2012, for a recent review). Type 3 FZ can be identified in the Gulf of represent conceptual models of the transform fault zone and fracture Aden (Fig. 18a). They are indeed not correlated to any structure in zone evolution for orthogonal and oblique rifts, respectively. Three dif- the continental margin or the OCT and formed after the onset of oce- ferent Type 1 FZ are detailed below. anic spreading (see also d'Acremont et al., 2006, 2010). N. Bellahsen et al. / Tectonophysics 607 (2013) 32–50 43

Fig. 11. Structural map of Socotra with bathymetry and OCT and oceanic domains offshore. Bathymetric map is built with the compiled data from all cruises in the database of Autin et al., in press. During the Oligo–Miocene rifting, the island is divided into three domains: the Qalansiya–Sherubrub domain (western domain), the Central domain, and the Mount Haggier domain (eastern domain). Two main transfer zones are defined: the Hadibo Transfer Zone, and the Balan–Kadarma transfer zone. An extra transfer zone is located east of the Island, the Ras Momi transfer zone that bounds eastward the Mount Haggier domain. The Qalansya–Sherubrub domain is mainly characterized by low-angle normal fault segments; the amount of extension is high. The eastern boundary of this domain, the Balan–Kadarma transfer zone is a rather diffuse one. It is clearly observed in the Allan–Kadarma basin. Further south, it may cross the Sherubrub–Balan basin. However, the continuity in the basin geometry at the suggested location of the transfer zone shows that there is a structural complexity that needs further work. The central domain is structured by steep normal faults and does not exhibit high amounts of extension. The Mount Haggier domain is weakly deformed except in the northern part, where low-angle segments are observed, but where probably most of the deformation is located further north offshore. Early depocentres are in light yellow. After the rifting stage, the OCT was emplaced and was strongly segmented, as was the oceanic domain. In the distal margins, there is a clear bathymetric signature of the HTZ. Surprisingly, the distal struc- ture bounds a structural high westwards and a relative low eastwards. Further work is needed to better characterize the structural evolution of this area.

Three cases of syn-rift Type 1 FZ can be distinguished, we describe genetically related to continental transfer zones, namely the Pernambu- them below. co (Basile et al., 2005; Moulin et al., 2010) and Patos shear zone (Darros de Matos, 1999; Françolin and Cobbold, 1994; Sénant 4.2.1. Type 1-C (syn-rift) FZ and Popoff, 1991) in and the Sanaga and N'Gaoundere The first Type 1 FZ is named Type 1-C (C for continental, see below): shear zones in West Africa (Basile et al., 2005; Moulin et al., 2010). For such FZ formed as a transform fault that is linked to a blockage in along the Ascension and the Romanche FZ, the structural evolution may be strike ridge propagation. In this case the transform fault is a “continental similar although less straight forward (see a discussion in Basile et al., transform fault” that, at some point, separated an oceanic domain from 2005 for Romanche FZ; compare Mohriak and Rosendahl (2003), a continental domain. Thus, it was located at the tip of a propagating Blaich et al. (2008), and Meyers et al. (1996) for Ascension FZ). spreading center (Figs. 15 and 16). Such a setting necessitates the pres- In the North , similarly, the Senja–Greenland FZ (see ence of “continental” transform fault zone to accommodate the differ- Mosar et al., 2002 and references therein) separates two oceanic do- ence in the amount of divergence. This type of FZ can be found in mains with different ages of spreading onset. In central Atlantic, the several locations globally. same is true for the Charlie-Gibbs, 15°20′N-Guinean, and Jacksonville- In the South Atlantic Ocean (Fig. 1), onset of seafloor spreading along Cap Vert FZ (see Labails et al., 2010). The Owen FZ (easternmost Gulf the ridge between the Ascension and the Rio Grande FZ is younger than of Aden, Fig. 1) is also most probably a Type 1-C FZ but with the notable between the Rio Grande and the Falkland FZ (see Moulin et al., 2010). difference that it likely formed within an oceanic domain (Mountain Thus, the Rio Grande FZ likely formed as a continental transform fault and Prell, 1990). These faults are parallel to divergence and their loca- zone, which eventually corresponded with the northern tip of the oce- tion and trend are probably (but not necessarily) determined by pre- anic spreading center. Such setting can last during a significant time existing structures. Indeed, as these faults are not particularly well ori- only if there are continental transverse structures that accommodate ented to be reactivated in an extensive stress field (because they trend the important differential opening (see reconstructions in Moulin parallel to the least compressive stress), it is more likely that they are et al., 2010; and references therein). The Rio Grande FZ is, for example, reactivated pre-existing faults or shear zones. Finally, these transform 44 N. Bellahsen et al. / Tectonophysics 607 (2013) 32–50

Fig. 13. The southern East African Rift System (after Rosendahl, 1987 and Corti et al., 2007). The global trend of the rift is N–S (as in Turkana, Gregory, Malawi rifts, for exam- ple), except in transfer zones where it trend NW–SE (as in the Rukwa rift, for example). Fig. 12. Structural map of the Gulf of Suez (after Younes and McClay, 2002). The main nor- The transfer zones reactivated pre-existing structures in the basement (such as the mal faults are trending around 150°E. Accommodation or transfer zones divide the rift into Aswa lineament). The direction of extension is still debated. Present-day divergence several domains with opposite sense of dip of half . The Zaafarana Accommodation based on GPS is around 100°E–110°E, close to the long term E–W extension (e.g. Zone (ZAZ, NW–SE) separates the Darag and the October basins and probably reactivates Bosworth, 1992; Corti et al., 2007; Ebinger, 1989; Morley, 1999). However, NW–SE exten- the Rihba shear zone; the Morgan Accommodation Zone (MAZ, N–S to NNE–SSW) sepa- sion has been proposed (e.g. Rosendahl, 1987; Chorowicz, 2005). Our study that shows rates the October and the Zeit basins. Many smaller transfer zones trending parallel to that transfer zones are commonly oblique to the divergence direction would suggest the MAZ can be observed in the Zeit basin. that the syn-rift extension was E–W to WNW–ESE.

faults clearly form during rifting (at least for the continental domain sit- reactivated pre-existing faults. In the East African Rift System, the uated on one side of the transform fault) and thus can be named syn-rift Rukwa rift zone (Figs. 13 and 15a) is an oblique rift zone where slip is FZ. occurring along old reactivated lineaments. If this zone eventually be- In the Gulf of Aden, the Shukra El Sheik FZ is a Type 1-C FZ from the comes an ocean basin, it is most likely that a Type 1-T FZ would form. onset of spreading (17.6 Ma) and until 10 Ma, there was no spreading In such case, one could not find any faults parallel to the divergence west of this fault (Leroy et al., 2012). Thus, motion on this fault activated within the continental margin. because the westward ridge propagation was blocked there. This major These Type 1-T FZ may form either during the latest rifting stages or fault extended southwestward into Afar where it was probably during earliest OCT times. If the OCT proximal boundary is clearly offset, connected to the other active faults (Audin et al., 2004)(Fig. 18). it probably means that transform fault formation is slightly older than the onset of OCT formation, thus late syn-rift (Fig. 17a, upper panel). If 4.2.2. Type 1-T (syn-rift to syn-OCT) fracture zones the continental transfer zone and the FZ join at the proximal boundary A second Type 1 FZ is named Type 1-T (T for transfer) when they of the OCT, the formation of the transform fault zone may be dated to formed at continental transfer zone, as in the Hadibo–Sharbithat FZ the age of the transition between syn-rift and syn-OCT (Fig. 17a, middle case (Figs. 2 and 11). In this case, a transform fault formed where a con- panel). If the offset of the OCT proximal boundary is controlled by the tinental transfer zone was active during rifting (this study) and proba- continental transfer zone (Fig. 17a, lower panel), it is most likely that bly during OCT formation (Pik et al., 2013-this volume), leaving a the transform fault formed during OCT times. In the Hadibo–Sharbithat fracture zone behind. The FZ and transfer zone have different trends FZ, the OCT proximal boundary is clearly offset (Figs. 2 and 13, Leroy (Figs. 11 and 17). In Figs. 15 and 16, the Type 1-T FZ are located in a et al., 2012) and this offset suggests that the paleo-transform fault zone that was acting as a transfer zone during rifting. The extension is formed during syn-rift to syn-OCT times (compare Figs. 11 and 17a). accommodated by faults oblique to both the extension and the main This provisional conclusion requires more data in the distal continental rift-related faults. As shown above, such transfer zones are very often domains in order to be confirmed. N. Bellahsen et al. / Tectonophysics 607 (2013) 32–50 45

Fig. 14. The Norwegian continental margin. Structural map simplified after Brekke (2000) and Olesen et al. (2007). The margin as well as a number of important normal faults trends NE–SW, except within the Jan Mayen accommodation zone, that are aligned along the Jan Mayen fracture zone. In this zone, the normal faults are oblique to both the divergence di- rection (Mosar et al., 2002, and references therein) and the future FZ.

4.2.3. Type 1-P (syn-rift) fracture zones On the Mid-Norwegian continental margin, a final Cretaceous– Tertiary rifting stage led to continental breakup and opening of the northeastern Atlantic Ocean during Early Eocene times (for recent stud- ies see for example Brekke, 2000; Skogseid et al., 2000; Mosar et al., 2002). The long Jan Mayen FZ separates the margin into two basins: the Vøring basin in the north (Brekke, 2000) and the Møre basin in the south (Gabrielsen et al., 1999)(Fig. 14). This FZ is thought to contin- ue onshore (Doré et al., 1997; Olesen et al., 2007) in the form of a Pro- terozoic lineament. The Norwegian continental margin is affected by numerous normal faults sub-perpendicular to the future transform fault, thus perpendicular to the Cretaceous–Tertiary divergence (see Mosar et al., 2002 for the various directions of divergence). However, in the accommodation zone, i.e. around the supposed trace of the Jan Mayen lineament, normal faults are strongly oblique to divergence (NW–SE) and strike N–S to NNE–SSW (Fig. 14). During rifting, the de- formation and the depocentre locations across the Jan Mayen transfer zone clearly evolved from a diffuse accommodation zone (with some orientations that are not perpendicular to the divergence) to more lo- Fig. 15. Conceptual model of oceanic FZ in relation with rifting geometry, for orthogonal calized transfer zone (with divergence-perpendicular faults, compare rifts. Three kinds of type 1 FZ are differentiated: Type 1-C when they derive from a “con- Figs. 7 and 13 in Brekke, 2000, for Cretaceous and Figs. 15 and 16 for tinental transform zone” that has been separating an from a continental early Tertiary times). This clearly indicates that an accommodation one (as the Levant fault zone for example), Type 1-T when their location is influenced zone existed during rifting and that this zone evolved into a transfer/ by an (often pre-existing) oblique continental transfer zone (re)activated at the onset of transform (see below) fault zone during the late rifting phase; i.e. (or during) the rifting, and Type 1-P when their location it due to a pre-existing continen- tal structure reactivated at the very end of the rifting. Type 2 FZ are defined as deforming when the amount of extension became large. Note that this is consistent only the OCT lithosphere: it means that the proximal boundary of the OCT is not offset, with the review of Acocella et al. (2005) on transfer zones, where it is while the distal one is offset. Type 3 FZ initiate after the oceanic spreading onset and are suggested that transfer zones form during late syn-rift phases, while ac- related to evolving stresses within the oceanic crust. commodation zones are active during early syn-rift times. However, here the late syn-rift structures that segment the margin are named 46 N. Bellahsen et al. / Tectonophysics 607 (2013) 32–50

Fig. 16. Conceptual model of oceanic FZ in relation with rifting geometry, in the specific case of oblique rifts. Same legend that is in Fig. 15. Three Type 1 FZ are differentiated: Type 1-C, Type 1-T, and Type 1-P. Type 2 and Type 3 FZ are also represented.

transform fault (Fig. 17b), as it divergence parallel and represent a In the Gulf of Aden, Alula Fartak FZ is a large-offset FZ. There is a main proto-plate boundary. coastal offset and early syn-rift normal faults in Qamar basin are also off- We name such faults syn-rift or Type 1-P FZ (P for pre-existing) set by the FZ. There is no clear continuity of the FZ in the proximal do- when the transform fault location is influenced by a pre-existing fault main (Fig. 2), near the Yemen/Oman border; this led Taylor et al. parallel to the divergence (Figs. 15 and 16). This does not necessarily (2009) to propose that it nucleated during the onset of spreading or mean that such a transform fault was active in the proximal margin. In- soon after. However, Autin et al. (2013-this volume) suggest that this deed, such a pre-existing fault may be reactivated only in the distal mar- FZ formed because of the presence of the pre-existing Jiza/Qamar– gin, where the continental crust is extremely thin within a narrow zone. Gardafui Mesozoic basin that implied a stronger crust. There, the strain This probably occurs during or slightly before OCT formation and/or localization occurred on each side of the pre-existing basin (d'Acremont mantle exhumation. The timing of paleo-transform fault zone formation et al., 2005) as a consequence of rift obliquity; this setting necessitated a depends on the structural relationships between the FZ and the proxi- transform fault when the stretching localized during the late syn-rift/ mal boundary of the OCT (Fig. 17b), similarly to the discussion about OCT. In this case, the FZ does not correspond to any continental Type 1-T FZ (see 4.2.2). In Norway, the Jan Mayen FZ clearly strongly off- transfer zone but its location is “inherited” from the rifted basin pattern. set the proximal boundary of the OCT (Fig. 14) and is most likely related The FZ may also be a Type 1-P FZ, in our classification, as NNE–SSW pre- to a paleo-transform fault zone that formed during the latest stages of existing structures were and are present in the Arabian shield (Al- rifting. Husseini, 2000; Denele et al., 2012; Stern and Johnson, 2010). N. Bellahsen et al. / Tectonophysics 607 (2013) 32–50 47

Fig. 17. Detail of a) Type 1-T and b) Type 1-P FZ in distal margins, structures parallel to the oceanic FZ and transform fault zone are explicitly named FZ (and not transfer zones as in Fig. 2). Continental transfer zones are usually oblique to both the divergence and the normal faults. See text Section 4.2. for a discussion of the timing of transform fault zones/oceanic FZ initiation.

5. Conclusions zones during late syn-rift to early syn-OCT times, and Type 1-P that form during late syn-rift to early syn-OCT at pre-existing structures that were In this geological study of Socotra Island (Yemen), we have de- not reactivated during early rifting. Type 2 FZ initiated during the OCT scribed in detail the structural evolution of an oblique transfer zone in emplacement; Type 3 FZ form after the onset of seafloor spreading. a continental margin. This zone was active as an oblique-slip normal Our classification requires the acquisition of more high-resolution data fault zone and is not parallel to the plate divergence, even if during its particularly in distal oceanic domains where transfer zones and fracture history, some pure strike-slip movements have occurred. We also zones connect, to better constrain the geometry, the structural evolu- show that the faults in this zone are pre-existing structures that were tion, and the timing of transform fault zone formation. reactivated at the onset of rifting. Similar features can be observed in many extensional settings such as in the Gulf of Suez and the East Acknowledgments African Rifts. On the basis of our findings, we propose a new classifica- tion of oceanic FZ based on their spatial-temporal relationships with The comments by C. Ebinger and an anonymous reviewer and the syn-rift structures. We distinguish three distinct Type 1 FZ, i.e. oceanic significant work of G. Peron-Pinvidic and P.T. Osmundsen as invited ed- FZ that are related to paleo-transform fault zones that deformed conti- itors greatly improved the initial version of this contribution. This work nental margins: Type 1-C that form at “continental transform fault benefited from financial support from GDR Marges, Action Marges zones” that are syn-rift in age, Type 1-T that form at continental transfer (CNRS-INSU, TOTAL, BRGM, IFREMER), ANR YOCMAL and Rift2Ridge. 48 N. Bellahsen et al. / Tectonophysics 607 (2013) 32–50

Fig. 18. Segmentation of the Gulf of Aden through time. a) The Gulf of Aden rift at around 20 Ma, at the time of OCT onset (purple line). Light orange color is for the pre-existing Mesozoic basins, bright orange represents the OCT. Type 1 FZ (in red) that initiated at that time are represented by: a Type 1-C FZ in the west (S. El S., Shukra El Sheik), a Type 1-T FZ in the east (S.H., Socotra Hadbeen and H.S. Hadibo–Sharbithat), and possibly a Type 1-P FZ in the center (A.F., Alula Fartak). b) The Gulf of Aden rift at the end of the OCT (yellow areas) formation at ~18 Ma. Type 2 FZ (syn-OCT, in blue). c) Present-day Gulf of Aden. Type 1 FZ are still active (except Shukra El Sheik). Some Type 2 FZ died out (in the eastern part). Type 3 FZ were initiated after the onset of oceanic spreading. N. Bellahsen et al. / Tectonophysics 607 (2013) 32–50 49

The authors thank Y. Denele for fruitful discussions. GSMRB is thanked d'Acremont, E., Leroy, S., Maia, M., Patriat, P., Beslier, M.-O., Bellahsen, N., Fournier, M., Gente, P., 2006. Structure and evolution of the eastern Gulf of Aden: insights from for technical support. magnetic and gravity data (Encens-Sheba MD117 cruise). Geophysical Journal Inter- national 165, 786–803. d'Acremont, E., Leroy, S., Maia, M., Gente, P., Autin, J., 2010. Volcanism, jump and propa- References gation on the Sheba ridge, eastern Gulf of Aden: segmentation evolution and implica- tions for oceanic accretion processes. Geophysical Journal International 180 (2), Acocella, V., Morvillo, P., Funiciello, R., 2005. What controls relay ramps and transfer faults 535–551. http://dx.doi.org/10.1111/j.1365-1246X.2009.04448.x. within rift zones? Insights from analogue models. Journal of 27, Dalstrom, C.D.A., 1969. Revue Canadienne des Sciences de la Terre 6, 743–757. 397–408. Darros de Matos, R.M., 1999. History of the northeast Brazilian rift system: kinematic im- Al-Husseini, M.I., 2000. Origin of the Arabian Plate structures: Amar collision and Najd rift. plications for the break-up between Brazil and West Africa. In: Cameron, N.R., Bate, GeoArabia 5, 527–554. R.H., Clure, V.S. (Eds.), The Oil and Gas Habitats of the South Atlantic. Geological So- Angelier, J., 1984. Tectonic analysis of fault slip data sets. J. Geophys. Res. 89, 5835–5848. ciety, London, Special Publications, 153, pp. 55–73. Audin, L., Quidelleur, X., Coulié, E., Courtillot, V., Gilder, S., Manighetti, I., Gillot, P., Denele, Y., Leroy, S., Pelleter, E., Pik, R., Talbot, J.Y., Khanbari, K., 2012. The Cryogenian arc Tapponnier, P., Kidane, T., 2004. and K–Ar and 40Ar/39Ar ages in formation and successive high-K calc-alkaline plutons of Socotra Island (Yemen). the Ali Sabieh area (Republic of Djibouti and Ethiopia): constraints on the mechanism Arabian Journal of Geosciences 5, 903–924. of Aden ridge propagation into southeastern Afar during the last 10 Myr. Geophysical Déprez, A., Doubre, C., Masson, F., Ulrich, P., 2013. Seismic and aseismic deformation along Journal International 158, 327–345. the East African Rift System from a reanalysis of the GPS velocity field of Africa. Geo- Autin, J., Bellahsen, N., Husson, L., Beslier, M.-O., Leroy, S., d'Acremont, E., 2010a. Analogue physical Journal International 193, 1353–1369. models of oblique rifting in a cold lithosphere. Tectonics 29, 1–23. Detrick, R.S., Needham, H.D., Renard, V., 1995. Gravity anomalies and crustal thickness Autin, J., Leroy, S., Beslier, M.-O., d'Acremont, E., Razin, P., Ribodetti, A., Bellahsen, N., variations along the Mid-Atlantic Ridge between 33°N and 40°N. Journal of Geophys- Robin, C., Al Toubi, K., 2010b. Continental break-up history of a deep -poor ical Research 100, 3767–3787. margin based on seismic reflection data (northeastern Gulf of Aden margin, offshore Doré, A.G., Lundin, E.R., Birkeland, Ø., Eliassen, P.E., Jensen, L.N., 1997. The NE Atlantic Oman). Geophysical Journal International 180 (2), 501–519. http://dx.doi.org/ Margin: implications of late Mesozoic and Cenozoic events for hydrocarbon 10.1111/j.1365-1246X.2009.04424.x. prospectivity. Petroleum Geosciences 3, 117–131. Autin, J., Bellahsen, N., Leroy, S., Husson, L., Beslier, M.-O., d'Acremont, E., 2013. The role of Ebinger, C.J., 1989. Tectonic development of the western branch of the East African rift structural inheritance in oblique rifting: insights from analogue models and applica- system. GSA Bulletin 101, 885–903. tion to the Gulf of Aden. Tectonophysics 607, 51–64 (this volume). Fernandes, R.M.S., Ambrosius, B.A.C., Noomen, R., Bastos, L., Combrinck, L., Miranda, J.M., Autin, J., et al., 2013. The role of structural inheritance in oblique rifting: insights from an- Spakman, W., 2004. Angular velocities of Nubia and Somalia from continuous GPS alogue models and application to the Gulf of Aden. Tectonophysics 607, 51–64. data: implications on present-day relative kinematics, Earth Planet. Sci. Lett. 222, Basile, C., Mascle, J., Guiraud, R., 2005. Phanerozoic geological evolution of the Equatorial 197–208. Atlantic domain. Journal of African Earth Sciences 43, 275–282. Fournier, M., Bellahsen, N., Fabbri, O., Gunnell, Y., 2004. Oblique rifting and segmentation of Behn, M.D., Lin, J., 2000. Segmentation in gravity and magnetic anomalies along the U.S. the NE Gulf of Aden . , , Geosystems 5 (11), 24. east coast passive margin: implications for incipient structure of the oceanic litho- Fournier, M., Huchon, P., Khanbari, K., Leroy, S., 2007. Segmentation and along-strike sphere. Journal of Geophysical Research 105, 25769–25790. asymmetry of the passive margin in Socotra, eastern Gulf of Aden: are they controlled Bellahsen, N., Daniel, J.M., 2005. Fault reactivation control on normal fault growth: an ex- by detachment faults? Geochemistry, Geophysics, Geosystems 8. http://dx.doi.org/ perimental study. Journal of Structural Geology 27, 769–780. 10.1029/2006gc001526. Bellahsen, N., Faccenna, C., Funiciello, F., Daniel, J.-M., Jolivet, L., 2003. Why did Arabia sep- Fournier, M., Chamot-Rooke, N., Petit, C., Huchon, P., Al-Kathiri, A., Audin, L., Beslier, M.-O., arate from Africa? Insights from 3-D laboratory experiments. Earth and Planetary d'Acremont, E., Fabbri, O., Fleury, J.-M., Khanbari, K., Lepvrier, C., Leroy, S., Maillot, B., Science Letters 216, 365–381. Merkouriev, S., 2010. Arabia–Somalia plate kinematics, evolution of the Aden–Owen– Bellahsen, N., Fournier, M., d'Acremont, E., Leroy, S., Daniel, J.-M., 2006. Fault reactivation Carlsberg , and opening of the Gulf of Aden. Journal of Geophysical Re- and rift localization: the northeastern Gulf of Aden margin. Tectonics 25. http:// search 115. http://dx.doi.org/10.1029/2008jb006257. dx.doi.org/10.1029/2004TC001626. Françolin, J.B.L., Cobbold, P.R., 1994. Faulting in the Early Cretaceous Rio do Peixe basin Bellahsen, N., Husson, L., Autin, J., Leroy, S., d'Acremont, E., 2013. The effect of thermal (NE Brazil) and its significance for the opening of the Atlantic. Journal of Structural weakening and buoyancy forces on rift localization: field evidences from the Gulf of Geology 16, 647–661. Aden oblique rifting. Tectonophysics 607, 80–97 (this volume). Freund, R., Merzer, A.M., 1976. Anisotropic origin of transform faults. Science 192, Birse, A.C.R., Bott, W.F., Morrison, J., Samuel, M.A., 1997. The Mesozoic and early tertiary 137–138. tectonic evolution of the Socotra area, eastern Gulf of Aden, Yemen. Marine and Pe- Gabrielsen, R.H., Odinsen, T., Grunnaleite, I., 1999. Structuring of the northern Viking troleum Geology 14 (6), 675–684. Graben and the Møre Basin; the influence of basement structural grain, and the par- Blaich, O.A., Tsikalas, F., Inge Faleide, J., 2008. Northeastern Brazilian margin: regional tec- ticular role of the Møre–Trøndelag Fault complex. Marine and Petroleum Geology 16, tonic evolution based on integrated analysis of seismic reflection and potential field 443–465. data and modeling. Tectonophysics 458, 51–67. Gente, P., et al., 1995. Characteristics and evolution of the segmentation of the Mid- Bosworth, 1992. Mesozoic and early Tertiary rift tectonics in East Africa. Tectonophysics Atlantic Ridge between 20°N and 24°N during the last 10 million years. Earth and 209, 115–137. Letters 129, 55–71. Bosworth, W., 1994. A model for the three-dimensional evolution of continental rift ba- Gerya, T., 2010. Dynamical instability produces transform faults at mid-ocean ridges. sins, north-east Africa. Geologische Rundschau 83, 671–688. Science 329, 1047–1050. Bosworth, W., McClay, K., 2001. Structural and stratigraphic evolution of the Gulf of Suez Gerya, T., 2012. Origin and models of oceanic transform faults. Tectonophysics 522–523, rift, Egypt: a synthesis. In: Ziegler, P.A., Cavazza, W., Robertson, A.H.F., Crasquin- 34–54. Soleau, S. (Eds.), Peri-Tethys Memoir 6: Peri-Tethyan Rift/Wrench Basins and Passive Hooft, E.E.E., Detrick, R.S., Toomey, D.R., Collins, J.A., Lin, J., 2000. Crustal thickness and Margins. Mémoires du Museum National d'Histoire Naturelle de Paris, 186, pp. 567–606. structure along three contrasting spreading segments of the Mid-Atlantic Ridge, Bosworth, W., Huchon, P., McClay, K., 2005. The Red Sea and Gulf of Aden Basins. Journal 33.5°–35°N. Journal of Geophysical Research 105, 8205–8226. of African Earth Sciences 43, 334–378. Huchon, P., Khanbari, K., 2003. Rotation of the syn-rift stress field of the northern Gulf of Brannan, J., Gerdes, K.D., Newth, I.R., 1997. Tectono-stratigraphic development of the Aden margin, Yemen. Tectonophysics 364 (3–4), 147–166. Qamar basin, eastern Yemen. Marine and Petroleum Geology 14, 701–730. Jarrigue, J.J., d'Estevou, P.O., Burollet, P.F., Montenat, C., Prat, P., Richert, J.P., Thiriet, J.P., Brekke, H., 2000. The tectonic evolution of the Norwegian Sea continental margin with 1990. The multistage tectonic evolution of the Gulf of Suez and northern Red Sea con- emphasis on the Vøring and Møre Basins. Geological Society, London, Special Publica- tinental rifting from field observations. Tectonics 9, 441–465. tion 167, 327–378. Jestin, F., Huchon, P., Gaulier, J.M., 1994. The Somalia plate and the East African Rift Sys- Calais, E., Ebinger, C.J., Hartnady, C., Nocquet, J.M., 2006. Kinematics of the East African rift tem: present-day kinematics. Geophysical Journal International 116, 637–654. from GPS and slip vector data, in the Afar Volcanic Province within the Katz, R.F., Ragnarsson, R., Bodenschatz, E., 2005. Tectonic microplates in a wax model of East African Rift System. In: Yirgu, G., Ebinger, C.J., Maguire, P.K.H. (Eds.), Geol. Soc. sea-floor spreading. New Journal of Physics 7. http://dx.doi.org/10.1088/1367-2630/ Spec. Publ., 259, pp. 9–22. 7/1/037. Choi, E., Lavier, L., Gurnis, M., 2008. Thermomechanics of mid-ocean ridge segmentation. Labails, C., Olivet, J.L., Aslanian, D., Roest, W.R., 2010. An alternative early opening scenario Physics of the Earth and Planetary Interiors 171, 374–386. for the Central Atlantic Ocean. Earth and Planetary Science Letters 297, 355–368. Chorowicz, J., 2005. The East African Rift System. Journal of African Earth Sciences 43, Lepvrier, C., Fournier, M., Bérard, T., Roger, J., 2002. Cenozoic extension in coastal Dhofar 379–410. (southern Oman): implications on the oblique rifting of the Gulf of Aden. Cochran, J.R., Martinez, F., 1988. Evidence from the northern Red Sea on the transition Tectonophysics 357, 279–293. from continental to oceanic rifting. Tectonophysics 153, 25–53. Leroy, S., et al., 2004. From rifting to spreading in the eastern Gulf of Aden: a geophysical Colletta, B., Le Quellec, P., Letouzey, J., Moretti, I., 1988. Longitudinal evolution of the Suez survey of a young oceanic basin from margin to margin. Terra Nova 16 (4), 185–192. rift structure (Egypt). Tectonophysics 153, 221–233. Leroy, S., et al., 2010a. Contrasted styles of rifting in the eastern Gulf of Aden: a combined Corti, G., van Wijk, J., Cloetingh, S., Morley, C.K., 2007. Tectonic inheritance and continen- wide-angle, MCS and heat flow survey. Geochemistry, Geophysics, Geosystems. tal rift architecture: numerical and analogue models of the East African Rift System. http://dx.doi.org/10.1029/2009GC002963. Tectonics 26. http://dx.doi.org/10.1029/2006TC002086. Leroy, S., d'Acremont, E., Tiberi, C., Basuyau, C., Autin, J., Lucazeau, F., Sloan, H., 2010b. d'Acremont, E., Leroy, S., Beslier, M.-O., Bellahsen, N., Fournier, M., Robin, C., Maia, M., Recent off-axis volcanism in the eastern Gulf of Aden: implications for plume–ridge Gente, P., 2005. Structure and evolution of conjugate passive margins of the eastern interaction. Earth and Planetary Science Letters 293 (1–2), 140–153. Gulf of Aden from seismic reflection data. Geophysical Journal International 160, Leroy, S., et al., 2012. From rifting to oceanic spreading in the Gulf of Aden: a synthesis. Ara- 869–890. bian Journal of Geosciences 5, 859–901. http://dx.doi.org/10.1007/s12517-011-0475-4. 50 N. Bellahsen et al. / Tectonophysics 607 (2013) 32–50

Lezzar, K.E., Tiercelin, J.J., LeTurdu, C., Cohen, A.S., Reynolds, D.J., LeGall, B., Scholz, C.A., 2002. Aden): stratigraphy and depositional environments. Tectonophysics 607, 1–16 (this Control of normal fault interaction on the distribution of major Neogene sedimentary volume). depocenters, Lake Tanganyika, East African Rift. AAPG Bulletin 86, 1027–1059. Roger, J., Platel, J.P., Cavelier, C., Bourdillon-de-Grisac, C., 1989. Données nouvelles sur la Macdonald, K., Scheirer, D., Carbotte, S., 1991. Mid-ocean ridges: discontinuities, seg- stratigraphie et l'histoire géologique du Dhofar (Sultanat d'Oman). Bulletin de la ments and giant cracks. Science 253 (5023), 986–994. Société Géologique de France 2, 265–277. McClay, K., Khalil, S., 1998. Extensional hard linkages, eastern Gulf of Suez, Egypt. Geology Rosendahl, B.R., 1987. Architecture or continental rifts with special reference to East 26, 563–566. Africa. Annual Review of Earth and Planetary Sciences 15, 445–503. Meyers, J.B., Rosendahl, B.R., GroscheI-Becker, H., Austin Jr., J.A., Rona, P.A., 1996. Deep Royer, J.Y., Gordon, R.G., Horner-Johnson, B.C., 2006. Motion of Nubia relative to penetrating MCS imaging of the rift-to-drift transition, offshore Douala and North Antarctica since 11 Ma: implications for Nubia–Somalia, Pacific-, and Gabon basins, West Africa. Marine and Petroleum Geology 13, 791–835. India–Eurasia motion. Geology 34, 501–504. http://dx.doi.org/10.1130/G22463.1. Mohriak, W.U., Rosendahl, B.R., 2003. Transform zones in the South Atlantic rifted conti- Ryan, W.B.F., Carbotte, S.M., Coplan, J., O'Hara, S., Melkonian, A., Arko, R., Weissel, R.A., nental margins. In: Storti, F., Holdsworth, R.E., Salvini, F. (Eds.), Intraplate Strike-slip De- Ferrini, V., Goodwillie, A., Nitsche, F., Bonczkowski, J., Zemsky, R., 2009. Global formation Belts. Geological Society, London, Special Publications, 210, pp. 211–228 Multi-Resolution Topography (GMRT) synthesis data set. Geochemistry, Geophysics, (0305-8719/03/$15). Geosystems 10, Q03014. http://dx.doi.org/10.1029/2008GC002332. Morley, C.K., 1989. Extension, detachments and sedimentation in continental rifts (with Sander, S., Rosendhal, B.R., 1989. The geometry of rifting in Lake Tanganyika, East Africa. particular reference to East Africa). Tectonics 8, 1175–1192. Journal of African Earth Sciences 8, 323–354. Morley, C.K., 1999. How successful are analogue models in addressing the influence Sandwell, D., 1986. Thermal stress and the spacings of transform faults. Journal of Geo- of pre-existing fabrics on rift structure? Journal of Structural Geology 21, physical Research 91, 6405–6417. 1267–1274. Scott, D.L., Etheridge, M.A., Rosendahl, B.R., 1992. Oblique-slip deformation in extensional Morley, C.K., Nelson, R.A., Patton, T.L., Munn, S.G., 1990. Transfer zones in the East African terrains: a case study in the lakes Tanganyika and Malawi rift zones. Tectonics 11 (5), Rift System and their relevance to hydrocarbon exploration in rifts. AAPG Bulletin 74, 998–1009. 1234–1953. Sénant, J., Popoff, M., 1991. Early Cretaceous extension in northeast Brazil related to the Morley, C.K., Cunningham, S.M., Harper, R.M., Wescott, W.A., 1992. Geology and geophys- South Atlantic opening. Tectonophysics 198, 35–46. ics of the Rukwa Rift, East Africa. Tectonics 11, 69–81. Skogseid, J., Planke, S., Faleide, J.I., Pedersen, T., Eldholm, O., Neverdal, F., 2000. NE Atlantic Mosar, J., Eide, E.A., Osmundsen, P.T., Sommaruga, A., Torsvik, T.H., 2002. Greenland– continental rifting and volcanic margin formation. Geological Society of London, Norway separation: a geodynamic model for the North Atlantic. Norwegian Journal Special Publication 167, 295–326. of Geology 82, 282–299. Smith, M., Mosley, P., 1993. Crustal heterogeneity and basement influence on the devel- Moulin, M., Aslanian, D., Unternehr, P., 2010. A new starting point for the South and Equa- opment of the Kenya rift, East Africa. Tectonics 12, 591–606. torial Atlantic Ocean. Earth-Science Reviews 98, 1–37. Specht, T.D., Rosendahl, B.R., 1989. Architecture of the Lake Malawi Rift, East-Africa. Journal Mountain, G.S., Prell, W.L., 1990. A multiphase plate tectonic history of the southeast con- of African Earth Sciences 8, 355–382. tinental margin of Oman. In: Robertson, A.H.F., Searle, M.P., Ries, A.C. (Eds.), The Ge- Stamps, D.S., Calais, E., Saria, E., Hartnady, C., Nocquet, J.-M., Ebinger, C.J., Fernandes, R.M., ology and Tectonics of the Oman Region. Geological Society Special Publication, 49, 2008. A kinematic model for the East African Rift. Geophys. Res. Lett. 35, L05304. pp. 725–743. http://dx.doi.org/10.1029/2007GL032781. Moustafa, A.R., 1997. Controls on the development and evolution of transfer zones: the Stern, R.J., Johnson, P., 2010. Continental lithosphere of the Arabian Plate: a geologic, pet- influence of basement structure and sedimentary thickness in the Suez rift and Red rologic, and geophysical synthesis. Earth-Science Reviews 101, 29–67. Sea. Journal of Structural Geology 19, 755–768. Taylor, B., Goodliffe, A., Martinez, F., 2009. Initiation of transform faults at rifted continen- Nocquet, J.M., Willis, P., Garcia, S., 2006. Plate kinematics of Nubia–Somalia using a com- tal margins. Comptes Rendus Geosciences 341, 428–438. bined DORIS and GPS solution. Journal of 80, 591–607. Turcotte, D., 1974. Are transform faults thermal contraction cracks? Journal of Geophysi- O'Bryan, J.W., Cohen, R., Gilliland, W.N., 1975. Experimental origin of transform faults and cal Research 79 (17), 2573–2577. straight spreading-center segments. GSA Bulletin 86, 793–796. Versfelt, J., Rosendahl, B.R., 1989. Relationships between pre-rift structure and rift archi- Oldenburg, D.W., Brune, J.N., 1972. Ridge transform fault spreading pattern in freezing tecture in Lakes Tanganyika and Malawi, East Africa. 337 (26), 354–355. wax. Science 178, 301. Vigny, C., Huchon, P., Ruegg, J.C., Khanbari, K., Asfaw, L.M., 2006. Confirmation of Arabia Oldenburg, D.W., Brune, J.N., 1975. An explanation for the orthogonality of ocean ridges plate slow motion by new GPS data in Yemen. Journal of Geophysical Research 111. and transform faults. Journal of Geophysical Research 80, 2575–2585. http://dx.doi.org/10.1029/2004JB003229. Olesen, O., Ebbing, J., Lundin, E., Mauring, E., Skilbrei, J.R., Torsvik, T.H., Hansen, E.K., Watchorn, F., Nichols, G.J., Bosence, D.W.J., 1998. Rift-related sedimentation and stratigraphy, Henningsen, T., Midbøe, P., Sand, M., 2007. An improved tectonic model for the southern Yemen (Gulf of Aden). In: Purser, B., Bosence, D. (Eds.), Sedimentation and Tec- Eocene opening of the Norwegian–Greenland Sea: use of modern magnetic data. tonics in the Rift Basins Red Sea — Gulf of Aden. Chapman & Hall, London, pp. 165–189. Marine and Petroleum Geology 24, 53–66. Watremez, L., Leroy, S., Rouzo, S., d'Acremont, E., Unternehr, P., Ebinger, C., Lucazeau, F., Pik, et al., 2013. Structural control of basement denudation during rifting revealed by low- Al-Lazki, A., 2011. The crustal structure of the north-eastern Gulf of Aden continental temperature (U–Th–Sm)/He thermochronology of the Socotra Island basement— margin: insights from wide-angle seismic data. Geophysical Journal International Southern Gulf of Aden margin. Tectonophysics 607, 17–31 (this volume). 184, 575–594. Ragnarsson, R., Ford, J.L., Santangelo, C.D., Bodenschatz, E., 1996. Rifts in spreading wax Watts, A.B., Stewart, J., 1998. Gravity anomalies and segmentation of the continental mar- layers. Physical Review Letters 76, 3456–3459. gin offshore West Africa. Earth and Planetary Science Letters 156, 239–252. Razin, P., Leroy, S., Robin, C., Robinet, J., Serra Kiel, J., Bellahsen, N., Grelaud, C., 2010. Wilson, J.T., 1965. A new class of faults and their bearing on . Nature 207, Dispositifs Tecto-Sédimentaires syn-rift et post-rift oligo-miocènes sur la marge sud 343–347. du golfe d'Aden–Ile de Socotra (Yémen). Paper presented at the RST, Bordeaux, Younes, A.I., McClay, K.R., 2002. Development of accommodation zones in the Gulf of Octobre 2010. Suez-Red Sea rift, Egypt. American Association of Petroleum Bulletin 86, Ring, U., 1994. The influence of preexisting structure on the evolution of the Cenozoic 1003–1026. Malawi rift (East African Rift System). Tectonics 13, 313–326. Young, M.J., Gawthorpe, R.L., Hardy, S., 2001. Growth and linkage of a segmented normal Robinet, J., Razin, P., Serra-Kiel, S., Gallardo Garcia, A., Leroy, S., Roger, J., Grelaud, C., 2013.The fault zone; the Late Murchison–Statfjord North Fault, northern North Sea. Paleogene pre-rift to syn-rift in the Dhofar margin (northeastern Gulf of Journal of Structural Geology 23, 1933–1952.