Continental Rift Evolution: from Rift Initiation to Incipient Break-Up in the Main Ethiopian Rift, East Africa

Home , Afar Triple Junction, East African Rift, Geography of Malawi, Kone (volcano), Volcanic plateau

Earth-Science Reviews 96 (2009) 1–53

Contents lists available at ScienceDirect

Earth-Science Reviews

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

Continental rift evolution: From rift initiation to incipient break-up in the Main Ethiopian Rift, East Africa

Giacomo Corti

Consiglio Nazionale delle Ricerche, Istituto di Geoscienze e Georisorse, via G. La Pira, 4, 50121, Firenze, Italy article info abstract

Article history: The Main Ethiopian Rift is a key sector of the East African Rift System that connects the Afar depression, at Received 15 January 2009 Red Sea–Gulf of Aden junction, with the Turkana depression and Kenya Rift to the South. It is a magmatic rift Accepted 23 June 2009 that records all the different stages of rift evolution from rift initiation to break-up and incipient oceanic Available online 3 July 2009 spreading: it is thus an ideal place to analyse the evolution of continental extension, the rupture of lithospheric plates and the dynamics by which distributed continental deformation is progressively focused Keywords: at oceanic spreading centres. continental rifting fi fl rift evolution The rst tectono-magmatic event related to the Tertiary rifting was the eruption of voluminous ood basalts continental break-up that apparently occurred in a rather short time interval at around 30 Ma; strong plateau uplift, which magmatism resulted in the development of the Ethiopian and Somalian plateaus now surrounding the rift valley, has deformation been suggested to have initiated contemporaneously or shortly after the extensive flood-basalt volcanism, East African Rift although its exact timing remains controversial. Voluminous volcanism and uplift started prior to the main Main Ethiopian Rift rifting phases, suggesting a mantle plume influence on the Tertiary deformation in East Africa. Different – Nubia Somalia kinematics plume hypothesis have been suggested, with recent models indicating the existence of deep superplume originating at the core-mantle boundary beneath southern Africa, rising in a north–northeastward direction toward eastern Africa, and feeding multiple plume stems in the upper mantle. However, the existence of this whole-mantle feature and its possible connection with Tertiary rifting are highly debated. The main rifting phases started diachronously along the MER in the Mio-Pliocene; rift propagation was not a smooth process but rather a process with punctuated episodes of extension and relative quiescence. Rift location was most probably controlled by the reactivation of a lithospheric-scale pre-Cambrian weakness; the orientation of this weakness (roughly NE–SW) and the Late Pliocene (post 3.2 Ma)-recent extensional stress field generated by relative motion between Nubia and Somalia plates (roughly ESE–WNW) suggest that oblique rifting conditions have controlled rift evolution. However, it is still unclear if these kinematical boundary conditions have remained steady since the initial stages of rifting or the kinematics has changed during the Late Pliocene or at the Pliocene–Pleistocene boundary. Analysis of geological–geophysical data suggests that continental rifting in the MER evolved in two different phases. An early (Mio-Pliocene) continental rifting stage was characterised by displacement along large boundary faults, subsidence of rift depression with local development of deep (up to 5 km) asymmetric basins and diffuse magmatic activity. In this initial phase, magmatism encompassed the whole rift, with volcanic activity affecting the rift depression, the major boundary faults and limited portions of the rift shoulders (off-axis volcanism). Progressive extension led to the second (Pleistocene) rifting stage, characterised by a riftward narrowing of the volcano-tectonic activity. In this phase, the main boundary faults were deactivated and extensional deformation was accommodated by dense swarms of faults (Wonji segments) in the thinned rift depression. The progressive thinning of the continental lithosphere under constant, prolonged oblique rifting conditions controlled this migration of deformation, possibly in tandem with the weakening related to magmatic processes and/or a change in rift kinematics. Owing to the oblique rifting conditions, the fault swarms obliquely cut the rift floor and were characterised by a typical right- stepping arrangement. Ascending magmas were focused by the Wonji segments, with eruption of magmas at surface preferentially occurring along the oblique faults. As soon as the volcano-tectonic activity was localised within Wonji segments, a strong feedback between deformation and magmatism developed: the thinned lithosphere was strongly modified by the extensive magma intrusion and extension was facilitated and accommodated by a combination of magmatic intrusion, dyking and faulting. In these conditions, focused melt intrusion allows the rupture of the thick continental lithosphere and the magmatic segments act as incipient slow-spreading mid-ocean spreading centres sandwiched by continental lithosphere.

E-mail address: giacomo.corti@uniﬁ.it.

Overall the above-described evolution of the MER (at least in its northernmost sector) documents a transition from fault-dominated rift morphology in the early stages of extension toward magma-assisted rifting during the ﬁnal stages of continental break-up. A strong increase in coupling between deformation and magmatism with extension is documented, with magma intrusion and dyking playing a larger role than faulting in strain accommodation as rifting progresses to seaﬂoor spreading. © 2009 Elsevier B.V. All rights reserved.

Contents

1. Introduction ...... 2 2. Physiography and different rift segments ...... 3 3. Brief summary of the pre-Tertiary geology of Ethiopia ...... 4 4. Flood-basalt volcanism and plateau uplift ...... 6 4.1. Flood-basalt volcanism ...... 6 4.2. Plateau uplift ...... 8 4.3. Dynamics of plateau uplift and ﬂood volcanism: Plume hypothesis for the East African Rift ...... 8 5. Tertiary rifting ...... 10 5.1. Plate kinematics setting ...... 10 5.1.1. Present-day kinematics ...... 10 5.1.2. Geological estimates of Nubia–Somalia kinematics ...... 11 5.1.3. Estimates of Nubia–Somalia separation ...... 11 5.2. Fault pattern ...... 11 5.2.1. Boundary fault system ...... 11 5.2.2. Wonji Fault Belt faulting ...... 15 5.2.3. Transverse structures ...... 17 5.2.4. Deformation outside the rift valley ...... 19 5.3. Summary of the evolution of volcanic activity ...... 19 5.3.1. Northern MER ...... 19 5.3.2. Central MER ...... 23 5.3.3. Southern MER ...... 24 5.3.4. Spatio-temporal distribution of volcanic activity ...... 24 5.4. Geophysical data constraining the crustal and mantle structure in the MER ...... 27 5.4.1. Crustal structure ...... 27 5.4.2. Upper mantle structure ...... 34 5.5. Seismicity and distribution of current deformation ...... 36 6. The evolution of the Main Ethiopian Rift: Continental rifting from initiation to incipient break-up ...... 38 6.1. Rift initiation: Localisation and propagation of extensional structures ...... 38 6.1.1. Rift localisation and propagation ...... 38 6.1.2. The Red Sea, Gulf of Aden, and Ethiopian rifts triple junction history ...... 40 6.1.3. Activation of boundary fault systems ...... 42 6.2. Rift maturity: Abandonment of boundary faults and development of Wonji segments ...... 43 6.3. Wonji magmatic segments and continental break-up ...... 44 7. Conclusions ...... 48 Acknowledgments ...... 49 References ...... 49

1. Introduction Building on previous synthetic works (e.g., Hayward and Ebinger, 1996; Ebinger, 2005), in this paper I will review these new geological The Main Ethiopian Rift is a key sector of the East African Rift and geophysical findings and integrate them with old data, to provide System that connects the Afar depression, at the Red Sea–Gulf of Aden a comprehensive analysis of the evolution continental rifting in the junction, with the Turkana depression and Kenya Rift to the south Main Ethiopian Rift, from early stages to break-up and incipient (e.g., Mohr, 1983; Rosendahl, 1987; Braile et al., 1995; Boccaletti and spreading. Peccerillo,1999; Chorowicz, 2005). It is a magmatic rift that records all After introducing the rift system and its physiography (Section 2), I the different stages of rift evolution from rift initiation to break-up and will briefly summarise the main aspects of the pre-Tertiary geological embryonic oceanic spreading (e.g., Ebinger, 2005), marking the history of Ethiopia that have relevance for the evolution of rifting incipient boundary between Nubia and Somalia plates. The Main (Section 3), and describe the flood-basalt volcanism and plateau uplift Ethiopian Rift it is thus an ideal place to analyse the evolution of (Section 4). Note that since it has been shown that the flood-basalt continental extension, the rupture of lithospheric plates and the event that affected Ethiopia at around 30 Ma is (temporally) unrelated dynamics by which distributed continental deformation is progres- to the Mio-Pliocene rifting (e.g., Wolfenden et al., 2004; Bonini et al., sively focused at oceanic spreading centres. Thanks to these ideal 2005), this magmatic pulse will be treated before the description of conditions, the Main Ethiopian Rift has focused the attention of many the rifting phases. The Tertiary extensional rifting will be indeed research teams and a large dataset of new geological and geophysical described in detail in Section 5. In this section, I will firstly review the data has been acquired in recent years, greatly improving our plate kinematics setting of the rifting process (Section 5.1), then will knowledge of the structure of the rift, its evolution and the discuss in detail the fault pattern in the different rift segments interactions between magmatism and deformation and how they (Section 5.2), summarise the main volcanic events accompanying evolve from early rifting to incipient break-up. extension (Section 5.3), present an overview of the geophysical data G. Corti / Earth-Science Reviews 96 (2009) 1–53 3 constraining the crust, lithospheric and mantle structure (Section 5.4) modated by the ~300 km-wide system of basins and ranges (referred and illustrate the characteristics of seismic activity and the distribu- to as Broadly Rifted Zone; Baker et al., 1972; Moore and Davidson, tion of the current extensional deformation (Section 5.5). 1978; Davidson and Rex, 1980; Ebinger et al., 2000) that characterises Finally, in Section 6, all the above data will be used to illustrate the the overlapping area between the Ethiopian and Kenya Rifts. To the history of extension in the MER, which provides a more general model north, the present-day Red Sea–Gulf of Aden–Ethiopian rift triple for the evolution of continental rifts, from early stages to incipient junction lies in a complex zone at ~11.5° N within the central Afar break-up. Discussion will be focused on the influence of the depression, where a left-lateral, oblique-slip, Quaternary fault zone modification of the lithospheric properties induced by magmatic (the Tendaho-Goba'ad Discontinuity) separates the roughly E–W processes on weakening and subsequent break-up of the lithosphere. extension in the south (Ethiopian Rift) from the NE–SW extension in It is out of the scope of this paper to summarise the geology of the the north (e.g., Wolfenden et al., 2004 and references therein). other sectors of the East African Rift and the Red Sea–Gulf of Aden The Ethiopian Rift can be divided into two main physiographic systems and only some general aspects that are of interest for the segments, namely southern Afar and the Main Ethiopian Rift evolution and structure of the MER (e.g., lithosphere–asthenosphere (Figs. 2, 3); the rift morphology is typically developed in this latter structure and general characteristics of the magmatism of the Afar segment, where a ~80 km-wide rift valley (Ethiopian Rift valley sensu depression) will be briefly introduced in the following sections. stricto of Mohr, 1983) separates the uplifted western (Ethiopian) and Further insight into these rift systems can be found in recent review eastern (Somalian) plateaus (Fig. 2b). Mohr (1962) pointed out that papers (e.g., Chorowicz, 2005; Bosworth et al., 2005; Beyene and the boundary between the MER and southern Afar does not Abdelsalam, 2005). correspond to any physiographic feature, as the rift valley gradually funnels outwards into the wide Afar depression north of Addis 2. Physiography and different rift segments Ababa. A limit can be placed in correspondence to an arcuate pattern of faults at latitude ~10° N (Tesfaye et al., 2003), which Wolfenden et The Ethiopian Rift extends for about 1000 km in a NE–SW to N–S al. (2004) interpreted as the southern termination of the Oligocene direction from the Afar depression, at the Red Sea–Gulf of Aden Red Sea rift. junction, southwards to the Turkana depression (Figs. 1–3). The In turn the MER, which is the focus of this paper, can be subdivided southern boundary may be traced at latitude ~5°N, south of the area into three segments that have been interpreted to reflect different where the rift is divided into two branches (Chamo basin to the west stages of the continental extension process, being characterised by and Galana basin to the east) by the Amaro Mts; southwards, the rift different fault architecture, timing of volcanism and deformation, zone widens and deformation becomes more complex being accom- crustal and lithospheric structure (Figs. 2, 3; see below; e.g., Hayward

Fig. 1. Digital elevation model (from Shuttle Radar Topography Mission – SRTM – data) showing the topographic expression of the East African Rift System. Inset shows a schematic plate kinematic setting of the area. BRZ: Broadly Rifted Zone; EAP: East African Plateau; ER: Ethiopian Rift; ESP: Ethiopian–Somalian plateaus; KR: Kenya Rift; MR: Malawi Rift; SAP: Southern African Plateau; TR: Tanganyika Rift. 4 G. Corti / Earth-Science Reviews 96 (2009) 1–53

Fig. 2. a) Digital elevation model (SRTM data) of the Ethiopian Rift showing the main rift segments: (from north to south) Southern Afar (Safar), Northern Main Ethiopian Rift (NMER), Central Main Ethiopian Rift (CMER) and Southern Main Ethiopian Rift (SMER). BRZ: Broadly Rifted Zone; CB: Chow Bahir Rift; GB: Gofa Basin and Range. b) Topographic profile across the rift valley. Note the rift depression separating the uplifted Ethiopian and Somalian Plateaus. c) Topographic profile along the rift valley. Note the decrease in elevation of the rift bottom from the lakes region in the CMER both southwards and northwards. See text for details. and Ebinger, 1996). The Northern MER extends from the MER–Afar and Awash rivers immediately north of Lake Ziway, where the rift boundary southwards to the Lake Koka area, where it is separated valley attains its maximum elevation at ~1700 m asl. Northwards the from the Central MER by the Boru Toru Structural High (Bonini et al., rift floor descends regularly into the Afar depression where, over 2005). To the south, the boundary between Central and Southern MER extensive areas, it lies below sea level. Local increases in the elevation can be placed at ~7° N latitude, in the Lake Awasa area, where the rift of the rift valley are generally due to volcanic edifices, as in the margins rotate from ~NE–SW to ~N–S, in correspondence to the Northern MER where several volcanoes raise from the flat rift floor Goba–Bonga transverse lineament (Bonini et al., 2005). (Fig. 2c). As stated above, the rift valley in the MER separates the Ethiopian and Somalian plateaus, made of uplifted basement rocks, and 3. Brief summary of the pre-Tertiary geology of Ethiopia overlying sedimentary sequences and Eocene–Recent flood basalts (Fig. 2). The plateaus rise to elevations N2000 m above sea level; north The pre-rift rocks of Ethiopia are composed by an extremely folded of latitude 9° the highest elevations are attained by the Ethiopian and foliated basement of pre-Cambrian rocks, overlain by sub- Plateau, whereas south of this latitude the Somalian plateau reaches horizontal Mesozoic marine sediments. The basement rocks, exposed the highest elevations (Fig. 2). The rift floor raises in elevation from in the extreme south of the MER, in northern Afar and in a small the Turkana depression up to the main watershed between the Meki outcrop at the base of the western rift escarpment of the Central MER G. Corti / Earth-Science Reviews 96 (2009) 1–53 5

Fig. 3. Three-dimensional representation of the rift topography in the different MER segments (SRTM data). Abbreviations in a): Ank: Ankober escarpment; Ar: Arboye escarpment; B: Boseti volcano; D: Dofen volcano; F: Fantale volcano; G: Gedemsa caldera; Gu: Guraghe escarpment; K: Kone caldera; Si: Sire escarpment; Zi: Ziquala volcano. Abbreviations in b): A: Lake Abiyata; Ab: Lake Abaya; AH: Amaro Horst; As: Asela escarpment; AS: Agere-Selam escarpment; Aw: Awasa; C: Chilalo volcano; Ch: Lake Chamo; Che: Chencha escarpment; Ga–Gi: Gamo–Gidole horst; Fo: Fonko escarpment; L: Lake Langano; S: Lake Shala; Z: Lake Ziway. Other abbreviations as in Fig. 2.

(Kella Horst), are part of the Arabian–Nubian Shield and record are mainly aligned N–StoNE–SW, although the trends of the sutures folding, compression and metamorphism during different deforma- are still debated (e.g., Vail, 1983; Kroner, 1985; Berhe, 1990; Church, tion episodes in the pre-Cambrian. The Arabian–Nubian Shield is the 1991; Abdelsalam and Stern, 1996). In the site of the future MER, northern part of the East African Orogen, which formed by collision differences in crustal and mantle properties between the eastern and between east and west Gondwana at the end of a Wilson Cycle that western plateaus (see Section 5.4) may provide evidence for the encompassed most of the Neoproterozoic and deﬁned the Pan-African location of a pre-Cambrian suture that has been interpreted to run Orogeny (Stern, 1994; Fig. 4). Collision is marked by north- to east- through this region by mapping of ophiolitic fragments (Vail, 1985; trending sutures, characterised by ophiolites complex, and post- Berhe, 1990; Stern et al., 1990), Nd isotopic data (Stern, 2002), and accretionary structures including north-trending shortening zones SKS-splitting patterns (Gashawbeza et al., 2004). and major northwest-trending sinistral and minor northeast-trending The Pan-African deformation was followed by a long period of dextral strike-slip faults (Abdelsalam and Stern, 1996). The ophiolites peneplanation in the Paleozoic, with the pre-Cambrian orogenic 6 G. Corti / Earth-Science Reviews 96 (2009) 1–53

Fig. 4. Cratons/shields and orogens forming East (yellow) and West (blue) Gondwana (modified from Meert and Lieberman, 2008). The heavy dashed box indicates the future area of the East African Rift. MB: Mozambique Belt, RP: Rio de la Plata Craton, TC: Tanzanian Craton. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) mountain ranges almost completely worn down by denudation during mould for the development of the East African Rift in Ethiopia. In this period (e.g., Mohr, 1962). particular, ancient NE–SW to NNE–SSW tectonic trends largely Extensional deformation associated to Karoo-type rifts developed controlled the location and directions of the Tertiary rift-faulting from Early Permian times in Eastern Africa and a narrow, NE–SW- (e.g., Mohr, 1962; Kazmin et al., 1980). Recent geophysical analysis trending shallow-sea arm invaded the depressed corridor developing (Keranen and Klemperer, 2008; Bastow et al., 2008; Keranen et al., between the Horn of Africa and the Madagascan–Indian promontory 2009) support that the pre-rift lithospheric structure controlled the of Eastern Gondwana. This episode was followed by development of initial rift evolution, with lithospheric-scale pre-existing weaknesses several NW–SE striking structural basins that served as active controlling rift location and propagation in a roughly NE–SW direction depositional basins during the Mesozoic (see Guiraud et al., 2005 (see also below Sections 5.4.1 and 6.1.1). for review). In the Jurassic–Cretaceous continental rifting was very At a more local scale, the Tertiary extensional reactivation of pre- active within the African plate, giving rise to the Central African Rift existing basement structures may have controlled local fault geome- System. Close to the area of the future MER, several NW–SE narrow try, interaction and propagation, as in many places the foliation of the troughs developed, as documented by the Blue Nile, Malut and pre-Cambrian rocks and the Tertiary extensional faulting are generally Muglad rifts in Sudan and by the Anza graben in Southern Ethiopia parallel (e.g., Mohr, 1962; see Section 5.2). and Northern Kenya (Fig. 5). The events were followed by local inversions and rift rejuvenation along these NW–SE striking basins up 4. Flood-basalt volcanism and plateau uplift to the Paleogene. This long pre-Tertiary history of tectonism created pre-existing 4.1. Flood-basalt volcanism heterogeneities with variable orientation that have represented a Volcanism in Ethiopia started during the Eocene–Late Oligocene with the eruption of the Ethiopia–Yemen flood-basalt province (Trap series), located at the Red Sea–Gulf of Aden–East African rift triple junction (Fig. 6). Prolific eruption of basalt and intercalated silicic volcanics built a subaerial volcanic pile typically 500–1500 m thick and locally attaining 3000 m (e.g., Mohr and Zanettin 1988). The total area covered by these volcanic rocks has been estimated as presently 600 000 km2, and not less than 750 000 km2 before erosion; these flood basalts contribute to an estimated volume of 300 000 km3 (e.g., Mohr, 1983; Mohr and Zanettin 1988). About 90% of the preserved volume forms the Ethiopian and Somalian plateaus, 10% the Yemen part and minor volumes have been also recognised in the Afar depression (Fig. 6; e.g., Coulié et al., 2003). This volcanic episode was characterised by the extrusion of flood tholeiitic to alkaline lava flows, with magmatic characteristics varying in different regions of the plateau (e.g. Kieffer et al., 2004); interbedded with the flood basalts, particularly at upper stratigraphic levels, are felsic lavas and pyroclastic rocks of rhyolitic, or less Fig. 5. NW–SE-trending Mesozoic basins developed in Ethiopia and surrounding areas commonly, trachytic compositions (e.g., Mohr and Zanettin, 1988). (after Gani et al., 2009). Eruption of flood basalts mostly occurred through fissures (e.g., Mohr G. Corti / Earth-Science Reviews 96 (2009) 1–53 7

Fig. 6. a) Schematic distribution of the flood-basalt volcanism (modified from Chorowitz, 2005). b) Geometry of flood basalts along the Guraghe escarpment (after Abebe et al., 2005). and Zanettin, 1988), at places controlled by pre-existing weaknesses Immediately after the peak of volcanic activity related to the flood- inherited from the Precambrian (Mege and Korme, 2004). Most of the basalt emplacement, a number of large shield volcanoes developed basalts and associated felsic rocks were apparently erupted in a rather from 30 Ma to about 10 Ma on the surface of the volcanic plateau (e.g., short time interval (b5 Ma) with the greatest eruption rates occurring Kieffer et al., 2004). This subsequent, less voluminous volcanic activity from 31 to 28 Ma (e.g., Baker et al., 1996; Hoffman et al., 1997; Pik formed some of the highest reliefs of the plateau (such as Mts. Simien, et al., 1998; Ukstins et al., 2002; Coulié et al., 2003), although 4533 m. asl and Choke, 4052 m. asl), rising 1000–2000 m above the radiometric dating results should be taken critically, as influenced by top of the surrounding flood volcanics and being characterised by a several factors (e.g., quantity of available datings compared with the basal diameter of 50–100 km (e.g., Mohr and Zanettin, 1988; Kieffer huge volume of the magmatism; rock chemistry, age and possible et al., 2004). The presence of these conspicuous shield volcanoes secondary processes; difficulties in exactly dating the beginning and above the flood basalts, together with a more complex space-time end of volcanic stages; etc.) and potentially modified by additional evolution of volcanism, a variable magmatic character within the datings. This strong eruption was concomitant with the onset of plateau and a high proportion of felsic pyroclastic rocks, differentiates continental rifting in the Red Sea–Gulf of Aden systems by 29 Ma the Ethiopian province from typical continental flood-basalt provinces (Wolfenden et al., 2005), but predates the main rifting phases (e.g., Deccan), which are characterised by thick, monotonous associated to the development of the MER (see Section 5.2). Limited sequences of thick, continuous, near-horizontal flows of tholeiitic volumes of basalts as old as 45 Ma have been described in southern basalt without overlying shield volcanoes (see Kieffer et al., 2004; Ethiopia, in the Broadly Rifted Zone separating the MER from the Yirgu et al., 2006). Kenya rift (e.g., Davidson and Rex, 1980; Ebinger et al., 1993; George The Ethiopian flood basalts have been attributed to melting et al., 1998). associated with the activity of one (Afar plume) or two (Afar and East 8 G. Corti / Earth-Science Reviews 96 (2009) 1–53

African or Kenyan plumes) mantle plumes impinging the base of the the Afar Depression, crustal modifications by magmatism, etc.; e.g., continental lithosphere (see Section 4.3; e.g., Ebinger and Sleep, 1998; Gani et al., 2007). In particular, a significant portion of plateau uplift in George et al., 1998). In addition, it has been suggested that the flood Ethiopia can be attributed to the thermal alteration of the crust, basalts of Ethiopia may result from multiple plume stems rising from a mantle lithosphere and deep upper mantle induced by a mantle single, broad and deep-seated mantle upwelling rising from the lower plume activity (see Dugda et al., 2007). The length-scale (a few tens of mantle (the African superplume; e.g., Furman et al., 2006; Meshesha kilometres) of rift-related flexural uplift is likely too short to have had and Shinjo, 2008;seeSection 4.3). The heterogeneity of the flood basalts a major role in uplifting the hundreds of kilometres-wide plateau (e.g., that built the Ethiopian plateau supports a magma genesis from a broad Pik et al., 2003; Gani et al., 2007); similarly, uplift induced by crustal region of mantle upwelling, heterogenous in terms of both temperature modifications by underplating (e.g., Maguire et al., 2006) seems to and composition (Kieffer et al., 2004). Within the lithosphere, slow or play a minor role, as crustal thickening by massive basal magmatic absent extension may have favoured the focussing of upwelling of large addition has been suggested to be mostly a local feature instead of volumes of magma, responsible for the formation of the flood-basalt widespread characteristic of the Ethiopian Plateau (see Section 5.4.1). province (Mulugeta et al., 2007). Uplift of the Ethiopian/Somalian and East African plateaus may have had a first-order impact on the climate evolution of East Africa in 4.2. Plateau uplift the last 8 Myr (e.g., Sepulcre et al., 2006), although the climatic evolution models are strongly dependent on the assumed timing of The Ethiopian and Somalian plateaus are part of the so-called topographic variations (see Pik et al., 2008). In turn, late Cenozoic African Superswell, a wide region of anomalously high topography environmental changes may have controlled the evolutionary paths of comprising the East African and southern African Plateaus as well as a East African hominins (see for instance Gani et al., 2007). bathymetric swell in the southeastern Atlantic Ocean basin (Figs. 1, 2; Nyblade and Robinson, 1994). Indeed, much of the African continent 4.3. Dynamics of plateau uplift and flood volcanism: Plume hypothesis for stands above 1 km, giving rise to the highest mean elevation of all of the East African Rift the continents. This anomalous topography results from strong uplift during the Tertiary; in Ethiopia, the high plateaus have experienced up The Tertiary rifting history of Ethiopia cannot be easily explained to ~2-km rock uplift since ~30 Ma (e.g., Pik et al., 2003; Gani et al., by simple passive rifting, where extensional stresses act in response to 2007). Early works (e.g., Dainelli, 1943; Beydoun, 1960) suggested this a regional stress field (usually originating from remote plate boundary uplift to be of Upper Eocene in age and to precede both the flood- forces). In this mechanism, indeed, volcanism (and local uplift) basalt event and the main rifting episodes (see Mohr, 1962). follows the extensional deformation of the lithosphere, since the Successive works (e.g., Mohr, 1967; Merla et al., 1979) argued against asthenospheric upwelling (giving rise to decompressional melting) is major uplift before the flood-basalt event (see also Menzies et al., a (passive) response to extensional-related lithospheric thinning. The 1997), and suggested that a complex history of uplift and volcanism development of the MER within broadly elevated regions (plateaus) may have occurred in Ethiopia during the Tertiary (e.g., Baker et al., and its development after eruption of large volumes of fissural basaltic 1972; McDougall et al., 1975; Berhe et al., 1987). lava flows are instead coherent with the thermal and/or dynamic More recent works attempted a more precise dating of the main consequences of mantle plumes acting at (or near) the base of the events; despite these attempts, no consensus exists on the exact lithosphere (e.g., Ebinger et al., 1989; Davies, 1998; Pik et al., 2006). timing of the uplift. For instance, based on thermochronological and In particular, rifting in East Africa has been typically associated to morphological analysis on the Blue Nile drainage network, Pik et al. the upwelling of starting thermal plumes, in which a heat source at a (2003) suggested that erosion initiated in the Blue Nile canyon as boundary layer (e.g., the core-mantle boundary) forms a thermally early as 25–29 Ma ago and that the elevated plateau physiography buoyant spherical head connected to a tail or conduit that feeds source existed since the Oligocene (20–30 Ma) predating the main tectonic material to the head (as in the classical model by Griffiths and events related to the Africa–Arabia separation (see also Menzies et al., Campbell, 1990). Geochemical analysis of rift basalts, volcanic gases 1997; Pik et al., 2008). In this interpretation, the morphology of the and geothermal fluids (e.g., Hart et al., 1989; Schilling et al., 1992; large marginal escarpments separating the Afar depression from the Marty et al., 1996; Scarsi and Craig, 1996) are consistent with a mantle uplifted plateau would be mainly the result of the collapse of the Afar plume influence in the Ethiopian Rift, but no consensus exists on the area from an initially elevated region since 20 Ma. Conversely, more number of mantle upwellings. Up to about 40 plumes beneath Africa recent analysis based on the incision of the 1.6-km-deep Gorge of the have been proposed in earlier works (e.g., Burke and Wilson, 1976); Nile (Gani et al., 2007) suggests a different history of the growth of the more recent works have proposed the existence of a single channeled Ethiopian Plateau. Although this model admits a possible, early plume (Ebinger and Sleep, 1998) or two starting plumes (e.g., George Oligocene domal uplift of ~1 km (an effect subsequently counter- et al., 1998; Rogers, 2006) impinging the base of the continental balanced by subsidence related to the flood-basalt emplacement), the lithosphere. In the channeled plume model, the starting plume analysis of the evolution of the Blue Nile drainage suggests that uplift impinges the base of the lithosphere beneath Central Ethiopia and of ~2 km occurred episodically in three different phases since ca. then propagates following the topography of the lithosphere/ 29 Ma. The plateau experienced a slow and steady uplift, if any, asthenosphere boundary, a process that can explain the distribution between ca. 29 and ca. 10 Ma (phase I), followed by an increase in and timing of magmatism and uplift throughout east Africa. In the uplift rates at ca. 10 Ma (phase II) and a further dramatic increase at ca. two-plume model, the occurrence of the oldest basalts in southern 6 Ma (phase III); according to this model, the majority of the ~2-km Ethiopia and their distinct geochemical signatures with respect to rock uplift of the Ethiopian Plateau occurred within a few m.y. after ca. flood basalts of the Ethiopian plateau argue for the activity of two 6 Ma. This evolution – with a polyphase history of doming and strong, separate mantle plumes (Afar and East African or Kenyan plumes). In recent (Plio-Pleistocene) increase in uplift rates of the plateau – is in this model, the old basalts of Southern Ethiopia represent the first line with predictions of previous works (e.g., Mohr, 1967; Baker et al., manifestation of the East African plume and, as the African plate 1972; McDougall et al., 1975; Almond, 1986 and references therein; migrated north, subsequent magmatic activity is represented by Adamson and Williams, 1987). progressively younger episodes further south through Turkana, Kenya Overall, of the total uplift of the plateau since ca. 30 Ma, only a and into northern Tanzania (e.g., George et al., 1998; Rogers, 2006). minor part seems to be related to isostatic uplift related to erosional Global seismic tomography studies have indicated the existence of unloading; the major contribute comes from rift-related activity (e.g., a large-scale low-velocity anomaly in the lower mantle beneath plume-related processes, flank uplift of the Main Ethiopian Rift and southern Africa that possibly connects with a low seismic velocity G. Corti / Earth-Science Reviews 96 (2009) 1–53 9 anomaly in the upper mantle beneath East Africa (e.g., Ritsema et al., 2000). Plume-like features extending to mid-mantle depths are 1999). This tomographic image has been interpreted as representing a imaged beneath Afar (Fig. 7b; Montelli et al., 2004), which may broad thermal upwelling (superplume) originating at the core-mantle actually reach the base of the mantle connecting to the African boundary beneath southern Africa, and rising in a north–north- superplume (Montelli et al., 2006). eastward direction toward eastern Africa, exhibiting an overall dip to Regional tomography shows a rather complex upper mantle the west or southwest (Fig. 7a; Ritsema et al., 1999). The mantle flow structure. Low-velocity anomalies in the upper mantle are imaged induced by the existence of this superplume has been shown to be beneath the MER (e.g., Bastow et al., 2005, 2008; Benoit et al., 2006a, able to dynamically support the excess elevation of the African b; see Section 5.4.2) and the characteristics of these anomalies (width superswell (e.g., Lithgow-Bertelloni and Silver, 1998; Gurnis et al., N500 km, depth≥660 km and an overall westward dip), which are not consistent with the classical starting plume hypotheses, have been interpreted to indicate a possible connection with the hypothesised superplume (e.g., Benoit et al., 2006a,b; Bastow et al., 2008). Similar low-velocity anomalies in the upper mantle have been indentified to the south under the East African Plateau and the Tanzanian craton (e.g., Weeraratne et al., 2003), suggestive of the existence of a separate plume or upper mantle instability (East African plume) that may have originally impacted beneath SW Ethiopia (e.g., George et al., 1998). Geochemical evidences from analysis of flood basalts in Ethiopia and northern Kenya have been used to suggest models in which these upper mantle anomalies (Afar and East African plumes) may represent multiple plume stems rising from the hypothesised African superplume (e.g., Furman et al., 2006; Meshesha and Shinjo, 2008). Similarly, Kendall et al. (2006) proposed a model, based on shear- wave splitting analysis indicating fast shear waves within the mantle at depths greater than 150 km parallel to the trend of the major thermal upwelling, in which the hypothesised African superplume feeds smaller plumes or upwellings that impinge the base of a heterogenous lithosphere. More complex scenario use geochemical data to suggest a deep-sited, superplume origin for the Ethiopian flood-basalt province only, whereas magmas from other African volcanic provinces are associated to distinct, shallower second-order upwellings, isolated from the main deep plume (Pik et al., 2006). Recent upper mantle tomographic analysis seems to support a different (deeper) origin for the Afar hot spot, with respect to the other (shallower) mantle upwellings in the African continent (Sicilia et al., 2008). According to these Authors, the Afar plume may represent a deep plume, whose connection with the lower mantle might have already disappeared. Geochemical analysis of Oligocene through recent basalt volcanism from Ethiopia, Yemen and Djibouti evidence a similar fundamental isotopic composition, suggestive of a common origin from the Afar plume that thus represents a long-lived feature of the mantle (Furman et al., 2006). The majority of the models illustrated above are modifications of the superplume hypothesis based on global seismic tomography studies. However, the existence of such a continuous hot feature penetrating the whole mantle is highly debated, based on the intrinsic limitations of global tomography studies and on the dependence of seismic velocity on other parameters (composition, phase) in addition to temperature (it is consequently not clear if the vast low-velocity anomaly in the lower mantle results from high temperatures or is controlled by other parameters such as density; see for instance Anderson, 2007 and references therein for discussion). In addition, it is also possible that the observed low-velocity structure in the lower mantle represent ‘normal’ thermal fluctuations in a stratified mantle (Ritsema and Allen, 2003). Even admitting the existence of this large- scale thermal upwelling, its connection with surface structures, volcanism and the uppermost mantle structure beneath the EARS remain poorly defined. Shear-wave tomography (Panza et al., 2007a) indicates the presence of highly stratified and heterogeneous upper mantle beneath the African continent, which – together with other evidences – Fig. 7. a) Cross-section of the mantle velocity structure beneath the African continent suggest shallow upper mantle convection/circulation (see also Panza showing the large-scale low-velocity anomaly (red areas) originating below southern et al., 2007b). Consistently, recent tomography reveals that at least Africa and rising towards East Africa (after Ritsema et al., 1999). b) Three-dimensional some Central Africa hotspots (Darfur, Tibesti, Hoggar) have a super- view showing the vertical extent of the Afar plume to mid-mantle depths (after fi Montelli et al., 2004). (For interpretation of the references to colour in this figure cial character and may be related to convective instabilities in the legend, the reader is referred to the web version of this article.) asthenosphere and/or edge-driven convection process (Sicilia et al., 10 G. Corti / Earth-Science Reviews 96 (2009) 1–53

2008), such that non-plume models for the EARS cannot be ruled out simplistic as extension occurs both along the Western and Eastern completely (see also Nyblade and Langston, 2002). branches and different microplates are present between the two major plates (Fig. 8). 5. Tertiary rifting 5.1.1. Present-day kinematics 5.1. Plate kinematics setting The most recent kinematics models based on analysis of space geodesy velocities and earthquake slip vectors (Stamps et al., 2008) The evolution of rifting in the MER is strictly related to the long- indicate the occurrence of at least three intervening microplates term kinematics of the major Nubia and Somalia plates, whereas south (Lwandle, Rovuma, Victoria) between Nubia and Somalia south of of the Turkana depression a two-plate model for the EARS is too latitude ~5°N. The relative Nubia–Somalia motion occurs with a

Fig. 8. a) Plate kinematics setting of the East African Rift System (EARS) (after Horner-Johnson et al., 2007). Black dots indicate earthquake epicentres (1964–1995); ABTFC: Andrew Bain Transform-Fault Complex; BTJ: Bouvet triple junction; RTJ: Rodrigues triple junction; CIR: Central Indian Ridge. b) Relative velocities along the EARS and adjoining areas in a Nubia-fixed reference frame (after Stamps et al., 2008). Thin black arrows indicate modelled velocities along plate or block boundaries; thick black arrows indicate motions at GPS sites (both modelled and measured velocities are reported). Relative rotation poles are shown with black stars. The first plate rotates counter-clockwise with respect to the second, except for VI–NU where Victoria rotates clockwise with respect to Nubia. c) Nubia–Somalia displacement vectors at the latitude of the Main Ethiopian Rift derived from different analyses, as follows: local geodetic observations (Bendick et al., 2006); analysis of elongation of caldera complexes (Casey et al., 2006); stress inversion from focal mechanisms (Keir et al., 2006a); palaeostress analysis of fault-slip data (Pizzi et al., 2006); plate motion data (Royer et al., 2006; Horner-Johnson et al., 2007); geodetic solution and earthquake slip vectors (Stamps et al., 2008). Light gray boxes in bottom left indicate the approximate timescale (reported in x axis) of the different approaches used to analyse the relative plate motion. G. Corti / Earth-Science Reviews 96 (2009) 1–53 11 rotation pole located at around 36°S, 35°E and gives rise to a roughly Nubia–Somalia separation. These values are in agreement with ESE–WNW-directed extension at the latitude of the MER, with rates of geological data predicting stretching values of β~1.5 in the Northern ~6–7 mm/yr (Fig. 8), similarly to previous GPS data (e.g., N108°E at MER, corresponding to ~30 km of Nubia–Somalia separation (e.g., ~7 mm/yr, Sella et al., 2002; N94°E at ~7 mm/yr, Fernandes et al., Wolfenden et al., 2004). Overall, stretching values increase towards 2004). These data are also consistent with local geodetic observations Afar along the rift axis: extension is thought to be limited in the across the rift valley (Billham et al., 1999; Bendick et al., 2006), which Southern MER (e.g., Ebinger et al., 1993) whereas stretching values indicate a similar direction of extension (N105°E to N108°E), although can exceed 100% (i.e., βN2) or maybe 200% (i.e., βN3) in the Afar at slower velocities (~4 mm/yr). Analysis of the current stress field depression (e.g., Berckhemer et al., 1975; Joffe and Garfunkel, 1987; derived from borehole breakout studies (Bosworth et al., 1992) Eagles et al., 2002). Rift-parallel seismic sections (Maguire et al., 2006; support a current ESE–WNW extension, similar to the analysis of the see also Keranen and Klemperer, 2008) confirm a northward increase current seismic activity in the rift (Keir et al., 2006a). Stress inversion in thinning, corresponding to an increase in stretching from β~1.1in using focal mechanisms of earthquakes recorded between October the Central MER to β~1.7 in the Northern MER–Southern Afar; this is 2001 and January 2003 (see Section 5.5) suggests indeed an extension consistent with gravity data by Tiberi et al. (2005) imaging an increase direction directed ~N103°E, which is strikingly similar to the in stretching from the Central MER (β~1.2) to the southern Afar stretching vector estimated from previous analysis of the rift seismic depression (β ~2). release (e.g., roughly N100°E extension in the MER; Foster and Jackson, 1998). 5.2. Fault pattern

5.1.2. Geological estimates of Nubia–Somalia kinematics 5.2.1. Boundary fault system These present-day data are consistent with the Quaternary The Main Ethiopian Rift is bounded by discontinuous boundary extension direction derived from elongation of caldera complexes faults that give rise to major fault-escarpments separating the rift (e.g., Boccaletti et al., 1998; Ebinger and Casey, 2001; Casey et al., depression from the Ethiopian and Somalian plateaus. These faults are 2006) and analysis of fault-slip and geological data (e.g., Boccaletti et normally long, widely spaced and characterised by large vertical al., 1998; Bonini et al., 2005; Pizzi et al., 2006). The above data are also offsets (N1 km; e.g., Boccaletti et al., 1998). Their orientation varies in consistent with 3.2 Myr average plate motion data based on a re- the different rift segments, as illustrated below. analysis of spreading rates and transform-fault azimuths along the Southwest Indian Ridge (Horner-Johnson et al., 2007). This latter 5.2.1.1. Northern MER. In the Northern MER, the boundary faults are analysis predicts indeed a rotation pole consistent with that obtained oriented ~N40°E (Figs. 9, 10). The southeastern basin margin is by Stamps et al. (2008) and a roughly ESE–WNW-directed extension marked by the major boundary fault systems of Arboye and Sire, with rates of ~7–8 mm/yr in MER (Fig. 8). Other plate kinematics which form a staircase pattern rising to the ~2600-m elevation of the models of Nubia–Somalia motion converge to similar extension uplifted rift flanks (Fig. 10; e.g., Wolfenden et al., 2004). The Arboye directions and velocities at the latitude of the MER (e.g., N101°E at and Sire border fault systems are separated by a right-lateral offset of ~5 mm/yr, Jestin et al., 1994; N96°E at ~6 mm/yr, Chu and Gordon, about 35 km at latitude ~8°20′ N; similarly, the Sire fault system dies 1999). Geological data from the Northern MER indicate a similar out to the southwest at latitude ~8°N and is separated by a right- extension direction (N105°E) in the last 3 Myr (Wolfenden et al., lateral offset of about 15 km from the Asela fault escarpment. Overall, 2004). This consistency between geologic and geodetic data suggests the southwestern margin of this MER sector is characterised by a that the Nubia–Somalia relative motion may have remained steady right-stepping en-echelon pattern; no major transversal faults are over the past 3 Myr. The pre-3 Myr Nubia–Somalia kinematics is less present in these offsets to connect the main fault systems; rather, they constrained. Field geological and structural data suggested a poly- are separated by gentle flexure of the Somalian plateau which deeps to phase history of rifting in Ethiopia related to a change in Nubia– the NNE down to the rift depression (e.g., Di Paola,1972). Kazmin et al. Somalia motion sometime in the interval 6.6 to 3 Ma (Wolfenden (1980) suggested these offsets to be related to an inherited structural et al., 2004) or at the Pliocene–Quaternary boundary (Bonini et al., grain reactivated in the course of rifting. The northwestern margin is a 1997; Boccaletti et al., 1998, 1999a; Bonini et al., 2005). These works faulted flexure that passes to the northeast to the major Ankober were supported by plate kinematics models indicating a shift in the border fault system, which in turn characterises the structurally pole of rotation between Nubia and Somalia (i.e., a change in complex ‘corner’ between the NE-trending MER and the N–S trending kinematics in the EARS) in the Pliocene (Lemaux et al., 2002). Red Sea rift (Wolfenden et al., 2004). To the southwest, this rift margin However, more recent plate kinematics models (Royer et al., 2006) curves into a roughly N–S-trending structural high (Boru Toru suggest that the roughly ESE–WNW direction of extension in MER structural high; see below) marking the boundary with the Central may have remained steady over the past 11 Myr, a finding that is MER (Bonini et al., 2005). Similarly to what is observed for the Arboye consistent with recent analysis of fault pattern evolution from and Sire border fault systems, structures of the northern margin are analogue modelling (Corti, 2008; see Section 6.2). en-echelon arranged and normally characterised by local complex geometries, with sinuous and curvilinear fault segments (Mohr, 1987) 5.1.3. Estimates of Nubia–Somalia separation and pull-aparts which are normally interpreted to reflect an oblique Plate kinematics models are also variable in constraining the total component of motion (e.g., Boccaletti et al., 1992, 1998). Marginal amount of Nubia and Somalia separation. End members of estimates grabens mark both sides of the Northern MER (e.g., WoldeGabriel may be considered the ~23 km since 11 Ma (giving rise to rates of et al., 2000), although their expression is overprinted by younger ~2 mm/yr) predicted by Lemaux et al. (2002) and the ~130 km since structures. 11 Ma (rates: ~12 mm/yr) predicted by Royer et al. (2006). Most Seismic refraction studies in the central part of this rift sector models converge to values of 30–40 km of Nubia–Somalia relative indicate that the general form of the rift basin is asymmetric with motion at the latitude of the MER (e.g., Garfunkel and Beyth, 2006), tilting to the southeast toward the large-offset Arboye normal faults which roughly fits the estimates based on extrapolation of current and maximum thickness of the basin fill estimated to be of about 5 km stretching velocities to the last 10–11 Myr. This amount of total (Fig. 13; MacKenzie et al., 2005). Magnetotelluric analysis (Whaler extension matches recent geophysical and geological data from the and Hautot, 2006) supports an asymmetric structure of the basin and MER. Seismic sections (e.g., Maguire et al., 2006) image a crustal tilting towards the Arboye fault. Scarce fault-slip data on the rift thinning in the Northern MER of N10 km, giving rise to stretching margins and local structural features indicate a roughly E–W direction values (β) of about 1.7 and roughly corresponding to ~40 km of of extension (Boccaletti et al., 1992). Field data indicate that the 12 G. Corti / Earth-Science Reviews 96 (2009) 1–53

Fig. 9. Tectonic sketch map of the Main Ethiopian Rift (modified from Boccaletti et al., 1998) superimposed on a digital elevation model (SRTM data). Inset shows the en-echelon, right-stepping arrangement of the volcano-tectonic segments of the Wonji Fault Belt. AAE: Addis Ababa Embayment; BT: Boru Toru structural high; MK: Midre Kebd structural high; WFZ: Woito fault zone; YTVL: Yerer–Tullu Wellel volcano-tectonic lineament. Other abbreviations as in Fig. 2. boundary faults of the Northern MER started to develop at around alocalNW–SE trend and the interaction between these two 11 Ma (Wolfenden et al., 2004); geological data and historic and local intersecting trends result in the typical S- or Z-shaped pattern of the seismicity patterns indicate these fault systems are largely inactive Langano (or Haroresa) Rhomboidal Fault System (Fig. 15; e.g., Mohr, (Wolfenden et al., 2004; Casey et al., 2006; Keir et al., 2006a; see 1987; Boccaletti et al., 1998; Le Turdu et al., 1999). Moreover, the Section 5.5). western rift margin is characterised by the presence of roughly N–S- trending structural highs (such as the Boru Toru and Midre Kebd 5.2.1.2. Central MER. In the Central MER, the rift valley orients structural highs; Fig. 9), where lowermost syn-rift volcanic units crop between N25° and N45° and is characterised by major rift escarp- out, which give rise to major embayments in the rift structure (Abebe ments on both western and eastern margins; boundary faults show an et al., 2005). This is evident in the Debre Zeyt area, where the NE–SW- average trend around N30°E (Figs. 9, 11). The western margin is well trending rift margin and the N–S Boru Toru structural high give rise to expressed by the N25°E–N35°E-trending and ESE-dipping Guraghe the major Addis Ababa embayment (Fig. 9). The presence of these N–S and Fonko faults, whereas the eastern margin is well represented by structural highs and embayments defines lacustrine sub-basins that the N30°E-trending and WNW-dipping Asela–Langano fault system are aligned along a N–S trend (Abebe et al., 2005). Fault-slip data on (Fig. 9). Both systems are characterised by high-angle (N60°) normal both rift margins in the Central MER indicate a stress field faults, with large cumulative vertical throw. Overall, these fault characterised by an extension direction oriented roughly ESE–WNW, systems give rise to a roughly symmetric rift valley (Figs. 11, 13), as with local variations between E–W and NW–SE (Fig. 14; Boccaletti supported by the 3D modelling of gravity data that indicate syn-rift et al., 1992, 1998; Abebe, 1993; Korme et al., 1997; Acocella and Korme, sediment thickening towards the central part of the graben where 2002; Bonini et al., 2005; Pizzi et al., 2006). This gives rise to a sinistral basin infill thickness can exceed 6 km (Mahatsente et al., 1999). component oblique-slip on the NE–SW-trending fault planes, as However, knowledge of the subsurface rift architecture in the Central supported by local structural features as pull-aparts, en-echelon MER remains poor. The border fault systems are normally segmented tensional fissures and the en-echelon right-stepping arrangements of and articulated, and characterised by the presence of minor the boundary fault segments observed in the field and on satellite transversal structures and local complex geometries. As an example, imagery (Boccaletti et al., 1992, 1998; Mazzarini et al., 1999; Abebe near Lake Langano the NE–SW-trending border faults curve to acquire et al., 2005; Bonini et al., 2005). According to Bonini et al. (2005) the G. Corti / Earth-Science Reviews 96 (2009) 1–53 13 rift margins in the central portion of the MER formed in the late boundary faults (e.g., Asela–Langano) still accommodate some Miocene–Pliocene (post ~6 Ma), although WoldeGabriel et al. (1990) extensional deformation (Pizzi et al., 2006), although most of active suggested that boundary faults had already formed at around 8 Ma. strain seems to be accommodated by the internal Wonji faults (e.g., Morphostructural and geological analysis together with consideration Boccaletti et al.,1998; Bonini et al., 2005; Abebe et al., 2005; Pizzi et al., on the local seismicity suggests that at least some segments of the 2006; Keir et al., 2006a).

5.2.1.3. Southern MER. The Southern MER is characterised by a rotation of the rift valley from N20-35° to N5-20°; accordingly, the orientation of the boundary faults is N0°E to ~N20°E in this MER sector (Figs. 9, 12). Rotation of the rift valley occurs south of latitude 7°20′ in correspondence to a major E–W-trending transverse lineament (Goba–Bonga lineament; see Section 5.2.3; Abbate and Sagri, 1980). The Southern MER is characterised by the major fault- escarpments of Chencha (western margin) and Agere Selam (eastern margin). The Chencha escarpment is marked by a curvilinear boundary fault having an orientation between N–S and N40°E; its northern termination corresponds to a monocline that extends northwards to the Fonko fault system. The Agere Selam escarpment is more linear in shape and is characterised by NNE–SSW trend. South of latitude ~6°20′N the Southern MER is characterised by a progressive narrowing accompanied by an increase in physiographical complexity; its end corresponds to a division of the rift valley into two near-parallel grabens, namely the Chamo (or Ganjuli) basin to the west and the Galana basin to the east. The two grabens are separated by the Amaro Horst, a narrow block of Precambrian basement that widens and declines southwards over a length of ~90 km (Levitte et al., 1974; Ebinger et al., 1993). In correspondence to the Amaro Horst, the rift structures attain an overall N–S trend; however, whereas the faults of the Galana Basin are linear and characterised by an almost pure N–S trend, the structures bounding the Chamo graben are more segmented and characterised by a zig-zag pattern, with different segments showing sharp changes in orientation (from NW–SE to NE– SW). The general rift topography is rather asymmetric: at ~7°N latitude the western margin lacks a major border fault system whereas the eastern margin corresponds to the Agere Selam escarpment (e.g., Woldegabriel et al., 1990); southwards, the western Chencha escarpment has no correspondence on the eastern margin. Geological cross- sections reconstructed using bedding and fault orientations measured in the ﬁeld (Ebinger et al., 1993) suggest a rather symmetric structure in the Chamo basin, with thickness of syn-rift deposits of ~2 km, and an asymmetric architecture in the Galana basin, with the master fault located on the western side and maximum thickness of the rift inﬁll that may exceed 4 km (Fig. 13). Gravity modelling (Mahatsente et al., 1999) does not evidence a clear asymmetry in the sediment thickness below this rift segment; however, since no detailed geophysical studies are available, the subsurface rift architecture in the Southern MER remains poorly constrained. Mesoscopic fault measurements as well as the local architecture of extensional structures suggest a sinistral component of motion along the NNE–SSW-trending border faults, compatible with a stretching direction variable between E–W and NW–SE (Fig. 14; Boccaletti et al., 1998; Bonini et al., 2005). According to WoldeGabriel et al. (1990) the rift margin at Agere Selam formed in the late Miocene, at around 10 Ma. South of Lake Abaya,

Fig. 10. Structural setting of the Northern MER illustrated as a digital elevation model (SRTM data, upper panel), schematic structural pattern (central panel) and topographic profiles (lower panels). In the central panel, the distribution of both boundary and Wonji faults is illustrated as plots of the fault azimuths, weighted for the fault length; also illustrated is the distribution of the trend volcanic centre alignments. In these plots, the weighting factor for each fault or volcanic alignment is the ratio between the length and the minimum length of the whole data set, such that long faults (volcanic alignments) have higher ratio (weight) than short ones. The frequency of the azimuth of a fault or volcanic alignment directly relates to this ratio, the longer the fault (volcanic chain) the higher its frequency. B: Boseti; DZ: Debre Zeyt; F: Fantale; G: Gedemsa; K: Kone; Ko: Lake Koka. In the topographic profiles the colors indicate the physiographic location within the rift system (green: plateau, red: rift escarpment, black: rift floor); m. s.: magmatic segment. Other symbols as in Fig. 9. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) 14 G. Corti / Earth-Science Reviews 96 (2009) 1–53

extensional deformation initiated at ~20–21 Ma (see Bonini et al., 2005 and references therein). Continental rifting appears to be well- established at 17–14 Ma, as indicated by interbedding of basaltic ﬂows with fossiliferous ﬂuvio-lacustrine sediments (WoldeGabriel et al., 1991, 2000) and by alignment of Mid-Miocene (~14 Ma) volcanic centres along the trace of boundary faults (Ebinger et al., 1993). Extensional structures remained active until ~11 Ma and were later reactivated in the Late Pliocene–Pleistocene after a long period of quiescence (Bonini et al., 2005). Morphostructural and geological analysis on the area suggests that this border fault system may be still active (Boccaletti et al., 1998).

5.2.1.4. Southern rift termination and Broadly Rifted Zone. South of the Amaro Horst, the rift loses its typical physiography and is believed to propagate into the Ririba rift (Fig. 12; WoldeGabriel and Aronson, 1987; Ebinger et al., 2000; Bonini et al., 2005; Vetel and Le Gall, 2006), which is part of the ~300 km-wide system of basins and ranges (Broadly Rifted Zone; Baker et al., 1972; Moore and Davidson, 1978; Davidson and Rex, 1980; WoldeGabriel and Aronson, 1987; Ebinger et al., 2000) that characterises the overlapping area between the Ethiopian and Kenya rifts. West and south of the Chamo basin, the area is dominated by N–S to N35° normal faults giving rise to the Gofa Basin and Range (or Gofa Province) and the Chew Bahir Rifts (Fig. 12). The Gofa Basin and Range consists of NE–SW-trending fault blocks of Tertiary lavas tilted to the northwest (Moore and Davidson, 1978). Displacement along the bounding high-angle faults decreases northward along strike and the Gofa Basin and Range loses its topographic identity near 6°30′N latitude (Omo valley); more to the north, this NE–SW trend is completely interrupted by the E–W-trending Gojeb graben (Moore and Davidson, 1978; Boccaletti et al., 1998). South of the Gofa Basin and Range, extensional deformation is accommodated by the Chew Bahir basin, which is interpreted as the northernmost propagation of the Kenya Rift (e.g., WoldeGabriel et al., 2000 and reference therein). This basin has an overall N–S trend but the boundary faults are highly segmented with sharp changes in orientation giving rise to a zig-zag pattern, which is also evident in the Gofa basin and Range and is suggestive of a strong inﬂuence of the pre-existing fabric in the development of the fault systems (see Fig. 24; e.g., Moore and Davidson, 1978; Vetel et al., 2005; Vetel and Le Gall, 2006). Though the Chew Bahir deep geometry is unknown, interpretation of gravity data together with geological considerations indicates depth values of 1 to 2 km (Vetel and Le Gall, 2006 and references therein). Fault-slip data in this area are limited to the NW– SE-trending Woito Fault Zone (Fig. 12), where measurement of a dextral component of motion indicates a roughly ESE –WNW extension direction (Fig. 14; Abebe, 1993; Boccaletti et al., 1998). Different studies (Morley et al., 1992; Ebinger et al., 2000; Vetel and Le Gall, 2006) proposed that that rifting migrated eastward across the BRZ, initiating along inherited Mesozoic structures of the Anza Rift in the west (Lotikipi and Lokichar basins) as early as ~35–45 Ma and gradually jumping to the Main Ethiopian Rift direction. Based on detailed determination of fault-related uplift along the Chow Bahir rift escarpments, Pik et al. (2008) constrained initial extensional deformation in this area at ~20 Ma, with continuous rifting and associated relief changes since that time. Extensional structures were well developed by Mid-Miocene time in the area (e.g., Ebinger et al., 2000; Vetel and Le Gall, 2006 and references therein). Most of the faults have typical morphostructural features indicating ongoing tectonic activity (Boccaletti et al., 1998); of interest in this respect is the Woito Fault Zone (Fig. 11), which is marked by strong seismicity (Gouin, 1979; Asfaw, 1990, 1992) and possible surface breaks

Fig. 11. Structural setting of the Central MER illustrated as in Fig. 9. A: Lake Abiyata; Aw: Lake Awasa; Bu: Butajira volcanic chain; L: Lake Langano; S. Lake Shala; Z: Ziway. Other symbols as in previous figs. G. Corti / Earth-Science Reviews 96 (2009) 1–53 15 produced during the earthquake of December 29, 1987 (Asfaw, 1990). 5.2.2.1. Northern MER. In the Northern MER, there are four major According to Moore and Davidson (1978) this fault has controlled the WFB segments (Gedemsa, Boseti, Kone and Fantale–Dofen; Fig. 16); recent evolution of the Woito River. the en-echelon segmentation continues northeastward into the southern Afar depression, where these quaternary tectono-magmatic 5.2.2. Wonji Fault Belt faulting segments are superposed on the older Red Sea and Gulf of Aden rift The rift floor of the MER is affected by widespread deformation structures (Hayward and Ebinger 1996; Wolfenden et al., 2004, 2005) related to faulting along the Wonji Fault Belt (WFB; Figs. 9–12; Mohr, and fault systems display complex geometrical characteristics (dis- 1962, 1967; Gibson, 1969; Gibson and Tazieff, 1970; Mohr and Wood, tribution of deformation, segmentation, linkage, length/displacement 1976; Kazmin et al., 1980; Mohr, 1983, 1987; Boccaletti et al., 1992; relations, etc.; Soliva and Schultz, 2008). Chorowicz et al., 1994; Boccaletti et al., 1998; Ebinger and Casey, 2001; The Wonji faults in the Northern MER are oriented ~N20°, forming Acocella et al., 2003; Williams et al., 2004; Wolfenden et al., 2004; an angle of ~20° with the roughly N40°-trending boundary faults Abebe et al., 2005; Bonini et al., 2005; Casey et al., 2006; Pizzi et al., (Figs. 10, 16). The Gedemsa, Boseti and Kone WFB segments are axial 2006; Kurz et al., 2007). This is an arrangement of overlapping, right- ridges, whereas the Fantale–Dofen magmatic segment is marked by a stepping en-echelon fault segments oblique to the main direction of graben that lies 550 m below the Boseti and Kone segments (Figs. 10, the rift margins, where adjustment of these fault zones with the 17; Ebinger and Casey, 2001; Casey et al., 2006). The dimensions of border faults gives rise to a typical S-shaped curvature resulting in an individual segments range between 40 and 70 km in length and overall sigmoidal geometry (Fig. 16; see below). The faults of the WFB between 10 and 15 km in width; they are separated by areas devoid of are normally short, closely spaced and display relatively small throws magmatism and brittle deformation (Casey et al., 2006; Kurz et al., (b100 m; e.g., Boccaletti et al., 1998); they are characterised by steep 2007), where the distances between segments in an E–W direction (sub-vertical) scarps and are commonly en-echelon and linear or vary from 2 km (Boseti segment–Kone segment) to 18 km (Gedemsa curved in plain view over distances of up to a few tens of kilometres, segment–Boseti segment). As stated above, each segment is com- thus delimiting several fault-bounded blocks. Associated with the posed of small-displacement faults and extensional fractures. faults are open fissures with or without vertical displacement (e.g., Structural analysis reveals no evidence for transcurrent faults Asfaw, 1998; Williams et al., 2004), splay patterns and complex linking right-stepping WFB segments; rather, the tips of segments rhomb-shaped structures (see Mohr, 1987; Boccaletti et al., 1998). overlap, thereby accommodating strain transfer (Casey et al., 2006). These features thus testify a fundamental change in the deformation According to Kurz et al. (2007), the WFB tectono-magmatic style from the rift margins, where extension is accommodated by few segments are symmetric around central acidic strato-volcanoes (see widely-spaced faults with large vertical displacements, to the rift floor, also Ebinger and Casey, 2001; Abebe et al., 2007), where the where deformation occurs through dense fault swarms with small deformation pattern is characterised by small faults and a low fault vertical offset (Fig.17; e.g., Boccaletti et al.,1998; Giaquinta et al.,1999; density (although fault number may locally increase around collapse Ebinger and Casey, 2001; Casey et al., 2006). calderas possibly due to magma withdrawal; e.g., Casey et al., 2006). The Most of the recent (Quaternary) volcanic activity in the MER is number of faults increases longitudinally away from the volcanic closely associated with the Wonji faults, as suggested by alignments of centres, such that maximal fault densities and the longest faults are caldera structures, cinder cones, volcanic fissures (see Section 5.3). observed at tip domains. This is also accompanied by a decrease in the Geophysical data in the Northern MER have shown the presence of volume of basaltic rocks towards the tip domains, where recent basalt magma throughout the lithosphere below the different fault segments fl ows are absent. These characteristics suggest that individual segments (see Section 5.4), so that the WFB segments represent tectono- are characterised by different domains where brittle or magmatic- magmatic (or magmatic) segments within the central rift valley (e.g., induced deformation dominate (Fig. 18; Kurz et al., 2007). The centre of Ebinger and Casey, 2001). each segment is characterised by significant magmatic deformation The Wonji Fault Belt developed in the last 2 my (Mohr, 1962, 1967; accommodated by dyking, whereas the tip domains display brittle Gibson, 1969; Meyer et al., 1975; Chorowicz et al., 1994; Boccaletti deformation (faulting). Intermediate domains are characterised by et al., 1998; Ebinger and Casey, 2001; Bonini et al., 2005). The recent interaction between magma injections and faults, with shallow dikes fed alluvial cover and volcanic rocks are affected by this fault system that, by the central magmatic centres triggering the generation of open therefore, accommodates Holocene deformation (e.g., Boccaletti et al., fissures, aligned basaltic cones and faults at the surface (Kurz et al., 1998). Geodetic data support that 80% of the present-day strain is 2007). In these domains, some fault segments display low displace- accommodated across these fault segments (Billham et al., 1999; see ment–length ratios, i.e. long faults associated with small displacement, Section 5.5). indicative of fault growth dominated by fault linkage; these low Although a roughly NW–SE extension has been deduced from the displacement–length ratios are considerably lower than in other analysis of open fractures within different Wonji segments (Acocella and extensional areas characterised by brittle deformation without magma- Korme, 2002), the right-stepping en-echelon fault arrays, the complex tism and are thus interpreted as caused by magma injection (Kurz et al., pull-apart structures and the results of structural analysis suggest that 2007). These findings are consistent with analysis by Casey et al. (2006) this fault system accommodates a left-lateral, oblique shearing along the suggesting that Wonji faults are primarily driven by magma intrusion rift axis resulting compatible with a roughly E–W oriented extension into the mid- to upper crust, which triggers upward propagation of ( Fig. 14;e.g.,Bonini et al., 1997; Boccaletti et al., 1998; Pizzi et al., 2006; faulting and dyke intrusion into the brittle upper crust. Dyke-induced Kurz et al., 2007). Fault length and segmentation of WFB faults show growth of normal faults and fractures has been observed during a major comparatively minor values than the boundary faults (Kurz et al., 2007); magmatic rifting episode at a nascent slow-spreading ridge in the Afar this accords with an oblique rifting-related origin for the WFB faults as depression (the 2005-present Dabbahu rifting episode; Fig. 19; see also suggested by experimental results reported in Clifton and Schlische below Section 6.3; Rowland et al., 2007; Ebinger et al., 2008). (2001).TheroughlyE–W direction is consistent with strain analysis on At tip domains, the WFB propagates into and interacts with the volcanic edifices within the fault zones (extension direction: N110°E± boundary fault systems (Casey et al., 2006; Kurz et al., 2007); intra-rift 3°; Casey et al., 2006) and the analysis of the general architecture of faults grow into the border faults and inherit their orientation, giving Wonji segments (extension direction variable between ~N90E° and rise to a sigmoidal bending of the WFB segments (Fig. 16). If, however, ~N105E°; e.g., Bonini et al., 1997; Corti et al., 2001; Wolfenden et al., interaction occurs at high angles (N30°), catchment of the Wonji faults 2004; Kurz et al., 2007; Corti, 2008). All these data are also consistent by the border faults does not occur and the younger intra-rift faults cut with geodetic, plate kinematics, seismic and current stress-field data the older boundary faults, giving rise to rhomboidal structures (e.g., (see Section 5.1.1). Mohr, 1983; Kurz et al., 2007). 16 G. Corti / Earth-Science Reviews 96 (2009) 1–53

Awasa–Shala fault zones. These Wonji segments are marked by grabens that branch-off from the N30°E-trending eastern rift escarpment (Figs. 9, 11). Close to the western margin, the Quaternary volcano-tectonic activity is expressed by the NNE–SSW-trending volcanic chains of Debre Zeyt (Bishoftu) and Butajira (Fig. 11; e.g. Mohr 1968; Abebe et al., 2005; Mazzarini et al., 1999, 2004; Abebe et al., 2007). These chains seem to be characterised by a lower degree of seismicity with respect to the other Wonji belt segments (Keir et al., 2006a; see Section 5.5) and show comparatively less developed structural or morphological evidence of active strain, expressed by sets of right-stepping en-echelon faults (Abebe et al., 2005). Overall, conversely to the Northern MER where the Wonji Fault Belt- related deformation affects the centre of the depression, the Quaternary faulting in the Central MER seems to be mostly localised at the rift margins. As in the Northern MER, the Wonji faults form a close network of faults characterised by very fresh, steep fault scarps evidence of recent activity and high Late Quaternary slip rates (up to 2mm yr−1 for some of the densely distributed faults between the Asela fault and Lake Ziway; Pizzi et al., 2006). Close to the eastern margin, the faults delimit closely spaced sets of horsts, grabens and half grabens; fault planes are characterised by high-angle geometry with a dip generally N65 ° (Fig. 17c,d). Local fault characteristics, such as right-stepping en- echelon arrays of both faults and open fissures and N–S faults constituting systems of splays that depart from a NE–SW boundary faults, testify a nearly E–W oriented Quaternary direction of extension, giving rise to left-lateral shear along the rift strike. In the southern part of this MER's sector, the Quaternary Awasa caldera (Di Paola, 1972; Mohr et al., 1980; WoldeGabriel et al., 1990) has been severely broken by N–S-trending normal faults, forming a down-faulted central sector in which Lake Awasa is located. The displacement between the eastern and western parts of the caldera, as well as the fault pattern, again suggests a roughly Quaternary E–W direction of extension (e.g., Boccaletti et al., 1998). In places (e.g., where the Gedemsa segment branch-off from the Asela escarpment), most of the active strain is accommodated aseismically by magma-induced faulting, similarly to what is described in the Northern MER (Pizzi et al., 2006; see Section 5.2.2.1). However, consistent with a southward decrease in crustal extension and magmatic modification along the rift, the Wonji faults in the Central MER are expected to be comparatively less affected by magmatic processes than in the Northern MER (see Section 5.4). Where not influenced by magmatism, normal faults display propor- tional relationships between length and throw (Acocella et al., 2003); they seem to form from the progressive grow of extension fractures that reach a critical depth where the tensile stresses can no longer be sustained and fractures turn into faults (Fig. 20; Acocella et al., 2003). Once the faults have formed, they grow further, propagating both downward and along strike. Notably, the WFB in the Central MER cuts in places lacustrine sediments that record the Late Quaternary–Holocene evolution of a complex lake system, which is now represented by the Ziway, Langano, Abijata and Shalla Lakes. These lakes are remnants of a much larger lake basin that reached its maximum size during the Late Pleistocene and Early Holocene (Grove and Goudie, 1971; Laury and Albritton, 1975; Grove et al.,1975; Gasse and Street,1978; Gillespie et al.,1983; Le Fig. 12. Structural setting of the Southern MER illustrated as in Fig. 9. Ab: Lake Abaya; Turdu et al., 1999; Benvenuti et al., 2002); the evolution of this lake BoG: Bridge of God; Ch: Lake Chamo; Du: Duguna. Other symbols as in previous figs. fl fl Note the high segmentation and the zig-zag pattern (due to sharp changes in system has been strongly in uenced by intense climate uctuations orientation) of the boundary faults in the Chamo and Chow Bahir basins and in the Gofa during the last 100,000 years, although a structural control on the basin and Range. This feature is suggestive of a strong influence of the pre-existing rearrangement of the fluvial drainage networks and hydrological fabric in the development of the fault systems. thresholds has been in some cases identified (e.g., Benvenuti et al., 2002; Sagri et al., 2008). 5.2.2.2. Central MER. In the Central MER, the Wonji faults are oriented ~N12°E and oblique of ~18° to the roughly N30°-trending 5.2.2.3. Southern MER. In the Southern MER, the Wonji faults are boundary faults (Figs. 11, 17). The main Wonji fault segments are oriented around N–S, i.e. roughly parallel to the boundary fault represented by the southern termination of the Gedemsa WFB systems (Fig. 12). In this rift's sector, the Wonji faults are compara- segment of the Northern MER, and by the Langano–Ziway and tively less developed than in the Central and Northern MER; collapse G. Corti / Earth-Science Reviews 96 (2009) 1–53 17

Fig. 13. Geological cross-sections across the different segments of the MER. Note the asymmetric structure of the Northern MER, the roughly symmetric structure of the Central MER and the complex architecture of the Southern MER with the roughly symmetric Chamo basin and the asymmetric Galana basin separated by the Amaro Horst. Modiﬁed after: A) Wolfenden et al. (2004);B)Abebe et al. (2005);C)Ebinger et al. (1993).

caldera and faults occur across a much broader zone within the rift The Quaternary volcano-tectonic activity in the rift floor can be traced and border faults are seismically active (Asfaw, 1992; Keir et al., southwards along the eastern border of Lake Abaya to the land bridge 2006a). These observations are consistent with the general south- (Tosa Sucha) separating Lake Abaya from Lake Chamo. In this area, wards decrease in thinning and rift-related volcano-tectonic deforma- roughly NNE–SSW-trending faults affect a Pleistocene volcanic complex tion (see Sections 5.3 and 6). The main volcano-tectonic segments and display strong morphotectonic signature of ongoing tectonic activity correspond to the southern termination of the Awasa–Shala fault (Fig. 22), which is occurring at rates around 1 mm yr−1 (Boccaletti et al., zone, that branch-off from the Agere Selam escarpment, and the 1998). This slip-rate is lower than the ~6 mm yr−1 predicted by geodetic Abaya–Duguna segment (Fig. 12). Quaternary faults in this latter analysis (see Section 5.1.1), suggesting that the border faults of the rift segment, appear to reactivate the northern segment of the Chencha may take up a considerable part of the extensional deformation in the escarpment (Boccaletti et al., 1998). The displacement has been taken Southern MER (see Boccaletti et al.,1998). No recent volcanism and rift- up by an array of sigmoidal, right-stepping en-echelon normal or floor deformation characterise the Galana basin, suggesting that the oblique faults that are particularly evident north of Lake Abaya WFB has not propagated into this rift branch. (Fig. 21), where they control the development of small Quaternary basins and spatter cones. The right-stepping arrangement of faults is 5.2.3. Transverse structures strongly suggestive of a sinistral shear component of motion along the The Main Ethiopian Rift is characterised by the presence of major rift trend, which is also testified by the development of small pull- structures transverse to the trend of the main extensional structures apart basins filled by Quaternary lacustrine deposits along these faults and that may extend for hundreds of kilometres from the rift margins (Boccaletti et al., 1998). Analogue modelling results (Corti, 2008) on both the Somalian and Ethiopian plateaus (Fig. 9). The transverse suggest that this fault pattern and kinematics are again compatible faults can be grouped in two main trends, oriented roughly E–Wto with a sub-E–W extension direction (Fig. 21). ENE–WSW and NW–SE. 18 G. Corti / Earth-Science Reviews 96 (2009) 1–53

Fig. 14. Representative fault kinematics data for the different MER sectors. Also reported is the stretching direction deduced from the elongation of caldera complexes (from Bonini et al., 2005; Casey et al., 2006), assuming that volcanoes with elliptical forms have a long axis that is roughly parallel to the extension direction. In the central and Northern MER, caldera complexes have a long axis oriented between E–W and ESE–WNW.

The roughly E–W structures are well represented by the Yerer–Tullu the transversal YTVL probably controlled the development of the Boru Wellel Volcano-tectonic Lineament (YTVL; e.g., Abebe et al., 1998), Toru structural high, the transfer zone separating the Northern MER which extends from the western rift margin to Tullu Wellel near the from the Central MER (Bonini et al., 2005). Ethiopia–Sudan border (Figs. 9, 23). This fracture system is about Another major E–W-trending transverse lineament is the Goba– 700 km long and about 80 km wide; its northern limit partly coincides Bonga lineament (e.g., Abbate and Sagri, 1980), which extends across with the Ambo Fault (Mohr, 1967; Meyer et al., 1975; Moore and the central part of the MER and is expressed in the Ethiopian Plateau Davidson,1978; Abebe et al.,1998; Korme et al., 2004), a well-developed by the Gojeb graben (Figs. 9, 23; Moore and Davidson, 1978; Boccaletti fault zone with vertical scarps up to 150 m in height. The YTVL consists et al., 1998). According to Bonini et al. (2005), the development of this of a tectono-magmatic system characterised by structural alignments of E–W depression may be related to an early deformation phase, normal faults and major volcanoes (Fig. 23;e.g.,Abebe et al., 1998; characterised by NNE–SSW-trending extension, resulting from the Tommasini et al., 2005). During rift evolution it has been characterised southernmost propagation of the extensional tectonics that affected by a complex volcano-tectonic evolution, with activation of several the boundary of the Afro-Arabian plates during Oligocene times, as structures with different trends and an overall eastward migration of supported by plate reconstructions (Collet et al., 2000) and physical volcanic activity in time (Abebe et al., 1998). Geophysical data (see models of Africa–Arabia separation (Bellahsen et al., 2003). The Goba– Section 5.4) indicate that the YTVL is associated to a low-velocity mantle Bonga lineament controlled the northward propagation of Kenya Rift- zone, suggestive of anomalously hot upper mantle and local areas of related deformation at around 20 Ma and it also controlled the melting or melt ponding (e.g., Bastow et al., 2005, 2008); this structure southward propagation of structures from Central MER to the also controls a sharp variation in crustal thickness of the Ethiopian Southern MER in the Mio-Pliocene (Bonini et al., 2005). Plateau (Keranen and Klemperer, 2008) and the distribution of the The NW–SE structures developed in correspondence to pre- Oligo-Miocene plateau volcanics (Abbate and Sagri,1980). All these data existing Precambrian major crustal weakness zones that parallelise suggest that the YTVL probably represents a pre-existing weakness the trend of the Red Sea (e.g., Korme et al., 2004). Some of these zone, parallel to the trend of the Gulf of Aden, that may correspond to an structures were reactivated during the Mesozoic to form NW–SE- ancient E–Wsuture(Stern et al.,1990), whose interpretation is however trending structural sedimentation basins, such as the Ogaden rift (e.g., controversial (e.g., Church, 1991). During the development of the MER, Korme et al., 2004 and references therein) and the Blue Nile Basin G. Corti / Earth-Science Reviews 96 (2009) 1–53 19

Chorowicz et al. (1998) suggested that the location of the Lake Tana occurs at the junction of three different grabens (the NNW–SSE to N–S-trending Gondar graben, the NE–SW to NNE–SSW-trending Dengel Ber graben and the E–W to ESE–WNW-trending Debre Gabor graben; Fig. 25), which were active during the build-up of the ﬂood- basalt pile and were subsequently reactivated during the Late Miocene– Quaternary. The complex history of local uplift, deformation and volcanism of the Lake Tana region has been associated to the inﬂuence of the Afar plume (e.g. Chorowicz et al., 1998); the area has been interpreted to possibly represent the impact area of the plume head from which the low-density material spreads (see Mege and Korme, 2004), although the current locus of plume material is interpreted to be in the Lake Abhe area (e.g., Schilling et al., 1992).

5.3. Summary of the evolution of volcanic activity

The tectonic development of the MER is matched by intense volcanic activity starting from the late Eocene (e.g., Mohr, 1970; WoldeGabriel et al., 1990; Ebinger et al., 1993; Boccaletti et al., 1995; Chernet et al., 1998; Wolfenden et al., 2004; Abebe et al., 2005). Starting from the early episodes of flood-basalt volcanism (see Fig. 15. a) Fault pattern in the Langano (or Haroresa) Rhomboidal Fault System. Note the Section 4.1), magmatic activity in the MER seems to have been interaction between NE–SW and NW–SE-trending structures giving rise to the typical S- episodic rather than continuous (Girdler, 1983; WoldeGabriel et al., or Z-shaped fault pattern. b) Examples of models that can explain the development of 1990) and showed different characters in the different rift segments, fl – the rhomboidal fault pattern: upper panel: in uence of a pre-existing NW SE-trending as outlined below. weakness zone causing a left-lateral offset of the rift margins and the NE–SW boundary faults (e.g., Mohr and Potter, 1976; Mohr, 1987; Korme et al., 2004); lower panel: pull- Note that, as stated in Section 4.1, our current knowledge of the apart structure related to the left-lateral component of motion along the border fault timing of this activity is limited by different problems (e.g., relatively system (Boccaletti et al., 1998). Note that the presence of a transverse NW–SE structure scarce available datings compared with the huge volume of the controlling the 5 km-left offset of the plateau rim in this area is supported by both field magmatism) and potentially modified by additional datings. and subsurface gravity data (e.g., Mohr, 1987; Korme et al., 2004; see Section 5.2.3.). 5.3.1. Northern MER (Fig. 5; e.g., Gani et al., 2009). These ancient rifts extend under the In the Northern MER, early eruption of the flood basalts was flood-basalt cover cutting the MER orthogonally. In places, the NW–SE followed by Mid-Miocene eruptions from shield volcanoes along the faults related to this ancient structure have influenced the develop- developing rift shoulders (Termaber–Megezez Formation of Chernet ment of the Tertiary rifting faults, as illustrated above for the Langano et al., 1998). Trachytic flows from that base of the Megezez volcano Rhomboidal Fault System whose architecture is probably controlled by have been dated at around 10.5 Ma (Wolfenden et al., 2004). In the one of these transversal structures (see Fig.15). Control of pre-existing Adama basin, these flows are locally onlapped by a contemporaneous transverse structures on rift structures close to the Lake Langano area sequence of basalts and ignimbrites (Kessem Formation of Wolfenden is also supported by gravity data (Korme et al., 2004) evidencing the et al., 2004), whose base is dated at ~10.5 Ma. According to Wolfenden presence of NW–SE graben below the rift depression filled with low- et al. (2004) this sequence shows growth and syn-sedimentary density sediments. Northwards, NW-trending faults delimit the faulting, thus placing the initiation of rifting in the area at ~10.5 Ma. northern and southern sides of the Boru Toru structural high (e.g., The top of syn-rift volcanism in the Adama basin is marked by Abebe et al., 2005) and, west of Welenchiti, they also border a major widespread ignimbritic episode dated at ~6.6 Ma (Wolfenden et al., rectangular depression filled by lacustrine sediments (e.g., Mohr, 2004); coeval widespread ignimbrites, with intercalated minor silicic 1987; Korme et al., 2004; Abebe et al., 2005). and mafic lavas, occur throughout the Northern MER (Nazret Group; see for instance Chernet et al., 1998 and references therein), but the details of the their absolute geochronologic placement and their time 5.2.4. Deformation outside the rift valley extent remain ambiguous. Indeed, whereas Wolfenden et al. (2004), Although most of the extensional deformation is accommodated recognise a volcanic hiatus in the Adama basin between 6.6 and within the rift valley, recent deformation and volcanism characterise 3.5 Ma, Chernet et al. (1998) report ages as young as 5.8 Ma for the limited portions of the surrounding plateaus. For instance, this is Nazret Group, whereas younger ages are reported close to the documented in the Lake Tana region by Late Miocene–Quaternary boundary between the Northern and Central MER (see Abebe et al., faulting, subsidence and predominantly basaltic volcanism (Fig. 25; 2005 and references therein; see Section 5.3.2). Contemporaneous e.g., Chorowicz et al., 1998). Recent geological analysis suggests a basaltic activity is subordinated, although its importance increases complex deformation pattern in the area, with superposition of northwards towards Afar where Late Miocene–Pliocene deformation differently-oriented systems of extensional structures (Gani et al., is associated to a phase of voluminous flood-basalt volcanism (Afar 2009). According to these studies, Neogene deformation localised in Stratoid Series; see below; e.g., Barberi and Santacroce, 1980; Chernet correspondence to an inherited NW–SE graben buried below the et al., 1998). Overall, syn-rift volcanics display a typical bimodal Oligocene flood basalts pile, as imaged by recent magnetelluric composition (Fig. 26). analysis (Hautot et al., 2006) and suggested by geological studies The sequence is followed by upper Pliocene basalts (known by (Gani et al., 2009); this pre-existing structure can be correlated to different names in different localities, e. g., Wolenchiti, Bishoftu, Bofa, the Mesozoic Blue Nile graben (Fig. 5; see also Gani et al., 2009). The Tullu Rie basalts, see Chernet et al., 1998; Abebe et al., 2005 and recent evolution of the area has been further complicated by the su- references therein), with ages ranging between 3.5 and 1.6 Ma (e.g., perposition of different directions of tensile stresses during the Qua- Kazmin et al., 1980; Chernet et al., 1998; Abebe et al., 2005). ternary in relation to tectonic events within the MER and Afar (Gani Contemporaneous are pyroclastic deposits made of volcanic ash et al., 2009). flow and fall deposits composed by mainly comenditic rhyolites 20 G. Corti / Earth-Science Reviews 96 (2009) 1–53

Fig. 16. a) Boundary faults of the Chow Bahir basin, displaying a typical angular pattern suggestive of a strong influence of the pre-existing fabric in the fault development. Inset shows the location of panel a (AH: Amaro Horst; Aw: Awasa caldera; Ch: Chow Bahir basin; LA: Lake Abaya). b) Details of the boundary fault system; c) structural sketch. d) Trend distribution of pre-existing basement fabrics and boundary faults, illustrated as plots of weighted fault azimuths (see Fig. 10 for details of calculations). Note the striking correspondence between the trend of basement fabrics and boundary faults, supporting that the development of the latter is strongly controlled by pre-existing weaknesses. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) possibly fed by large caldera structures (Chefe Donsa pyroclastics of flows, scoria cones and phreatomagmatic deposits mostly aligned in a Abebe et al., 2005), with ages ranging between 2.5 and 1.7 Ma (e.g., NNE–SSW-direction along the faults of the Wonji Fault Belt (e.g., Abebe Morton et al., 1979; Mazzarini et al., 1999; Boccaletti et al., 1999a,b; et al., 2005; Fig. 27). Specific relations between the Wonji Fault Belt Abebe et al., 2005). Wolfenden et al. (2004) described the occurrence segments and magmatism have been proposed (see Section 5.2.2), in the Adama basin of a sequence of predominantly felsic volcanism with each segment being characterised by one or more silicic centre, (Balchi Formation) between 3.5 and 2.5 Ma. and associated basaltic eruptions (e.g., Chernet et al.,1998; Ebinger and The subsequent, Quaternary bimodal volcanic activity (lava, Casey, 2001; Acocella et al., 2002; Abebe et al., 2007; Kurz et al., 2007; pyroclastics and volcanoclastic strata; Wonji Group; e.g., Meyer et al., see also below Section 6.3). Radiometric ages of this activity are 1975; Kazmin et al., 1980; WoldeGabriel et al., 1990 and references generally Quaternary (b1.8 Ma; e.g., WoldeGabriel et al.,1990; Chernet therein) is spatially associated with the oblique faults of the Wonji et al., 1998; Wolfenden et al., 2004; Abebe et al., 2005), whereas the Fault belt affecting the rift floor (Figs. 9, 10, 27). Silicic rocks (Wonji last eruptions are estimated to be Holocene-historical in age (see Group Silicics of Chernet et al., 1998; Bora–Bericha Rhyolites of Abebe Abebe et al., 2005 and references therein). Ongoing volcanic activity is et al., 2005) form large central volcanoes, some characterised by well- testified by historic fissural basalt flows at Fantale and Kone volcanoes developed calderas, which give rise to ignimbrites, lava flows, domes as recently as 1810 (Harris, 1844; Gibson, 1969). and phreatomagmatic deposits, with compositions ranging from The genetic relationships between the basalts and rhyolites and the trachyte to peralkaline rhyolites. Basalts (Wonji Group Basalts of absence of intermediate products (the so-called Daly Gap) are still Chernet et al.,1998; Wonji Basalts of Abebe et al., 2005) form small lava debated (e.g., Peccerillo et al., 2003). Many petrological indications G. Corti / Earth-Science Reviews 96 (2009) 1–53 21

Fig. 17. Geometry of WFB segments in the different MER sectors. The WFB is oblique to the rift trend; at the rift margins, the Wonji faults curve to adjust to the boundary faults giving rise to a typical S-shaped curvature and an overall sigmoidal geometry. a) Fantale–Dofen WFB segment (Northern MER); b) Boseti and Kone WFB segments (Northern MER); c) Gedemsa WFB segments (boundary between Northern and Central MER); d) Awasa–Shala and Langano–Ziway WFB segments (Central MER). Insets in the top, left corners schematically illustrate the local architecture of the WFB segments; insets in the bottom, right corners display a schematic structural pattern illustrating the adjustment of the Wonji faults to the trend of the boundary faults (trends of both structures and the obliquity angles between them are also reported). Inset in bottom, right corner modified from Kurz et al. (2007). suggest that silicic rocks may have been generated by fractional (Ayalew et al., 2006), as imaged by geophysical data (see Section 5.4). At crystallization of transitional basaltic magmas in shallow-level magma shallow levels, magmas can fractionate in predominantly small magma chambers, with some degree of crustal assimilation (e.g., Gasparon et al., bodies (dikes) and some larger magma chambers (e.g., nested calderas), 1993; Boccaletti et al.,1995; Trua et al.,1999; Peccerillo et al., 2003, 2007; where zoned reservoirs are produced with a peralkaline silicic upper Rooney et al., 2007). The chemistry of 28 to 2.5 Ma-old silicic rocks layer and basalts at the bottom (e.g., Peccerillo et al., 2007). In this indicate a possible increase in crustal involvement with time, suggestive system, eruption preferentially taps the silicic layer, giving rise to the of a progressive thermal and mechanical weakening of the crust in abundant silicic activity, whereas mafic melts reach the surface when response to thinning and magmatic modification of the lithosphere fractures intersect the lower layer of the shallow chamber or reach some 22 G. Corti / Earth-Science Reviews 96 (2009) 1–53

Fig. 18. a) Photo of the Asela escarpment showing the large vertical offset of the border fault system (indicated in the close up of panel b). c) Photo a Wonji fault northeast of Lake Ziway showing the small vertical displacement (indicated in the close up of panel d) typical of the WFB. Note the difference in deformation style. Photo A. Agostini. deep underplated basalt reservoirs (Fig. 28a; Peccerillo et al., 2007). indicate that each cinder cone has magmas derived from slightly Remnants of the crystallized magmatic material, basically consisting of different sources precluding large magma chambers feeding multiple maﬁc magmas and their cumulates, are imaged by geophysical data cones. Rooney et al. (2007) indicate that Wonji magmatic segments in beneath the rift (see Section 5.4). Rooney et al. (2005) and Furman et al. the Northern MER represent well-developed magmatic system that (2006) suggest that this model may be applicable for the large centres allow magmas to rise quickly through existing conduits from melt but cannot work for cindercones and ﬁssures; their petrological data generation zones to shallow levels (Fig. 28b); geochemical analysis G. Corti / Earth-Science Reviews 96 (2009) 1–53 23

over a low-relief landscape and preceded the main rifting phases (Bonini et al., 2005; Abebe et al., 2005). This volcanic event was followed by minor local volcanic activity such as some basaltic flows characterising the Addis Ababa area (Addis Ababa basalts; Abebe et al., 2005), with ages ranging between 7.5 and 5 Ma (Morton et al., 1979, Chernet et al., 1998). As in the Northern MER, the onset of major deformation is accompanied by volcanism with fundamentally bimodal character, with predominant acid products and associated basalts (Fig. 26). In particular, the syn-rift volcanism starts with eruption of voluminous silicic pyroclastic material (Butajira Ignimbrites of WoldeGabriel et al., 1990; Nazret pyroclastics of Abebe et al., 2005)thatlargely characterises the rift floor and attains a maximum estimated thickness of about 700 m. This silicic unit consists of peralkaline pantelleritic ignimbrites with minor intercalated basaltic lava flows, and rhyolitic and trachytic lava domes associated with lava flows. Major ignimbritic layers are believed to have formed by cataclysmic eruption related to the collapse of large caldera structures (such as the 3.5 Ma-old “Munesa crystal tuff” Caldera now buried beneath the Ziway–Shala lake basin system; see Le Turdu et al., 1999). The ages of the Nazret pyroclastics typically range between 5.2 and 2.6 Ma, with a maximum concentration at about 3.5 Ma (e.g., Abebe et al., 2005 and references therein). The sequence is followed by upper Pliocene basalts and pyroclastic deposits analogous to those described in the Northern MER. Fig. 19. Symmetrical spatial distribution of surface deformation type and style within During the Pliocene the volcanic activity also characterises the the different WFB tectono-magmatic segments in the Northern MER (after Kurz et al., eastern and western plateaus, close to the rift margin (Fig. 29). On the 2007). western margin, volcanic activity gives rise to some central silicic volcanoes (such as Wechacha, Furi and Yerer) that are preferentially indicate that Quaternary magmas are generated at ~15–25 kbar (~50– aligned in a E–W trend and are intimately related to the deformation 90 km; Rooney et al., 2005; Furman et al., 2006). along the transversal Yerer–Tullu volcano-tectonic lineament (Abebe A temporal evolution from silicic volcanism from the shallow et al.,1998; Chernet et al.,1998; Mazzarini et al.,1999;seeSection 5.2.3.). magma chambers to basaltic activity within the Wonji segments has Radiometric ages of 4.6–3.3 Ma were obtained for the western marginal also been suggested (Abebe et al., 2007). In this model, the successive volcanoes (Miller and Mohr, 1966; Kunz et al., 1975; Mohr and Potter, cooling, faulting and fracturing of the silicic centres may allow the 1976; Morton et al., 1979; Kazmin et al., 1980; WoldeGabriel et al., 1990; upraising and eruption of basaltic lava flows, which may have led to an Mazzarini et al., 1999). On the Somalian plateau, trachytic and basaltic increase in basaltic activity in the last 0.65 Ma (Abebe et al., 2007). central volcanoes grew during Pliocene times, giving rise to an This temporal trend mimics the northward increase in basaltic important phase of off-axis volcanism (see Section 5.3.4). volcanism toward Southern Afar (see Hayward and Ebinger, 1996). The subsequent volcanic activity is spatially associated with the In Afar, during the last 4 Myr magmatism has been bimodal, but with oblique faults of the Wonji Fault belt (Wonji Group) and is predominance of basaltic rocks over silicic products (e.g., Lahitte et al., characterised by a typical bimodal association, as explained in 2003). Volcanism indeed mainly consisted in the eruption of a thick Section 5.3.2. Major ignimbritic eruptions, related to large caldera stratiform sequence of fissure-fed basaltic to hawaiite lava flows (‘Stratoid Series’; e.g., Barberi et al., 1975). Silicic (trachy-rhyolites to pantellerites) central volcanoes locally grew on the upper part of the Stratoid Series (see Lahitte et al., 2003 and references therein). Further to the north, in Central and Northern Afar, the zone of Quaternary extension and magmatism is marked by clusters of voluminous fissure basalts and basaltic shield volcanoes of a transitional alkali/tholeiitic composition. These axial basalt ranges are interpreted to be sub-aerial equivalents of oceanic spreading centres (e.g., Barberi and Varet, 1977; Hayward and Ebinger, 1996).

5.3.2. Central MER In the Central MER, early eruption of the ~30 Ma-old flood basalts was locally followed by undersaturated intermediate and acidic rocks (at ~17–12 Ma; Shebele Trachyte of WoldeGabriel et al., 1990) and by – eruption of Late Miocene (~11 8 Ma) basalts and trachybasalts with fi fi fl Fig. 20. Simpli ed cartoon showing the development of faults and ssures during a subordinate silicic ows (Guraghe basalts of WoldeGabriel et al., 1990; single dyke injection event (after Tentler 2005 and Rowland et al., 2007). I: A dyke Guraghe–Anchar basalts of Abebe et al., 2005 and Bonini et al., 2005; injection event occurs at depth at a critical moment governed by the balance between see also Morton et al., 1979; Kazmin et al., 1980; WoldeGabriel et al., the magma pressure and the tectonic stress state. II: Normal faults are initiated or 1990; Abebe et al., 1998; Chernet et al., 1998; Mazzarini et al., 1999). reactivated ahead of and above the laterally and possibly upward propagating dyke. III: Pre-existing sub-vertical cooling joints are reactivated as opening mode fissures in the Comparison with time-correlative basaltic units in both the western region of induced stress above the upper tip line of the normal faults. Once the upward and eastern plateaus and correlation with volcanic sections away from propagating normal faults and the fissures are fully linked, vertical displacement the rift, suggest that the volcanics of this Late Miocene event erupted accrues at the surface. 24 G. Corti / Earth-Science Reviews 96 (2009) 1–53

Fig. 21. Simplified model of formation and growth of the normal faults in the axial part of the Ethiopian Rift (after Acocella et al., 2003). a) Initial development and growth of extension fractures. b) Fractures reach a critical depth is reached where the tensile stresses can no longer be sustained. c) At this stage, a failure behaviour controls the further propagation of the ruptures at depth and the extensional fractures turn into normal faults. collapses (e.g., Gademotta, O'a, Corbetti) blanketed the rift floor, After these events, volcanic activity drastically decreased in the especially at around 0.20–0.30 Ma, when a regional paroxism of silicic Upper Miocene –Lower Pliocene, with only a limited eruption of and basaltic volcanism is recognised (Fig. 30; e.g., Morton et al., 1979; ~7 Ma-old basalts that crop out scarcely in the plateau above the Le Turdu et al., 1999). Radiometric ages of this activity are generally Chencha escarpment. Volcanic activity was then resumed in the Late Quaternary (b1.6 Ma; e.g., WoldeGabriel et al., 1990; Chernet et al., Pliocene(?)–Early Pleistocene (Zanettin et al., 1978), with bimodal 1998; Abebe et al., 2005), whereas the last eruptions are estimated to volcanism on the rift floor. This late activity started with the eruption be Holocene-historical in age (see Abebe et al., 2005 and references of widespread Pleistocene ignimbrites (1.6–0.5 Ma) that, although therein). younger, may be correlated to the Nazret unit that largely char- As in the Northern MER, the bimodality of magmatic products with acterises the Northern MER (Bonini et al., 2005). The volcanic predominance of acidic products over mafic volcanics has been succession is closed by the Nech Sar olivine basalts (1.34–0.77 Ma; explained in terms of zoned magma chambers containing rhyolitic Ebinger et al., 1993), by 0.66 Ma-old trachybasalts of the Bobem melt at their top and erupting preferentially acid magmas, with volcano, and by pumiceous tuffs, obsidian flows and basalts in the subordinate mafic melts erupting when fractures reach the lower Bridge of God area (Ebinger et al., 1993; George and Rogers, 1999; layer of the chamber or deep underplated basalt reservoirs (Fig. 28a; Bonini et al., 2005). Some olivine basalts derived from N10°E-striking e.g., Peccerillo et al., 2007). Rooney et al. (2007) suggested a more fissures and cinder cones located along the eastern Chamo basin and complex magmatic system in the Central MER, characterised by two Bridge of God may have erupted in historic times (Ebinger et al., 1993). different sub-parallel tectono-magmatic zones (the volcanic chains of The Nech Sar basalts and successive volcanic activity are coeval with Debre Zeyt and Butajira to the West, and the Wonji Fault Belt the Wonji basalts of the Northern MER, although they show a higher segments to the East) differing in terms of depths of melt fractiona- alkalinity suggesting that the Southern MER underwent a compara- tion, which in turn reflect the different maturity of the magmatic tively minor degree of extension (Zanettin et al., 1978). plumbing system. In this model, the Wonji Belt is characterised by a well-developed magmatic system, analogous to that of the Northern 5.3.4. Spatio-temporal distribution of volcanic activity MER and possibly representing its southwards propagation; con- The spatio-temporal distribution of volcanic activity in the MER is versely, the Debre Zeyt and Butajira volcanic chains display a less characterised by a two-phase evolution that mimics the two-phase evolved magmatic system, with less defined conduits and melt evolution of faulting, with the transition from large boundary faults to fractionation at various depths within the crust, which may be related en-echelon Wonji faults. to the northwards propagation of the EARS (see Section 6.1.2). During the Mio-Pliocene, concomitant with the activity of the large boundary fault systems, there is widespread volcanic activity, with volcanism encompassing the whole rift depression (Fig. 31). Localisation 5.3.3. Southern MER of volcanic centres is mainly controlled by boundary fault systems and The Tertiary volcanic activity in the Southern MER started earlier by pre-existing fabrics both parallel and transversal to the rift. Influence than in the other MER sectors, since the pre-rift flood-basalt event is of boundary faults on localisation of volcanism in these initial stages of dated at about 45 Ma (Amaro–Gamo basalts of Ebinger et al., 1993; see rifting is well testified in the different rift segments (e.g., WoldeGabriel also Davison and Rex, 1980; WoldeGabriel et al., 1991; George et al., et al., 1990, 1999; Ebinger et al., 1993; Wolfenden et al., 2004). Similarly, 1998; Yemane et al., 1999; Ebinger et al., 2000). This initial mainly pre-existing fabrics represented important features controlling the basaltic activity ended around 30 Ma (Zanettin et al., 1978). According localisation of volcanism during initial rifting, as they acted as weakness to Levitte et al. (1974) and Zanettin et al. (1978) a second phase of zones and represented preferential pathways for ascending magmas, mainly basaltic volcanism started in the Early Miocene with the whereas during progressive rift evolution their influence is expected to eruption of stratoid basalts (equivalent to the Getra–Kele basalts of have decreased (e.g., Mohr, 1983). Transversal fabrics corresponded to Ebinger et al., 1993). Dykes intruding the Amaro–Gamo sequence on NW–SE-trending pre-existing structures (see Section 5.2.3)andgave the Chencha escarpment and on the opposite rift margin (the western rise to the localisation of magmatic activity at major transfer zones, as margin of the Amaro horst) have been dated at about 21 Ma (Levitte et currently observed in the weakly extended Western Branch of the EARS al., 1974) and may represent feeding dykes for the Getra–Kele basalts. (e.g., Ebinger et al., 1989; Ebinger, 1989; Corti et al., 2003) and other rift Volcanic activity continued in this area during the Miocene up to zones worldwide (e.g., Wilson, 1993). The influence of rift-parallel pre- 11 Ma, with eruption of basalts, trachytes and rhyolites (Zanettin et al., existing fabrics is still testified by the alignment of off-axis volcanoes 1978; Ebinger et al., 1993). that grew on the Somalian plateau during the Mio-Pliocene (Fig. 29; e.g., G. Corti / Earth-Science Reviews 96 (2009) 1–53 25

Fig. 22. a) Landsat image and structural interpretations of Wonji fault segments north of Lake Abaya. b) Detail of small-scale analogue model of oblique rifting (after Corti, 2008) with structural scheme showing the en-echelon fault pattern resulting from the left-lateral transcurrent component of motion. Note the similarity between the model and natural structural pattern.

Mohr and Potter,1976). Although different hypothesis have been applied extensional stress field developing on the plateaus flanking the rift for explaining the development of these off-axis volcanic centres (see (Ellis and King, 1991). The close correspondence between the volcanic Corti et al., 2003 and references therein), Corti et al. (2004) suggested trend and the basement fabric (although locally obscured by the flood- that lateral magma migration may have favoured magma accumulation basalt cover) supports a strong control exerted by pre-existing below the rift shoulders with the final emplacement controlled by rift- weaknesses on the development of off-axis volcanism (Fig. 29; Mohr parallel pre-existing weaknesses (Mohr and Potter, 1976) aided by an and Potter, 1976). 26 G. Corti / Earth-Science Reviews 96 (2009) 1–53

Fig. 23. Recent and active tectonics in the Bridge of God area (from Bonini et al., 2005). (a) Satellite Landsat TM image. (b) View from the south of the axial graben, marking a riftward deformation, extending from Lake Abaya to Lake Chamo (Bridge of God area). (c) View from the west of the fault bounding to the east the fault-controlled axial basin. Ongoing deformation of this structure is demonstrated by the sharp and fresh west-dipping fault scarp, indicated by the tip of black triangles.

At the end of the Pliocene there is a substantial change in the (compare plots of faults/eruptive centres distribution in Figs. 10–12, distribution of volcanic activity, concomitant with the activation of the 27); at depth, geophysical data indicate a strong magma intrusion Wonji fault belt. Quaternary volcanism is indeed focused within the below the deformation segments, with trend and en-echelon rift depression and strongly localised along the Wonji faults, giving architecture mimicking that of Wonji structures (e.g., Keranen et al., rise to magmatic segments with only minor activity outside the en- 2004; see Section 5.4.1). The strong association between deformation echelon deformation belt (Fig. 31). At surface, the alignment of and magmatism along the Wonji Fault belt gives rise to magmatic volcanic centres strikingly mimics the trend of the Wonji faults segments (see Ebinger and Casey, 2001) that are well developed in the G. Corti / Earth-Science Reviews 96 (2009) 1–53 27

Fig. 24. Schematic representation of the main volcano-tectonic features along the Yerer-Tullu Wellel and Goba–Bonga transversal structures.

Northern MER; conversely, consistent with a southwards decrease in the effective elastic thickness of the lithosphere (see Section 5.4; Ten extension and rift evolution, magmatic segments are less expressed in Brink, 1991). the Central MER where extension and magmatic modification seem to be less developed (e.g., Rooney et al., 2007). 5.4. Geophysical data constraining the crustal and mantle structure in Similarly, a northward increase in basaltic activity is observed, the MER with fissural basalts dominating over silicic products in the Afar depression (see Section 5.3.1). Mohr and Wood (1976) also noted a 5.4.1. Crustal structure northward decrease in the separation between large eruptive centres Knowledge of the crustal and lithospheric structure of the Main in the MER and Afar, a pattern that may be controlled by a decrease in Ethiopian Rift has increased significantly in recent years. This is 28 G. Corti / Earth-Science Reviews 96 (2009) 1–53

Fig. 25. Simplified structural sketch of the Lake Tana region (modified from Mege and Korme, 2004) superimposed on a SRTM digital elevation model. Location indicated in the inset. mainly due to the large amount of data collected for the Ethiopian Geoscientific Lithospheric Experiment; Maguire et al., 2003) pro- Broadband Experiment (Nyblade and Langston, 2002; Dugda et al., jects. This latter in particular collected new seismic (Keranen et al., 2005;Benoitetal.,2006a,b), and the EAGLE (Ethiopia–Afar 2004; Bastow et al., 2005; MacKenzie et al., 2005; Kendall et al.,

Fig. 26. Total alkali silica diagram for volcanic products from the MER (after Abebe et al., 2005). Ages and main lithologic characteristics (see main text; Abebe et al., 2005): Guraghe– Anchar Basalts (basaltic and trachybasaltic lava flows, intercalated with minor rhyolitic tuffs; Late Miocene), Addis Ababa Basalts (basaltic lava flows; Late Miocene), Nazret Pyroclastic Rocks (ignimbrites, lava flows and domes; Late Miocene–Pliocene), Gash Megal Rhyolites (lava flows and pyroclastic deposits; Late Miocene?–Pliocene), Wechacha Trachytes (lava flows and lava domes with minor pyroclastics; Pliocene), Bofa Basalts (basaltic lava flows; Pliocene–Pleistocene), Chilalo Volcanics (basaltic to trachytic lava flows, Pliocene–Pleistocene), Chefe Donsa Pyroclastic Deposits (volcanic ash flow and fall deposits; Pliocene–Pleistocene), Akaki Basalts (basaltic lava flows and scoria cones; Pliocene), Zikwala Trachytes (lava flows and subordinate pyroclastic deposits; Pleistocene), Bora–Bericha Rhyolites (ignimbrites, lava flows, domes and phreatomagmatic deposits; Pleistocene–Holocene), Wonji Basalts (basaltic lava flows, scoria and phreatomagmatic deposits; Pleistocene–Holocene). G. Corti / Earth-Science Reviews 96 (2009) 1–53 29

Fig. 27. Satellite (Landsat) image (left panel) and digital elevation model (right panel) of the Boseti volcano, showing the alignment of Quaternary volcanic centres along the Wonji Fault Belt. For the location of the volcano see Fig. 10.

2005, 2006; Maguire et al., 2006; Stuart et al., 2006; Daly et al., be caused by high temperatures and the presence of melt (up to 20% of 2008), gravity (Cornwell et al., 2006; Mickus et al., 2007), seismicity melt; Keranen et al., 2009), as also supported by analysis of electrical (Keir et al., 2006a,b), geodetic (Bendick et al., 2006), magnetotelluric conductivity (Whaler and Hautot, 2006), shear-wave splitting (Keir et (Whaler and Hautot, 2006), and geochemical (Rooney et al., 2005; al., 2005), bulk-crustal Vp/Vs ratios (Dugda et al., 2005; Stuart et al., Furman et al., 2006; Rooney et al., 2007) data, which – although 2006) and also consistent with the presence of Quaternary eruptive unevenly distributed along the MER and mainly concentrated in centres as far north as Lake Tana and with an off-rift low-velocity the Northern MER–Central MER and adjacent plateaus – allowed anomaly in the upper mantle (Bastow et al., 2005, 2008). to greatly improve the knowledge of the crustal and lithospheric A high-velocity layer in lower crust, evidenced by controlled- structure of the rift. source seismic survey from the EAGLE experiment has been Although it is outside the scope of this paper to review all the interpreted to be an underplated maficmaterialemplacedin details of the geophysical data acquired in the MER, I will summarise association with the voluminous flood basalts and thickening the below the main findings of these geophysical projects that are crust to 40–50 km (MacKenzie et al., 2005; Maguire et al., 2006). This relevant for the process of rifting and break-up. layer is believed to be a localised feature and not to be a widespread characteristic of the NW shoulder (Keranen et al., 2009). 5.4.1.1. Ethiopian and Somalian Plateaus. The crust beneath the Based on the geophysical observations, Keranen and Klemperer Ethiopian and Somalian Plateaus has not been modified significantly (2008) suggested that the Ethiopian and Somalian Plateaus corre- by the Cenozoic rifting and magmatism; both plateaus are char- spond to two major Proterozoic basement terranes (Fig. 34). The acterised by thick crust, which are generally N38–40 km (Figs. 32, 33; difference in geophysical characteristics is evident in crustal thickness Berckhemer et al., 1975; Kebede et al., 1996; Mackenzie et al., 2005; (variable in a north–south direction in the Ethiopian Plateau, rather Dugda et al., 2005, 2007; Stuart et al., 2006; Keranen and Klemperer, constant in the Somalian Plateau), in an observed difference in Vp/Vs 2008; Keranen et al., 2009). The Somalian Plateau is of relatively ratio between the two plateaus that may reflect a difference in pre-rift constant thickness from north to south whereas the Ethiopian Plateau crustal composition (Stuart et al., 2006), and in upper-mantle has a strong north–south decrease in crustal thickness. This variation velocities (e.g., Bastow et al., 2005). The difference in crustal structure occurs rather sharply at 9°N latitude in correspondence to the YTVL from shoulder to shoulder implies a pre-existing boundary between transverse structure where the western shoulder thins rapidly from the two distinct basement terranes, which probably corresponds to a N38–40 km thick to 34–36 km thick (Berckhemer et al., 1975; Dugda roughly NE–SW-trending Proterozoic suture (Fig. 34; Keranen and et al., 2005; Stuart et al., 2006; Dugda et al., 2007; Keranen and Klemperer, 2008). Coincidence of the MER extensional structures with Klemperer, 2008; Keranen et al., 2009). In both the Somalian and this terrane boundary strongly suggests that the location and initial Ethiopian (northwest of latitude 9°N) plateaus, there is a rather sharp evolution of the extensional process in the MER was strongly transition to a thinner crust (of ~5 km) below the rift (Fig. 33). controlled by the presence of this ancient lithospheric-scale hetero- Rayleigh wave/receiver function joint inversion analysis by Keranen geneity (see also Section 6.1; Keranen and Klemperer, 2008). et al. (2009) suggests the presence of a widespread, low lower-crust (and uppermost mantle, see Section 5.4.2) shear-wave velocity beneath the rift, the NW shoulder and the region of the Somalian Plateau covered 5.4.1.2. Northern MER. The Northern MER is characterised by a large by flood-basalt volcanism. This velocity anomaly has been interpreted to dataset of geophysical data (source and passive seismic, gravity and 30 G. Corti / Earth-Science Reviews 96 (2009) 1–53

rather constant thickness. Locally (as beneath the Fantale–Dofen magmatic segment) the crust is thicker than the surroundings; this is associated with the highest upper-crustal seismic velocity and highest Poisson's ratios, suggesting that magmatic processes may have locally thickened the crust (with respect to the rest of the Northern MER) in a zone of overall extension (see below; Keranen and Klemperer, 2008). Crustal tomography reveals the presence of anomalously fast, elongate bodies in the mid- to lower crust extending along the rift axis and rising to a ~10 km subsurface (Fig. 35; Keranen et al., 2004; Daly et al., 2008). These 20-km-wide and 50-km-long bodies are separated and laterally offset from one another in a right-stepping en-echelon pattern, approximately mimicking surface segmentation of Quatern- ary volcanic centres. The anomalously fast bodies are also characterised by a high Vp/Vs ratio (Daly et al., 2008), which increase from beneath the margin of the northwestern plateau into the rift (and northward along the length of the MER; Dugda et al., 2005; Stuart et al., 2006), and by relative positive Bouguer anomalies (Mahatsente et al., 1999, 2000; Tessema and Fontaine, 2004; Tiberi et al., 2005; Cornwell et al., 2006); both features are interpreted as indicating the presence of mid- to lower-crustal cooled mafic intrusions (Keranen et al., 2004; Daly et al., 2008). Results of gravity analysis by Cornwell et al. (2006) suggest that these intrusions contain at least 40% gabbro. The magnitude of the Vp/Vs ratios and P-wave velocity anomalies along the rift axis could also indicate the presence of a molten fraction in fractures within the solidified mafic intrusions (Dugda et al., 2005; Stuart et al., 2006; Daly et al., 2008), as suggested on the basis of petrological evidences (Rooney et al., 2005). Geophysical data support the presence of melt in the crust and upper mantle beneath Wonji segments. Patterns of seismic velocity anisotropy determined from shear-wave splitting studies (Kendall et al., 2005; Keir et al., 2005; Kendall et al., 2006) suggest magma intrusion throughout the continental lithosphere beneath Wonji segments. In particular, the magnitude of anisotropy and its parallelism with aligned chains of eruptive centres and fissures suggest the presence of melt-filled cracks and/or dykes that penetrate both the upper crust and the uppermost mantle to a depth of ~75 km Fig. 28. a) Schematic cross-section showing a possible model for the distribution of magma chambers along the northern sector of the main Ethiopian rift (after Peccerillo (Kendall et al., 2005; Keir et al., 2005; Kendall et al., 2006). Similarly, et al., 2007). b) Schematic representation of the magmatic plumbing system beneath the low lower-crust (and uppermost mantle, see Section 5.4.2) shear- the Wonji Fault Belt and Debre Zeyt–Butajira volcanic chains (after Rooney et al., 2007). wave velocity beneath the rift imaged by Keranen et al. (2009) This cartoon shows that the magmatic plumbing system beneath the Wonji Faults is indicates high temperatures and the presence of melt. Magneto- more developed than that beneath the Debre Zeyt–Butajira volcanoes. This results in magmas within the Wonji Fault Belt rising toward the surface more rapidly, telluric analysis (Whaler and Hautot, 2006) image a low resistivity fractionating close to the surface. Beneath the Debre Zeyt–Butajira volcanic chains the zone at ~1 km depth beneath Boseti volcano, which may represent a magmas may fractionate at various crustal depths prior to eruption. Note that this shallow magma body (Fig. 36); presence of magma bodies in the cartoon best explains the crustal structure of the Central MER, where the volcano- uppermost crust is also supported by gravity modelling (Cornwell tectonic activity is localised at the two rift margins; conversely, the Northern MER is et al., 2006). characterised by a single rift-centred zone of magmatism (beneath the Wonji Fault fi Belt). See text for further details. The above ndings indicate that the crust beneath the rifted regions in Ethiopia has been extensively modified by magmatic processes and by the addition of mafic rock in the mid- to lower crust. Geophysical data support that magmatic modification increases magnetotelluric experiments), mainly collected in the frame of the northwards from the Northern MER to the Southern Afar, i.e. towards EAGLE project. the more oceanic part of the rift (e.g., Kendall et al., 2006), consistent The cross-sectional crustal thickness profile is typical for a with a northward increase in thinning. Notably, this thinning is mostly continental rift profile, with large border faults bounding the rift, accommodated in the upper to middle crust, whereas the lower crust and rift shoulders that are ~5 km thicker than the crust beneath the has relatively constant thickness along the length of rift suggestive of a rift valley (Mackenzie et al., 2005; Dugda et al., 2005; Stuart et al., low strength and viscous behaviour, possibly caused by high 2006; Maguire et al., 2006; Dugda et al., 2007; Mickus et al., 2007; temperatures and presence of melt (Keranen et al., 2009). Keranen et al., 2009); the seismic and structural data show large asymmetries in the basin infill, with tilting towards the large-offset 5.4.1.3. Central MER. The MER in its central sector is characterised border fault system (see Section 5.2.1.1). Along the rift axis, the crust by a thicker crust than the Northern MER with a clear and abrupt in this rift sector gradually thins to the north, from about 33–35 km transition in crustal properties coincident with the Boru Toru at the Northern MER–Central MER boundary to about 24–26 km in structural high at the Northern MER–Central MER boundary (Bonini southern Afar (Fig. 32; Kebede et al., 1996; Dugda et al., 2005; et al., 2005; Maguire et al., 2006; Keranen and Klemperer, 2008). In Maguire et al., 2006; Mickus et al., 2007; Stuart et al., 2006). North- particular, crustal thickness increases rapidly from 33–35 km in the ward thinning of the crust is mostly accommodated in the upper southernmost Northern MER to 38–40 km in the Central MER crust (e.g., Mackenzie et al., 2005); the lower crust is instead of (Figs. 32, 33; Dugda et al., 2005; Maguire et al., 2006; Mickus et al., G. Corti / Earth-Science Reviews 96 (2009) 1–53 31

Fig. 29. Digital elevation model (left panel) and structural scheme (right panel) showing the main off-axis volcanoes that developed during the Mio-Pliocene on the Somalian Plateau. Dashed lines indicate the trend of pre-existing fabrics that possibly feed the volcanoes.

Fig. 30. Schematic time chart of volcanism in the Lakes region of the Central MER from middle Pliocene to present-day. After Le Turdu et al. (1999). 32 G. Corti / Earth-Science Reviews 96 (2009) 1–53

Fig. 31. Distribution of volcanic rocks in the MER in the Mio-Pliocene and Quaternary.

2007; Stuart et al., 2006; Keranen et al., 2009), although local the Central MER than in the Northern MER. The upper crust has a estimates in volcanic areas indicate thicknesses b30 km (Mazzarini, significantly lower seismic velocity (Maguire et al., 2006). All these 2004). The majority of the crustal thickness variation is accommo- observations point to a significant change in crustal (and upper dated in the upper crust, whereas the lower crust is of relatively mantle, see below) structure between the Northern MER and the constant thickness. In cross-section, crustal thickness gradients Central MER, which may be interpreted as reflecting a discontinuity in between the plateaus and the rift depression are sensibly lower in the amount of extension and magmatism between the two rift sectors

Fig. 32. Cross-sections across (a) and along (b) the rift axis from the EAGLE controlled-source experiment (after Mackenzie et al., 2005; Maguire et al., 2006; Keranen and Klemperer, 2008). P-wave velocities are given in km/s. The superjacent elevation profiles are plotted with respect to sea level. HVLC – high-velocity lower crust; solid black lines – reflectors; vertical lines – segment boundaries (dashed at the poorly localised monoclinal boundary between the Ethiopian Plateau and the NMER); solid gray line – Moho; dashed gray line – approximate location of Moho. The dashed black line in (b) shows the boundary between an upper mantle with density of 3.25 kg/m3 and 3.15 kg/m3 from the model of Mickus et al. (2007). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) G. Corti / Earth-Science Reviews 96 (2009) 1–53 33

Notably, a high P-wave anomaly and high Vp/Vs ratio is observed beneath the Debre Zeyt volcanic field, at the intersection between the MER faults and the transversal YTVL. This area is underlain by hot asthenosphere (Bastow et al., 2005, 2008) and is probably associated to the presence of melt in the crust (e.g., Whaler and Hautot, 2006; Stuart et al., 2006). Seismic tomography (Keranen et al., 2004) has identified a high-velocity anomaly underlying the Debre Zeyt region that may represent a cooled intrusion; xenolith evidence (Rooney et al., 2005) suggests that this large geophysical anomaly may reflect pervasive dyking/veining (to a depth of at least 30 km) rather than a single large intrusion. Magnetotelluric data (Whaler and Hautot, 2006) highlight a zone of conductive material at lower-crustal depths (beginning at ~15 km and becoming more conductive and wider at 20–25 km) indicating the presence of melt within the cooled mafic intrusion or dyking region. These observations suggest that the Debre Zeyt region may be a zone of extension and focused decompressional melting (Rooney et al., 2005), although the lack of typical deformation features (Wolfenden et al., 2004; Casey et al., 2006) and seismic activity (Keir et al., 2006a, see Section 5.5) that characterise the Wonji segments in the Northern MER led Keir et al. (2006a) to suggest that the Butajira and Debre Zeyt volcanic chains are either unfavourably oriented ‘failed’ magmatic segments, or incipient zones of strain related to Wonji segments branching off from the western rift margin (Bonini et al., 2005; Abebe et al., 2005).

Fig. 33. Contour map of Moho depth after Keranen et al. (2009). The dashed gray lines show the rift outlines. Solid black lines are the rift border faults. For the mapped area, 5.4.1.4. Southern MER. Few geophysical data are available for the sensitivity testing indicates an uncertainty in the depth of the Moho discontinuity of about 2.5 km; comparison with active source wide-angle profiles from the EAGLE southern sector of the MER. Receiver function analysis from the experiment (Maguire et al., 2006)confirms the reliability of the Moho depth estimates Ethiopian Broadband Seismic Experiment (Dugda et al., 2005) suggest (Keranen et al., 2009). (For interpretation of the references to colour in this figure a crustal thickness decrease in the Southern MER to values of ~30 km legend, the reader is referred to the web version of this article.) (Fig. 33), consistent with gravity modelling (Mahatsente et al., 1999, 2000). The low Poisson's ratio at one station in the Southern MER and the lower surface volcanic activity within the rift depression suggest the absence of significant magmatic processes in this rift's sector. (e.g., Keranen and Klemperer, 2008). In particular, the differences in These observations point to a further decrease in extension and geophysical characteristics (thicker crust, lower magnitude upper- magmatic modification in the Southern MER. The lower-crustal crustal velocities and higher uppermost mantle density, see below) in thickness in this rift sector with respect to the Central MER is probably the Central MER reflect the lower amount of extension and magmatic related to the superposition of multiple phases of rifting in the modification in the crust (and upper mantle). Mesozoic and Cenozoic that created a thinner-than-normal crust in the

Fig. 34. Pre-existing lithospheric domains, crustal thickness, and locations of Cenozoic extension and volcanism in the MER (after Keranen and Klemperer, 2008). Cenozoic extension follows the rheological boundaries between two distinct Proterozoic basement terranes, underlying the Ethiopian and Somalian plateaus; this lithospheric weakness zone corresponded to a suture zone. Crustal modiﬁcation occurred within the MER and the Yerer–Tullu Wellel volcano-tectonic lineament; this latter developed in correspondence to the sharp north–south decrease in crustal thickness which characterises the Ethiopian plateau at 9°N latitude. 34 G. Corti / Earth-Science Reviews 96 (2009) 1–53

Fig. 35. Three-dimensional seismic velocity model of Keranen et al. (2004). Horizontal slice at 10 km depth (left) and rift-perpendicular cross-sections (right) highlighting the occurrence of high-velocity (Vp) bodies below rift axis interpreted as solidified magmatic intrusions. Magmatic segments are shown by dotted lines in left panel. H – Hertale, LHF – Liado Hayk field, D – Dofan, F – Fantale, K – Kone, B – Boseti, Y – Yerer, Z – Zikwala, G – Gedemsa, TM – Tullu Moje, Al – Aluto, Sh – Shala. The vertical arrow in the right-upper panel indicates the high-velocity body. Depth of slice (DS) in left panel is marked by horizontal line in the cross-sections. ABF–Arboye border fault. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

area (Benoit et al., 2006c). The first rifting event at 20 Ma (caused by (Keranen et al., 2009). The thinned lithosphere beneath the Ethiopian the northward propagation of the Kenya rift) further thinned the crust Plateau can be explained by a mantle plume model, if a plume rapidly before the Pliocene southward propagation of the Ethiopian rift (see thinned the lithosphere at the time of the flood-basalt event (at Bonini et al., 2005). around 30 Ma), and if warm plume material has remained beneath the lithosphere since then (Dugda et al., 2007). The removal of the mantle 5.4.2. Upper mantle structure lithosphere beneath the rift is attributable to the extension-related The magmatic modification that is evident in the crust beneath the thinning, in addition to the plume-related effect (Dugda et al., 2007). rift is also mimicked in the upper mantle. In the Northern MER, a low- Beneath Afar, the lithosphere/asthenosphere structure approaches velocity (Mackenzie et al., 2005) and low-density (Cornwell et al., that of an active spreading ridge system (see for instance exempli- 2006) uppermost mantle characterises the region below the rift, ficative cross-sections in Panza, 1980), and the crust has been indicating the presence of hot rocks containing a small percentage (3– interpreted to consist predominantly of new igneous rock, emplaced 5%) of melt in an upwelling mantle immediately beneath the Moho. through dyking, sill formation and underplating during the late syn- Shear-wave splitting studies (Kendall et al., 2005, 2006) support the rift stage (Dugda and Nyblade, 2006), although the nature of the presence of melt-filled cracks and/or dykes in the upper mantle that lithosphere is still debated (e.g., Beyene and Abdelsalam, 2005). penetrate to a depth of ~75 km. According to Keranen et al. (2009) the Southwards, influence of magmatic activity seems to decrease Northern MER corresponds (together with YTVL) to the region with towards the Central MER, consistent with what suggested for the the lowest upper-mantle shear-wave velocity in the 5 km below crustal structure. A segmentation between the Northern MER and the Moho, further supporting the presence of hot temperatures and Central MER is modelled in the along-axis EAGLE gravity model, which melts. shows a step in upper-mantle density of ~100 kg/m3 at the Northern Joint inversion of receiver functions and Rayleigh wave group MER–Central MER transition (Mickus et al., 2007). velocities by Dugda et al. (2007) suggest that due to these magmatic Rooney et al. (2005) suggest, on the basis of analysis of xenoliths modifications, the mantle lithosphere is nonexistent of greatly from locations in and along the sides of the MER, that the lithospheric thinned beneath the MER and the Afar depression. This is consistent mantle beneath Ethiopia has been modified significantly by silicate with estimates of a very low elastic thickness of the lithosphere of ~8 melts. Modification of the lithosphere decreases southwards, as (±2) km in the Northern MER and Central MER (Ebinger and indicated by the general increase in the strength of the lithosphere Hayward, 1996), with values further decreasing to ~6 (±2) km in (the elastic thickness of the lithosphere is estimated to be of ~17 (±2) Southern Afar. A thin lithospheric mantle extending from the Moho to km in Southern MER; Ebinger and Hayward, 1996). 60–80 km depth is identified below the Ethiopian and Somalian Analysis of the P- and S-wave velocity structure beneath the rift plateaus by Dugda et al. (2007); below the Ethiopian plateau, the and plateaus indicates that the low-velocity mantle zone continues to a mantle lid is indentified in the southwestern part only, whereas a depth ≥75 km (Bastow et al., 2005; Benoit et al., 2006a,b; Pasyanos similar layer is not observed beneath the northwestern portion and Nyblade, 2007; Bastow et al., 2008). Tomographic images reveal G. Corti / Earth-Science Reviews 96 (2009) 1–53 35

Fig. 36. 2D model of resistivity structure across the MER (after Whaler and Hautot, 2006). Note the presence of red areas, which indicate low resistivity zones. These are interpreted as indicating the presence of a shallow magma body beneath the Boseti volcano, and melts within the cooled mafic intrusions at lower-crustal depths. Station numbers, the extent of the Boseti magmatic segment and Adama basin and a projection of the Silti Debre Zeyt Fault Zone (SDFZ) onto the model are indicated along the profile. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

indeed a ~500 km wide low P- and S-wave velocity zone at 75 to this anomaly extends to ~150 km depth and then it broadens laterally ≥400 km depth in the upper mantle that extends from close to the with depth. Southwards, the narrow, tabular low-velocity region eastern edge of the Main Ethiopian Rift (MER) westward beneath the extends to ≥300 km depth (Fig. 37). At 75 km depth, a limb of low- Ethiopian Plateau (Fig. 37). Within the broad anomaly, the lowest velocity structure extends from the NNE-trending Ethiopian rift more velocity zones are not centred beneath the highly thinned Afar but than 100 km in a westerly direction from the Boseti magmatic beneath the MER at ~9° N, 39°E; these low velocities are interpreted as segment to the YTVL fault zone and the Debre Zeyt volcanic field evidence for the presence of melt beneath the MER and adjacent NW (e.g., Bastow et al., 2005, 2008). Plateau (Bastow et al., 2008) to depths in excess of those inferred for At depths ≥150 km the centre of the anomalous upper mantle is the source of Quaternary lavas in the MER (50–90 km; Rooney et al., located beneath the Ethiopian plateau, not the MER or Afar, indicating 2005; Furman et al., 2006). It is unclear whether these low-velocity that rifting developed at the Eastern edge of the low-velocity zone. zones result from focused mantle upwelling and/or enhanced Overall, the structure of the broad low-velocity anomaly has been decompressional melting. Significant lateral smaller-scale heteroge- interpreted to be consistent with the upper-mantle continuation of neity at sub-lithospheric depths also characterise the rift shoulders the African Superplume found in the lower mantle beneath southern where a relatively faster velocity structure, down to depths of ~100 km Africa (e.g., Benoit et al., 2006a,b; Bastow et al., 2008); the (Fig. 37), may indicate a strong/old/cold lithosphere whose greater heterogeneities imaged in the low-velocity region at 200–400 km residual strength compared to the surrounding lithosphere may have depth could be indicative of smaller upwellings from the underlying played an important role in rift development (Bastow et al., 2008). superplume as suggested on the basis of geochemical findings (e.g., Beneath the axis of the Northern MER and Southern Afar, the low- Furman et al., 2006; see Sections 4.1 and 4.3). However, the existence velocity region is narrow and tabular, and is offset toward the side of of this large-scale whole-mantle structure and its connection with the the rift with the higher rift flank topography (Ankober Border Fault); upper-mantle structures are highly debated (see Section 4.3). 36 G. Corti / Earth-Science Reviews 96 (2009) 1–53

5.5. Seismicity and distribution of current deformation

The MER is characterised by diffuse seismic activity, which is usually represented by small to intermediate size events. Historical records spanning the past 150 years show that large magnitude (MN6) earthquakes are rare throughout the rift (e.g., Gouin, 1979; Asfaw, 1990; Kebede and Kulhanek, 1991; Foster and Jackson, 1998; Ayele and Kulhanek, 2000; Hofstetter and Beyth, 2003; Keir et al., 2006a,b), although seismic analysis suggests a maximal expected magnitude of ~7, with a return period of about 80 yr (Horsfetter and Beyth, 2003). Despite the widespread seismic activity, a comparison of the expected released seismic moment and observed seismic moment for the period 1960–2000 suggests that less than 50% of extension across the MER may be accommodated by rapid slip on faults (Hofstetter and Beyth, 2003). Thus most of the extensional deformation (N50%) seems to be accommodated aseismically, although the time window over which seismicity data are analysed is rather short to justify statistically significant conclusions since earthquake catalo- gues of very limited duration may be not representative of long-term tectonic processes. Detailed data about the seismicity of the Northern and Central MER have been acquired in the period from October 2001 to January 2003 by the EAGLE network of seismic stations (Keir et al., 2006a,b) and during 2000–2002 by the stations of the Ethiopia Broadband Seismic Experiment (Brazier et al., 2008). These data show that the seismic activity in the Northern and Central MER is typically localised within Wonji segments (Keir et al., 2006a; Fig. 38), pointing to a close correlation between seismicity and alignments of Quaternary volcano-tectonic zones. However, this does not occur for the Debre Zeyt and Butajira chains of eruptive centres that have been largely aseismic during the period of data acquisition. Major boundary faults are largely aseismic, although cluster of events may characterise the border escarpments in structurally complex region that still experi- ences some strain (as the Arboye border fault system, at the intersection between the north striking Red Sea rift and the NE striking MER; Keir et al., 2006a). Historical data support the inactivity of border faults, highlighting the lack of large magnitude earthquakes on these fault systems over the last ~50 years in the Northern and Central MER (Ayele and Kulhanek, 1997). Conversely, although less constrained, seismicity in the southern part of the MER seems to be mostly concentrated in the escarpments bounding the rift (e.g., Kebede and Kulhanek, 1991). Moreover, historical data suggest that large magnitude events may also affect the plateaus surrounding the rift, as suggested by a 1906 major (M=6.8) event that ruptured the eastern shoulder of the Central MER (Ayele and Kulhanek, 2000). Clusters of earthquakes are located in narrow regions within Wonji Fig. 37. Vertical cross-sections through the P-wave velocity model of Bastow et al. fi segments, with elongation paralleling the surface expression of major (2008). (For interpretation of the references to colour in this gure legend, the reader is referred to the web version of this article.) Quaternary faults and volcanic fissures (Keir et al., 2006a). In the Northern MER, the area affected by seismic events is only 10-km wide, whereas in the Central MER seismicity is comparatively more diffuse than to the north and located within a 30- to 40-km wide zone (Fig. 38). Fantale and Dofen volcanoes (Fig. 39d; Keir et al., 2006a), although the The regions connecting the right-stepping en-echelon WFB zones are short time window of observations does not allow to draw solid largely aseismic. Most earthquake hypocentres are located at a depth of conclusions on the (long-term) characteristics of seismic activity. ~8–16 km (Fig. 39a–c), which coincides with the top of the middle- to Focal mechanism solutions indicate predominantly normal dip- lower-crustal extensive mafic intrusions (see Section 5.4); hypocentre slip on steep faults that strike approximately north to approximately depths increase with increasing distance from major volcanoes. The NNE (e.g., Foster and Jackson, 1998; Ayele, 2000; Hofstetter and Beyth, temporal distribution of seismicity in the Fantale–Dofen WFB segment is 2003; Keir et al., 2006a; Fig. 38), sub-parallel to the dominant characterised by earthquake swarms that punctuate largely aseismic orientation of WFB faults. Exceptions to these normal dip-slip focal periods; conversely, in the Central MER seismicity is relatively low and mechanisms are strike-slip or oblique-slip events, interpreted as lacks the periods of swarm activity observed further north close to the resulting from a left-lateral motion on approximately NE to

Fig. 38. Upper panel: seismic activity of the Horn of Africa since 1960. Triangles indicate Quaternary volcanoes. Lower panel: Seismicity of the Northern and Central MER from October 2001 to January 2003, as recorded from the EAGLE network of seismic stations. Heavy black lines show major border faults; dashed lines mark Quaternary volcano-tectonic segments. The star shows the location of the October 2003 earthquake swarm near Dofen volcano. Fault plane solutions are lower hemisphere projections. The size of the solution is scaled to magnitude between ML 1.17 and 5.3. All panels are modified from Keir et al. (2006a). G. Corti / Earth-Science Reviews 96 (2009) 1–53 37 38 G. Corti / Earth-Science Reviews 96 (2009) 1–53 approximately ENE striking faults (Foster and Jackson, 1998; Ayele, reactivation of the weakness zone occurred at the eastern edge of the 2000; Keir et al., 2006a). Left-lateral strike motion is also observed in upwelling mantle not above its centre (e.g., Bastow et al., 2008), as focal mechanisms located in between magmatic segments, near documented in other flood-basalt provinces (Courtillot et al., 1999) segment tips (especially near Fantale and Boseti; D. Keir, personal and suggested by theoretical modelling (Tommasi and Vauchez, communication); this kinematics is consistent with the transcurrent 2001). The rheological and dynamical effects of the warm plume component of motion in the regions of complex deformation connects material may have persisted in the last 30 Ma (e.g., Dugda et al., 2007), Wonji segments predicted by analogue models (Corti, 2008). Stress thinning and weakening the lithosphere and leading to the reactiva- inversion from these focal mechanisms indicates that the trend of the tion of the pre-existing weakness when external boundary conditions minimum principal stress is N103°E, coherent with sub-E–W direction (e.g., plate-boundary forces) allowed the initiation of extensional of extension (see Section 5.1; Keir et al., 2006a). deformation (e.g., Courtillot et al., 1999; Coulié et al., 2003). Notably, Laser ranging and GPS (geodetical) observations in the rift zone reactivation of an inherited weakness can explain the development of support the seismic observations and indicate that about 80% of the a narrow rift (sensu Buck, 1991; i.e., extensional deformation localised extensional deformation is currently accommodated in correspondence to in regions 80–100 km wide) in a weak lithosphere characterised by the narrow regions of seismic activity, along the en-echelon faults of the presence of a hot lower crust and uppermost mantle, with the absence Wonji fault belt (Fig. 40; Billham et al., 1999). The measured opening of a strong lithospheric mantle (Fig. 41). This rheological layering, velocity (~4 mm/yr) is nearly a factor of two slower than the opening rate indeed, should prevent the localisation of strain and result in a estimated from global plate motion models (see Section 5.1). Geodetical distributed deformation over very wide regions, giving rise to the so- observations by Bendick et al. (2006) suggest that rift opening near their called wide rifts (i.e., extensional deformation distributed in regions geodetic array during the period 1992–2003 was accommodated by a 1000 km wide; Buck, 1991; Brun, 1999). A pre-existing weakness can single dike injection event in 1993, spatially coincident with active solve this mechanical inconsistency by localising deformation since magmatic segments. This supports that extension within the rift is the early stages of extension and leading to a lithosphere necking (e.g., accommodated at least partially by aseismic dike injection. Sokoutis et al., 2007; Dyksterhuis et al., 2007). The development of the All these observations suggest that slip on faults appears to have EARS as a narrow rift in regions characterised by vastly different slowed in favour of dike injection associated with magmatic segments in lithospheric strength profiles (i.e., weak lithosphere in the MER and the past ~1.8 Ma (e.g., Keir et al., 2006a) and that a combination of strong lithospheres in the eastern and western branches of the EARS dyking and normal faulting within the Wonji segments appears to be the to the south; Fig. 41) indicates that inherited structure, not rheological predominant mechanism of extension in the MER (see Section 6.3). layering, may be the primary control on the mode of extension in these continental rifts (Keranen et al., 2009). 6. The evolution of the Main Ethiopian Rift: Continental rifting Results of extensive field work, summarised in Section 5.2.1., from initiation to incipient break-up suggest a diachronous development of the different MER segments (e.g., Bonini et al., 2005): rift propagation was not a smooth process 6.1. Rift initiation: Localisation and propagation of extensional structures but rather a process with punctuated episodes of extension and relative quiescence (see also Keranen and Klemperer, 2008). 6.1.1. Rift localisation and propagation Deformation started in the Early Miocene (20–21 Ma) in the Southern Analysis of continental rifts worldwide suggests that “in the same MER, where activation of the N–S trending structures was most probably way that tensile cracks in the side of a brick building generally follow related to a northward propagation of the Kenya Rift-related deformation the mortar between bricks, rifts initially follow the weakest pathways (e.g., Bonini et al., 2005); this event probably lasted until 11 Ma, then in the pre-rift materials” (Versfelt and Rosendahl, 1989). Thus, since deformation in the area decreased radically. No major extensional continental rifting results from the application of extensional stresses deformation apparently affected the Central and Northern MER segments to a pre-deformed, anisotropic lithosphere, rift structures are not between 21 and 11 Ma, whereas Africa–Arabia separation in the Red Sea randomly distributed but tend to follow the trend of pre-existing and Gulf of Aden system was active since about 30 Ma (e.g., Wolfenden et weaknesses (such as ancient orogenic belts) avoiding stronger regions al., 2005). Deformation in the Northern MER started in the Late Miocene (such as cratons; e.g., Dunbar and Sawyer, 1989; Versfelt and at ~11 Ma (Wolfenden et al., 2004). In the Central MER, recent analysis by Rosendahl, 1989; Tommasi and Vauchez, 2001; Corti et al., 2003; Bonini et al. (2005) indicate that no major extensional faulting was active Ziegler and Cloetingh, 2004). Recent geophysical investigations in the intheLateMiocene,astestified by the emplacement at 12–8Maof MER support this observation and suggest that rift location and initial extensive basalts currently outcropping on both rift margins and evolution has been probably controlled by a NE–SW-trending litho- interpreted as representing a pre-rift flood-basalt phase. According to spheric-scale pre-existing heterogeneity (Bastow et al., 2005, 2008; this study, extensional deformation in the Central MER started at about Keranen and Klemperer, 2008; Corti, 2008; Keranen et al., 2009). This 5–6 Ma, although WoldeGabriel et al. (1990) suggest an earlier lithospheric weakness zone corresponded to a suture zone separating development of boundary faults in the area in the Late Miocene (at 8– two distinct Proterozoic basement terranes, underlying the Ethiopian 10 Ma). Further to the south, in the Southern MER, the early (mid- and Somalian plateaus (Fig. 34; Keranen and Klemperer, 2008). Miocene) extensional structures were probably reactivated during Late Extensional deformation occurred in correspondence to an upraising Pliocene–Pleistocene times roughly concomitant with (or slightly hot mantle material, possibly connected to the African Superplume preceding) the development of the Ririba and Marsabit volcanic found in the lower mantle beneath southern Africa. Typically, the lineaments between the Ethiopian and Kenyan domes (WoldeGabriel

Fig. 39. Characteristics of October 2001–January 2003 seismic events in the Northern Main Ethiopian Rift (from Keir et al., 2006a). a) Earthquakes in the Ankober region and Fantale– Dofen magmatic segment, plotted on 90 m resolution SRTM topographic data. b) profiles A–A′ and B–B′ project earthquakes within 30 km of the line of section onto the profile. The thickened portions of the profiles show where the profile crosses the Fantale–Dofen magmatic segment. c) Histograms of number of earthquakes per 4 km depth interval for the Fantale – Dofen magmatic segment (left) and Ankober region (right), redrawn from figure 7 of Keir et al. (2006a). Of the 2139 local earthquakes recorded at four or more EAGLE stations, only 280 well-located events were used for the plots and statistical analysis; tests on hypocenter accuracy gave estimates of about ±600 m in horizontal direction and ±2000 m in depth for these earthquakes (Keir et al., 2006a). Note that the graph of number of earthquakes with depth may not be an entirely representative measure of depth-distribution of seismicity, as for instance may be the total energy released by the earthquakes with depth (e.g., Panza and Raykova, 2008). d) Cumulative number of earthquakes versus recording time of the regions 1 (Ankober), 2 (Fantale–Dofen), and 3 (south of Aluto–Gedemsa), indicated in the inset of panel a. e) Results of the stress inversion using 36 focal mechanisms in MER from

EAGLE stations. Circle shows σ3, the minimum compressive stress. Square shows σ2, the intermediate compressive stress. Triangle shows σ1, the maximum compressive stress. The 95% conﬁdence limits are shown by regions of gray shading. G. Corti / Earth-Science Reviews 96 (2009) 1–53 39 40 G. Corti / Earth-Science Reviews 96 (2009) 1–53

Southern MER during the Late Pliocene–Pleistocene (Fig. 42). This process, which is in agreement with the north–south decrease in extension-related crustal and lithospheric modification imaged by geophysical–geological data, has been interpreted as triggered by a counterclockwise rotation of the Somalian Plate starting at around 10 Ma (Fig. 42; e.g., Bonini et al., 2005). Keranen and Klemperer (2008) build on this model to suggest a more complex southwards propagation of extensional deformation (Fig. 43b). According to these Authors, rift propagation from north to south began smoothly in the Northern MER at ~11 Ma, then stalled upon reaching the Northern–Central MER boundary and was diverted Fig. 40. Velocity in the N135°E direction and distribution of current deformation across westwards along the YTVL. Extension eventually affected the Central the rift from geodetic data (after Billham et al., 1999). MER in the Late Miocene/Pliocene, then connecting the Northern MER with the Southern MER. A rotation of the stress field (i.e., of the direction of plate motion) would have made extension along the YTVL and Aronson,1987; WoldeGabriel et al., 1990; Ebinger et al., 2000; Bonini less favourable and extension along the Central MER more favourable et al., 2005; Vetel and Le Gall, 2006). Currently, the MER and the Kenya (see Keranen and Klemperer, 2008). Rift remain kinematically linked across a 200-km-wide zone represent- Alternatively, Rooney et al. (2007) suggest that the MER records ing the eastern part of the Broadly Rifted Zone of south Ethiopia (Ebinger the unification of two rift systems, the Red Sea Rift and the East et al., 2000), which has been interpreted as a regional transfer zone African Rift System (Fig. 43c). The Red Sea-related deformation affects separating the two major rifts (Corti et al., 2002). the Northern MER and propagates southwards from the Afar region, Overall, the diachronous development of the different MER whereas the East African-related affects the Southern and Central MER segments has been interpreted in terms of different models of rift and propagates to the north. These two different structural and propagation (Figs. 42, 43). tectonic domains (Red Sea and East African Rifts) now link and At one end-member, Wolfenden et al. (2004) suggest that the interact within the Central MER (Rooney et al., 2007). MER-related deformation initiated at ~18 Ma in southwestern Further analysis is thus required to better define the timing of the Ethiopia, and propagated northwards through the Central and North- main volcano-tectonic events in the different rift segments and the ern MER into the Afar depression after ~11 Ma (Fig. 43a). modalities of rift development in the MER. At the other end-member, Bonini et al. (2005) suggest a Miocene- recent southwards rift propagation from the Afar depression, with 6.1.2. The Red Sea, Gulf of Aden, and Ethiopian rifts triple junction history deformation affecting the Northern MER in the Late Miocene, the The present-day Red Sea–Gulf of Aden–Main Ethiopian rift triple Central MER in the Pliocene and then further propagating into the junction lies in a complex zone at ~11.5°N along the Tendaho-Goba'ad

Fig. 41. Graphs showing the thermal structure (a), strength proﬁles (b) and integrated resistance of the lithosphere (c) of the Eastern branch of the East African Rift System (blue curves) and adjacent to the Main Ethiopian Rift (red curves), beneath the Ethiopian Plateau (from Keranen et al., 2009). Note the difference in geotherms and strength envelopes, which results in an integrated strength of the lithosphere surrounding the Eastern branch an order of magnitude higher than surrounding the MER. Also note that no lithospheric mantle is predicted below the Ethiopian Plateau. See Keranen et al. (2009) for details about the calculations. (For interpretation of the references to colour in this ﬁgure legend, the reader is referred to the web version of this article.) G. Corti / Earth-Science Reviews 96 (2009) 1–53 41

Discontinuity (TDG) within the central Afar depression (e.g., Tesfaye incipient Nubia–Somalia plate boundary. Nubia–Arabia separation et al., 2003). Models for the formation of this archetypal rift–rift–rift and Nubia–Somalia rifting are thus distinct and diachronous events, triple junction have assumed the synchronous development of the thus ruling out the application of the archetypal rift–rift–rift triple three rift arms soon after flood-basaltic magmatism at ~30 Ma. junction model above the Afar mantle plume (see Wolfenden et al., Tesfaye et al. (2003) suggested that a paleo-triple junction first 2004). formed in an arcuate accommodation zone at latitude of ~10° N in the Thus, although Tesfaye et al. (2003) individuated an early (pre- Oligo-Miocene, then migrated to the NE in the TDG at ~11.5°N latitude. 11 Ma) triple junction at 10°N latitude, the data by Wolfenden et al. The chronology of initial rifting in the different rift segments sum- (2004) and Bonini et al. (2005) argue against the development of a marised above suggests however a more complex scenario with a triple junction in the Afar depression before ~11 Ma (Fig. 44). Rather, diachronous development of the link area between the different Wolfenden et al. (2004, 2005) showed that the southern Red Sea rift systems (Fig. 44). Particularly, development of southern Afar at existed as far south as 10°N at 27 Ma, suggesting that the arcuate ~11Ma (Wolfenden et al., 2004) and Main Ethiopian Rift at ~5–6Ma accommodation zone (paleo-triple junction) of Tesfaye et al. (2003) (Bonini et al., 2005)or8–10 Ma (WoldeGabriel et al., 1990), occur- marked the southern termination of the Oligocene Red Sea rift red well after the flood-basaltic magmatism and the beginning of (Fig. 44). Later development of the MER created a triple junction at rifting in the Gulf of Aden–Red Sea systems at ~30 Ma (Watchtorn 11 Ma in this area; then the triple junction migrated north–eastwards et al., 1998; Wolfenden et al., 2005). Consequently, the flood-basaltic to the present-day TDG (Wolfenden et al., 2004), most probably in magmatism and separation of Arabia from Africa are widely separated response to the along-axis propagation of the Gulf of Aden and Red in time from the opening of the Main Ethiopian rift, which marks the Sea spreading centres.

Fig. 42. Cartoon showing the schematic evolution of the main rift segment structures at the northern termination of the East African Rift System according to Bonini et al. (2005). Black arrows indicate the direction of propagation of the extensional deformation. In (e) and (f), the red numbers indicate the approximate timing ofonsetof extensional activation, or reactivation, of structures. MER: Main Ethiopian Rift; SA: Southern Afar; SMER: Southern Main Ethiopian Rift; YTVL: Yerer Tullu–Wellel volcano- tectonic lineament. Modiﬁed from Bonini et al. (2005). (For interpretation of the references to colour in this ﬁgure legend, the reader is referred to the web version of this article.) 42 G. Corti / Earth-Science Reviews 96 (2009) 1–53

Fig. 43. Models for the propagation of extensional deformation in the Main Ethiopian Rift. a) Model of northward propagation of deformation from southwestern Ethiopia to the Afar depression (Wolfenden et al., 2004). b) Complex rift propagation as hypothesised by Keranen and Klemperer (2008). In this model, rifting propagated smoothly from Afar into the Northern MER at ~11 Ma, then stalled upon reaching the Northern–Central MER boundary and was diverted westwards along the Yerer–Tullu Wellel Volcanotectonic Lineament (YTVL). Extension eventually affected the central MER in the Late Miocene/Pliocene, connecting the Northern MER with the Southern MER. In this model, a rotation of the stress ﬁeld is responsible for the migration of extensional deformation from the YTVL to the central MER. c) Alternative model proposed by Rooney et al. (2007) in which the MER records the propagation of two rift systems, the Red Sea Rift and the East African Rift System. At ~10 Ma the Red Sea-related deformation (represented by the Wonji Fault Belt) is propagating southwards from Afar into the Northern MER, whereas the East African-related deformation (represented by the Silti–Debre Zeyt volcano-tectonic alignment) affects the Southern and Central MER and propagates to the north. At this stage, deformation has not affected the Boru Toru Structural High (BTSH) that separates the central from the Northern MER. Further extension led to breaching of the BTSH by the Wonji Fault Belt and to interaction of the two different structural and tectonic domains (Red Sea and East African Rifts) within the central MER (Rooney et al., 2007).

6.1.3. Activation of boundary fault systems characterised by a general en-echelon arrangement. These faults gave The initial phases of continental rifting in the MER corresponded to rise to major fault-escarpments separating the rift depression from the activation of long, widely-spaced and large-offset boundary faults, the Ethiopian and Somalian plateaus (Fig. 45). Deformation on these G. Corti / Earth-Science Reviews 96 (2009) 1–53 43

(Fig. 45). At about 2 Ma, the deformation style changed dramatically: extension shifted from the few widely-spaced boundary faults with large vertical displacements, to the right-stepping arrangement of dense fault swarms with small vertical offset of the Wonji segments obliquely affecting the rift ﬂoor (Fig. 45). This shift was concomitant with a focusing of Quaternary volcanic activity within the rift depression along the Wonji faults, giving rise to magmatic segments with only minor activity outside the en-echelon deformation belt these fault zones. This process is well expressed in the Northern MER, where ﬁeld and geophysical data suggest that large-offset boundary faults are no longer active and most of deformation is accommodated within the oblique fault segments, where focusing of magmatic activity is observed at all lithospheric levels in the upper ~75 km. In the Central MER, this process may be in a less evolved stage, with Wonji segments that start to propagate from the rift margins towards the rift centre, as illustrated by analogue models (Corti, 2008). Several models have been applied to explain the transition from the boundary to the Wonji faults (Figs. 46, 47).

(1) Chorowicz et al. (1994) related the development of the Wonji faults as due to the reactivation of pre-existing fabrics (Fig. 46a). However, as noted by Bonini et al. (1997), with this hypothesis it is difficult to explain why the weakness were not reactivated during the initial stages of rifting but later during rift evolution. The rift lithosphere was indeed already strongly attenuated at the time of WFB strain initiation, and any pre-existing crustal structures would largely have been modified or obliterated by “pre-stressing” (Mohr, 1983). Moreover, as noted by Mohr (1983) the uniformity of the WFB orientation over a distance of 1000 km argue against a reactivation of inherited fabrics but rather supports its development as a response of the lithosphere to a regional stress field. Fig. 44. Tectonic development of the Afar triple junction, illustrated with snapshots of (2) Polyphase kinematics models (e.g., Bonini et al.,1997; Boccaletti representative stages (after Wolfenden et al., 2004). a) Between 35 and 27 Ma, et al.,1999a,b; Wolfenden et al., 2004) suggested a change in the continental rifting commences in Red Sea and Gulf of Aden. Arcuate accommodation Nubia–Somalia kinematics to explain the change in deformation zone (AAZ) of Tesfaye et al. (2003) marks previous location of southwestern Arabia. style with the activation of the differently-oriented boundary b) Rifting continues in the Red Sea, and seafloor spreading has commenced in the eastern Aden rift. Extension between Nubia and Danakil microplate may have initiated. and Wonji fault systems (Fig. 46b). In these models, a Mio- c) After 11 Ma, extension in the Main Ethiopian rift initiates to form a triple junction for Pliocene phase of NW–SE-directed orthogonal extension con- the first time. Greatest stretching has occurred in southern Afar, where some oceanic trolled the development of the boundary faults, whereas a – crust may have been created by 8 Ma. d) Then the triple junction migrated north Quaternary, roughly E –W extension, producing sinistral oblique eastwards to the present-day Tendaho-Goba'ad Discontinuity (TDG; Wolfenden et al., 2004), most probably in response to the along-axis propagation of the Gulf of Aden and rifting, activated the en-echelon Wonji segments; localisation of Red Sea spreading centres. magmatic activity along the Wonji faults is normally interpreted as a passive feature in these models (e.g., Corti et al., 2003). Geodetic, seismic and stress-field data, as well as field geological faults was accompanied by progressive subsidence of the rift investigations, confirm a current oblique rifting kinematics (see depression and the development of locally asymmetric basins (as Sections 5.1 and 5.2.2.), but recent plate kinematics models shown for instance by seismic refraction studies in the Northern MER) (Royer et al., 2006) suggest that the roughly ESE–WNW with up to 5 km of syn-rift sediments accumulation. As summarised direction of extension in MER may have remained steady over above, the large-offset border fault system developed diachronously the past 11 Myr, thus apparently contradicting the polyphase along the rift between 11 and 6 Ma, has been mostly active until about kinematics models. However, the Mio-Pliocene kinematical 2 Ma and is now considered (at least in the Northern MER) largely boundary conditions of rifting remain poorly defined. inactive. This first (Mio-Pliocene) rift phase was associated to diffuse (3) According to magma-assisted rifting models (e.g., Kendall et bimodal volcanic activity, which encompassed the whole rift depres- al., 2005), the strong alteration of the lithosphere exerted by sion. Eruption of widespread volcanic products mainly occurred magmatic processes suggest a magmatic control on the through boundary faults and pre-existing weaknesses (e.g., transfer change in deformation style from the boundary faults to the zones); contemporaneously, lateral magma migration favoured magmatic Wonji segments (Fig. 46c). With the ready supply magma accumulation outside the rift valley, with development of of hot and weak magma, strain accommodation by magma large volcanic centres on the eastern and western rift shoulders (off- intrusion and dyking occurs at lower stress levels than are axis volcanism). needed to activate the large displacement faults (Buck, 2004, 2006). In these conditions, magmatic processes induce the 6.2. Rift maturity: Abandonment of boundary faults and development of abandonment of boundary faults and the activation (and Wonji segments architecture) of oblique volcano-tectonics Wonji segments. Thus, in these models, magmatism has an active role in the Once rift structure was developed in the whole MER, deformation evolution of the MER. continued (during Pliocene times) along large boundary fault systems, (4) More recently, Corti (2008) showed by using analogue models with slip on these faults giving rise to a deepening of the rift floor that the two-phase evolution with the activation of the two fault 44 G. Corti / Earth-Science Reviews 96 (2009) 1–53

Fig. 45. Two-phase volcano-tectonic evolution of the Main Ethiopian Rift (after Corti, 2008). Left panel: Mio-Pliocene deformation along the boundary fault systems generates a subsiding rift depression and asymmetric basins (as shown in the Northern MER); this deformation phase is associated to widespread volcanic activity encompassing the whole rift depression. Right panel: abandonment of boundary faults and localisation of deformation and volcanism along the Wonji segments, obliquely affecting the rift ﬂoor.

systems do not require neither a change in plate kinematics, nor mafic intrusions below major Wonji volcanoes with shape, elongation magma weakening. In these models, the reactivation of a NE– and en-echelon arrangement that follow the Quaternary tectono- SW-trending pre-existing weakness under a constant, roughly magmatic zones at surface (Keranen et al., 2004; Daly et al., 2008). ESE–WNW (~N100°E) extension induces oblique rifting condi- Petrologic data (Rooney et al., 2005) argue for the presence of a tions (Fig. 47). Oblique rifting evolves from fault-bounded molten fraction in fractures within the solidified mafic intrusions, as asymmetric basins in an initial rifting stage to later en-echelon also supported by several geophysical data (e.g., seismic analysis, regions of faulting and lithospheric thinning obliquely cutting magnetotelluric data; Dugda et al., 2005; Stuart et al., 2006; Whaler the rift floor, with trend, architecture and kinematics matching and Hautot, 2006; Daly et al., 2008). The patterns of seismic velocity those of Wonji segments in the Northern and Central MER. The anisotropy (Kendall et al., 2005; Keir et al., 2005; Kendall et al., 2006) modellingresultssuggestthusthatriftarchitectureand indicate a pervasive dyking in the upper crust and the uppermost evolution are controlled by rift obliquity and independent mantle to a depth of ~75 km. The low lower-crust and uppermost from magmatic processes: comparison of the results with mantle shear-wave velocity beneath the rift suggest high tempera- numerical models of oblique rifting with similar initial tures and the presence of melts (Keranen et al., 2009). In the boundary conditions (Van Wijk, 2005) suggests instead that uppermost crust, shallow magma bodies below major volcanoes are the en-echelon regions of lithospheric thinning are the locus of imaged by magnetotelluric analysis (Whaler and Hautot, 2006) and upraising of warm mantle material and enhanced melt produc- supported by gravity modelling (Cornwell et al., 2006). Geochemical/ tion, with ascending magmas then focused by the oblique fault petrological data (Rooney et al., 2005, 2007) have been used to segments (Figs. 45, 47; Corti et al., 2003, 2004). suggest a well-developed magmatic system beneath the Wonji segments in the Northern MER, where magmas can rise quickly It is likely that the progressive thinning of the continental lithosphere from melt generation zones at ~50–90 km before undergoing more under constant, prolonged oblique rifting conditions may have significant fractionation at near surface levels; however, as stated controlled this migration of deformation and the resulting Quaternary above (see Section 5.3.1) the modalities of magma migration are still volcano-tectonic segmentation, although it cannot be excluded that the not well defined. weakening related to magmatic processes and/or a change in rift Overall, the extensive magma intrusion imaged by EAGLE kinematics may have contributed to the change in deformation style geophysical data suggests that as soon as the deformation shifts (see above points). As the magmatic activity focused on Wonji faults, the from boundary to Wonji faults and the volcano-tectonic activity lateral magma migration feeding off-axis volcanism ceased consistent focuses on the en-echelon magmatic segments, a strong interaction with the cessation (or strong reduction) of this volcanic activity at the between magmatic and deformation processes develops. Extension end of the Pliocene (Corti et al., 2003, 2004; Bonini et al., 2005), is – within the Wonji magmatic segments – accommodated by a although some boundary faults may remain important in controlling the combination of magmatic intrusion, dyking and faulting (e.g., Ebinger location of mantle upwellings and melt migration pathways, as recently and Casey, 2001; Keranen et al., 2004; Kendall et al., 2005; Keir et al., imaged by geophysical data (Bastow et al., 2008). 2006a), as supported by several evidences. A comparison of the expected released seismic moment and observed seismic moment in 6.3. Wonji magmatic segments and continental break-up the period 1960–2000 shows that less than 50% of extension across the Northern MER is accommodated by rapid displacement on faults Geophysical data from the Northern MER evidence the extensive (Hofstetter and Beyth, 2003), although the seismic activity is analysed presence of magma beneath the Wonji segments from melt genera- over a too short time interval and thus may be not necessarily tion zones up to the surface (see Section 5.5). Magma intrusion has a representative of ongoing tectonic processes. GPS measurements segmented nature and occurs in right-stepping en-echelon pattern, show that approximately 80% of current extensional deformation approximately mimicking the 20-km wide, 60-km-long surface across the Northern MER is accommodated in a ~20-km-wide zone of segmentation of Quaternary volcanic centres and faults. Seismic faulting and volcanism within Wonji segments (Billham et al., 1999). tomography images the presence of mid- to lower-crustal cooled The measured velocity is however much lower than that predicted by G. Corti / Earth-Science Reviews 96 (2009) 1–53 45

Fig. 46. Different models applied to explain the transition from the boundary to the Wonji faults: a) reactivation of NNE–SSW-trending pre-existing fabrics (Chorowicz et al., 1994); b) polyphase plate kinematics (e.g., Bonini et al., 1997; Boccaletti et al., 1999a; Wolfenden et al., 2004); c) magma-assisted rifting (modified after Ebinger, 2005). See text for details. plate kinematic models. Seismicity data support localisation of Keir et al., 2006a) from magma generation zones in the upper mantle. deformation in narrow zones in the brittle upper crust: elongate Seismicity is mostly concentrated in the upper ~10 km above axial clusters of earthquakes are associated with observed faults, fissures mafic intrusions, where extension is accommodated by both dyke and active eruptive centres in the Wonji segments (Keir et al., 2006a). intrusion and faulting (Fig. 48; Keir et al., 2006a). Earthquakes and Low-magnitude earthquakes are concentrated at 8–14 km depth rapid fault slip may be induced by injection of dykes, with swarms of which coincides with the top of the ~20- to 30-km-wide zone of seismicity spatially reflecting areas of magma intrusion (Keir et al., extensive mafic intrusions imaged beneath Wonji segments at 8– 2006a). Analysis of the spatio-temporal distribution of seismic events 10 km depth (Keranen et al., 2004). Seismic anisotropy of the upper suggests that episodic rifting events within one magmatic segment crust is highest in the magmatic segments and attributed to melt- are independent of other magmatic segments, in turn arguing for filled cracks and dikes aligned perpendicular to the minimum stress spatially and temporally discrete magma source regions (Keir et al., (Keir et al., 2005), supporting petrological data (Rooney et al., 2005, 2006a). This is in agreement with models of dykes fed by magma 2007). Overall, these data indicate that below a depth of ~10 km, chambers and propagating sub-parallel to the segment axis and extension is mostly aseismic and controlled by magma injection in a normal to the extension direction (Ebinger and Casey, 2001), as also hot, ductile middle to lower crust (Fig. 48; e.g., Keranen et al., 2004; indicated by characteristics of Quaternary faults within Wonji 46 G. Corti / Earth-Science Reviews 96 (2009) 1–53

Fig. 47. Analogue modelling results of fault evolution during oblique rifting (30° of obliquity; after Corti, 2008). a) Initial boundary conditions and rheological stratification of the model (for details see Corti, 2008). b–c) Top-view photos of the experiment during the initial phases of rifting corresponding to the activation of large boundary faults. Reported are the plots of weighted fault azimuths; εh1, εh2: direction of the maximum and secondary horizontal extensional strain. d–e) Top-view photos of the experiment during the subsequent phases of rifting corresponding to the activation of internal faults. Fault distribution illustrated as above. Note that the two-phase evolution of rifting does not require a change in rift kinematics but results from a constant roughly ESE–WNW (~N100°E) extension. f) Structural scheme. g) Results of numerical models of oblique rifting with initial conditions similar to those of analogue models (30° of obliquity and similar rheological layering; after Van Wijk, 2005), illustrated as crustal thinning factors (calculated as ratio between the initial and final model thickness) in map view after 4 My of model evolution. Boundaries of the weak trend are super-positioned (black dashed lines). Similarly to analogue models, numerical modelling results show the development of en-echelon regions of maximum lithospheric thinning; these regions follow the weak trend location as a group, but are individually oriented according to the extension direction (i.e., rotated with respect to the weak trend). As in the analogue models, rift segmentation results from the oblique rifting kinematics and is not influenced by magmatic processes; rather, warm mantle material wells up below the en-echelon segments of strong lithospheric thinning (Van Wijk, 2005).

segments (e.g., Casey et al., 2006; Kurz et al., 2007). A similar process The combination of magmatic intrusion, dyking and faulting at all has been observed in southern Afar (2005-present Dabbahu dyking lithospheric levels within magmatic segments fundamentally modifies episode), where injection of lateral dikes fed from magma reservoirs the rheological properties of the crust and uppermost mantle beneath in middle to lower crust or upper-mantle beneath rift segment centres the rift. Buck (2004, 2006) has shown that dyke intrusion throughout is a key component in creating and maintaining the regular along-axis the lithosphere is able to reduce the yield stress by an order of rift segmentation during the final stages of continental rupture magnitude (Fig. 49), leading to the strong strain localisation within (Wright et al., 2006; Keir et al., 2009). Repeated episodes of magma magmatic segments supported by the geodetical and seismicity data injection into the crust may explain the strain deficit indicated by the summarised above. With the ready supply of hot magma, the process gap between expected and observed seismic moment and by the of strain localisation, rift axial injection of magma and accompanying discrepancy in plate separation velocity predicted by plate kinematic lithospheric strength reduction via localised heating is self-reinforcing, models and measured with GPS. facilitating the break-up of the continental lithosphere (e.g., Hayward G. Corti / Earth-Science Reviews 96 (2009) 1–53 47

Fig. 48. Combination of magmatic intrusion, dyking and faulting within magmatic segments and relations with the depth-distribution of seismic activity. Below a depth of ~10 km, extension is mostly aseismic and controlled by magma injection in a hot, ductile middle to lower crust and uppermost mantle (e.g., Keranen et al., 2004; Keir et al., 2006a). Seismicity is mostly concentrated in the upper ~10 km above axial maﬁc intrusions, where extension is accommodated by both dyke intrusion and faulting (Keir et al., 2006a).

and Ebinger, 1996; Ebinger, 2005). The characteristics of magmatic between the MER and the Woodlark Rift in Papua-New Guinea, a segments outlined above point indeed to a new, shorter, narrower and highly evolved backarc rift where break-up has occurred without magmatic along-axis segmentation evident at all lithospheric level extrusive volcanism, suggests a similar pattern of extensive mafic developed within the broad rift depression in the last ~2 Ma. The addition to the crust within a narrow zone of localised strain, despite morphology and architecture of this new tectono-magmatic segmen- the crustal stretching is much more limited in MER than in the tation, defined by 60–80 km long, 20 km-wide zones of magmatic Woodlark basin (see Daly et al., 2008 and references therein). The MER intrusions at depth and faulting and dyke intrusion in the upper crust magmatic segments, therefore, suggest that transitional crust can be in the centre of the rift valley, is comparable to slow-spreading mid- produced at relatively small stretching factors where basaltic melt is oceanic ridges (e.g., Ebinger and Casey, 2001; Keranen et al., 2004; available. The uppermost mantle beneath the MER has however not yet Ebinger, 2005; Daly et al., 2008). For instance, the crustal structure organised itself into a series of punctuated magmatic upwellings as is beneath the magmatic segments in the Northern MER shares striking expected beneath a fully developed seafloor spreading centre; similarities with that shown by the Northern Neovolcanic Zone in punctuated upwellings in the mantle are probably only occurring Iceland, consisting of rift segments arranged en echelon along the Mid northwards beneath the Afar depression where break-up and spread- Atlantic ridge plate boundary (see Daly et al., 2008 and references ing are in a more advanced stage (Bastow et al., 2008). therein). This suggests that the Wonji magmatic segments in the Rooney et al. (2007) relate this incipient spreading in the Northern Northern MER may represent a precursor to seafloor spreading MER to the southwards propagation of the Red Sea Rift system, now centres, developing within a lithosphere that is transitional between interacting with the EARS-related deformation in the Central MER. continental and oceanic (e.g., Ebinger and Casey, 2001; Keranen et al., Irrespective of the of the propagation model (see Section 6.1.1), 2004; Rooney et al., 2007; Daly et al., 2008). Comparison of the crustal geophysical indications (such as thicker crust, lower magnitude

Fig. 49. Different ways to extend normal-thickness continental lithosphere. Upper panel: tectonic stretching; lower panel: magmatic stretching. Note the large difference in yield stress (the stress difference needed to get extensional separation of two lithospheric blocks) with and without magma injection. After Buck (2004, 2006) and Ebinger (2005). 48 G. Corti / Earth-Science Reviews 96 (2009) 1–53 upper-crustal velocities and higher uppermost mantle density; see geophysical data (see Sections 5.2–5.4), together with a systematic Section 5.4) together with rift marginal axes of extension (instead of a variation of magmatism (northward increase in volume of fissural single rift-centred volcano-tectonic axis) and – possibly – a more basalts, with onset of voluminous basaltic volcanism in Afar) suggests complex magmatic system (see Section 5.3.2) suggest that the Central indeed a transition from continental rifting stages in the Southern/ and Southern MER are characterised by a lower degree of extension Central MER to oceanic spreading in Afar (e.g., Hayward and Ebinger, and magmatic modification of a lithosphere with continental affinities 1996). (Rooney et al., 2007). This in turn suggests that the temporal evolution of deformation (from incipient rifting and boundary fault activity to 7. Conclusions magmatic segments and continental break-up) summarised above for the Northern MER is also currently spatially observable in the different Fig. 50 summarises the different stages of the evolution of the Main MER/Afar segments, whose characteristics of structure/magmatism Ethiopian Rift (MER), from rift initiation to break-up and incipient reflect different stages in the development of a continental rift. The oceanic spreading (see also Ebinger, 2005). The first tectono- general south to north decrease in the amount of tectono-magmatic magmatic event related to the Tertiary rifting was the eruption of modification of the crust and the lithosphere indicated by geological/ voluminous flood basalts, which seems to have occurred in a rather short time interval at around 30 Ma; strong plateau uplift, resulting in the development of the Ethiopian and Somalian plateaus now surrounding the rift valley, has been suggested to have initiated contemporaneously or shortly after the extensive flood-basalt volcanism, although its exact timing remains controversial. Volumi- nous volcanism and uplift started prior to the main rifting phases, suggesting a mantle plume influence on the Tertiary deformation in East Africa, possibly in connection with a hypothetical African superplume. The main rifting phases started diachronously along the MER in the Mio-Pliocene with a discontinuous rift propagation interpreted in terms of southwards or northwards rift propagation or more complex evolutive scenarios. Rift location was most probably controlled by the reactivation of a lithospheric-scale pre-existing weakness; the orientation of the weakness (roughly NE–SW) and the Late Pliocene (post 3.2 Ma)-recent Nubia–Somalia relative motion (roughly ESE–WNW) suggests that oblique rifting conditions have controlled rift evolution. However, it is still unclear if these kinematical boundary conditions have remained steady since the initial stages of rifting or the kinematics has changed during the Late Pliocene or at the Pliocene–Pleistocene boundary. Continental rifting in the MER evolved in two different phases. An early (Mio-Pliocene) continental rifting stage (Fig. 50b) was characterised by displacement along large boundary faults, subsidence of rift depression with local development of deep (up to 5 km) asymmetric basins and diffuse magmatic activity. In this initial phase, magmatism encompassed the whole rift, with volcanic activity affecting the rift depression, the major boundary faults and limited portions of the rift shoulders (off-axis volcanism). Progressive extension led to the second (Pleistocene) rifting stage (Fig. 50c–d), characterised by a riftward narrowing of the volcano- tectonic activity. In this phase, which is well expressed in the Northern MER the main boundary faults were deactivated and extensional deformation was accommodated by dense swarms of faults (Wonji

Fig. 50. Schematic model of rift evolution in the Main Ethiopian Rift (modified after Ebinger, 2005). a) Flood-basalt event affecting Ethiopia before the main extensional events. Note the presence of an inherited weakness zone that localises rifting in the subsequent deformation phases. b) Activation of large boundary faults (11–2 Ma), giving rise to major fault-escarpments and rift-floor subsidence. This first (Mio- Pliocene) rift phase was associated to diffuse volcanism, which encompassed the whole rift depression. c) Abandonment of large boundary faults and shift of deformation within the rift valley, with activation of the oblique Wonji Fault Belt (~2 Ma). This riftward narrowing of deformation does not require a change in Nubia–Somalia kinematics and is independent of magmatic processes (Corti, 2008). Rather, the oblique deformation zones focus magmatic activity that becomes then localised along Wonji segments (d). As soon as, the volcano-tectonic activity is localised within Wonji segments, a strong feedback between deformation and magmatism develops: the thinned lithosphere is strongly modified by the extensive magma intrusion and extension is facilitated and accommodated by a combination of magmatic intrusion, dyking and faulting. e) With further thinning, heating and magma intrusion, the tectonically and magmatically thinned lithosphere may rupture in the heavily intruded zones, and new oceanic lithosphere created. The solidified mafic intrusions in the mid- to lower crust and the thick piles of lavas in the magmatic segments load the weak plate, flexing it towards the new ocean basin to form seaward-dipping lavas. The passive margin subsides as heat transferred from the asthenosphere dissipates (Ebinger, 2005). G. Corti / Earth-Science Reviews 96 (2009) 1–53 49 segments) in the thinned rift depression. The progressive thinning of the Abebe, B., 1993. Studio geologico-strutturale del Rift Etiopico a sud di Asela. Ph. D dissertation, University of Firenze, Firenze, Italy, 153 pp. continental lithosphere under constant oblique rifting conditions, Abebe, T., Mazzarini, F., Innocenti, F., Manetti, P., 1998. The Yerer–Tullu Wellel possibly aided by magmatic processes and/or a change in rift volcanotectonic lineament: a transtentional structure in Central Ethiopia and the kinematics, may have controlled this migration of deformation and associated magmatic activity. Journal of African Earth Sciences 26, 135–150. Abebe, T., Manetti, P., Bonini, M., Corti, G., Innocenti, F., Mazzarini, F., Pecskay, Z., 2005. the resulting Quaternary volcano-tectonic segmentation. Owing to the Geological map (scale 1:200000) of the northern main Ethiopian rift and its oblique rifting conditions, the fault swarms obliquely cut the rift floor implication for the volcano-tectonic evolution of the rift. Geological Society of and are characterised by a typical right-stepping arrangement. Warm America, Boulder Colorado, USA, Maps and Charts series, MCH094. mantle material uprose below these en-echelon segments of strong Abebe, B., Acocella, V., Korme, T., Ayalew, D., 2007. Quaternary faulting and volcanism in the Main Ethiopian Rift. Journal of African Earth Sciences 48, 115–124. lithospheric thinning; ascending magmas were focused by the Wonji Acocella, V., Korme, T., 2002. Holocene extension direction along the Main Ethiopian segments, with eruption of magmas at surface preferentially occurring Rift, East Africa. Terra Nova 14, 191–197. along the oblique faults. As soon as the volcano-tectonic activity was Acocella, V., Korme, T., Salvini, F., Funiciello, R., 2002. Elliptical calderas in the Ethiopian Rift: control of pre-existing structures. Journal of Volcanology and Geothermal localised within Wonji segments, a strong feedback between deforma- Research 119, 189–203. tion and magmatism developed: the thinned lithosphere was strongly Acocella, V., Korme, T., Salvini, F., 2003. Formation of normal faults along the axial zone modified by the extensive magma intrusion and extension was of the Ethiopian rift. Journal of Structural Geology 25, 503–513. Adamson, D., Williams, M.A.J., 1987. Geological setting of Pliocene rifting and deposition facilitated and accommodated by a combination of magmatic intrusion, in the Afar Depression of Ethiopia. Journal of Human Evolution 16, 597–610. dyking and faulting. At this stage, magmatic segments in the Northern doi:10.1016/0047-2484(87)90015-7. MER act as incipient slow-spreading mid-ocean spreading centres Almond, D.C., 1986. Geological evolution of the Afro-Arabian dome. Tectonophysics 131, 301–332. sandwiched by continental lithosphere (e.g., Ebinger, 2005). With Anderson, D.L., 2007. Is there convincing tomographic evidence for whole mantle further thinning, heating and magma intrusion, a new oceanic litho- convection? Available at: http://www.mantleplumes.org/TomographyProblems. sphere will be created along the magmatic segments (Fig. 50e). html. Asfaw, L.M., 1990. Implication of shear deformation and earthquake distribution in the Overall the above-described evolution of the MER documents a East African Rift between 4 N and 6 N. Journal of African Earth Sciences 10, 745–751. transition from fault-dominated rift morphology in the early stages of Asfaw, L.M., 1992. Constraining the African pole of rotation. Tectonophysics 209, 55–63. extension toward magma-assisted rifting during the final stages of Asfaw, L.M., 1998. Environmental hazard from fissures in the Main Ethiopian Rift. – continental break-up. A strong increase in coupling between deforma- Journal of African Earth Sciences 27, 481 490. Ayalew, D., Ebinger, C., Bourdon, E., Wolfenden, E., Yirgu, G., Grassineau, N., 2006. tion and magmatism with extension is documented, with magma Temporal compositional variation of syn-rift rhyolites along the western margin of intrusion and dyking playing a larger role than faulting in strain the southern Red Sea and northern Main Ethiopian Rift. In: Yirgu, G., Ebinger, C.J., accommodation as rifting progresses to seafloor spreading (e.g., Maguire, P.K.H. (Eds.), The Afar Volcanic Province within the East African Rift System: Geological Society Special Publication, vol. 259, pp. 121–130. Ebinger and Casey, 2001; Ebinger, 2005). Ayele, A., 2000. Normal left-oblique fault mechanisms as an indication of sinistral deformation between the Nubian and Somalian plates in the Main Ethiopian Rift. Acknowledgments Journal of African Earth Sciences 31, 359–367. Ayele, A., Kulhanek, O., 1997. Spatial and temporal variations of seismicity in the Horn of Africa from 1960 to 1993. Geophysical Journal International 130, 805–810. Part of the material presented in this paper was collected for three Ayele, A., Kulhanek, O., 2000. Reassessment of source parameters for three major solicited talks given at the European Geosciences Union General earthquakes in East African rift system from historical seismograms and bulletins. fi – – Annali di Geo sica 43, 81 94. Assembly 2008 (Vienna, Austria, 13 18 April 2008), the RiftLink Baker, B.H., Mohr, P.A., Williams, L.A.J., 1972. Geology of the Eastern Rift System of Africa. Workshop (Neustadt/Weinstr., Germany, 24–2 June 2008) and the Geological Society of America Special Paper, vol. 136. 67 pp., Boulder Colorado. First Joint Meeting of S.I.M.P. – Italian Society of Mineralogy and Baker, J., Snee, L., Menzies, M., 1996. A brief Oligocene period of flood volcanism in – Petrology – and A.I.C. – Italian Crystallographic Association – (Sestri Yemen. Earth and Planetary Science Letters 138, 39 55. Barberi, F., Santacroce, R., 1980. The Afar stratoid series and the magmatic evolution of Levante, Genova, Italy, 7–12 September 2008). I thank Daniel Koehn, the East African Rift System. Bulletin de la Societe Geologique de France 22, Friederike Bauer, Ulrich Glasmacher, Giovanni Piccardo, Georg 891–899. Rümpker for these invitations. I warmly thank Marco Bonini, Fabrizio Barberi, F., Varet, J., 1977. Volcanism of Afar: small-scale plate tectonics implication. Geological Society of America Bulletin 88, 1251–1266. Innocenti, Francesco Mazzarini, Piero Manetti and Tsegaye Abebe for Barberi, F., Ferrara, G., Santacroce, R., Varet, J., 1975. Structural evolution of the Afar greatly contributing to my background on the Main Ethiopian Rift and junction. In: Pilger, A., Rosler, A. (Eds.), Afar Depression of Ethiopia. Schweizerbart, – for the stimulating discussions concerning its evolution. This research Stuttgart, pp. 38 54. fi Bastow, I.D., Stuart, G.W., Kendall, J.M., Ebinger, C.J., 2005. Upper mantle seismic also bene ted from discussions with Andrea Agostini, Marco structure in a region of incipient continental breakup: northern Ethiopian rift. Benvenuti, Derek Keir, Genene Mulugeta, Federico Sani. I express my Geophysical Journal International 162 (2), 479–493. gratitude to Antonio Zeoli for the patience he demonstrated by Bastow, I.D., Nyblade, A.A., Stuart, G.W., Rooney, T.O., Benoit, M.H., 2008. Upper mantle seismic structure beneath the Ethiopian hot spot: rifting at the edge of the African answering a large number of questions related to the elaboration of low-velocity anomaly. Geochemistry, Geophysics, Geosystems 9, Q12022. digital elevation models. Simon Klemperer and Katie Klemperer are doi:10.1029/2008GC002107. thanked for providing me a pre-print of their 2009 work and for Bellahsen, N., Faccenna, C., Funiciello, F., Daniel, J.M., Jolivet, L., 2003. Why did Arabia separate from Africa? Insights from 3-D laboratory experiments. Earth and discussions. I also thank Tyrone Rooney, an anonymous Reviewer and Planetary Science Letters 216, 365–381. the Editor G. Panza for the detailed, constructive comments and Bendick, R., Bilham, R., Asfaw, L., Klemperer, S., 2006. Distributed Nubia–Somalia suggestions that helped to improve the manuscript. relative motion and dyke intrusion in the main Ethiopian rift. Geophysical Journal International 165 (1), 303–310. Research on the Main Ethiopian Rift is supported by CNR Funds Benoit, M.H., Nyblade, A.A., VanDecar, J.C., 2006a. Upper mantle P wave speed variations (RSTL no. 105 “Evoluzione della parte Nord del rift Afroarabico e beneath Ethiopia and the origin of the Afar Hotspot. Geology 34, 329–332. distribuzione regionale delle georisorse”). Benoit, M.H., Nyblade, A.A., Owens, T.J., Stuart, G., 2006b. Mantle transition zone During the writing of this paper, the picture of my little Alessandro structure and upper mantle S velocity variations beneath Ethiopia: evidence for a broad, deep-seated thermal anomaly. Geochemistry, Geophysics, Geosystems 7, on my desktop always reminded me that “three things remain with us Q11013. doi:10.1029/2006GC001398. from Paradise: stars at night, flowers in the day and the eyes of Benoit, M.B., Nyblade, A.A., Pasyanos, M.E., 2006c. Crustal thinning between the children” (attributed to Dante Alighieri). Ethiopian and East African Plateaus from modeling Rayleigh wave dispersion. Geophysical Research Letters 33, L13301. doi:10.1029/2006GL025687. Benvenuti, M., Carnicelli, S., Belluomini, G., Dainelli, N., Di Grazia, S., Ferrari, G.A., Iasio, – References C., Sagri, M., Ventra, D., Balemwald, Atnatu, Seifu, Ke, 2002. The Ziway Shala lake basin (main Ethiopian rift, Ethiopia): a revision of basin evolution with special reference to the Late Quaternary. Journal of African Earth Sciences 35, 247–269. Abbate, E., Sagri, M., 1980. Volcanites of Ethiopian and Somali Plateaus and major Berckhemer, H., Baier, B., Bartelsen, H., Behle, A., Burckhardt, H., Gebrande, H., Makris, J., tectonic lines. Atti Convegni Lincei 47, 219–227. Menzel, H., Miller, H., Vees, R., 1975. Deep seismic soundings in the Afar region and Abdelsalam, M.G., Stern, R.J., 1996. Sutures and shear zones in the Arabian–Nubian on the highland of Ethiopia. In: Pilger, A., Rosler, A. (Eds.), Afar Depression of Shield. Journal of African Earth Sciences 23, 289–310. Ethiopia. Schweizerbart, Stuttgart, pp. 89–107. 50 G. Corti / Earth-Science Reviews 96 (2009) 1–53

Berhe, S.M., 1990. Ophiolites in northeast and East Africa: implications for Proterozoic Collet, B., Taud, H., Parrot, J.F., Bonavia, F., Chorowicz, J., 2000. A new kinematic approach crustal growth. Journal of the Geological Society of London 147, 41–57. for the Danakil block using a Digital Elevation Model representation. Tectonophy- Berhe, S.M., Desta, B., Nicoletti, M., Teferra, M., 1987. Geology, geochronology and sics 316, 343–357. geodynamic implications of the Cenozoic magmatic province in W and SE Ethiopia. Cornwell, D.G., Mackenzie, G.D., England, R.W., Maguire, P.K.H., Asfaw, L., Oluma, B., Journal of the Geological Society of London 144, 213–226. 2006. Northern main Ethiopian rift crustal structure from new high-precision Beydoun, Z.R., 1960. Synopsis of the Geology of East Aden Protectorate. XXI gravity data. In: Yirgu, G., Ebinger, C.J., Maguire, P.K.H. (Eds.), The Afar Volcanic International Geological Congress, Copenhagen 21, 131–149. Province within the East African Rift System: Geological Society Special Publication, Beyene, A., Abdelsalam, M.G., 2005. Tectonics of the Afar Depression: a review and synthesis. vol. 259, pp. 307–321. Journal of African Earth Sciences 41, 41–59. doi:10.1016/j.jafrearsci.2005.03.003. Corti, G., 2008. Control of rift obliquity on the evolution and segmentation of the main Billham, R., Bendick, R., Larson, K., Braun, J., Tesfaye, S., Mohr, P., Asfaw, L., 1999. Secular Ethiopian rift. Nature Geoscience 1, 258–262. and tidal strain across the Ethiopian rift. Geophysical Research Letters 27, Corti, G., Bonini, M., Mazzarini, F., Boccaletti, M., Innocenti, F., Manetti, P., Mulugeta, G., 2789–2984. Sokoutis, D., 2002. Melt-induced strain localization and magma emplacement in Boccaletti, M., Getaneh, A., Tortorici, L., 1992. The Main Ethiopian Rift: an example of centrifuge models of transfer zones. Tectonophysics 348, 205–218. oblique rifting. Annales Tectonicae 6, 20–25. Corti, G., Bonini, M., Conticelli, S., Innocenti, F., Manetti, P., Sokoutis, D., 2003. Analogue Boccaletti, M., Getaneh, A., Mazzuoli, R., Tortorici, L., Trua, T., 1995. Chemical variations modelling of continental extension: a review focused on the relations between the in a bimodal magma system: the Plio-Quaternary volcanism in the Dera Nazret area patterns of deformation and the presence of magma. Earth Science Reviews 63, 169–247. (Main Ethiopian Rift, Ethiopia). Africa Geoscience Review 2, 37–60. Corti, G., Bonini, M., Sokoutis, D., Innocenti, F., Manetti, P., Cloetingh, S., Mulugeta, G., 2004. Boccaletti, M., Bonini, M., Mazzuoli, R., Abebe, B., Piccardi, L., Tortorici, L., 1998. Continental rift architecture and patterns of magma migration: a dynamic analysis Quaternary oblique extensional tectonics in the Ethiopian Rift (Horn of Africa). based on centrifuge models. Tectonics 23, TC2012. doi:10.1029/2003TC001561. Tectonophysics 287, 97–116. Coulié, E., Quidelleur, X., Gillot, P.Y., Coutillot, V., Lefevre, J.C., Chiessa, S., 2003. Boccaletti, M., Peccerillo, A., 1999. Foreword to The Ethiopian Rift System. In: Boccaletti, Comparative K–Ar and Ar/Ar dating of Ethiopian and Yemenite Oligocene M., Peccerillo, A. (Eds.), Acta Vulcanologica, 11, pp. V–VII. volcanism: implication for timing and duration of the Ethiopian traps. Earth and Boccaletti,M.,Mazzuoli,R.,Bonini,M.,Trua,T.,Abebe,B.,1999a.Plio-Quaternary Planetary Science Letters 206, 477–492. doi:10.1016/S0012-821X(02)01089-0. volcano-tectonic activity in the northern sector of the Main Ethiopian Rift Courtillot, V., Jaupart, C., Manighetti, I., Tapponnier, P., Besse, J., 1999. On causal links (MER): relationships with oblique rifting. Journal of African Earth Sciences 29, between fl ood basalts and continental breakup. Earth and Planetary Science Letters 679–698. 166, 177–195. Boccaletti, M., Bonini, M., Mazzuoli, R., Trua, T., 1999b. Pliocene-Quaternary volcanism Dainelli, G., 1943. Geologia dell'Africa Orientale. Reale Accademia d'Italia. 3 vols. and faulting in the northern Main Ethiopian Rift (with two geological maps at scale Daly, E., Keir, D., Ebinger, C.J., Stuart, G.W., Bastow, I.D., Ayele, A., 2008. Crustal 1:50000). Acta Vulcanologica 11, 83–97. tomographic imaging of a transitional continental rift: the Ethiopian rift. Bonini, M., Souriot, T., Boccaletti, M., Brun, J.P., 1997. Successive orthogonal and oblique Geophysical Journal International 172, 1033–1048. extension episodes in a rift zone: laboratory experiments with application to the Davidson, A., Rex, D.C., 1980. Age of volcanism and rifting in south-western Ethiopia. Ethiopian Rift. Tectonics 16, 347–362. Nature 283, 654–658. Bonini, M., Corti, G., Innocenti, F., Manetti, P., Mazzarini, F., Abebe, T., Pecskay, Z., 2005. Davies, G., 1998. A channelled plume under Africa. Nature 395, 743. Evolution of the Main Ethiopian Rift in the frame of Afar and Kenya rifts Di Paola, G.M., 1972. The Ethiopian Rift Valley (between 7 and 8 40' lat. North). Bulletin propagation. Tectonics 24, TC1007. doi:10.1029/2004TC001680. of Volcanology 36, 517–560. Bosworth, W., Strecker, M.R., Blisniuk, P.M., 1992. Integration of East African paleostress Dugda, M.T., Nyblade, A.A., 2006. New constraints on crustal structure in eastern Afar from and present-day stress data: implications for continental stress field dynamics. the analysis of receiver functions and surface wave dispersion in Djibouti. In: Yirgu, G., Journal of Geophysical Research 97, 11851–11865. Ebinger, C.J., Maguire, P.K.H. (Eds.), The Afar Volcanic Province within the East African Bosworth, W., Huchon, P., McClay, K., 2005. The Red Sea and Gulf of Aden Basins. Journal Rift System: Geological Society Special Publication, vol. 259, pp. 239–251. of African Earth Sciences 43, 334–378. doi:10.1016/j.jafrearsci.2005.07.020. Dugda, M.T., Nyblade, A.A., Jordi, J., Langston, C.A., Ammon, C.J., Simiyu, S., 2005. Crustal Braile, L.W., Keller, G.R., Wendlandt, R.F., Morgan, P., Khan, M.A., 1995. The East African structure in Ethiopia and Kenya from receiver function analysis: implications for rift Rift System. In: Olsen, K.H. (Ed.), Continental Rifts: Evolution, Structure, Tectonics: development in Eastern Africa. Journal of Geophysical Research 110 (B1), B01303. Developments in Geotectonics, vol. 25, pp. 213–231. doi:10.1029/2004JB003065. Brazier, R.A., Miao, Q., Nyblade, A.A., Ayele, A., Langston, C.A., 2008. Local magnitude scale Dugda, M., Nyblade, A.A., Julia, J., 2007. Thin lithosphere beneath the Ethiopian Plateau for the Ethiopian Plateau. Bulletin of the Seismological Society of America 98, revealed by a joint inversion of Rayleigh Wave Group velocities and receiver 2341–2348. functions. Journal of Geophysical Research 112. doi:10.1029/2006JB004918. Brun, J.-P., 1999. Narrow rifts versus wide rifts: inferences for the mechanics of rifting Dunbar, J.A., Sawyer, D.S., 1989. Continental rifting at pre-existing lithospheric from laboratory experiments. Philosophical Transactions Royal Society London Ser. weaknesses. Nature 242, 565–571. A 357, 695–712. Dyksterhuis, S., Rey, P., Muller, R.D., Moresi, L., 2007. Effects of initial weakness on rift Buck, W.R., 1991. Modes of continental lithospheric extension. Journal of Geophysical architecture. In: Karner, G.D., Manatschal, G., Pinhiero, L.M. (Eds.), Imaging, Research 96, 20,161–20,178. mapping and modelling continental lithosphere extension and breakup: Geol. Buck, W.R., 2004. Consequences of asthenospheric variability on continental rifting. In: Soc. London Spec. Pub., vol. 282, pp. 443–455. Karner, G.D., Taylor, B., Droscoll, N.W., Kohlstedt, D.L. (Eds.), Rheology and Eagles, G., Gloaguen, R., Ebinger, C.J., 2002. Kinematics of the Danakil microplate. Earth Deformation of the Lithosphere at Continental Margins. Columbia Univ. Press, and Planetary Science Letters 203, 607–620. New York, pp. 1–30. Ebinger, C.J., 1989. Tectonic development of the western branch of the East African Rift Buck, W.R., 2006. The role of magma in the development of the Afro-Arabian Rift System. Geological Society of America Bulletin 101, 885–903. System. In: Yirgu, G., Ebinger, C.J., Maguire, P.K.H. (Eds.), The Afar Volcanic Province Ebinger, C., 2005. Continental breakup: the East African perspective. Astronomy and within the East African Rift System: Geological Society Special Publication, vol. 259, Geophysics 46, 2.16–2.21. pp. 43–54. Ebinger, C.J., Casey, M., 2001. Continental breakup in magmatic provinces: an Ethiopian Burke, K., Wilson, J., 1976. Hot spots on the earth's surface. Scientific American 235, example. Geology 29, 527–530. 46–57. Ebinger, C.J., Hayward, N.J., 1996. Soft plates and hot spots: views from Afar. Journal of Casey, M., Ebinger, C.J., Keir, D., Gloaguen, R., Mohamad, F., 2006. Strain accommodation Geophysical Research 101, 21859–21876. in transitional rifts: extension by magma intrusion and faulting in Ethiopian rift Ebinger, C., Sleep, N.H., 1998. Cenozoic magmatism in central and east Africa resulting magmatic segments. In: Yirgu, G., Ebinger, C.J., Maguire, P.K.H. (Eds.), The Afar from impact of one large plume. Nature 395, 788–791. Volcanic Province within the East African Rift System: Geological Society Special Ebinger, C.J., Bechtel, T.D., Forsyth, D.W., Bowin, C.O., 1989. Effective elastic plate Publication, vol. 259, pp. 143–163. thicknesses beneath the East African and Afar domes. Journal of Geophysical Chernet, T., Hart, W.K., Aronson, J.L., Walter, R.C., 1998. New age constraints on the Research 94, 2883–2990. timing of volcanism and tectonism in the northern Main Ethiopian Rift-southern Ebinger, C.J., Yemane, T., WoldeGabriel, G., Aronson, J.L., Walter, R.C., 1993. Late Eocene- Afar transition zone (Ethiopia). Journal of Volcanology and Geothermal Research Recent volcanism and faulting in the southern main Ethiopian rift. Journal of the 80, 267–280. Geological Society of London 150, 99–108. Chorowicz, J., 2005. The East African Rift System. Journal of African Earth Sciences 43, Ebinger, C.J., Yemane, T., Harding, D.J., Tesfaye, S., Kelley, S., Rex, D.C., 2000. Rift 379–410. doi:10.1016/j.jafrearsci.2005.07.019. deflection, migration, and propagation: linkage of the Ethiopian and Eastern rifts, Chorowicz, J., Collet, B., Bonavia, F., Korme, T., 1994. Northwest to North–Northwest extension Africa. Geological Society of America Bulletin 112, 163–176. direction in the Ethiopian rift deduced from the orientation of extension structures and Ebinger, C.J., Keir, D., Ayele, A., Calais, E., Wright, T.J., Belachew, M., Hammond, J.O.S., fault-slip analysis. Geological Society of America Bulletin 105, 1560–1570. Campbell, E., Buck, W.R., 2008. Capturing magma intrusion and faulting processes Chorowicz, J., Collet, B., Bonavia, F.F., Mohr, P., Parrot, J.-F., Korme, T., 1998. The Tana during continental rupture: seismicity of the Dabbahu (Afar) rift. Geophysical basin, Ethiopia: intraplateau uplift, rifting and subsidence. Tectonophysics 295, Journal International 174, 1138–1152. 351–367. Ellis, M., King, G., 1991. Structural control of flank volcanism in continental rifts. Science Chu, D., Gordon, R.G., 1999. Evidence for motion between Nubia and Somalia along the 254, 839–842. Southwest Indian Ridge. Nature 398, 64–67. Fernandes, R.M.S., Ambrosius, B.A.C., Noomen, R., Bastos, L., Combinck, L., Miranda, J.M., Church, W.R., 1991. Discussion of “Ophiolites in northeast and East Africa: implications Spakman, W., 2004. Angular velocities of Nubia and Somalia from continuous GPS for Proterozoic crustal growth”. Journal of the Geological Society of London 148, data: implications on present-day relative kinematics. Earth and Planetary Science 600–606. Letters 222, 197–208. Clifton, A.E., Schlische, R.W., 2001. Nucleation, growth, and linkage of faults in oblique Foster, A.N., Jackson, J.A., 1998. Source parameters of large African earthquakes: rift zones: results from experimental clay models and implications for maximum implications for crustal rheology and regional kinematics. Geophysical Journal fault size. Geology 29, 455–458. International 134, 422–448. G. Corti / Earth-Science Reviews 96 (2009) 1–53 51

Furman, T., Bryce, J., Rooney, T., Hanan, B., Yurgu, G., Ayalew, D., 2006. Heads and tails: Keir, D., Kendall, J.-M., Ebinger, C., 2005. Variations in late syn-rift melt alignment 30 million years of the Afar plume. In: Yirgu, G., Ebinger, C.J., Maguire, P.K.H. (Eds.), inferred from shear-wave splitting in crustal earthquakes beneath the Ethiopian The Afar Volcanic Province within the East African Rift System: Geological Society rift. Geophysical Research Letters 32. doi:10.1029/2005GL024150. Special Publication, vol. 259, pp. 95–119. Keir, D., Ebinger, C.J., Stuart, G.W., Daly, E., Ayele, A., 2006a. Strain accommodation by Gani, N.D., Abdelsalam, M.G., Gani, M.R., 2007. Blue Nile incision on the Ethiopian magmatism and faulting as rifting proceeds to breakup: seismicity of the northern Plateau: pulsed plateau growth, Pliocene uplift, and hominin evolution. GSA Today Ethiopian rift. Journal of Geophysical Research 111 (B5), B05314. doi:10.1029/ 17, 4–11. 2005JB003748. Gani, N.D.S., Abdelsalam, M.G., Gera, S., Gani, M.R., 2009. Stratigraphic and structural Keir, D., Stuart, G.W., Jackson, A, Ayele, A., 2006b. Local earthquake magnitude scale and evolution of the Blue Nile Basin, Northwestern Ethiopian Plateau. Geological seismicity rate for the Ethiopian rift. Bulletin of the Seismological Society of Journal 44, 30–56. America 96, 2221–2230. Garfunkel, Z., Beyth, M., 2006. Constraints on the structural development of Afar Keir, D., et al., 2009. Evidence for focused magmatic accretion at segment centers from imposed by the kinematics of the major surrounding plates. In: Yirgu, G., Ebinger, C.J., lateral dike injections captured beneath the Red Sea rift in Afar. Geology 37, 59–62. Maguire, P.K.H. (Eds.), The Afar Volcanic Province within the East African Rift System: Kendall, J.M., Stuart, G.W., Ebinger, C.J., Bastow, I.D., Keir, D., 2005. Magma assisted Geological Society Special Publication, vol. 259, pp. 23–42. rifting in Ethiopia. Nature 433, 146–148. Gashawbeza, E.M., Klemperer, S.L., Nyblade, A.A., Walker, K.T., Keranen, K.M., 2004. Shear- Kendall, J.M., Pilidou, S., Keir, D., Bastow, I.D., Stuart, G.W., Ayele, A., 2006. Mantle wave splitting in Ethiopia: Precambrian mantle anisotropy locally modified by upwellings, melt migration, and the rifting of Africa: insights from seismic anisotropy. Neogene rifting. Geophysical Research Letters 31, L18602. doi:10.1029/2004GL020471. In: Yirgu, G., Ebinger, C.J., Maguire, P.K.H. (Eds.), The Afar Volcanic Province within the Gasparon, M., Innocenti, F., Manetti, P., Peccerillo, A., Abebe, T., 1993. Genesis of the East African Rift System: Geological Society Special Publication, vol. 259, pp. 55–72. Pliocene to Recent bimodal mafic–felsic volcanism in the Debre–Zeyt area, central Keranen, K., Klemperer, S.L., 2008. Discontinuous and diachronous evolution of the Ethiopia: volcanological and geochemical constraints. Journal of African Earth Main Ethiopian Rift: implications for the development of continental rifts. Earth Sciences 17, 145–165. and Planetary Science Letters 265, 96–111. doi:10.1016/j.epsl.2007.09.038. Gasse, F., Street, F.A., 1978. Late Quaternary lake-level fluctuations and environments of Keranen, K., Klemperer, S.L., Gloaguen, R., Eagle working group, 2004. Three-dimensional the Northern Rift Valley and Afar region (Ethiopia and Djibuti). Palaeogeography, seismic imaging of a protoridge axis in the Main Ethiopian rift. Geology 32, 949–952. Palaeoclimatology, Palaeoecology 24, 279–325. Keranen, K., Klemperer, S.L., Julia, J., Lawrence, J.L., Nyblade, A., 2009. Low lower-crustal George, R.M., Rogers, N.W., 1999. The petrogenesis of Plio-Pleistocene alkaline volcanic velocity across Ethiopia: is the Main Ethiopian Rift a narrow rift in a hot craton? rocks from the Tosa Sucha region, Arba Minch, southern Main Ethiopian Rift. Acta Geochemistry, Geophysics, Geosystems 10, Q0AB01. doi:10.1029/2008GC002293. Volcanologica 11, 121–130. Kieffer, B., Arndt, N., LaPierre, H., Bastien, F., Bosch, D., Pecher, A., Yirgu, G., Ayalew, D., George, R., Rogers, N., Kelley, S., 1998. Earliest magmatism in Ethiopia: evidence for two Weis, D., Jerram, D., Keller, F., Meugniot, C., 2004. Flood and shield basalts from mantle plumes in one flood basalt province. Geology 26, 923–926. Ethiopia: magmas from the African Superswell. Journal of Petrology 45, 793–834. Giaquinta, A., Boccaletti, S., Boccaletti, M., Piccardi, L., Arecchi, F.T., 1999. Investigating Korme, T., Chorowicz, J., Collet, B., Bonavia, F.F., 1997. Volcanic vents rooted on extension the fractal properties of geological fault system: the Main Ethiopian Rift case. fractures and their geodynamic implications in the Ethiopian Rift. Journal of Geophysical Research Letters 26, 1633–1636. Volcanology and Geothermal Research 79, 205–222. Gibson, I.L., 1969. The structure and volcanic geology of an axial portion of the Main Korme, T., Acocella, V., Abebe, B., 2004. The role of pre-existing structures in the origin, Ethiopian Rift. Tectonophysics 8, 561–565. propagation and architecture of faults in the Main Ethiopian Rift. Gondwana Gibson, I.L., Tazieff, H., 1970. The structure of the Afar and the northern part of the Research 7, 467–479. Ethiopian Rift. Philosophical Transactions. Royal Society of London 267, 331–338. Kroner, A., 1985. Ophiolites and evolution of tectonic boundaries in the Iate Proterozoic Gillespie, R., Street-Perrot, F., Switsur, R., 1983. Post-glacial arid episodes in Ethiopia Arabian Nubian Shield of north–east Africa and Arabia. Precambrian Research 27, have implications for climate prediction. Nature 306, 680–683. 277–300. Girdler, R.W., 1983. Processes of rifting and break up of Africa. In: Morgan, P., Baker, B.H., Kunz, K., Krewzer, H., Muller, P., 1975. K/Ar determination of the trap basalts of the (Eds.), Processes of Continental Rifting. Tectonophysics 94, 241–252. southeastern part of the Afar Rift. In: Pilger, A., Rosler, A. (Eds.), Afar Depression of Gouin, P., 1979. Earthquake history of Ethiopia and the Horn of Africa. IDRC, Ottawa. Ethiopia. Schweizerbart, Stuttgart, pp. 370–374. 258 pp. Kurz, T., Gloaguen, R., Ebinger, C., Casey, M., Abebe, B., 2007. Deformation distribution Griffiths, R.W., Campbell, I.H., 1990. Stirring and structure in mantle starting plumes. and type in the Main Ethiopian Rift (MER): a remote sensing study. Journal of Earth and Planetary Science Letters 99, 66–78. African Earth Sciences 48, 100–114. Grove, A.T., Goudie, A.S., 1971. Late Quaternary lake levels in the Rift Valley of Southern Lahitte, P., Gillot, P.Y., Courtillot, V., 2003. Silicic central volcanoes as precursors to rift Ethiopia and elsewhere in Tropical Africa. Nature 234, 403–405. propagation: the Afar case. Earth and Planetary Science Letters 207, 103–116. Grove, A.T., Street, F.A., Goudie, A.S., 1975. Former lake levels and climatic changes in the Laury, R.L., Albritton, C.C., 1975. Geology of the Middle Stone Age archaeological sites in rift valley of Southern Ethiopia. Geographical Journal 141, 177–202. the Main Ethiopian Rift Valley. Geological Society of America Bulletin 86, 999–1011. Guiraud, R., Bosworth, W., Thierry, J., Delplanque, A., 2005. Phanerozoic geological Le Turdu, C., Tiercelin, J.J., Gibert, E., Travi, Y., Lezzar, K.E., Richert, J.P., Massault, M., evolution of Northern and Central Africa: an overview. Journal of African Earth Gasse, F., Bonnefille, R., Decobert, M., Gensous, B., Jeudy, V., Tamrat, E., Mohammed, Sciences 43, 83–143. M.U., Martens, K., Atnafu, B., Cherent, T., Williamson, D., Taieb, M., 1999. The Ziway– Gurnis, M., Mitrovica, J.X., Ritsema, J., van Heijst, H.J., 2000. Constraining mantle density Shala lake basin system, Main Ethiopian Rift: influence of volcanism, tectonics and structure using geological evidence of surface uplift rates: the case of the African climatic forcing on basin formation and sedimantation. Palaeogeography, Palaeo- Superplume. Geochemistry, Geophysics, Geosystems 1 1999GC000035. climatology, Palaeoecology 150, 135–177. Harris, W.C., 1844. The Highlands of Ethiopia, vol. 3. Longman, New York. Lemaux, J., Gordon, R., Royer, J.-Y., 2002. Location of the Nubia–Somalia boundary along Hart, W.K., Woldegabriel, G., Walter, R.C., Mertzman, S.A., 1989. Basaltic volcanism in the Southwest Indian Ridge. Geology 30, 339–342. Ethiopia: constraints on continental rifting and mantle interactions. Journal of Levitte, D., Columba, J., Mohr, P., 1974. Reconnaissance geology of the Amaro horst, Geophysical Research 94 (B6), 7731–7748. southern Ethiopian rift. Geological Society of America Bulletin 85, 417–422. Hautot, S., Whaler, K., Gebru, W., Desissa, M., 2006. The structure of a Mesozoic basin Lithgow-Bertelloni, C., Silver, P., 1998. Dynamic topography, plate driving forces and the beneath the Lake Tana area, Ethiopia, revealed by magnetotelluric imaging. Journal African superswell. Nature 395, 269–272. of African Earth Sciences 44, 331–338. Mackenzie, G.H., Thybo, G.H., Maguire, P., 2005. Crustal velocity structure across the Hayward, N.J., Ebinger, C.J., 1996. Variations in the along-axis segmentation of the Afar Main Ethiopian Rift: results from 2-dimenional wide-angle seismic modeling. Rift system. Tectonics 15, 244–257. Geophysical Journal International 162, 994–1006. Hoffman, C., Courtillot, V., Féraud, G., Rochette, P., Yirgu, G., Ketefo, E., Pik, R., 1997. Maguire, P.K.H., Ebinger, C.J., Stuart, G.W., Mackenzie, G.D., Whaler, K.A., Kendall, J.-M., Timing of the Ethiopian basalt event and implications for plume birth and global Khan, M.A., Fowler, C.M.R., Klemperer, S.L., Keller, G.R., Harder, S., Furman, T., Mickus, change. Nature 389, 838–841. K., Asfaw, L., Ayele, A., Bekele, A., 2003. Geophysical project in Ethiopia studies Hofstetter, R., Beyth, M., 2003. The Afar Depression interpretation of the 1960–2000 continental breakup. Eos Transaction on American Geophysical Union 84, 337–340. earthquakes. Geophysical Journal International 155, 715–732. doi:10.1046/j.1365- Maguire, P.K.H., Keller, G.R., Klemperer, S.L., Mackenzie, G.D., Keranen, K., Harder, S., 246X.2003.02080.x. O'Reilly, B., Thybo, H., Asfaw, L., Khan, M.A., Amha, M., 2006. Crustal structure of the Horner-Johnson, B.C., Gordon, R.G., Argus, D.F., 2007. Plate kinematic evidence for the Northern Main Ethiopian Rift from the EAGLE controlled source survey; a snapshot existence of a distinct plate between the Nubian and Somalian plates along the of incipient lithospheric break-up. In: Yirgu, G., Ebinger, C.J., Maguire, P.K.H. (Eds.), Southwest Indian Ridge. Journal of Geophysical Research 112, B05418. doi:10.1029/ The Afar Volcanic Province within the East African Rift System: Geological Society 2006JB004519. Special Publication, vol. 259, pp. 269–291. Jestin, F., Huchon, P., Gaulier, M., 1994. The Somalia plate and the East African Mahatsente, R., Jentzsch, G., Jahr, T., 1999. Crustal structure of the Main Ethiopian Rift Rift System: present-day kinematics. Geophysical Journal International 116, from gravity data: 3-dimensional modeling. Tectonophysics 313, 363–382. 637–654. Mahatsente, R., Jentzsch, G., Jahr, T., 2000. Three- dimension inversion of gravity data Joffe, S., Garfunkel, Z., 1987. Plate kinematics of the circum Red Sea; a re-evaluation. from the Main Ethiopian Rift. Journal of African Earth Science 31, 451–466. Tectonophysics 141, 5–22. Marty, B., Pik, R., Gezahegn, Y., 1996. Helium isotopic variations in Ethiopian plume Kazmin, V., Seife, M.B., Nicoletti, M., Petrucciani, C., 1980. Evolution of the northern part lavas: nature of magmatic sources and limit on lower mantle contribution. Earth of the Ethiopian Rift. Geodynamic Evolution of the Afro-Arabian Rift System, and Planetary Science Letters 144, 223–237. Accademia Nazionale Dei Lincei, Atti dei Convegni Lincei, vol. 47, pp. 275–292. Mazzarini, F., 2004. Volcanic velt self-similar clustering and crustal thickness in the Kebede, F., Kulhanek, O., 1991. Recent seismicity of the East African Rift system and its northern Main Ethiopian Rift. Geophysical Research Letters 31, L04604. doi:10.1029/ implications. Physics of Earth Planetary Interiors 68, 259–273. 2003GL018574. Kebede, F., Vuan, A., Mammo, T., Costa, G., Panza, G.F., 1996. Shear wave velocity Mazzarini, F., Abebe, T., Innocenti, F., Manetti, P., Pareschi, M.T., 1999. Geology of the structure of Northern and North-eastern Ethiopia. Acta Geodaetica et Geophysica Debre Zeyt area (Ethiopia) (with a geological map at scale 1:100.000). Acta 31, 145–159. Vulcanologica 11, 131–141. 52 G. Corti / Earth-Science Reviews 96 (2009) 1–53

Mazzarini, F., Corti, G., Manetti, P., Innocenti, F., 2004. Strain rate bimodal volcanism in Pik, R., Deniel, C., Coulon, C., Yirgu, G., Hofmann, C., Ayalew, D., 1998. The Northwest the continental rift: Debre Zeyt volcanic field, northern MER, Ethiopia. Journal of Ethiopian plateau flood basalts: classification and spatial distribution of magma African Earth Sciences 39, 415–420. doi:10.1016/j.jafrearsci.2004.07.025. types. Journal of Volcanology and Geothermal Research 81, 91–111. McDougall, I., Morton, W.H., William, M.A.J., 1975. Ages and rates of denudation of trap Pik, R., Marty, B., Carignan, J., Lavé, J., 2003. Stability of the Upper Nile drainage network series basalts at the Blue Nile Gorge, Ethiopia. Nature 254, 207–209. (Ethiopia) deduced from (U–Th)/He thermochronometry: implications for uplift Meert, J.G., Lieberman, B.S., 2008. The Neoproterozoic assembly of Gondwana and its and erosion of the Afar plume dome. Earth and Planetary Science Letters 215, relationship to the Ediacaran–Cambrian radiation. Gondwana Research 14, 5–21. 73–88. doi:10.1016/S0012-821X(03)00457-6. Mege, D., Korme, T., 2004. Dyke swarm emplacement in the Ethiopian Large Igneous Pik, R., Marty, B., Hilton, D.R., 2006. How many mantle plumes in Africa? The geochemical Province: not only a matter of stress. Journal of Volcanology and Geothermal point of view. Chemical Geology 226, 100–114. doi:10.1016/j.chemgeo.2005.09.016. Research 132, 283–310. Pik, R., Marty, B., Carignan, J., Yirgu, G., Ayalew, T., 2008. Timing of East African Rift Menzies, M., Gallagher, K., Yelland, A., Hurford, A.J., 1997. Volcanic and nonvolcanic development in southern Ethiopia: implication for mantle plume activity and rifted margins of the Red Sea and Gulf of Aden: crustal cooling and margin evolution of topography. Geology 36, 167–170. doi:10.1130/G24233A.1. evolution in Yemen. Geochimica et Cosmochimica Acta 61, 2511–2527. Pizzi, A., Coltorti, M., Abebe, B., Disperati, L., Sacchi, G., Salvini, R., 2006. The Winji fault Merla, G., Abbate, E., Canuti, P., Sagri, M., Tacconi, P., 1979. Geological map of Ethiopia belt (main Ethiopian Rift): structural and geomorphological constraints and GPS and Somalia and comment with a map of major landforms (scale 1:2, 000, 000). monitoring. In: Yirgu, G., Ebinger, C.J., Maguire, P.K.H. (Eds.), The Afar Volcanic Consiglio Nazionale delle Ricerche, Rome. 95. Province within the East African Rift System: Geological Society Special Publication, Meshesha, D., Shinjo, R., 2008. Rethinking geochemical feature of the Afar and Kenya vol. 259, pp. 191–. 207 mantle plumes and geodynamic implications. Journal of Geophysical Research 113, Ritsema, J., Allen, R.M., 2003. The elusive mantle plume. Earth and Planetary Science B09209. doi:10.1029/2007JB005549. Letters 207, 1–12. Meyer, W., Pilger, A., Rosler, A., Stets, J., 1975. Tectonic evolution of the northern part of Ritsema, J., van Heijst, H.J., Woodhouse, J.H., 1999. Complex shear wave velocity the Main Ethiopian Rift in Southern Ethiopia. In: Pilger, A., Rosler, A. (Eds.), Afar structure imaged beneath Africa and Iceland. Science 286, 1925–1928. Depression of Ethiopia. Schweizerbart, Stuttgart, pp. 352–362. Rogers, N.W., 2006. Basaltic magmatism and the geodynamics of the East African Rift Mickus, K., Tadesse, K., Keller, G.R., Oluma, B., 2007. Gravity analysis of the main System. In: Yirgu, G., Ebinger, C.J., Maguire, P.K.H. (Eds.), The Afar Volcanic Province Ethiopian rift. Journal of African Earth Sciences 48, 59–69. within the East African Rift System: Geological Society Special Publication, vol. 259, Miller, J.A., Mohr, P.A., 1966. Age of the Wechacha trachyte-carbontite volcanic centre. pp. 77–93. Bulletin of the Geophysical Observatory of Addis Ababa 9, 1–6. Rooney, T., Furman, T., Yirgu, G., Ayelew, D., 2005. Structure of the Ethiopian lithosphere: Mohr, P., 1962. The Ethiopian Rift System. Bulletin of the Geophysical Observatory of evidence from mantle Xenoliths. Geochemica et Cosmochimica Acta 69, 3889–3910. Addis Ababa 5, 33–62. Rooney, T., Furman, T., Bastow, I., Ayalew, D., Yirgu, G., 2007. Lithospheric modification Mohr, P., 1967. The Ethiopian Rift System. Bulletin of the Geophysical Observatory of during crustal extension in the Main Ethiopian Rift. Journal of Geophysical Research Addis Ababa 11, 1–65. 112, B10201. doi:10.1029/2006JB004916. Mohr, P., 1968. Transcurrent faulting in the Ethiopian Rift System. Nature 218, 938–941. Rosendahl, B.L., 1987. Architecture of continental rifts with special reference to east Mohr, P., 1970. The Geology of Ethiopia. Addis Ababa University Press, Addis Ababa. Africa. Annual Reviews of Earth and Planetary Sciences 15, 445–503. Mohr, P., 1983. Volcanotectonic aspects of the Ethiopian Rift evolution. Bulletin Centre Rowland, J.R., Baker, E., Ebinger, C., Keir, D., Kidane, T., Biggs, J., Hayward, N., Wright, T., Recherches Elf Aquitaine Exploration Production 7, 175–189. 2007. Fault growth at a nascent slowspreading ridge: 2005 Dabbahu rifting episode, Mohr, P., 1987. Patterns of faulting in the Ethiopian Rift Valley. Tectonophysics 143, Afar. Geophysical Journal International 171 doi:10.1111/j.1365-246X.2007.03584.x. 169–179. Royer, J.-Y., Gordon, R.G., Horner-Johnson, B.C., 2006. Motion of Nubia relative to Mohr, P.A., Potter, E.C., 1976. The Sagatu Ridge dike swarms, Ethiopian rift margin. Antarctica since 11 Ma: implications for Nubia–Somalia, Pacific–North America, and Journal of Volcanology and Geothermal Research 1, 55–71. India–Eurasia motion. Geology 34, 501–504. doi:10.1130/G22463.1. Mohr, P.A., Wood, C.A., 1976. Volcano spacing and lithospheric attenuation in the Sagri, M., Bartolini, C., Billi, P., Ferrari, G., Benvenuti, M., Carnicelli, S., Barbano, F., 2008. Eastern Rift of Africa. Earth and Planetary Science Letters 33, 126–144. Latest Pleistocene and Holocene river network evolution in the Ethiopian Lakes Mohr, P., Zanettin, B., 1988. The Ethiopian food basalt province. In: Macdougall, J.D. Region. Geomorphology 94, 79–97. (Ed.), Continental flood basalts. Kluwer Academic Publishers, pp. 63–110. Scarsi, P., Craig, H., 1996. Helium isotope ratios in Ethiopian Rift basalts. Earth and Mohr, P., Mittchell, J.G., Raynolds, R.G.H., 1980. Quaternary volcanism and faulting at O'a Planetary Science Letters 144, 505–516. caldera, central Ethiopian Rift. Bulletin of Volcanology 43, 173–189. Schilling, J.G., Kingsley, R.H., Hanan, B.B., McCully, B.L., 1992. Nd–Sr–Pb isotopic Montelli, R., Nolet, G., Dahlen, F.A., Masters, G., Engdahl, E.R., Hung, S.H., 2004. Finite- variations along the Gulf of Aden: evidence for Afar mantle plume-continental frequency tomography reveals a variety of plumes in the mantle. Science 303, lithosphere interaction. Journal of Geophysical Research B 97 (7), 10,927–10,966. 338–343. Sella, G., Dixon, T.H., Mao, A., 2002. REVEL: a model for recent plate velocities from space Montelli, R., Nolet, G., Dahlen, F., Masters, G., 2006. A catalogue of deep mantle plumes: new geodesy. Journal of Geophysical Research 107. doi:10.1029/2000JB000033. results from finite-frequency tomography. Geochemistry, Geophysics, Geosystems 7, Sepulcre, P., Ramstein, G., Fluteau, F., Schuster, M., Tiercelin, J.-J., Brunet, M., 2006. Q11007. doi:10.1029/2006GC001248. Tectonic uplift and eastern Africa aridification. Science 313, 1419–1423. Moore, J.M., Davidson, A., 1978. Rift structure in Southern Ethiopia. Tectonophysics 46, Sicilia, D., Montagner, J.-P., Cara, M., Stutzmann, E., Debayle, E., Lépine, J.-C., Lévêque, J.-J., 159– 173. Beucler, E., Sebai, A., Roult, G., Ayele, A., Sholan, J.M., 2008. Upper mantle structure of Morley, C.K., Wescott, W.A., Stone, D.M., Harper, R.M., Wigger, S.T., Karanja, F.M., 1992. shear-waves velocities and stratification of anisotropy in the Afar Hotspot region. Tectonic evolution of the northern Kenyan Rift. Journal of the Geological Society of Tectonophysics 462, 164–177. London 149, 333–348. Sokoutis,D.,Corti,G.,Bonini,M.,Brun,J.-P.,Cloetingh,S.,Mauduit,T.,Manetti,P.,2007. Morton, W.H., Rex, D.C., Mitchell, J.G., Mohr, P.A., 1979. Rift ward younging of volcanic Modelling the extension of heterogeneous hot lithosphere. Tectonophysics 444, 63–79. units in the Addis Ababa region, Ethiopian rift valley. Nature 280, 284–288. Soliva, R., Schultz, R.A., 2008. Distributed and localized faulting in extensional settings: Mulugeta, G., Abebe, B., Korme, T., Sokoutis, D., 2007. Emplacement mechanisms for insight from the North Ethiopian Rift–Afar transition area. Tectonics 27, TC2003. continental flood basalts and implications for plume activity during incipient doi:10.1029/2007TC002148. continental breakup. Journal of African Earth Sciences 48, 137–146. Stamps, D.S., Calais, E., Saria, E., Hartnady, C., Nocquet, J.-M., Ebinger, C.J., Fernandes, R.M., Nyblade, A.A., Langston, C.A., 2002. Broadband seismic experiments probe the East 2008. A kinematic model for the East African Rift. Geophysical Research Letters 35, African rift. EOS Transaction on American Geophysical Union 83, 405–408. L05304. doi:10.1029/2007GL032781. Nyblade, A.A., Robinson, S.W., 1994. The African superswell. Geophysical Research Stern, R.J., 1994. Arc assembly and continental collision in the Neoproterozoic East Letters 21, 765–768. African orogen. Annual Review of Earth and Planetary Sciences 22, 319–351. Panza, G.F., 1980. Evolution of the Earth's lithosphere. NATO Adv. Stud. Inst. Newcastle, Stern, R.J., 2002. Crustal evolution in the East African Orogen: a neodymium isotopic 1979. In: Davies, P.A., Runcorn, S.K. (Eds.), Mechanisms of Continental Drift and perspective. Journal of African Earth Sciences 34, 109–117. Plate Tectonics. In: Mechanisms of Continental Drift and Plate Tectonics. Academic Stern, R.J., Nielsen, K.C., Best, E., Sultan, M., Arvidson, R.E., Kroner, A.,1990. Orientation of Press, pp. 75–87. the late Precambrian sutures in the Arabian–Nubian Shield. Geology 18, 1103–1106. Panza, G.F., Raykova, R.B., 2008. Structure and rheology of lithosphere in Italy and Stuart, G.W., Bastow, I.D., Ebinger, C.J., 2006. Crustal structure of the northern Main surrounding. Terra Nova 20, 194–199. Ethiopian Rift from receiver function studies. In: Yirgu, G., Ebinger, C.J., Maguire, P.K.H. Panza, G.F., Raykova, R.B., Carminati, E., Doglioni, C., 2007a. Upper mantle flow in the (Eds.), The Afar Volcanic Province within the East African Rift System: Geological western Mediterranean. Earth and Planetary Science Letters 257, 200–214. Society Special Publication, vol. 259, pp. 253–267. Panza, G.F., Peccerillo, A., Aoudia, A., Farina, B., 2007b. Geophysical and petrological Ten Brink, U., 1991. Volcano spacing and plate rigidity. Geology 19, 307–400. modeling of the structure and composition of the crust and upper mantle in Tentler, T., 2005. Propagation of brittle failure triggered by magma in Iceland. complex geodynamic settings: the Tyrrhenian Sea and surroundings. Earth Science Tectonophysics 406, 17–38. Reviews 80, 1–46. Tesfaye, S., Harding, J.D., Kusky, T.M., 2003. Early continental breakup boundary and Pasyanos, M.E., Nyblade, A.A., 2007. A top to bottom lithospheric study of Africa and migration of the Afar triple junction, Ethiopia. Geological Society of America Arabia. Tectonophysics 444, 27–44. Bulletin 115, 1053–1067. Peccerillo, A., Barberio, M.R., Yirgu, G., Ayalew, D., Barberi, M., Wu, T.W., 2003. Tessema, A., Fontaine, L.A.G., 2004. Processing and interpretation of the gravity field of Relationships between mafic and acid peralkaline magmatism in continental rift the East African Rift: implication for crustal extension. Tectonophysics 394, 87–110. settings: a petrological, geochemical and isotopic study of the Gedemsa volcano, Tiberi, C., Ebinger, C., Ballu, V., Stuart, G., Oluma, B., 2005. Inverse models of gravity data central Ethiopian Rift. Journal of Petrology 44, 2003–2032. from the Red Sea–Aden–East African rifts triple junction zone. Geophysical Journal Peccerillo,A.,Donati,C.,Santo,A.P.,Orlando,A.,Yirgu,G.,Ayalew,D.,2007. International 163, 775–787. doi:10.1111/j.1365-246X.2005.02736.x. Petrogenesis of silicic peralkaline rocks in the Ethiopian rift: geochemical Tommasi, A., Vauchez, A., 2001. Continental rifting parallel to ancient collisional belts: an evidence and volcanological implications. Journal of African Earth Sciences 48, effect of the mechanical anisotropy of the lithospheric mantle. Earth and Planetary 161–173. Science Letters 185, 199–210. G. Corti / Earth-Science Reviews 96 (2009) 1–53 53

Tommasini, S., Manetti, P., Innocenti, F., Abebe, T., Sintoni, M.F., Conticelli, S., 2005. The WoldeGabriel, G., Aronson, J., 1987. The Chow Bahir Rift: a “failed” rift in Southern Ethiopian subcontinental mantle domains: geochemical evidence from Cenozoic Ethiopia. Geology 15, 430–433. mafic lavas. Mineralogy and Petrology 84, 259–281. WoldeGabriel, G., Aronson, J.L., Walter, R.C., 1990. Geology, geochronology, and rift basin Trua, T., Deniel, C., Mazzuoli, R., 1999. Crustal control in the genesis of Plio-Quaternary development in the central sector of the Main Ethiopia Rift. Geological Society of bimodal magmatism of the Main Ethiopian Rift (MER): geochemical and isotopic America Bulletin 102, 439–458. (Sr, Nd and Pb) evidence. Chemical Geology 155, 201–231. WoldeGabriel, G., Yemane, T., White, T., Asfaw, B., Suwa, G., 1991. Age of volcanism and Ukstins, I.A., Renne, P.R., Wolfenden, E., Baker, J., Ayalew, D., Menzies, M., 2002. fossils in the Burji-Soyoma area, Amaro Horst, southern Main Ethiopian Rift. Journal Matching conjugate volcanic rifted margins: 40Ar39Ar chrono-stratigraphy of pre- of African Earth Sciences 13, 437–447. and syn-rift bimodal flood volcanism in Ethiopia and Yemen. Earth and Planetary WoldeGabriel, G., Walter, R.C., Hart, W.K., Mertzman, S.A., Aronson, J.L., 1999. Temporal Sciences Letters 198, 289–306. relations and geochemical features of felsic volcanism in the central sector of the Vail, J.R., 1983. Pan-African crustal accretion in the north–east Africa. Journal of African Main Ethiopian Rift. Acta Volcanologica 11, 53–67. Earth Sciences 1, 285–294. WoldeGabriel, G., Heiken, G., White, T.D., Asfaw, B., Hart, W.K., Renne, P.R., 2000. Vail, J.R., 1985. Pan-African (late Precambrian) tectonic terrains and the reconstruction Volcanism, tectonism, sedimentation, and the paleoanthropological record in the of the Arabian–Nubian Shield. Geology 13, 839–849. Ethiopian Rift System. In: McCoy, F.W., Heiken, G. (Eds.), Volcanic Hazards and Van Wijk, J.W., 2005. Role of weak zone orientation in continental lithosphere Disasters in Human Antiquity: Geological Society of America Special Paper, vol. 345, extension. Geophysical Research Letters 32. doi:10.1029/2004GL022192. pp. 83–99. Versfelt, J., Rosendahl, B.R., 1989. Relationships between pre-rift structure and rift Wolfenden, E., Ebinger, C., Yirgu, G., Deino, A., Ayale, D., 2004. Evolution of the northern architecture in Lakes Tanganyika and Malawi: East Africa. Nature 337, 354–357. Main Ethiopian rift: birth of a triple junction. Earth and Planetary Science Letters Vetel, W., Le Gall, B., 2006. Dynamics of prolonged continental extension in magmatic 224, 213–228. rifts: the Turkana Rift case study (North Kenya. In: Yirgu, G., Ebinger, C.J., Maguire, Wolfenden, E., Ebinger, C., Yirgu, G., Renne, P., Kelley, S.P., 2005. Evolution of the P.K.H. (Eds.), The Afar Volcanic Province within the East African Rift System: southern Red Sea rift: birth of a magmatic margin. Geological Society of America Geological Society Special Publication, vol. 259, pp. 209–233. Bulletin 117, 846–864. Vetel, W., Le Gall, B., Walsh, J.J., 2005. Geometry and growth of an inner rift fault pattern: Wright, T.J., Ebinger, C., Biggs, J., Ayele, A., Yirgu, G., Keir, D., Stork, A., 2006. Magma- the Kino Sogo Fault Belt, Turkana Rift (North Kenya). Journal of Structural Geology maintained rift segmentation at continental rupture in the 2005 Afar dyking 27, 2204–2222. episode. Nature 442, 291–294. Weeraratne, D.S., Forsyth, D.W., Fischer, K.M., Nyblade, A.A., 2003. Evidence for an upper Yemane, T., WoldeGabriel, G., Tesfaye, S., Berhe, S.M., Durary, S., Ebinger, C., Kelley, S., mantle plume beneath the Tanzanian craton from Rayleigh wave tomography. 1999. Temporal and geochemical characterstics of Tertiary volcanic rocks and Journal of Geophysical Research 108 (B9), 2427. doi:10.1029/2002JB002273. tectonic history in the southern Main Ethiopian Rift and the adjacent fields. Acta Whaler, K.A., Hautot, S., 2006. The electrical resistivity structure of the crust beneath the Vulcanologica 11, 99–119. northern Main Ethiopian Rift. In: Yirgu, G., Ebinger, C.J., Maguire, P.K.H. (Eds.), The Yirgu, G., Ebinger, C.J., Maguire, P.K.H. (Eds.), 2006. The Afar Volcanic Province within Afar Volcanic Province within the East African Rift System: Geological Society the East African Rift System: Geological Society Special Publication, vol. 259. Special Publication, vol. 259, pp. 293–305. Zanettin, B., Justin-Visentin, E., Nicoletti, M., Petrucciani, C., 1978. Evolution of the Williams, F.M., Williams, M.A.J., Aumento, F., 2004. Tensional fissures and crustal Chencha escarpment and the Ganjiuli graben (Lake Abaya) in the southern extension rates in the northern part of the Main Ethiopian Rift. Journal of African Ethiopian rift. Neues Jahrbuch fur Geologie und Palaontologie. Monatshefte 8, Earth Sciences 38, 183–197. 473–490. Wilson, M., 1993. Magmatism and the geodynamics of basin formation. Sedimentary Ziegler, P.A., Cloetingh, S.A.P.L., 2004. Dynamic processes controlling evolution of rifted Geology 86, 5–29. basins. Earth Science Reviews 64, 1–50.