Dyke swarm emplacement in the Ethiopian Large Igneous Province: not only a matter of stress

Abstract

In the Tana–Belaya area, western Ethiopia, field data and satellite imagery reveal the existence of two dyke swarms, the NE–SW Serpent-God dyke swarm, and the NW–SE Dinder dyke swarm. Both swarms are thought to have the same age, 30 Ma, and are likely to have contributed to feeding the traps. After a description of the swarms, this paper examines their relationships with the basement structures. The two dyke swarms follow major lithospheric weakness zones. The Serpent-God dyke swarm follows the Pan-African Tulu Dimtu ductile shear zone, and the Dinder dyke swarm follows a large NW–SE-trending Precambrian fracture zone already reactivated during the Mesozoic and Cenozoic as the northern boundary of the Blue Nile Rift. Because the dyke swarms are adjacent but their orientation differs, the stress trajectory patterns during their emplacement were spatially variable at local scale. Therefore, rather than plate-boundary processes, the origin of stress is thought to be primarily related to the Ethiopian plume. Postulating (in the absence of more data relating to the magma chambers that fed the traps) that dyke orientation is the result of an axisymmetric stress field, the location of the stress source can be placed close to Lake Tana, which is the centre of the Ethiopian broad negative regional Bouguer anomaly. The dykes in the Tana–Belaya area provide the first clues to the orientation of the stress field that prevailed in the early history of the Ethiopian mantle plume, and to some of the factors that guided the distribution of the trap feeders.

Introduction

Continental mafic dyke swarms in Large Igneous Provinces (LIP) transmit magma from reservoirs located at the asthenosphere–lithosphere boundary or at shallower depth to the surface where they feed voluminous traps, on the order of 10⁵–10⁶ km³ (e.g. Ernst et al., 2001). The composition, age, geometry, palaeomagnetism, flow pattern, and tectonics of dyke swarms in LIPs have been investigated in many other parts of the continental world (review in Ernst et al., 1995). Dyke swarms have been studied not only for locating the eruption sites of flood basalts (e.g. Swanson et al., 1975), but also for the identification of reservoirs, through magma flow direction analysis (e.g. Ernst and Baragar, 1992, Callot et al., 2001) and overgrowth texture patterns (Ohnenstetter and Brown, 1992), as well as for palaeostress trajectory retrieval at various scales on various planets (May, 1971, Féraud et al., 1987, Baer and Reches, 1991, Cadman, 1994, Grosfils and Head, 1994, Mège and Masson, 1996). Dyke swarms are thus key contributors to plume tectonics analysis.

As far as the Ethiopian LIP is concerned, the term flood basalt province should be avoided because it incorrectly reflects the composition and emplacement of the lavas erupted in response to the impingement of the Ethiopian plume on the base of the Ethiopian lithosphere, an event that took place at 30 Ma (Hofmann et al., 1997). The whole lava pile includes basaltic lava flows, basaltic tuffs, as well as a considerable volume of rhyolitic, trachytic, and phonolitic products (e.g. Mohr and Zanettin, 1988). Intermediate volcanic products are scarce but not absent (Mohr, 1963, Mohr and Zanettin, 1988). In this work we use the generic term Trap Series instead of flood basalts for describing the erupted volcanic products at the onset of mantle plume activity.

In the Ethiopian LIP a body of work has been done recently on Trap Series geochemistry and age determination (Hart et al., 1989, Marty et al., 1996, Stewart and Rogers, 1996, Hofmann et al., 1997, Pik et al., 1998, Pik et al., 1999, Ayalew and Yirgu, 2003, Coulié et al., 2003). Several local dyke swarms are exposed (Fig. 1), some of which might belong to the same, thus larger, swarm, but due to the still huge surface area covered by the Trap Series as well as the vegetation cover, exposures are usually not followed over distances exceeding kilometres. East of the East Ethiopian Rift, local swarms are observed on the rift margin and the Bale Mountains. On the rift margin, the Sagatu Ridge dyke swarm has been studied by Mohr and Potter (1976), Mohr (1980), and Kennan et al. (1990). On the northern plateau (Abyssinia), the best known dyke swarm exposures are on the Afar margin and the plain area southwest of Lake Tana, which we call the Tana–Belaya area (Fig. 2). Dykes exposed across the tilted blocks of the Afar margin along the Kombolcha–Bati Road were investigated by Abbate and Sagri (1969), Justin-Visentin and Zanettin (1974), and Mohr (1983). Reconnaissance mapping of dykes in the Tana–Belaya area, western Ethiopia, has been carried out by Jepsen and Athearn (1963a), and satellite imagery interpretation was proposed by Chorowicz et al. (1998).

Other investigated dyke swarms in Ethiopia and Eritrea include the dykes from the Angareb ring complex (Hahn et al., 1977) and the Asmara dyke swarm (Mohr, 1999). Other dyke swarms have been identified (Mohr, 1971, Mohr and Zanettin, 1988, and personal observations), but are still scientifically pristine (Fig. 1).

It is to be expected that most dyke swarms are related to one of the following events: trap eruption, Red Sea opening, or opening of the East African Rift. However, few works have attempted to correlate the dykes with their regional tectonic setting. In this paper, we report on the first field observations of the dykes in the Tana–Belaya area, western Ethiopia, that we complement with satellite imagery. This area was selected due to excellent dyke exposure compared with most other dyke swarms in Ethiopia (Fig. 2). The dyke swarms are identified and described, and their emplacement is investigated, especially in relation to basement fabric. Basement fabric, stress patterns inferred from dyke orientation, and gravity data are combined to put dyke emplacement in the geodynamic context of the Ethiopian plume.

The Tana–Belaya area is at the convergence of three geologic provinces, the uplifted Trap Series of Abyssinia to the east, the Pan-African orogen, observed to the south, and the Mesozoic–Cenozoic basins in Sudan to the west. The dykes are observed on the Ethio–Sudanese Plain, below the Abyssinian plateau (Plate I). The plain is mainly represented by the Pan-African basement south of Mount Belaya, and the base of the Tertiary Trap Series, made of basaltic flow breccias to the north (e.g. Merla et al., 1979). The plain gradually enters the Sudanese rift domain as one approaches the Ethio–Sudanese border. Alluvial sediments have also been deposited by the Nile and its tributaries.

The Pan-African basement is composed of magmatic rocks (granites, syenites), metamorphosed magmatic rocks (basic metavolcanites) metamorphosed sediments (graphitic schists, phyllites, quartzites, marbles), and also arkoses (Kazmin, 1975, Kazmin et al., 1978, Berhe, 1990, Braathen et al., 2001). It displays a ductile shear zone, the Tulu Dimtu shear zone, of mean orientation NNE (Fig. 3). The shear zone exhibits several ophiolite exposures, usually called the Tulu Dimtu ophiolite belt, one of the several Pan-African ophiolite alignments identified from Tanzania to Egypt and Sudan to Arabia (Shackleton, 1979, Shackleton, 1986, Vail, 1985, Berhe, 1990, Abdelsalam and Stern, 1996), and dated 800 Ma or younger (Berhe, 1990). Elevation maps of the Precambrian basement in Ethiopia were published by Dainelli (reproduced in Baker et al., 1972, p. 11) and, more recently, Beyth (1991). Quaternary lava flows have been observed to lie directly on the Precambrian basement (Jepsen and Athearn, 1962), and on the bottom levels of the Trap Series (field observations by the authors).

The Abyssinian plateau is composed of the Tertiary Trap Series, Miocene shield volcanoes such as the Semien (Mohr, 1967) and Choke Mountains, and Quaternary volcanics. A review of trap stratigraphy throughout Ethiopia can be found in Pik et al. (1998). The total volume of erupted traps in the Ethiopian LIP was estimated to be on the order of 400 000 km³, with an initial surface area of 750 000 km² (Mohr and Zanettin, 1988). Trap thickness reaches up to 2000 m in some areas (Jepsen and Athearn, 1962, Hofmann et al., 1997). Recently determined ⁴⁰Ar/³⁹Ar dating in northern Ethiopia has yielded 30 Ma±a few ka for most of the Trap Series (Hofmann et al., 1997, Coulié et al., 2003, Touchard et al., 2003). This very short time span is consistent with the eruption time span at most other flood basalt provinces. However, the origin of the Ethiopian LIP is not fully understood yet. The age of volcanism has been shown to be as old as 45 Ma in Kenya, an age that appears to decrease northward, where ages as young as 19–12 Ma have been found on the southern Ethiopian plateau (George et al., 1998). Late Miocene volcanism has also been documented southeast of the study area, north of Addis Ababa (Zanettin and Justin-Visentin, 1974). Hydrovolcanic craters, Quaternary cinder cones, and related lava flows are also observed (Comucci, 1950, Jepsen and Athearn, 1961). Quaternary cinder cones are observed on the plateau southwest of Lake Tana at a short distance from the Tana–Belaya area.

The plateau has undergone intense fracturing ascribed to uplift associated with trap emplacement, as well as normal faulting. Field evidence shows waning structural deformation after flood basalt emplacement, although some areas underwent Pliocene–Quaternary uplift exceeding 2000 m (Mohr and Zanettin, 1988). A couple of high intensity historical earthquakes have been reported (Jepsen and Athearn, 1963b).

Coblentz and Sandiford (1994) showed that lithospheric density variations are presently the most likely stress sources in the area. They suggested that the present-day state of stress is extensional, and that the magnitude of extensional stress in the Abyssinian plateau is the highest of the whole African plate. Bosworth and Strecker (1997) determined from borehole breakouts that the minimum stress trajectory in the Sudanese Plain west of the study area is NNE-oriented, perpendicular to the stress field predicted by Coblentz and Sandiford (1994). Inversion of fault slip data sets on faults located around Lake Tana has suggested that the maximum principal stress axis is vertical and the other principal stresses are horizontal and of equal magnitude, which was interpreted as evidence that the region undergoes active subsidence with little contribution of the regional geodynamics (Chorowicz et al., 1998). Tilted blocks observed on the western edge of Lake Tana (Jepsen and Athearn, 1963b, and Fig. 2, Fig. 11) have been interpreted as a consequence of this subsidence (Chorowicz et al., 1998).