January 2005 LIP of the Month

Corresponds to event #2 in LIP record database.

The Ethiopian Large Igneous Province

Nicholas Arndt
LGCA, Univ. Grenoble, Grenoble, France
write Firstname.Lastname_AT_ujf-grenoble.fr
(substitute "Nicholas.Arndt@" for "Firstname.Lastname_AT_")

Martin A. Menzies
Dept. of Geology, Royal Holloway, University of London, Egham, UK
m.menzies@gl.rhul.ac.uk

Introduction
The Ethiopian flood basalts and associated felsic pyroclastic rocks erupted mainly around. 30 Ma ago during a relatively short, <5 Myr period (Figs. 1, 2) (McDougall et al., 1975; Berhe et al., 1987; Hoffman et al., 1997; Chernet et al., 1998; George et al., 1998; Rochette et al., 1998; Ukstins et al., 2002; Kieffer et al., 2004). Following this peak of activity, large shield volcanoes developed on the surface of the volcanic plateau and after this volcanism was confined to regions of rifting or local volcanic centres (Mohr, 1983; Mohr and Zanettin, 1988). More complex bimodal (i.e. mafic and felsic volcanism) magmatic systems were associated with such centres, in some cases involving caldera-forming eruptions. The rift that opened along the Red Sea and Gulf of Aden separating the Arabian and African continents, isolated part of the volcanic plateau in Yemen and Saudi Arabia (Chazot and Bertrand, 1993; Baker et al., 1996a; Menzies et al., 2001). Syn-rift volcanic activity continues to the present day on the uplifted rift shoulders of the Red Sea/Gulf of Aden and on ocean islands in the Red Sea and Gulf of Aden (Vidal et al., 1991; Schilling et al., 1992; Deniel et al., 1994).

The Ethiopian flood basalts are the youngest example of a major continental volcanic plateau. Because of their relatively young age and their eruption in a region where plate movements are slow, a complete record of volcanic and tectonic activity is preserved. The nature of volcanism has changed markedly during the Oligo-Miocene-Quaternary and includes : (a) high-flux, fissure-fed (Hawaiian), mafic flood volcanism (29-31 Ma) producing 1-2 km of mafic eruptives; (b), low flux, bimodal mafic-felsic volcanism from stratovolcanoes with associated caldera-forming (Ultra-Plinian) eruptions (26-29 Ma), producing ca. 1 km thickness of mafic-felsic eruptives. (c) minor mafic volcanism during a volcanic hiatus coincident with continental rifting and break-up (<19-26 Ma); (d) syn-rift bimodal mafic-felsic volcanism associated with rift valley formation and magmatic segmentation (Maguire et al., 2003) (<5 Ma).

The region represents an ideal situation to study the dynamics of mantle sources involved in continental volcanism and the manner in which sources interacted with continental lithosphere (e.g., Mohr and Zanettin, 1988; Berhe et al 1987; Baker et al., 1996a; Hofmann et al., 1997; Pik et al., 1998, 1999; Baker et al., 2000, 2002; Ukstins et al., 2002; Ukstins-Peate et al., 2003; Kieffer et al., 2004).

Geological Background

The Ethiopian flood basalts (or traps) cover an area of about 600,000 km2 forming a layer of mafic and felsic volcanic rocks (Fig. 1). The thickness of this layer is highly variable but reaches 2 km in some regions. The total volume of extrusive and hypabyssal rocks has been estimated by Mohr (1983) and Mohr and Zanettin (1988) at about 350,000 km3.

The mineralogical and chemical composition of the flood basalts is relatively uniform. Most are aphyric to sparsely phyric, and contain phenocrysts of plagioclase and clinopyroxene with or without olivine. Most have tholeiitic to transitional compositions (Mohr, 1983; Mohr and Zanettin, 1988; Pik et al., 1998) (Figs. 3, 4). Inter-layered with the flood basalts are alkali picrites, and, particularly at upper stratigraphic levels, felsic lavas and pyroclastic rocks of rhyolitic, or less commonly, trachytic compositions (Ayalew et al., 2002; Ukstins et al., 2002; Ukstins-Peate et al., 2003).

Pik et al. (1998, 1999) and Keiffer et al. (2004) divided the basaltic rocks into several types on the basis of trace-element and Ti concentrations. They recognized a suite of “low-Ti” basalts (LT) characterized by relatively flat rare-earth element (REE) patterns and low levels of Ti and incompatible trace elements (Figs. 5, 6). These rocks are largely restricted to the northwestern part of the province, as shown in Fig. 1. Alkali basalts with higher concentrations of incompatible elements and more fractionated REE patterns — the so-called “high-Ti” basalts (HT1 and HT2) — are found to the south and east. The HT2 basalts are slightly more magnesian than the HT1 basalts and commonly are rich in olivine ± clinopyroxene phenocrysts. They have higher concentrations of incompatible elements and show extreme fractionation of the REE. Although most of the volcanic plateau erupted in a short period between 31 and 30 Ma (Hofmann et al., 1997), the upper parts of the flood volcanic pile have ages to around 25 Ma, suggesting a relatively protracted period of flood volcanism (Ukstins et al., 2002).

Shield volcanoes (Figs. 7, 8), a conspicuous feature of the Ethiopian plateau, are composed predominantly of alkaline basalts, with minor felsic lavas and pyroclastic rocks near their summits (Mohr and Zanettin, 1988; Keiffer et al., 2004). The basal diameters of the shields range from 50 to 100 km and the highest point in Ethiopia, the 4533 m high peak of Ras Dashan, is the present summit of the eroded Simien shield. The ages of the shield volcanoes range from 30-31 Ma for the Simien shield in the north of the province, essentially the same as that of underlying flood basalts, to less than 3 Ma for shields in the south of the province (Zanettin, 1992; Ukstins et al., 2002). The tholeiitic Simien shield surmounts tholeiitic flood basalts and the two probably form parts of the same magmatic system. The 23 Ma Choke and Guguftu shields in the central northern part of the plateau are alkaline and they overlie a thin sequence of 23 Ma alkaline flood basalts. Finally, still younger silica-undersaturated lavas erupted from 11 Ma (Guna) to recent times, in shields or small cones, in various parts of province.

Thick sheets of felsic pyroclastic rocks, dacites and rhyolites, make up a significant proportion, up to 30%, of the upper parts of the volcanic plateau. The geology, geochemistry and petrogenesis of these felsic rocks has been described recently by Ayalew et al. (2002), and Ukstins-Peate et al. (2003, 2005). Volcano-, chemo-, and chrono-stratigraphic correlations between Yemen and Ethiopia indicates that the source of these felsic rocks may relate to volcanic centres (e.g. calderas) now located in Yemen (Ukstins-Peate et al., 2005) (Fig. 9). Pyroclastic fall deposits from Plinian to UtraPlinian eruptions during the Oligo-Miocene in Ethiopia-Yemen are found 3000 km from their source (Touchard et al., 2003; Ukstins-Peate et al., 2003).

The post-trap, syn-rift volcanism in the regions of active rifting around Addis Ababa and Djibouti and in the Arabian Peninsula has been the subject of numerous publications (e.g. Justin Visentin et al., 1974; Zanettin et al., 1978; Barrat et al., 1990; Deniel et al., 1994; Baker et al., 1998, 2000, 2002). With the exception of the stratoid unit, a small sequence of basaltic sheet flows, the volcanic activity is largely confined to small bimodal, mafic-felsic stratovolcanoes.

Is the Ethiopian Plateau a Typical Flood Basalt Province?
Type examples of continental flood basalts, such as those of the Deccan and Karoo provinces, are described as thick, monotonous sequences of thick, continuous, near-horizontal flows of tholeiitic basalt. In contrast, the Ethiopian province is made up of a series of flood basalts overlain by large and conspicuous shield volcanoes. Are the differences real, or are they a consequence of differing extents of erosion and degrees of preservation, compounded by incomplete sampling of these vast volcanic structures?

Ethiopia is the youngest major flood basalt province. It is one of the least deformed, and probably the only one in which the uppermost volcanic units are well preserved. Although the upper levels of other, older provinces such as the Deccan and Karoo are strongly eroded, there is solid evidence that flat-lying tholeiitic basalt flows erupted from start to finish of volcanic episode, and that shield volcanoes never formed. In these examples, the main phase of flood volcanism produced large horizontal flows that can be followed for many kilometres, and are difficult to relate with any specific volcanic centre. No direct link between flood and shield volcanism can be established. The situation is different in Ethiopia where the upper part of the volcanic sequence is dominated by the large shield volcanoes. Keiffer et al. (2004) present geological and geochemical data that indicates that in many cases the shield volcanism was comagmatic with the flood volcanism that formed the main volcanic plateau.

Distinctive petrological features of the Ethiopian plateau are the transitional tholeiitic to alkaline magmatic character of the mafic lavas, in contrast with the tholeiitic character of most continental flood basalts, and the high proportion of felsic pyroclastic rocks.

Origin of the lavas
The concentrations of incompatible trace elements vary enormously in lavas from the Ethiopian plateau. At one extreme, the LT tholeiites have flat normalized trace element patterns, with mantle normalized Nb contents (NbN) around 10 times primitive mantle; at the other, the HT2 type alkali picrites have steeply dipping patterns and NbN contents more than 100 times primitive mantle. The various lava types have similar Nd and Sr isotopic compositions but very different Pb isotopic compositions (Fig. 10). Some have a weak “HIMU” (high 206Pb/204Pb and low 87Sr/86Sr) character. Lavas in a large region surrounding the volcanically active Afar region have high 3He/4He ratios (Marty et al., 1996).

In the literature, these variations are interpreted in terms of derivation of the parental magmas from various parts of the mantle; some from “the Afar plume”, or from various plumes, others from the lithospheric mantle, and still others from the asthenosphere. Keiffer et al. (2004) present another model. The major and trace element composition of the magmas are explained in terms of melting, to various degrees, of different components of a common, deep mantle source. Whether this source should be described as a plume is a matter of debate: it was large and hotter than ambient mantle; it was heterogeneous, both compositionally and thermally; and it was capable of producing large volumes of magma over a protracted period. Comparison can be made with volcanism in Polynesia, where magmas of diverse chemical and isotopic compositions are derived from a heterogeneous mantle source within the Pacific “superswell”. The high MgO contents, the high trace element contents and the steeply dipping HREE patterns of the HT2 picrites reflect low-degree melting of a peridotite source at depths well below the base of the lithosphere.

Baker et al. (2000) showed that zoned phenocrysts in highly magnesian mafic volcanic rocks (e.g., picrites/ankaramites) display marked variations in Sr-Nd-O isotope composition. Such highly magnesian rocks can become highly contaminated with crustal rocks. However once the effects of minor crustal contamination are stripped from the compositions of the LT lavas, the composition of the parental magma is seen as well to be picritic and consistent with derivation from a sub-lithosphere source. Although xenoliths from the lithospheric mantle below Ethiopia and the Arabian Peninsular have a weak HIMU character like that of some of the lavas, this feature is thought to have been transferred to the lithosphere from the deep mantle source of the lavas: it does not indicate that the lithosphere was the source of these magmas (Baker et al., 1998, 2002).

The felsic volcanic rocks in Ethiopia-Yemen, occurring as pyroclastic flow and fall deposits, are highly fractionated with rhyolitic-trachytic-dacitic compositions (Fig. 4). Some of the most evolved felsic rocks occur as co-ignimbrite fall deposits and can be matched with proximal ignimbrites in Ethiopia and Yemen (Ukstins-Peate et al., 2003, 2005). A silica gap is apparent between the mafic (basaltic) and felsic (rhyolitic) volcanic units indicative of a complex petrogenetic process, as yet not fully understood. The presence of proximal and distal pyroclastic flow and fall deposits has important implications for magmatic systems within flood basalt provinces. Clearly major magmatic centres (e.g. stratovolcanoes) had developed by 29 Ma and the collapse of these volcanic structures to produce calderas, fuelled Plinian to Ultraplinian eruptions.

Relationship between magmatism and tectonics

Systematic dating of volcanic and tectonic activity provides strong constraints on how the mantle source of the Ethiopian plateau interacted with the lithosphere, and how this interaction was linked to rifting (Davison and Rex 1980). Menzies et al. (1992, 1997, 2001) have used geological observations, detailed geochronology and fission track studies to establish the timing of magmatism and rifting. Chazot et al. (1998) demonstrated that the spatial distribution of pre-rift alkaline and tholeiitic volcanism from Ethiopia to Jordan mimicked the orientation of the “proto-Red Sea rift”. This may indicate an early (é30 Ma) control on rift orientation and formation. In Ethiopia, magmatism (ca. 31 Ma) predated break-up extension by è5 Myr (Baker et al., 1996a; Hoffman et al., 1997; Menzies et al., 1997, 2001; Ukstins et al., 2002). A significant volcanic hiatus occurred in Ethiopia (ca. 22-26 Ma) and coincided with a period of erosion and extension (19-26 Ma) on the conjugate margin in Yemen. In the main part of the Ethiopian (and Yemeni) plateau (the SW portion is older, George and Rogers, 2002) metre-scale uplift seems to have preceded the onset of flood volcanism, whereas kilometer-scale uplift was synchronous with main phase of flood volcanism. Studies of the absolute and relative timing of rift processes in Ethiopia and Yemen indicates that uplift and magmatism were followed by break-up extension and erosion. There is no unequivocal evidence for either widespread extension or kilometer-scale uplift significantly prior to magmatism. As proposed by Courtillot et al. (1999) the plume heated and weakened the lithosphere, processes that eventually led to localized rifting. Since the break-up of Africa and Arabia, active magmatism and tectonics has been centred in the northern Ethiopian rift valley. In northern Ethiopia, Maquire et al. (2003) used geophysical data to illustrate the switch from border fault-dominated rifting to magmatic segmentation, an important part of the continent-ocean transition. Similarly Klausen and Larsen (2002) demonstrated a switch from tectonic rifting to magmatic segmentation on the Greenland margin (associated with the North Atlantic Igneous Province).

Subduction to the northeast provided drag that led eventually to successful rifting, first westward along the Gulf of Aden, then southward along the Red Sea: the rift propagated towards the weakened lithosphere below the centre of the volcanic province.

References