On causal links between flood basalts and continental breakup
Introduction
Within the framework of plate tectonics, seafloor spreading at ridges is taken to be the surface expression of `normal' mantle convection. In contrast, hotspot volcanism is often interpreted as the surface expression of a different, deep-seated form of mantle convection, linked to some more localized instabilities. It is generally assumed that there is little interaction between the two modes of mantle convection and related volcanism. In his landmark 1971 paper, Morgan [1]pointed out the close temporal coincidence between flood basalt (FB) volcanism and continental breakup. He was the first to propose a connection between transient stages of plume activity and the onset of normal seafloor spreading. Based on laboratory experiments, Morgan [2]next associated FB volcanism with the formation of a new plume head and subsequent hotspot volcanism with continued activity from its tail.
Some of these ideas were based on rather scant data, particularly regarding the age and duration of FB volcanism. Work on the Deccan traps of India established that several million km3 of basaltic lava had erupted in a million years or even less, about 65 Ma ago [3]. Richards et al. [4]recognized some ten flood basalt provinces as the traces of the heads of an equal number of plumes, which they associated with continental breakup, suggesting that rifting occurred shortly after trap emplacement and that lithospheric thinning was not a prerequisite for FB eruption. This is a member of a family of `active' rifting models 5, 6, 7, 8. Active rifting is supposed to be due to a combination of several processes, including impingement of the plume head at the base of the lithosphere, stripping of the thermal boundary layer, lateral spreading and conductive heating of the overlying mechanical boundary layer, small-scale convection in the plume head eroding the lithosphere, and finally uplift and tensional failing of the upper, colder mechanical lithosphere.
In contrast, `passive' models assume that rifts are primarily the result of regional stress distributions related to plate boundary forces and geometry, acting for instance on pre-existing zones of weakness either in the crust, or the mantle part of the lithosphere or both. White and McKenzie [9]proposed such a `passive' model, based on the physics of partial melting at rift zones. In cold rifts [10], continental crust is extensively stretched, with little accompanying volcanism, until it breaks and a new oceanic spreading center is formed. A hot rift is created when stretching and rifting occur over the hot, steady state, tail of a mature plume. Independently driven extension is a prerequisite for production of abnormal volumes of volcanic material by passive decompression melting of upward rising asthenospheric material. White and McKenzie [9]proposed that hot and cold rifts are equally likely. They argued that many hotspots are not initially linked with rifting (including Hawaii, although traces of the plume head, if any, have likely disappeared in the Kurile–Aleutian subduction zones) and that many continental rifts have been generated away from the influence of hotspots.
However, in a comprehensive review, White and McKenzie [11]concluded “the peak of extrusive activity in FB provinces on rifted continental margins shortly predates the oldest magnetic anomaly, which marks the onset of seafloor spreading”. High resolution mapping and dating of both FB provinces and rifted margins should provide further constraints on whether this is a general situation.
Anderson [12]has recently summarized very different views. He questions key hypotheses he believes underlie the plume model. He concludes that flood basalt volcanism is entirely explainable by near-surface processes, and that related mantle upwelling and melting are due to large lateral thermal gradients at the edges of the cratons. This model requires relatively large mantle temperatures, or a volatile-rich mantle in order to generate the observed volumes of basaltic melts. The alternative models available have strong implications for mantle convection and for extension processes. Key observations are the magnitude of extension and the temporal relationship between volcanism and extension.
The purpose of this paper is to bring together recent data on the spatial association and temporal sequence of rifting and trap emplacement, which may shed additional light on a causal relationship. We begin with the most recent large flood basalt event, the Ethiopian traps. Because they are still young and the African plate moves only slowly over the mantle, the plume head and tail should still be close to one another, and the associated rifts are still young and narrow. Moreover, the supposed location of the plume tail in the Afar depression is not yet covered by water. Hence, this is a unique region in which all relevant geological features are readily observable across a rather limited area. Next, we briefly review other pairs of FB provinces and rifted margins (Fig. 1) and conclude that the models for their generation should include both `active' (in the sense defined above, i.e. associated with a local mantle upwelling) and `passive' aspects.
Section snippets
The Ethiopian/Yemen traps, Afar plume and East African, Red Sea and Gulf of Aden rifts
Since the early years of plate tectonics, the Afar area has been considered as one of the best examples of a triple rift junction (Fig. 2). The western Gulf of Aden opened by westward rift propagation, with formation of oceanic crust in a single phase since 10–15 Ma 13, 14. Girdler and Styles [15]had interpreted magnetic and other geophysical data in the Red Sea in terms of two distinct phases of seafloor spreading, one between 30 and 15 Ma, and a younger one since about 5 Ma. These were partly
Other plume–rift (or trap-breakup) associations
An extensive summary can be found in White and McKenzie [9], supplemented by updates on trap ages and durations by Courtillot [18]. Omitting the Columbia FB, which are an order of magnitude smaller than others, we go backwards through the last 250 Myr and summarize our present state of knowledge on trap ages and durations, ages and locations of continental breakup and subsequent seafloor spreading.
Discussion
There is now enough evidence that the association of flood basalts with continental rifting is the rule rather than the exception. Opening of the Red Sea and Gulf of Aden, North Atlantic, Arabian Sea, South Atlantic, Southwest Indian Ocean and Central Atlantic can all be linked, and occur in close time association with emplacement of a new flood basalt province. As a result, all known major traps in the last 200 Ma can be associated with a new ocean basin. Even the older Siberian and Emeishan
Acknowledgements
We thank Ian Campbell, Geoff Davies and Bob Duncan for careful, constructive reviews and Francis Albarède and Anny Cazenave for editorial assistance and comments. We thank L.E. Ricou and J. Marcoux for comments on some geological dating aspects of the older traps. We would like to extend particular thanks to Don Anderson for a long, detailed and very critical review, which we solicited, yet could not in this paper do full justice to (given the legitimate but drastic reduction requested by the
References (70)
- et al.
Deccan flood basalts at the Cretaceous/Tertiary boundary?
Earth Planet. Sci. Lett.
(1986) - et al.
Implications of mantle plume structure for the evolution of flood basalts
Earth Planet. Sci. Lett.
(1990) Starting plumes and continental breakup
Earth Planet. Sci. Lett.
(1991)Opening of the Gulf of Aden and Afar by progressive tearing
Phys. Earth Planet. Inter.
(1980)- et al.
A brief Oligocene period of flood volcanism in Yemen: implications for the duration and rate of continental flood volcanism at the Afro-Arabian triple junction
Earth Planet. Sci. Lett.
(1996) - et al.
The tectonic development of the western margin of the Gulf of Elat (Aqaba) rift
Tectonophysics
(1981) - et al.
40Ar/39Ar dating of alkaline and tholeiitic magmatism of Saudi Arabia related to the early Red Sea Rifting
Earth Planet. Sci. Lett.
(1991) - et al.
Did Deccan volcanism pre-date the Cretaceous/Tertiary transition?
Earth Planet. Sci. Lett.
(1993) - et al.
Comment on: “Did Deccan volcanism pre-date the Cretaceous–Tertiary transition?”
Earth Planet. Sci. Lett.
(1994) - et al.
Early spreading and continental to oceanic basement transition beneath the Indus deep-sea fan: northeastern Arabian Sea
Mar. Geol.
(1997)