Rifting of continents is usually explained by one of two mechanisms based on effects that originate far from the zone of rifting. Laboratory experiments show that this geodynamic process can also be caused by local effects.
Vast continental regions have experienced volcanism precisely where 1,000-kilometre-scale crustal blocks were pulling apart. For example, such rifts began to cut across much of Africa about 140 million years ago, and distributed, low-flux volcanism continues in that region today (Fig. 1). Such broadly distributed, synchronous activity is hard to fit into standard theories of rifting and volcanism. Writing in the Journal of Geophysical Research, Fourel et al.1 suggest an explanation for this activity based on laboratory experiments with fluids whose densities depend on temperature and composition.
Radiation of heat to space cools the strong outer layer of the Earth, called the lithosphere, which overlies the hot, convecting interior. Minerals contract as they cool, and this can make the cold lithosphere denser than the interior. This thermal density contrast is what makes sub-oceanic lithosphere sink and so drive the motion of the planet's tectonic plates2. Continental lithosphere is also cold and yet it does not sink, and this may be because it is composed of intrinsically lighter minerals. As long as compositional density differences between the lithosphere and the interior are greater than thermal density differences, the lithosphere will float on top of the hot, fluid interior.
Fourel et al. discuss cases in which the bottom part of the continental lithosphere cools enough for thermal density differences to dominate compositional ones. The dense lower lithosphere then becomes unstable and begins to sink into the hot interior. Between sinking lithospheric blobs, melting of hot upwelling mantle generates magma that can feed volcanoes. The intrusion of this magma into the lithosphere would also allow rifting to proceed even at the moderate extensional stress levels produced by the density-driven lithospheric flow.
In their elegant laboratory models, Fourel and colleagues1,3 use two viscous fluids to simulate possible interactions between a compositionally buoyant lithosphere overlying a weaker mantle layer. Diffusion of heat across the thin high-viscosity layer eventually causes thermal density differences to exceed the compositional density differences. This drives an oscillatory instability at the interface between the two fluid layers. Their analysis of these and other experiments indicates that the development of this instability on Earth requires a large region of fairly uniform, cooling lithosphere. The required size depends on the thickness of the lower lithosphere that can flow under modest stress levels, and, for reasonable values of this thickness, the region becoming unstable must be at least 1,000 km in radius. This is about the size of Australia, the smallest present-day continent, and, as noted by Fourel et al., it is a region where distributed rifting occurred about 800 million years ago.
The authors' analysis offers an explanation for why rifting does not seem to affect extremely old continental regions such as the Tanzanean craton. Evidence suggests that there has been a steady change in the composition and density of the lithosphere with time4. Lithosphere that formed in the first half of Earth's history seems to be too buoyant to sink, even though such old lithosphere can be extremely thick, as much as 250 km (ref. 5). Thus, only lithosphere formed in about the past 2 billion years seems to have the correct composition to undergo density-driven rifting and distributed volcanism.
In this new model, the stresses that drive rifting arise locally from the density structure of the lithosphere that is rifted. By contrast, the two most widely discussed mechanisms for continental rifting call on processes that originate far from the zone of rifting. In the passive rifting model6, stresses transmitted laterally from the lithospheric plate edge cause local weak spots to extend and thin. A major problem with the passive model is that the lithosphere may be too strong to extend, given reasonable magnitudes of stress7. In the active rifting model6, plumes of hot material from deep in the Earth, perhaps from the core–mantle boundary, rise and push the surface up, causing extensional stress over the hot upwelling8.
The association of most major continental break-up events with a massive outpouring of magma, which has become clearer with more precise dating of magmas and improved geophysical imaging of buried bodies of magma, favours the active model of rifting. The small rifts discussed by Fourel et al. are associated with much smaller magmatic output, but in both cases the magma may be the key to allowing rifting to happen at all. The presence of magma should allow the lithosphere to rift at much lower stress levels than without magma. Small magma fluxes may not allow enough heating and weakening of the lithosphere to lead to continental break-up9. This may be why these small intracontinental rifts are sometimes called failed rifts.
Volcanism that occurs away from plate boundaries is usually attributed to upwelling and melting of mantle plumes. Plumes are thought to be associated with a fairly high rate of magma production, and thus volcanism, in a localized zone10. Therefore, the extremely low rate of volcanism in multiple, widely distributed locations across West Africa is a problem for the plume model.
Instability of cool lower lithosphere offers an explanation for how distributed rifting and volcanic activity have affected many parts of the continents. However, the laboratory experiments that inspired this model avoided using strong variations in viscosity with temperature that are a key feature of Earth's lithosphere. The fact that the coldest and most negatively buoyant parts of the lithosphere are also the strongest may act to mute the instability. Therefore, the concept of a lower lithospheric instability needs to be investigated further using numerical techniques that can handle the kinds of temperature-dependent viscosity changes that are difficult to simulate in the laboratory.
Fourel, L., Milelli, L., Jaupart, C. & Limare, A. J. Geophys. Res. http://dx.doi.org/10.1002/jgrb.50218 (2013).
Forsyth, D. W. & Uyeda, S. Geophys. J. R. Astron. Soc. 43, 163–200 (1975).
Jaupart, C., Molnar, P. & Cottrell, E. J. Fluid Mech. 572, 433–469 (2007).
Djomani, Y. H. P., O'Reilly, S. Y., Griffin, W. L. & Morgan, P. Earth Planet. Sci. Lett. 184, 605–621 (2001).
Jordan, T. H. Nature 274, 544–548 (1978).
Sengör, A. M. C. & Burke, K. Geophys. Res. Lett. 5, 419–421 (1978).
Hopper, J. R. & Buck, W. R. J. Geophys. Res. 98, 16213–16221 (1993).
Hill, R. I. Earth Planet. Sci. Lett. 104, 398–416 (1991).
Bialas, R., Buck, W. R. & Qin, R. Earth Planet. Sci. Lett. 292, 68–78 (2010).
Sleep, N. H. Annu. Rev. Earth Planet. Sci. 20, 19–43 (1992).
About this article
Chapter 19 Geological history of the Kaladgi–Badami and Bhima basins, south India: sedimentation in a Proterozoic intracratonic setup
Geological Society, London, Memoirs (2015)