Plume tectonics

The reassessment of the mantle plume hypothesis has become the most exciting current debate in Earth science (Foulger 2005). To appreciate the dynamics of the debate, it is useful to consider the mantle plume model before exploring the reasons for its possible demise and replacement with a plate model.

The plume hypothesis

Mantle plumes may start growing the core-mantle boundary. The mechanisms by which they form and grow are undecided. They may involve rising plumes of liquid metal and light elements pumping latent heat outwards from the inner-core boundary by compositional convection, the outer core then supplying heat to the core-mantle boundary, whence giant silicate magma chambers pump it into the mantle, so providing a plume source (Morse 2000). W. Jason Morgan (1971) was the first to propose mantle plumes as geological features. Morgan extended J. Tuzo Wilson's (1963) idea of hotspots, which Wilson used to explain the time-progressive formation of the Hawaiian island and seamount train as the Pacific sea-floor moved over the Hawaiian hotspot lying atop a 'pipe' rooted to the deep mantle. Mantle plumes may be hundreds of kilometres in diameter and rise towards the Earth's surface from the core-mantle boundary or from the boundary between the upper and lower mantle. A plume consists of a leading 'glob' of hot material followed by a 'stalk'. On approaching the lithosphere, the plume head mushrooms beneath the lithosphere, spreading sideways and downwards a little. The plume temperature is 250-300 C hotter than the surrounding upper mantle, so that 10-20 per cent of the surrounding rock melts. This melted rock may then run onto the Earth's surface as flood basalt.

Researchers disagree about the number of plumes, typical figures being twenty in the mid-1970s, 5,200 in 1999 (though these include small plumes that feed seamounts), and nine in 2003 (see Malamud and Turcotte 1999; Courtillot et al. 2003; Foulger 2005). Plumes come in a range of sizes, the biggest being megaplumes or superplumes. A superplume may have lain beneath the Pacific Ocean during the middle of the Cretaceous period (Larson 1991). It rose rapidly from the core-mantle boundary about 125 million years ago. Production tailed off by 80 million years ago, but it did not stop until 50 million years later. It is possible that cold, subducted oceanic crust on both edges of a tectonic plate accumulating at the top of the lower mantle causes superplumes to form. These two cold pools of rock then sink to the hot layer just above the core, squeezing out a giant plume between them (Penvenne 1995).

Some researchers speculate that plume tectonics may be the dominant style of convection in the major part of the mantle. Two super-upwellings (the South Pacific and African superplumes) and one super-downwelling (the Asian cold plume) appear to prevail (Figure 2.6), which influence, but are also influenced by, plate tectonics. Indeed, crust, mantle, and

Asia cold supe

Asia cold supe

Oceanic Lithosphere Composition

Figure 2.6 A possible grand circulation of Earth materials. Oceanic lithosphere, created at mid-ocean ridges, subducts into the deeper mantle, stagnating at around 670 km and accumulating for 100-400 million years. Eventually, gravitational collapse forms a cold downwelling onto the outer core, as in the Asian cold superplume, which leads to mantle upwelling elsewhere, as in the South Pacific and African hot plumes. Source: Adapted from Fukao et al. (1994).

Figure 2.6 A possible grand circulation of Earth materials. Oceanic lithosphere, created at mid-ocean ridges, subducts into the deeper mantle, stagnating at around 670 km and accumulating for 100-400 million years. Eventually, gravitational collapse forms a cold downwelling onto the outer core, as in the Asian cold superplume, which leads to mantle upwelling elsewhere, as in the South Pacific and African hot plumes. Source: Adapted from Fukao et al. (1994).

core processes may act in concert to create 'whole Earth tectonics' (Kumazawa and Maruyama 1994; Maruyama et al. 1994). Whole Earth tectonics integrates plate tectonic processes in the lithosphere and upper mantle, plume tectonics in the lower mantle, and growth tectonics in the core, where the inner core slowly grows at the expense of the outer core. Plate tectonics supplies cold materials for plume tectonics. Sinking slabs of stagnant lithospheric material drop through the lower mantle. In sinking, they create super-upwellings that influence plate tectonics, and they modify convection pattern in the outer core, which in turn determines the growth of the inner core.

The plate hypothesis

A minority of rebellious voices have always spoken out against plumes, but, since about the turn of the millennium, the number of voices has swollen and the validity of the plume model has emerged as a key debate in Earth science. Gillian Foulger (2005) gives four chief reasons for the debate that concern a mismatch between observations and prediction, a question mark over the convectional mechanism for plume generation, the lack of a testable plume hypothesis, and a limited awareness of alternative models in the Earth science community. It will pay to explore these points in turn.

1 Many observations, though by no means all, fail to support the predictions of the original plume model in at least five particulars, this despite three decades of rigorous work (Foulger 2005). First, the classic plume model predicts that volcanic tracks will extend away from the locus of current active volcanism (the 'hotspot') progressively through time, but only a few locations, including Iceland and Ascension, show this pattern. A second prediction is that hotspots will hold a fixed position relative to each other through time. However, the degree of locational fixity seems variable, some hotspots moving relative to one another at a few centimetres per year, and many island chains originally assumed time-progressive not being so (Koppers et al. 2001). Third, narrow, vertical, cylinder-like bodies of anomalously hot rock should traverse the whole mantle, linking surface hotspots to the core-mantle boundary (see Figure 2.1). Seismic tomography of mantle at putative plume locations, such as Yellowstone, Tristan da Cuhna, and the Azores (Montagner and Ritsema 2001; Christiansen et al. 2002), reveals anomalies confined to the upper mantle, or even to the lower lithosphere. Heat-flow measurements and petrology, for example at Hawaii, Louisville, and Iceland, provide little evidence of the high magma temperatures predicted for deep plumes (Breddam 2002; Stein and Stein 2003). Fourth, the chemical character of lava at hotspots should mirror their high-temperature provenance. Petrological evidence for such an origin is ambiguous, Hawaii being the only currently active hotspot associated picrite glass, which is indicative of high temperatures. Most other hotspots have no petrological evidence for high temperatures. Fifth, large igneous provinces (LIPs) should represent 'plume heads' and they should contain volcanic tracks that represent the 'plume tail'. In fact, some plumes, such as Hawaii, lack an LIP and others, including the Ontong Java Plateau and the Siberian Traps, lack time-progressive volcanic tracks. Moreover, there is no evidence that uplift predicted by the plume hypothesis preceded the emplacement of the Ontong Java Plateau, which, with a volume of 60 million km3, is the largest LIP on Earth. And, in the case of the Siberian Traps, subsidence appears to have preceded emplacement.

2 The kind of convection needed to engender mantle plumes may not occur. Physical models suggest that the formation of classical plumes may be impossible, owing to the huge pressure in the deep mantle suppressing the buoyancy of hot material (Anderson 2001). That is not to say that no convection occurs in the mantle, but it does question the mantle's ability to produce coherent, narrow convective structures that pass through its full depth and deliver samples of the core-mantle boundary layer to the Earth's surface (Foulger 2005).

3 The plume hypothesis is no longer testable because, to accommodate conflicting data, a plethora of models, all modifications of the original model, now exists. It is natural that a hypothesis will evolve to embrace new findings, but it should remain open to refutation. A growing body of geologists feel that the plume hypothesis, at least in its current elastic form, is not susceptible of disproof.

4 Little known but workable alternative models, not involving plumes, are available. Such models include edge convection, plate-tectonic processes, melt focussing, large-scale ponding, continental lithospheric delimitation and slab break-off, rifting decompression melting, and meteorite impacts. Edge convection takes the observation further that vigorous, time-dependent magmatism results from small-scale convection at continental edges where thick and cold lithosphere abuts hot oceanic lithosphere, as in the north Atlantic (King and Anderson 1998). Plate-tectonic processes provide a means of cycling crustal and mantle lithosphere materials at shallow depths, rather than via the core-mantle boundary. Melt focussing centres around the tendency of melt to concentrate in a cone-shaped region beneath some plate boundaries, including ridge-transform and ridge-ridge-ridge triple junctions. Large-scale melt ponding is the speculation that huge reservoirs of melt, capable of producing the largest of the LIPs, may form over long periods before eruption occurs, despite the usual assumption that melt is extracted from its source region as it forms, at a relatively low degree of melting. Continental lithospheric delamination and slab break-off may explain the lack of uplift before LIP emplacement reported at some sites. The idea is that the continental lithosphere may thicken, transform to dense phases such as eclogite, and catastrophically sink and detach, a process that should produce surface subsidence followed by extensive magmatism (Elkins-Tanton 2005). Similarly, if a slab should break off, it would soon alter mantle flows patterns in the collision zone and create a burst of magmatism (Keskin 2003). Rifting decompression melting is the notion that the volumes of melt produced by rifting as a continental breaks up suffice to produce the material erupted at LIPs and volcanic passive margins (Corti et al. 2003). Meteorite impacts have long been recognized as a candidate for rapidly generating the large volumes of magma in LIPs (p. 41). A key factor here seems to be pressure-release (decompression) melting that would follow the sudden excavation of a huge crater (Jones et al. 2005).

A study testing the likely origin of hotspots by scoring them according to deep plume-related and shallow plate-related criteria added to the problems with the plume hypothesis (Anderson 2005). Some 'primary' (potentially deep-seated) hotspots - Iceland, Hawaii, Easter Island, Louisville, Afar, Reunion, and Tristan da Cunha - scored well with plume criteria, but they scored poorly with criteria more appropriate for deep or thermal processes, such as magma temperature, heat flow, transition zone thickness, and highresolution upper and lower mantle seismic tomographic results. In particular, tomography failed to confirm Iceland, Easter Island, Afar, Tristan da Cunha, and Yellowstone as plume-related, revealing them as shallow features with well-defined plate tectonic explanations. For most melting anomalies ('hotspots') the plume hypothesis scored poorly against competing hypotheses such as stress- and crack-controlled magmatism, which mechanisms are associated with plate tectonics. The scoring results suggested that thermal plumes from deep thermal boundary layers are an unlikely cause of most 'hotspots'.

The anti-plume lobbyists offer an alternative explanation for volcanism in a plume-free world. Foulger (2002) notes the two basic requirements for volcanism - a source of melt (apparently without exceptionally high temperatures) and extension of the Earth's surface to allow the melt to escape. Basalt reintroduced into the shallow mantle at subduction zones causes inhomogeneity and locally enhanced fertility in the form of eclogite, which can generate exceptionally large volumes of melt at relatively low temperatures (Cordery et al. 1997). Intraplate deformation causes crustal extension far away from plate boundaries. Such deformation often occurs along such pre-existing lines of weakness as transform zones and old sutures. The latter probably are also the sites of old eclogite-bear-ing slabs trapped in the lithospheric sutures formed when continents collided. Anomalous volcanism traditionally attributed to plumes commonly occurs at such locations. Examples include volcanism in Tristan da Cunha, the Deccan Traps, Yellowstone, Iceland, and many of the Pacific volcanic chains (Smith 1993; Christiansen et al. 2002; Foulger 2002). Findings such as these, reported in a host of papers, mainly published since 1997,

Plume model

Plate model

Plume model

Plate model

Plume Model

Louisville Réunion Afar

Low-velocity Relative velocities

Louisville Réunion Afar

Low-velocity Relative velocities form the observational backbone of what Don L. Anderson (2005) calls the 'plate model', which stands in contradistinction to the 'plume model' (Figure 2.7).

Sometimes called 'platonics' to distinguish it from the kinematic theory of rigid plate tectonics, and to emphasize its shallow and ephemeral nature (Anderson 2002), the plate model offers an alternative explanation for intraplate and mid-ocean ridge volcanism. Whereas the plume hypothesis invokes concentrated and hot upwellings from the deepest mantle, the plate hypothesis involves shallow processes dominated by stress, by plate tectonics, by mantle heterogeneity, and by fertility variations (composition, volatile content, solidus), along with an asthenosphere that is near the melting point (Anderson 2005). The hope is that the plate hypothesis will unify plate tectonics, plate boundaries, global plate reorganization, normal magmatism, melting anomalies, volcanic chains, and mantle geochemistry in a single theory. Undoubtedly, the plate hypothesis simplifies views of convec-

Figure 2.7 The plume model and plate model contrasted. The schematic cross-section of the Earth shows the plume model to the left (modified from Courtillot et al. 2003 with additions from other sources) and the plate model to the right. The left half illustrates three proposed kinds of hotspots and plumes. In the deep mantle, narrow tubes (inferred) and giant upwellings coexist. Narrow upwelling plumes, which bring material from great depth to the volcanoes, localize melting anomalies. In the various plume models, the deep mantle provides the material and the deep mantle or core provides the heat for hotspots; large isolated but accessible reservoirs, rather than dispersed components, and sampling differences account for geochemical variability. Deep slab penetration, true polar wander, core heat, and mantle avalanches are important. Dark regions are assumedly hot and buoyant; lighter grey regions in the upper mantle (and the slabs subducting into the lower mantle) are cold and dense. Only a few hotspots are claimed to be the result of deep narrow plumes extending to the core-mantle boundary - different authors have different candidates. The schematic is based on fluid dynamic experiments that ignore pressure effects and, of necessity, have low viscosity relative to conductivity. The right half indicates the important attributes of the plate model: variable depths of recycling, migrating ridges and trenches, concentration of volcanism in tensile regions of the plates, inhomogeneous and active upper mantle, isolated and sluggish lower mantle, and pressure-broadened ancient features in the deep mantle. Low-density regions in both the shallow and deep mantle produce uplift and extension of the lithosphere. Stress conditions and fabric of the plate and fertility of the mantle localize melting anomalies. Large-scale features are consistent with the viscosity-conductivity-thermal expansion relations of the mantle. In the plate model, the upper mantle (down to about 1,000 km - the Repetti Discontinuity) contains recycled and delaminated material of various ages and dimensions. These materials equilibrate at various times and depths. Migrating ridges, including incipient ridges and other plate boundaries, sample the dispersed components in this heterogeneous mantle. The upper 1,000 km (Bullen's Regions B and C) is the active and accessible layer. The deep mantle (Regions D' and D"), although interesting and important, is sluggish and inaccessible. The geochemical components of mid-ocean ridge basalts, oceanic island basalts, and so forth are in the upper mantle and are mainly recycled surface materials. Dark and light grey regions in the upper mantle are respectively low and high seismic velocity regions, not necessarily hot and cold, although some of the dark regions at the top and base of the mantle are due to the presence of a melt. Source: After Anderson (2005).

tion in the Earth (Foulger 2003). The plume hypothesis demands two independent modes of convection - plate tectonics and plumes. Ridge push and slab pull forces at plate boundaries drive plate tectonics; heat from the Earth's core powers plumes. Platonics only requires plate tectonics, with volcanism that appears anomalous in its location, in its distribution, or in its volume rate explained by inhomogeneity imparted to the mantle by subducting plates and by intraplate deformations that occur preferentially along pre-existing lines of weakness.

Was this article helpful?

+1 0


  • Tarquinia
    How many Pacific hotspots are fed by deepmantle plumes?
    9 years ago
  • Lucille Morris
    What plumes in convectional cells are?
    9 years ago

Post a comment