Roadmap | The review process | Home

Discussion of :

Boundary layer fabric, dynamics and the origin of midplate volcanoes,

by D.L. Anderson


16th March, 2011, Gillian R. Foulger

Don: Thank you for this contribution. Can you explain where MORB comes from, in your LLAMA model?

18th March, 20011, Don L. Anderson

Ocean island basalts (OIB) derive from the boundary layer (BL), between the plate and the so-called “convecting mantle”. Although it is never precisely defined the latter is what the conventional wisdom source of midocean ridge basalts (MORB) is called. In the BL model the source of MORB remains “the convecting mantle”, also called DMM (depleted MORB mantle), or DUM (depleted upper mantle). This is the mantle that flows in as plates diverge at ridges; it is not part of the upper BL of the mantle. LLAMA is neither the “convecting mantle” nor is it DUM. It is the sublayer beneath the plate (e.g., Elder, 1976). It is a laminated, laterally advecting mantle assembage that includes the lithosphere, lid, asthenosphere and melt-rich arrays. Its presence resolves the paradoxes with conventional 1D geochemical and thermal history models (e.g., Anderson, 2001, 2007, 2010; Hart et al., 1992; Korenaga, 2003).

BLs are broken up, diluted or dissipated in the vicinity of ridges where deeper mantle flows in. At the onset of rifting, or seafloor spreading, enriched material from the BL provides the initial magmas. Ancient isotopic signatures are more evident than at ridges where such evidence is overwhelmed by the more voluminous MORB magmas. This explains why MORB signatures are missing in isotope mixing curves, and why they are not the Common, or FOZO component of midplate magmas (e.g., Hart et al., 1992). Although the refractory lamellae in LLAMA are virtually U-free and can retain ancient helium isotope signatures (high 3He/4He) because they are cold (low diffusivities), the enriched lamellae may contain most of the “missing” (in the standard model) heat-producing elements. Since LLAMA is the shallowest layer in the mantle, away from ridges, it provides the ubiquitous (albeit minor) components (e.g., FOZO) in oceanic basalts. MORB only appears in quantity at mature ridges and is purest along fast spreading ridges. Recycling, delamination and slab fluids contaminate the shallowest mantle but not, to the same extent, the deeper NMORB source.

A basic question in mantle dynamics is whether BL flow is mainly buoyancy-driven or shear-driven. The fabric–i.e. the scales and statistical properties–of BLs are different for the two cases and thus should be resolvable by seismological techniques. Shear-driven flows tend to be stratified and laminar in the vertical direction and streaky in the horizontal direction. Buoyancy-driven flows are dominated by a few strong upwellings and the familiar Rayleigh-Benard hexagonal patterns in plan view. In this case, one does not expect long-range coherent anisotropy, a defining characteristic of LLAMA, with its laminated lithology and mesoscale anisotropy.

In buoyancy-dominated BL flow the primary flow structure is determined by updrafts (plumes) while in shear-driven BLs there are long streaks or “tracks” roughly aligned with the mean motion direction (Moeng & Sullivan, 1994). Shear-flows tend to be layered. In buoyancy driven BLs the structures generally span the entire depth of the fluid while in the shear-driven case the vertical extent of coherent structures is much less. In both cases, the fluctuations in all properties and motions are much larger in the BL than below it. Seismic heterogeneity is therefore a defining characteristic of a BL. Seismic anisotropy is a characteristic of a shear-driven BL.

The azimuthal component of mantle anisotropy has its fast axis aligned with the flow direction, which is roughly parallel to surface fracture zones. This direction is mainly EW in the eastern and central Pacific mantle. These trends are interpreted as the long streaks evident in shear-driven laboratory flows and are more robust indicators of mantle flow directions that so-called hotspot tracks. Seismic anisotropy does not confirm that the famous Hawaiian-Emperor bend is a deep feature or that it reflects an abupt change in plate motion.

BLs can be neutral, stable, shear-driven or buoyancy driven. Seismology clearly indicates that LLAMA is a ~200 km thick shear BL, and one therefore expects shear-driven upwellings (and downwellings). Most models of small-scale convection and midplate magmatism invoke unstable or buoyancy-driven BLs. However, shear-driven and passive upwellings dominate the dynamics  of LLAMA. These are mesoscale side-effects of the global or synoptic buoyancy-driven plate-slab system. Just as in the atmosphere and the oceans, upwellings do not need to be driven locally by their own buoyancy. BLs have their own convective patterns and fabrics, and can also differ in composition from the main convecting system since they collect the flotsam from the cooling mantle, from the tops of slabs and from foundering crust.

Basic questions in mantle geochemistry are, what is ambient mantle, and what is anomalous? Non-MORB are usually treated as anomalous and “therefore” not from the upper mantle. However, “normal” midocean ridge basalts (NMORB) may not be from “ambient” or “normal” mantle except in regions where existing BLs have been entrained and carried away by diverging plates.

Mantle potential temperatures are maximum near the base of a convective BL (e.g., Elder, 1976) and this effect is accentuated by internal heating, secular cooling and recumbent cold slabs. Ridges appear to tap colder mantle than exists deep in a BL (Anderson, 2010). Recall that the lower parts of a shear-driven BL are almost stationary and sources there define a relatively fixed reference system. Since mid-ocean ridges are fuelled from “the convecting mantle” they are not expected to be as stationary, or even as hot, as volcanoes that are fuelled from deep in the surface BL. LLAMA is a low-viscosity layer that decouples plate motions from the interior. This low viscosity probably makes plate tectonics possible. Narrow buoyant upwellings from deep in the mantle, if they exist, are blown around by winds in “the convecting mantle” and will not be vertical or fixed, and will not create parallel volcanic chains. These are three of the defining characteristics of the original deep mantle plume hypothesis but are more representative of shallow sources.

In summary, the source of normal midocean ridge basalt may be mantle displaced by slabs or subplate mantle that flows laterally toward the thinnest spots in the plate, e.g., midocean ridges. This upward and lateral flow provides the mass balance to diverging and subducting plates, and completes the plate tectonic cycle.


  • Anderson, D. L. 2001. Topside Tectonics?, Science, 293, 2016-2018.
  • Anderson, D. L. 2007. New Theory of the Earth 2nd Edition. Cambridge: Cambridge University Press, 384 pp; doi: 10.2277/0521849594.
  • Anderson, D.L. 2010. Hawaii, boundary layers and ambient mantle–geophysical constraints, J. Petrology, published online December 2, 2010, doi: 10.1093/petrology/egq068, available on-line at Journal of Petrology, Advanced Access.
  • Elder, John 1976. The Bowels of the Earth, John Elder, Oxford University Press, pp. 222.
  • Hart, S. R. et al. (1992), Mantle plumes and entrainment—Isotopic evidence, Science, 256, 517–520.
    Korenaga, J., 2003. Energetics of mantle convection and the fate of fossil heat, Geophys. Res. Lett., 30, 1437, doi:10.1029/2003GL016982.
  • Moeng, C.-H., and Sullivan, P.P. 1994. A comparison of shear and buoyancy driven planetary-boundary-layer flows. J. Atmospheric Sciences, 51, 999–1022.

31 July, 2013, Don L. Anderson

It has long been taught in geophysics and planetary physics courses, but not in mantle geochemistry courses, that the Earth started hot and was extensively differentiated during accretion. This knowledge goes back to the extremely influential papers of Francis Birch, including his classic 1952 paper, his 1965 Presidential Address and his energetics-of-core-formation papers. These papers form the foundations of modern geophysics but they are the antithesis of the Urey school of geochemistry (Urey, 1952) which produced many advocates of cold accretion, primordial Earth, crustal growth, continuous differentiation  and undegassed mantle. Thus, 1952 was a pivotal year for both mantle geophysics and mantle geochemistry. The two sciences diverged from that point on.

Schilling (Nature 242, 1973, page 565) states "Contrary to earlier views (Birch 1965), the model implies that plumes are transporting to the Earth's surface more primordial mantle material(s) than present in the low-velocity layer lying beneath those mid-ocean ridge segments remote from plumes." This was the nucleus of the geochemical version of the plume hypothesis. Tatsumoto (1978) and O'Hara (1973) almost immediately showed the flaws in Schilling's argument, and Tozer (1973 ) demonstrated the fluid dynamic implausibility of the geochemical model. Birch’s papers had already demonstrated that classical physics ruled out the assumptions in what became the canonical model of geochemistry.

The Birch ideas were extended to mantle geochemistry by Tatsumoto, Armstrong and Kay who developed top-down models of geochemistry. The mass balance associated with hot accretion and early differentiation was developed in many early papers and summarized in Chapter 8 of Theory of the Earth and Chapter 13 of New Theory of the Earth. The mass balance shows clearly that the whole mantle had to be processed to form the crust plus the kimberlites and carbonatites.

last updated 31 July, 2013