The mantle plume concept is now widely accepted and a limited understanding of the physics and chemistry of mantle plumes has been achieved. We know, for example, that plumes are compositionally distinct, at least in isotope and trace element composition, from the depleted upper mantle. It is also well established that although the composition of each plume is somewhat unique, plumes can be divided into a small number of classes (3 to 5). One of the most difficult geochemical problems has been deciphering the geochemical histories of plumes, i.e., determining the processes that have produced their distinct geochemical signatures. There have been three primary hypotheses:
(1) Plumes consist of primitive mantle (i.e., chemically unprocessed material) (Schilling, 1973; Wasserburg and DePaolo, 1977).
(2) Plumes consist of delaminated subcontinental mantle (McKenzie and O'Nions, 1983). Because it is relatively cold, this material, once decoupled from buoyant continents, might also sink to a thermal boundary layer.
(3) Plumes consist of subducted oceanic crust, perhaps along with subducted sediment in some cases (Hofmann
and White, 1982). Tectonic erosion in subduction zones
and delamination of lower continental crust may also add a continental "flavor" to this material. The cold, iron-rich nature of this material would cause it to sink to a thermal boundary layer, where it would eventually be reheated.
Both hypotheses (1) and (2) above can now be firmly rejected. Most plumes do not have radiogenic isotope compositions that match those expected of primitive mantle, nor do they define mixing arrays that pass through primitive mantle compositions. Even those that do have near-primitive compositions, such as Hawaii, have Pb/Ce and Nb/U ratios that differ significantly from primitive values (which are well constrained from isotope systematics or chondritic values). High 3He/4He ratios in some plumes require only a less degassed component, not a primitive one.
Recent studies of peridotite xenoliths (e.g. Carlson and Irving, 1994) in continental volcanics demonstrate that although much of the subcontinental lithospheric mantle is characterized by negative eNd, apparently the result of fluid or melt metasomatism, gOs values are almost invariably negative (range: -18 to +2; mean ~ -7). In contrast, the depleted upper mantle appears to have a mean gOs of about -2 (e.g. Roy-Barman and Allègre, 1994), and most oceanic island basalts have gOs in the range of 0 to +2 (e.g. Reisberg et al., 1993). Thus the subcontinental lithosphere appears to have a strongly depleted osmium isotope signature that is not shared by either the depleted upper mantle or mantle plumes. This distinction in osmium isotopic composition argues strongly against the mantle plumes originating from delaminated subcontinental lithosphere.
In contrast the Nd-Os isotopic systematics (as well as Pb-Os and Sr-Os systematics) of many oceanic island basalts are readily modeled by mixing of ancient (1-2 Ga) oceanic crust and depleted mantle. The composition of other oceanic island basalts matches that expected of ancient oceanic crust plus sediment that has been diluted by depleted mantle.
Further evidence in support of the oceanic crust recycling hypothesis comes from variations of Nb/U and Pb/Ce ratios that correlate with radiogenic isotope ratios (e.g., White
and Duncan, 1986). Neither Nb/U nor Pb/Ce appears to
be significantly fractionated by magmatic processes. Variations of these ratios observed in some oceanic island basalts therefore strongly suggests the presence of crustal material (e.g., subducted sediment). The amount of sediment required to explain Pb/Ce variations is, however, quite small (< 1%), which may explain why oceanic island basalts usually have d18O values that are close to the depleted mantle value of +5.7.
Carlson, R.W. & Irving, A.J., Earth Planet. Sci. Lett. 126, 457-472 (1994).
Hofmann, A.W. & White, W.M., Earth Planet. Sci. Lett. 57, 421-436 (1982).
McKenzie, D.P. & O'Nions, R.K., Nature 301, 229-231 (1983).
Reisberg, L. et al., Earth Planet. Sci. Lett. 120, 149-167 (1993).
Roy-Barman, M. & Allegre, C.J., Geochim. Cosmochim. Acta 58, 5043-5054 (1994).
Schilling, J.-G., Nature 242, 565-571 (1973).
Wasserburg, G.J. & DePaolo, D.J., Proc. Natl. Acad. Sci. USA 76, 3594-3598 (1977).
White, W.M. & Duncan, R.A., In Isotope Studies of Crust-Mantle Evolution, an AGU Monograph Honoring M. Tatsumoto and G. Tilton (Hart, S.R. & Basu, A., eds.) (AGU, Washington, 1996, in press).