Exchange Between Surface Reservoirs and the Mantle

Plate tectonic processes cycle water into and out of the mantle changing the mass of water at the Earth's surface (mainly the ocean but also pore waters of sediments). The sign and vigor of this recycling is very important to the evolution of life. With more surface water, there would be no dry land. With less, open water bodies would fill with sediments.

Subduction of sediments and hydrated oceanic crust is a sink to the shallow reservoir of water which is potentially large over geologic time. For example, hydrated basalt or greenstone has about with 3% water by mass or 10% by volume. It would take 2 billion years to subduct the global average 2.5 km thickness of water on the Earth assuming that an average thickness of 2 km of greenstone is present in subducted crust and that global spreading rate is the current rate 3 kmy.

The extent to which subducted water quickly returns to the surface in island arc volcanics is poorly constrained. The most voluminous primitive magma at island arcs, arc tholeiite, is formed in the asthenospheric wedge above the slab, rather than as direct melt of subducted sediment and oceanic crust. Trace element and isotopic studies are thus necessary to detect contributions from the slab. Unfortunately there are alterative sources for some of the components that might come from the slab. Contamination at crustal levels is a major problem because hydrothermally altered oceanic crust and arc volcanics, serpentinized mantle, and sedimentary rocks are assimilated into long-lived magma chambers beneath arcs. In addition, previously subducted material is a ubiquitous heterogeneity in the mantle. Direct measurements of volatiles in arc volcanics on land have not provided useful constraints because these rocks degas before they erupt.

However, recent studies of lavas erupted in deep water in the back-arc basin behind the Mariana arc have provided good evidence that water from the slab has a major role in generating island arc volcanics [ Stolper and Newman, 1994]. These rocks appear to have not degased significantly and thus retain their ratios of water to other elements. This information is used to deduce the role that water in the source region had in increasing the fraction of melting above that of a dry system. One measures the concentration of a ``moderately incompatible'' element like Na that preferentially enters the initial melt and is diluted in the melt by subsequent melting. In the case that additional water in the source region increases the fraction of melting, lavas with high water concentrations formed by high fractions of partial melting and thus have low Na concentrations. Conversely, fractional melting of an initially homogeneous source region concentrates both water and Na in the first melt and subsequent melting dilutes the concentration of both elements. Stolper and Newman [1994] found the negative correlation between Na and water which is expected if water derived from the slab promotes a higher fraction of partial melting.

The water-rich fluid from the slab efficiently leaches material from the subducted oceanic crust and from the asthenospheric wedge [ Stolper and Newman, 1994]. The heat-producing radioactive elements K, U, and Th are leached from the asthenospheric wedge, which, over time, has caused a significant part of the Earth's total abundance of these elements to accumulate in the continental crust. Copper, an important ore metal in arc regions, appears to have been derived from the slab.

Studies of the cosmogenic isotope Be in arc lavas provide strong evidence that water from the slabs is derived in part from subducted sediments and that some sediments are subducted to great depths [ Morris et al., 1990; Edwards et al., 1993; Gill et al., 1993; Leeman et al., 1994]. The half life of Be, 1.5 million years is short enough that the isotope is present in young sediments but not in other crust and mantle reservoirs that might enter the magma. The Be in sediments needs to be subducted and reach the source region within a few million years to be present in the observed concentrations. The concentration of Be, normalized to the concentration of the stable isotope Be, shows the variations implied by the geometry of subduction. The ratio Be/ is low for lavas derived from young hot slabs which dehydrate at depths shallower than the source region compared with lavas derived from old cold slabs that dehydrate at source region depths [ Leeman et al., 1994] and decreases for volcanoes further from the trench. As expected, detectable Be is absent in mid-ocean ridge and oceanic island lavas where no slab is present.

Boron is a good tracer for subducted sediments because it is very abundant there but scarce in the asthenosphere. The observed positive correlation of B with Be is evidence that water leaving the slab carries dissolved material into the source region in the asthenospheric wedge. Apparently, B is efficiently removed in the first fluid from subducted sediments because it is quite enriched at the volcanic front but weakly enriched in lavas farther from the trench which have Be. No B-rich Be-poor lava, which might be produced from aged subducted sediments, are known. Uranium series isotopic disequilibrium indicate that a few hundred thousand years were required for the fluid to flow from the slab to the asthenospheric source region [ Gill et al., 1993].

Overall the Na-water correlation in the Marianas and the Be data from several arcs indicate the subducted water returns to the surface at island arcs and that this process enriches the continental crust in various trace elements. Global element fluxes and the rate at which the oceans are growing or shrinking are less well constrained.