Mid-ocean ridges

One more time, from the top

Article metrics

One of the commonest geological processes on Earth is the partial melting of rocks. It is responsible for the creation of Earth's crust, and largely occurs at the network of undersea volcanoes that make up the mid-ocean ridge system. From studies of these ridges and their fragments now exposed on land (ophiolites), a well-accepted model for the creation of oceanic crust has emerged. As tectonic plates spread apart at mid-ocean ridges, the underlying mantle rocks (called peridotites) rise up to fill the gap, and decompression during this upward flow causes the mantle to melt and erupt the basalt magmas that form the overlying oceanic crust.

On page 514 of this issue1, however, Benoit et al. report evidence of a very different type of magma (andesite) from an ophiolite in Oman. They suggest that andesite could be generated by repetitive melting of seawater-altered rocks at the very top of the mantle at a mid-ocean ridge, and implicate a process quite different from normal melting.

The results of ocean-based studies provide a good explanation for the chemical composition of mid-ocean ridge basalt (MORB): decompression melting of mantle peridotite, followed by crystal separation during cooling of MORB magma. The result is a layered crustal structure, depicted in all three parts of Fig. 1, in which the MORB lavas overlie a coarse-grained layer of crystals (gabbroic cumulates) with a layer of magma feeder channels (dikes) in between.

Figure 1: General structure of ocean crust (ac), and the mantle-alteration model of Benoit et al.1.

a, Decompression melting produces a layered oceanic crust, and an underlying mantle that is highly depleted in sodium, titanium, strontium and light rare-earth elements at the top. b, Hydrothermal circulation of sea water (blue) adds water and seawater strontium to the top of the mantle. c, Later mantle intrusions heat the altered mantle, forming highly depleted, silica-rich melts which crystallize in veins within the oceanic plate (red). MORB, mid-ocean ridge basalt.

Geochemical analyses have further shown that the mantle rocks at the top of the mantle, just beneath the oceanic crust, have undergone large amounts of melting, removing as much as 30–35% of the original mass as MORB magma at high temperatures2,3. As a result, the top of the mantle at mid-ocean ridges is very depleted in elements that preferentially partition into basalt magma, such as sodium (Na), titanium (Ti), strontium (Sr), and light rare-earth elements (LREE) such as neodymium (Nd). In consequence peridotite rocks at the top of the mantle can generate little additional magma, even if reheated to high temperatures; and the magma that does form would have very low concentrations of Na, Ti, Sr and LREE, and have a very different overall chemical composition to typical MORB. The extremely depleted chemical signature would also be transferred to any crystals that formed from such a magma.

It is precisely this chemical signature that Benoit et al.1 have found in a subset of gabbroic rocks, called D (for depleted) gabbronorites, from the Oman ophiolite. They show that the depleted gabbros also contain unusually large amounts of the mineral orthopyroxene, which they suggest can only be the result of crystallization from a magma (andesite) with much more silica than normal MORB basalts.

How can such depleted magmas be generated beneath mid-ocean ridges? Presumably, the top of the mantle is so depleted in Na, Ti, Sr and LREE that it cannot yield additional magma. Nonetheless other examples of such rocks exist — as the authors say, gabbroic rocks with similarly depleted signatures have been recovered in drill cores from a site (DSDP 334) on the mid-Atlantic ridge, and low-Na plagioclase crystals similar to those in the Oman D-gabbronorites are a very minor but widespread component of otherwise normal MORB samples. Studies of the site 334 gabbros4 and low-Na plagioclase crystals5 have suggested that they are the crystal products of magmas formed at the very end of the melting process that produces normal MORB. The final fraction of magma formed at the top of the mantle during decompression melting would be depleted in Na, Ti, Sr and LREE because of the preceding extraction of magma earlier in the melting event.

Rather than accept this model, Benoit et al.1 have appealed to a second melting event to create the Oman D-gabbronorites. The authors went to great effort to separate clean, unaltered pyroxene crystals from these rocks and determine their 87Sr/86Sr isotopic ratio (this ratio is a direct tracer of the composition of the magma source because, unlike element abundances, it is unchanged by the melting process). The results are surprising — the Sr isotope ratios of the N (normal) gabbronorites are similar to those in typical MORB magmas, but those of the D-gabbronorites are much higher. This measurement indicates that the highly depleted rocks were not formed by separation of magma fractions late in the melting process. Rather, they must have been formed in a separate event involving a different peridotite source to that which generates normal MORB.

To identify this source, the authors look to another fundamental process — the circulation of sea water through the oceanic crust at regions of hydrothermal activity. As it seeps deeper into the cooling oceanic crust, sea water becomes heated to the point where it will react with the surrounding rocks, resulting in their partial dissolution and the precipitation of hydrothermal minerals. In the process, Sr is exchanged between sea water and the crustal rocks, and the very high Sr isotope ratio of sea water can be imparted to those parts of the crust that become highly altered. It is thought (but not known) that sea water may seep all the way through the seven kilometres of crust to alter even the top of the mantle.

Benoit et al. propose that hydrothermal circulation adds water and seawater-derived Sr to the highly depleted peridotite at the top of the mantle. The water decreases the melting temperature of the peridotite, which as a result melts easily during conductive heating resulting from later decompression melting events. That is, the mantle initially melts by decompression melting, yielding typical MORB, and the top of the mantle becomes highly depleted in Na, Ti, Sr and LREE. The top of the mantle then undergoes hydrothermal alteration by sea water, and is re-melted by subsequent conductive heating. This feature of the model stems from the restriction of the D-gabbronorites in the Oman ophiolite to the margins of a large mantle intrusion — an observation that could have been made only at an ophiolite, where the geological features of the oceanic plate are clearly exposed.

One nagging uncertainty about this model remains. The highly depleted chemical signatures of the D-gabbronorites have never been recognized in actual lava samples from the mid-ocean ridges, and only rarely as isolated inclusions of melt trapped within crystals6. Perhaps this process is not common and the highly depleted magmas are mixed with large amounts of normal MORB during eruption. If so, then how important is this process in the larger view of magmatism at mid-ocean ridges?

The answer may lie in further study of the site 334 gabbros and the widespread low-Na plagioclase crystals that are sometimes found in normal MORB. In particular, melt inclusions trapped in low-Na plagioclase may have retained other signatures of sea water hydrothermal alteration, such as increased water and chlorine content. Oxygen isotope studies of these and similar samples could also confirm the role of water in the generation of highly depleted ocean-ridge magmas. Meantime we are left to ponder the involvement of hydrothermal circulation in repetitive melting of the mantle, and its influence on magma composition at mid-ocean ridges.


  1. 1

    Benoit, M., Ceuleneer, G. & Polvé, M. Nature 402, 514–518 (1999).

  2. 2

    Dick, H. J. B., Fisher, R. L. & Bryan, W. B. Earth Planet. Sci. Lett. 69, 88–106 (1984).

  3. 3

    Johnson, K. T. M., Dick, H. J. B. & Shimizu, N. J. Geophys. Res. 95, 2661–2678 (1990).

  4. 4

    Ross, K. & Elthon, D. Nature 365, 826–829 (1993).

  5. 5

    Meyer, P. S. & Shibata, T. ODP Sci. Res. 106/109, 123–142 (1990).

  6. 6

    Sobolev, A. V. & Shimizu, N. Nature 363, 151–154 (1993).

Download references

Author information

Correspondence to Erik Hauri.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Hauri, E. One more time, from the top. Nature 402, 469–471 (1999) doi:10.1038/44970

Download citation

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.