The controversy over plumes: Who is actually right?

The current state of the theory of mantle plumes and its relation to classic plate tectonics show that the plume line of geodynamic research is in a period of serious crisis. The number of publications criticizing this concept is steadily increasing. The initial suggestions of plumes' advocates are disputed, and not without grounds. Questions have been raised as to whether all plumes are derived from the mantle-core interface; whether they all have a wide head and a narrow tail; whether they are always accompanied by uplifting of the Earth's surface; and whether they can be reliably identified by geochemical signatures, e.g., by the helium-isotope ratio. Rather convincing evidence indicates that plumes cannot be regarded as a strictly fixed reference frame for moving lithospheric plates. More generally, the very existence of plumes has become the subject of debate. Alternative ideas contend that all plumes, or hot spots, are directly related to plate-tectonic mechanisms and appear as a result of shallow tectonic stress, subsequent decompression, and melting of the mantle enriched in basaltic material. Attempts have been made to explain the regular variation in age of volcanoes in ocean ridges by the crack propagation mechanism or by drift of melted segregations of enriched mantle in a nearly horizontal asthenospheric flow. In the author's opinion, the crisis may be overcome by returning to the beginnings of the plume concept and by providing an adequate specification of plume attributes. Only mantle flows with sources situated below the asthenosphere should be referred to as plumes. These flows are not directly related to such plate-tectonic mechanisms as passive rifting and decompression melting in the upper asthenosphere and are marked by time-progressive volcanic chains; their subasthenospheric roots are detected in seismic tomographic images. Such plumes are mostly located at the margins of superswells, regions of attenuation of seismic waves at the mantle-core interface.