A significant portion of the Earth's lithosphere is recycled into the deeper mantle, as required by mass balance considerations in orogenic environments. The two principal mechanisms for recycling are subduction at plate margins and delamination. Subduction is a well-understood process that is essential to the plate tectonic engine of planet Earth. Delamination, on the other hand, requires recycling via convective removal of the lower parts of the lithosphere, and is more difficult to detect. One chief argument for delamination comes from extreme shortening at continental convergent margins, which requires far thicker mantle lithospheres than observed (DeCelles et al., 2009). The second argument comes from the intermediate average composition of the continental crust (Rudnick, 1995), which requires a large ultramafic complementary residue at the bottom of the continental crust; such a reservoir has not been identified over large portions of continental areas. Delamination (Bird, 1979), convective removal, foundering, and lithospheric dripping are terms used for the process of detachment and sinking of the lower parts of the continental lithosphere other than those that may have been buried into the mantle via continental subduction. Most researchers using the term “delamination” refer to a density-driven process of foundering, and do not imply its original “peeling-off” significance as defined by Bird (1979), which is closer to tectonic underplating in shallow subduction environments. Delamination is a form of vertical and spatially localized tectonics often generating amoeba-like or circular surface effects that are regional results of tectono-magmatic processes at convergent plate margins.
Mantle delamination is the process of foundering of dense, unstable mantle lithosphere into the asthenosphere until it reaches thermal equilibrium with the surrounding asthenosphere (Houseman et al., 1981; Houseman and Molnar, 1997). Crustal delamination requires that the lowermost crust undergoes mineral equilibration under eclogite facies, and becomes denser that the underlying mantle (Arndt and Goldstein, 1989). Eclogite formation in the lowermost crust can be a solid-state process only, e.g., in areas thickened by continental collision (Leech, 2001), or can involve separation of intermediate composition arc-forming melts in Cordilleran subduction environments (Ducea, 2002; Lee et al., 2006). Trace element data from most modern Andean arc and back arc volcanoes (Chiaradia et al., 2009; Mamani et al., 2010) provide evidence that andesitic and dacitic magmas formed in such tectonic settings are, to a first order, extracted out of an eclogitic residue.
Snapshots of modern delamination are obtained using indirect imaging of the deeper Earth. Tomographic images of dense lithospheric bodies located within the asthenosphere (e.g., Zandt et al.  for the Sierra Nevada region in California, and Fillerup et al.  for the eastern Carpathians) provide compelling documentation of areas undergoing delamination/dripping at the present time. Both examples (and others) provide evidence for vertical tectonics at the scale of 50–100 km within the asthenosphere. Proposed geologic consequences of large-scale delamination are magmatic and geomorphic (Kay and Kay, 1993).
Gutierrez-Alonso et al. (2011, p. 155 this issue of Geology), use isotopic tracers to detect an important Paleozoic delamination event in Western Europe. In this new paper, the authors argue that a change from continental lithospheric to asthenospheric mantle isotopic signatures took place at the end of the Variscan orogen. This observation is used to postulate lithospheric delamination under the Iberian Massif. Magmatism formed in response to delamination can be either from the upwelling asthenosphere or from the downgoing drip (Elkins-Tanton, 2007). Adiabatic upwelling of asthenospheric mantle has long been the most significant expected geologic product in response to delamination (Kay and Kay, 1993; Ducea and Saleeby, 1998). Surprisingly, unless major flood basalt provinces are products of delamination (Bedard, 2006), most areas suspected to have undergone recent delamination have only minor associated mafic magmatism. For example, the Puna region in the central Andes (Kay et al., 1994; Drew et al., 2009) and the southern Sierra Nevada in California (Ducea and Saleeby, 1996, 1998; Farmer et al., 2002), two areas most likely subject to recent delamination, are characterized by volumetrically insignificant mafic magmatism at the time of delamination. This observation suggests that perhaps the size of drips is small (few kilometers), therefore their ability to sink is limited, and the corresponding ascending asthenospheric column is short and unlikely to melt extensively (Drew et al., 2009). Furthermore, smaller convective instabilities develop over larger time scales, perhaps in the range of tens of million years, in contrast to larger drips that can appear as catastrophic events in the geologic record. An even more puzzling observation is that the syn-delamination “bulls-eye”−shaped volcanism in map view of the Sierra Nevada is ultra-potassic and has isotopic signatures expected of the partial melting of the downgoing drip, the lower crustal root of the Sierra Nevada batholith (Manley et al., 2000). Indeed, amphibole or phlogopite-bearing eclogite facies rocks or peridotites will likely undergo limited dehydration melting upon adiabatic descent (Elkins-Tanton, 2007), providing an explanation for syndrip alkalic lithologies.
Areas that potentially lost their crustal and mantle roots via delamination and continental extension following extreme thickening, such as the Basin and Range (Meissner and Mooney, 1998) and the central part of the Coast Mountains batholith (western Canada) (Calkins et al., 2010) appear to have associated mafic magmatism that transitioned from “older” lithospheric mantle (characteristic of the pre-delamination and extension mantle) to “young” asthenospheric mantle over tens of millions of years, e.g., 40 m.y. for the Coast Mountains region (Manthei et al., 2010). Specifically for the Coast Range region, there is a marked change of εNd isotopic signatures of primitive, non-contaminated basaltic magmas from values of +3 to +5 in basalts older than 20 Ma to >+7 in basalts younger than 10 Ma (Manthei et al., 2010). This observation is again consistent with smaller drips and a rather long-lived process of “cleansing” of the uppermost mantle of the excess lithosphere at the end of the orogenic cycle.
In the case of the Iberian Massif (Guttierez-Alonso et al., 2011), the identity of the two different mantle reservoirs, pre- and post delamination, is also defined via the εNd of whole-rock basaltic samples and their corresponding model ages (age of extraction from a depleted mantle reservoir, assuming a single melting event), prescreened for possible crustal contamination effects. The transition from lower to higher εNd values took place over a period of 20 m.y. (from 305 to 285 Ma), similar to the Cordilleran cases described above. It is well established that Variscan orogeny in Europe culminated with a Himalayan-style continental collision during the early to middle Carboniferous, and was followed by extensional collapse (Menard and Molnar, 1988) that led to the formation of the classic European coal-bearing continental basins of the Carboniferous. Lithospheric delamination associated with orogenic collapse is entirely plausible, especially given that the crustal thickness of the Variscan orogen may have been more than double the normal crustal thickness in some parts of Europe (Medaris et al., 2003). One of the intriguing aspects of Gutierrez-Alonso et al.’s paper is that it suggests that delamination took place in a core region that underwent severe oroclinal bending, similar to the modern south and eastern Carpathians and the adjacent Transylvanian basin.
A temporal change in Nd isotopes of basaltic magmas is in itself insufficient to fingerprint delamination, as extension alone, for example, can produce the same result. However, such observations, when taken into account together with other lines of evidence, can be extremely powerful in detecting past delamination events, as demonstrated by Gutierrez-Alonso et al. (2011).
A regional increase in elevation is predicted by buoyancy considerations in areas undergoing delamination, because the densest part of the lithosphere is replaced. There are numerous areas geologically active in the Cenozoic in which rapid elevation increase is attributed to post-delamination surface uplift (e.g., Garzione et al., 2008). Geodynamic modeling is another avenue for future research to break new ground and make specific geologic predictions in delamination science; the most promising geomorphic/magmatic targets for future, modern, or recent delamination are depressions/basins in orogenic plateaus, such as the Altiplano-Puna in the central Andes.
I gratefully acknowledge support by the U.S. National Science Foundation under grants EAR-0910941, EAR-0907880, EAR-0606967, and EAR-0309885.