Welcome to the new version of CaltechAUTHORS. Login is currently restricted to library staff. If you notice any issues, please email coda@library.caltech.edu
Published July 2015 | metadata_only
Journal Article

Continental crust formation at arcs, the arclogite "delamination" cycle, and one origin for fertile melting anomalies in the mantle


The total magmatic output in modern arcs, where continental crust is now being formed, is believed to derive from melting of the mantle wedge and is largely basaltic. Globally averaged continental crust, however, has an andesitic bulk composition and is hence too silicic to have been derived directly from the mantle. It is well known that one way this imbalance can be reconciled is if the parental basalt differentiates into a mafic garnet pyroxenitic residue/cumulate ("arclogite") and a complementary silicic melt, the former foundering or delaminating into the mantle due to its high densities and the latter remaining as the crust. Using the Sierra Nevada batholith in California as a case study, the composition of mature continental arc crust is shown in part to be the product of a cyclic process beginning with the growth of an arclogite layer followed by delamination of this layer and post-delamination basaltic underplating/recharge into what remains of the continental crust. A model is presented, wherein continuous arc magmatism and production of arclogites in continental arcs are periodically punctuated by a delamination event and an associated magmatic pulse every ~10–30 My. The recycling flux of arclogites is estimated to be ~5 %–20 % that of oceanic crust recycling by subduction. Delaminated arclogites have the necessary trace-element compositions to yield time-integrated isotopic compositions similar to those inferred to exist as reservoirs in the mantle. Because of their low melting temperatures, such pyroxenites may be preferentially melted, possibly forming a component of some hotspot magmas.

Additional Information

© 2015 Science China Press and Springer-Verlag Berlin Heidelberg. Received: 9 February 2015; Accepted: 27 April 2015; Published online: 30 June 2015 This work was supported by NSF (EAR-0309121, 0440033). The models were developed by Lee in 2000 as Chapter 7 of his PhD thesis [122]. In 2005, the paper was revived when D. Anderson was a Wiess Visiting professor at Rice University, but we never finished the paper. After D. Anderson's death in December 2014 and a once in a decade effort to clean his office, Lee stumbled upon this paper again. The resurrected manuscript has been expanded and revised to bring it up to date. We thank H. Stone, M. Manga and R. J. O'Connell for inspiring simplicity. Conflict of interest: The authors declare that they have no conflict of interest.

Additional details

August 22, 2023
August 22, 2023