Metasediment-derived Melts in Subduction-zone Magmas and their Influence on Crustal Evolution

Abstract

Abstract Subduction is a major process cycling material through Earth’s geochemical reservoirs. Although trends in chemical composition of arc magmas imply assimilation of metasediment, the degree of such assimilation and the loci of that metasediment contamination (whether via subducted sediment or country rock assimilation) are poorly understood. To address these issues, we explore compositional data of oceanic and continental arc systems from circum-Pacific subduction zones. We find that high-silica continental arc rocks of the circum-Pacific are associated with higher aluminium saturation indices interpreted to reflect higher degrees of metasediment assimilation, with Sr/Y suggestive of shallow emplacement levels within the crust. In contrast, high-silica oceanic rocks of the circum-Pacific display lower aluminosity and equilibrated at deeper levels within the crust. Continental arc basalts are often assumed to be the source of high-silica continental arc rocks. However, phase equilibrium modelling of partial melting and crystal fractionation of continental arc basalts yield results that question this assumption. Furthermore, continental arc rock compositions show that the assimilated metasediments have protoliths that are most probably felsic greywacke and pelite rather than mafic greywacke. These findings are consistent with the hypothesis that high-silica rocks in continental arcs are directly influenced by anatexis of metasediment at shallow crustal levels (<20 km). Based upon a new method of discriminating the contribution of metasediment-derived melt, approximately one-third of felsic rocks in continental arcs have a demonstrable and unambiguous metasedimentary component. The degree of metasedimentary reworking in continental arc magmas plays an important role in the evolution of the continental crust and highlights the importance of using sediment-sensitive geochemical proxies and a magma’s petrological history when deconvolving the histories of magmatic arcs. This study also underlines the caveats associated with the calculation of depleted mantle model ages, where traditional techniques may lead to discrepancies of the order of 0·5 billion years.