Highly heterogeneous mantle caused by recycling of oceanic lithosphere from the mantle transition zone

Abstract

Geochemical heterogeneities observed in the mantle are usually attributed to recycling of oceanic lithosphere through subduction. However, it remains hotly debated where recycled material stagnates, and how quickly it can be liberated back to surface. This knowledge gap hinders our understanding of mantle circulation and the chemical evolution of the Earth. Here we address these questions using a combination of geochronology and geochemistry from South China Sea (SCS) seamounts. The Shixingbei seamount lavas formed during active seafloor spreading at c. 19.1 Ma show limited geochemical variability, whereas the Zhenbei-Huangyan seamount chain formed during the post-spreading stage at c. 7.8 Ma and displays a wide range of compositions. However, melt inclusions in olivine and plagioclase from the Zhenbei-Huangyan basalts show considerably greater isotopic variability than seen in the whole rock compositions of both the SCS syn- and post-spreading lavas. A previously unidentified third mantle source component (FOZO) revealed by olivine-hosted melt inclusions along with both depleted (DMM) and enriched (EMII) mantle components is required in the source region to explain the observed isotopic and chemical variability. On the basis of our results, the age of the recycled ocean crust and sediments in this region are estimated to be c. 120 – 350 Ma. We infer that these enriched components in the SCS lavas come from the mantle transition zone. Variations in mantle source heterogeneity coupled with melting process control spatial–temporal (spreading vs. post-spreading stage) geochemical variations of lavas from the SCS and surrounding areas. Together with the results from published studies, we propose that marginal basins are one of the major locations on Earth where oceanic and/or continental lithosphere is transferred into the upper mantle and transition zone, representing an important source of upper mantle heterogeneity. We provide a simple conceptual model linking plate subduction and upper mantle heterogeneity and the volcanism in the SCS and surrounding areas.