Global structure of mantle isotopic heterogeneity and its implications for mantle differentiation and convection

In order to further our understanding of the global geochemical structure and mantle dynamics, a global isotopic data set of oceanic basalts was analyzed by Independent Component Analysis (ICA), a relatively new method of multivariate analysis. The data set consists of 2773 mid-ocean ridge basalts (MORB) and 1515 ocean island basalts (OIB) with five isotopic ratios of Pb, Nd and Sr. The data set spatially covers the major oceans and enables us to compare the results with global geophysical observations. Three independent components (ICs) have been found, two of which are essentially identical to those previously found for basalts from the Atlantic and Indian Oceans. The two ICs (IC1 and IC2) span a compositional plane that accounts for 95.7% of the sample variance, while the third IC (IC3) accounts for 3.7%. Based on the geochemical nature of ICs and a forward model concerning trace elemental and isotopic compositions, the origin of the ICs is discussed. IC1 discriminates OIB from MORB, and may be related to elemental fractionation associated with melting and the subsequent radiogenic in growth with an average recycling time of 0.8 to 2.4 Ga. IC2 tracks the regional provenance of both MORB and OIB and may be related to aqueous fluid–rock interaction and the subsequent radiogenic ingrowth with an average recycling time of 0.3 to 0.9 Ga. IC3 fingerprints upper continental crustal material and its high value appears in limited geographical and tectonic settings. Variations in the melt component (IC1) and in the aqueous fluid component (IC2) inherited in the mantle most likely reflect mid-ocean ridge and subduction zone processes, respectively. Long-term accumulation of dense materials rich in the IC1 melt component at the base of the convective mantle accounts for its longer recycling time with respect to that for less dense materials rich in the aqueous fluid component (IC2). IC2 broadly correlates with the seismic velocity structures of the lowermost mantle and electric conductivity around the mantle transition zones. We propose that IC2 reflects hydrogen distribution within the mantle and that several global domains enriched in hydrogen could exist as vertical sectors extending all the way down to the core–mantle boundary.