Trace Element and Isotope Geochemistry of the Northern and Central Tongan Islands with an Emphasis on the Genesis of High Nb/Ta Signatures at the Northern Volcanoes of Tafahi and Niuatoputapu

Abstract

We present a new comprehensive isotope and trace element dataset for north and central Tongan lavas including high-precision High Field Strength Element measurements. The emphasis of our study is on lavas from the northernmost volcanic islands of Tafahi and Niuatoputapu that exhibit unique compositions compared with other Tongan volcanoes, in particular elevated high-precision Nb/Ta ratios as high as similar to 28. These lavas are extremely depleted in incompatible, fluid-immobile elements such as Ti, Zr, Sc and Yb, suggesting about 6% depletion of their mantle wedge source relative to average upper mantle prior to 20-30% partial melting. The extremely depleted mantle wedge was re-enriched in highly incompatible elements including Nb and Ta from the subducting slab. Three slab components-fluids from altered oceanic crust, pelagic sediments, and Louisville Seamount Chain lavas-have contributed to the source region of the magmas. A fluid from the subducted Louisville lavas has been added after the other two slab components and has radiogenic Pb-206/Pb-204. Components from the altered oceanic crust and Louisville sea-mounts exhibit little Nb/Ta fractionation, as demonstrated by the nearly uniform and mid-ocean ridge basalt-like Nb/Ta ratios. We suggest that the apparently large slab fluid flux beneath Tafahi and Niuatoputapu results from focused flow through subducting Louisville Seamount crust that had no or little sediment cover, allowing the fluids to pass through the subducting lithosphere more efficiently. Thus, the high Nb/Ta ratios indicate that the Louisville-derived fluids had probably equilibrated with residual rutile. Our results imply that significant amounts of Nb and Ta can be mobilized from distinct subducting oceanic lithologies-a process that may probably also operate beneath other island arcs. We propose a model in which the appearance of the subducted seamount signatures some 3-4Myr after their actual subduction is explained by storage in shallow metasomatized lithospheric mantle. This relic Louisville seamount signature is subsequently reactivated and added to transiting magmas 4 Myr later. This reactivation may result from slab rollback of the subducting Pacific plate towards the east. Our new temporal and spatial model also provides a viable explanation for the conundrum of the occurrence of Louisville signatures even though the Louisville seamount chain is, at present, being subducted several hundred kilometres further south.