The terrestrial silicate mantle is the largest geochemical reservoir of the Earth and its chemical and isotopic inventories record the time-integrated history of differentiation. It is essential, for the understanding of the dynamic evolution of the Earth, to decode the message from mantle-derived melts. If mantle-derived melts are to be used to infer details of the chemical evolution of the mantle, their petrogenesis must be fully understood in the geodynamic context of plate tectonics. A wide variety of mantle-derived melts exist from different tectonic settings and their messages are not always unequivocal. As a result, petrogenetic relationships between different types of mantle-derived melts are hotly debated in the geochemical literature.
The extremely rare mantle derived melt called lamproite has anomalous major element, trace element and Sr, Nd, and Pb isotope chemistry, which indicate derivation from a rarely sampled and chemically very distinct portion of the mantle. Lamproites have traditionally been interpreted to be derived from the sub-continental lithospheric mantle. A lamproite source situated in the lithospheric mantle, however, requires an unlikely succession of very complex and mutually exclusive processes. In this thesis, an alternative petrogenetic model is proposed in which lamproites are derived from deeply subducted continent derived sediment and mid-ocean ridge basalt (MORB) sequestered in the mantle transition zone (410-670 km depth) for more than 109 years. The possible existence of such a source is found to have profound implications for the petrogenesis of other mantle-derived melts and more importantly, for the evolution of the mantle and the differentiation history of the Earth as a whole.
The mantle source region that is evident in the chemistry of lamproites can also be detected (albeit only as a mixing end member) in the chemistry of more typical mantle derived mafic melts like those that occur in the Newer Volcanic Province of southeast Australia. This igneous province contains melts with Pb isotope and trace element characteristics very similar to certain enriched mantle 1 (EM1) type ocean island basalts (OIB) such as those from Hawaii, indicating derivation from a mantle plume originating in the lower mantle. In close association with the EM1 type melts are basalts whose trace element chemistry is interpreted to require a small but significant source component of subducted continent-derived material. It is envisaged that this component was entrained into the rising mantle plume as it passed through the upper levels of the lower mantle and the transition zone. Similar basalts with comparable source requirements are demonstrated to exist in other continental settings such as the Etendeka igneous province, Namibia and the Deccan Traps, India. They also occur in oceanic settings such as on the Island of Pitcaim. These observations indicate that deeply subducted continent-derived material and MORB might be widespread in the deep mantle and not restricted to areas of recent subduction, implying persistence and isolation of recycled slabs in a laterally heterogeneous transition zone.
The existence of subducted sediment and MORB in the deep mantle is not just important as a source component of particular mafic mantle-derived melts but is far more significant for the understanding of the geodynamic evolution of the entire Earth. This is because the Pb isotopes chemistry of many of these mantle-derived melts indicates that their sources have evolved for a considerable time (up to 3 x 109 years) in isolation from the convecting mantle with a low 238U/204Pb ratio (μ). A low μ reservoir has long been known to exist from mass imbalance of Pb isotopes in the Earth (the first terrestrial Pb isotope paradox), which refers to the fact that on average the rocks that occur at the Earth surface (i.e. accessible Earth) plot significantly to the left of the meteorite isochron in a common Pb-isotope diagram. The identification of the transition zone as the low \i reservoir (as proposed in this thesis) is a solution that links the three major reservoirs of the silicate Earth (i.e. continental crust, MORB source mantle, and slabs in the deep mantle) with the metamorphic processes known to occur during subduction.
Finally, a review of further geochemical constraints shows that subduction and subsequent storage of slabs in the transition zone and/or lower mantle also potentially provide solutions to the apparent imbalance in Nb and Ta inventories and in Nb and Hf isotope compositions of accessible Earth relative to meteorites. These imbalances (including the Pb isotope paradox) indicate that the chemical evolution of the Earth cannot be approximated with the standard two-component system of complementary continental crust and depleted MORB-source mantle but that a third deep mantle reservoir is require. The geochemical and isotopic imbalances also highlight the importance of future research into a better approximation of the composition of the bulk Earth from meteoritic comparison.