Linking Ocean Island Basalt Geochemistry with Mantle Plume Dynamics

Abstract

Two of Earth's three principal modes of volcanism are directly linked to plate tectonic: decompression melting at mid-ocean ridges and fluid-induced melting above subduction zones. In contrast, intra-plate volcanism --- manifested in ocean islands and continental flood provinces --- cannot be explained by plate boundary processes. It is generally attributed to the surface expression of mantle plumes: hot, buoyant regions from the core-mantle boundary (CMB) that rise through the mantle and undergo decompression melting in the upper mantle. As plumes ascend, the cold, rigid, and refractory lithosphere imposes a mechanical and thermal barrier that sets the minimum pressure of melting. This lid effect, together with plume temperature and source composition, governs plume melting and thereby the first-order geochemistry of intra-plate magmas.

While mantle plume theory is well established, the details of plume-lithosphere interaction and its translation into geochemical signatures remain uncertain. This thesis integrates geochemical analyses, probabilistic inference, and geodynamical modelling to advance understanding of plume dynamics and their geochemical expression through three complementary studies.

First, I test the statistical evidence for the lid effect in ocean island basalt (OIB) geochemistry using a Bayesian framework. Lithospheric thickness emerges as the primary control on OIB compositions, reflecting its influence on mineral stability and degrees of melting. Nevertheless, two additional processes are required to explain observed deviations: (i) preferential sampling of hotter plumes beneath thick lithosphere (temperature filter); and (ii) direct ascent of melts generated at different pressures without equilibration at the base of the lithosphere.

Second, I reconcile these observations with geodynamical predictions by coupled mantle convection and melting models with geochemical calculations. The results validate both the lid effect and the melt-flux filter (a further manifestation of the previously described temperature filter), and show that the trace-element diversity of lavas from individual islands can be reproduced by sampling across the melt region of a single-lithology plume. The models also demonstrate that primitive mantle alone is insufficient to generate some OIB signatures, requiring contributions from recycled basaltic components in plume sources.

Finally, I develop an inverse method, grounded in optimal transport theory, to infer lithospheric thickness, plume excess temperature, and plausible degrees of source enrichment or depletion directly from lava geochemistry. This approach translates surface geochemical variability into constraints on the thermal and compositional structure of mantle plumes.

Together, these studies refine the conceptual framework of plume-lithosphere interaction and illuminate the processes that drive geochemical diversity in ocean island lavas. The results enhance our ability to interpret mantle plume evolution and its role in Earth's long-term geodynamic and geochemical cycles. Show more