CONSTRAINING ABSOLUTE PLATE MOTIONS WITH SUBDUCTED SLABS - LESSONS FROM THE AEGEAN TETHYS

Subduction of the ancient Tethys Ocean underneath the Eurasian margin is both complex and long-lived, offering a unique insight into subduction history and absolute plate motion. Tomographically fast material inferred to represent subducted slabs is observable in multiple P- and S-wave models between depths of 1900-2150km. Geological evidence of the onset of this subduction is apparent in the Vardar ophiolite complex, which obducted between 166-176 Ma, linking the deep mantle with the surface in the geological past. This information offers additional absolute plate motion constraints independent of hotspot and palaeomagnetic data, and combined with plate kinematic-mantle convection models provides us with the means to test alternative absolute reference frames. We connect a rheological model of the mantle, plate motions, and dynamic plate boundaries from three alternate absolute reference frames, including a published subduction reference frame, to test the implied subduction history against known mantle structure. We find that while none of the models reproduce the interpreted Aegean Tethys slab in the correct latitudinal position, the subduction reference frame approach based on aligning surface geological evidence of subduction above tomographically interpreted slabs better reproduces the longitudinal extent of Tethyan material beneath the southern Eurasian margin at mid mantle depths (~1000km). Furthermore, the subduction reference frame reproduces slabs globally at this depth, producing better fits than hybrid palaeomagnetic approaches in the Indian Tethys, East Australia, the Caribbean and North America, indicating that the longitudinal corrections of 2-4° at the corresponding time (40-50 Ma) are robust. At greater depths however, we find the longitudinal corrections of the subduction reference frame model to be too large, indicating that regional rather than global sinking rates, and iterative geodynamic models should be used to increase model robustness.