Plate tectonics characterises the complex and dynamic evolution of the outer shell of the Earth in terms of rigid plates. These tectonic plates overlie and interact with the Earth's mantle, which is slowly convecting owing to energy released by the decay of radioactive nuclides in the Earth's interior. Even though links between mantle convection and plate tectonics are becoming more evident, notably through subsurface tomographic images, advances in mineral physics and improved absolute plate motion reference frames, there is still no generally accepted mechanism that consistently explains plate tectonics and mantle convection in one framework. We will integrate plate tectonics into mantle dynamics and develop a theory that explains plate motions quantitatively and dynamically. This requires consistent and detailed reconstructions of plate motions through time (Objective 1).A new model of plate kinematics will be linked to the mantle with the aid of a new global reference frame based on moving hotspots and on palaeomagnetic data. The global reference frame will be corrected for true polar wander in order to develop a global plate motion reference frame with respect to the mantle back to Pangea (ca. 320 million years) and possibly Gondwana assembly (ca. 550 million years). The resulting plate reconstructions will constitute the input to subduction models that are meant to test the consistency between the reference frame and subduction histories. The final outcome will be a novel global subduction reference frame, to be used to unravel links between the surface and deep Earth (Objective 2).