Earth's tectonic history revisited in the light of episodic misfits between plate network and mantle convection

Episodic plate reorganisations abruptly change plate boundary configurations. To illustrate their role, we review the plate reorganisations that appear in the present-day oceans and in the reconstructed Tethys ocean. These time periods cover the dispersal of the Pangea super-continent and the collisions with Eurasia that foreshadow a new super-continent. Plate reorganisations have played a fundamental role in the tectonic history of the Earth, being responsible for continental break-up and, after oceanic spreading, for continental collisions. As a result, they governed the formation and dispersal of super-continents. We observe a bulk polarity in plate motion that governs continental collision and the opposite bulk polarity in plate reorganisation that governs continental break-up. Such opposite polarities show in the tectonic history that we follow since the 550 Ma formation of the Gondwana super-continent.In order to decipher the rules that govern plate reorganisation, we investigate the distribution of spreading and subduction that derives from the current plate motion. We observe a mismatch between the evolution tendency of the plate boundary network and convection in the deep mantle. The actual network of plate boundaries illustrates a compromise between the two. Based on the opposite polarities in plate motion and plate reorganisation, we propose that this compromise is maintained by plate reorganisations that counterbalance free evolution of the network in abruptly changing its boundaries. We propose that plate reorganisations are basically caused by the mismatch between the free evolution of the plate boundary network and the current convection pattern in the deep mantle.Evidence on Proterozoic rifting and continent collisions allows dating the oldest known plate reorganisation around 2 Ga, which is the age of the oldest known super-continent. Based on the geology of the Archean before 3 Ga, mantle convection appears limited under a greenstone cover and different from the current mantle convection. The distribution of the diapiric granitoids that intrude this cover points to a honeycomb convection centred on downwelling sites separated by diffuse upwelling, which fits the theory on the early Earth mantle convection when plates did not cover the globe. We propose that the plate reorganisation regime appeared sometime between 3 and 2 Ga.