Does unsteady overriding plate kinematicsexplain variability of slab dip and upper platedeformation?

National audiencePlates reorganization associated with mantle convection leadsto changes in absolute plate velocities which may in turn impactthe geometry of slabs and the upper plate regime of deformation.Analogue experiments suggest a very good correlation betweenoverriding plate absolute motion and, slab dip and upperplate deformation: steep slab and extension are associated withoverriding plate moving away from the trench while shallow slaband shortening occur when the upper plate goes the other way.However, these correlations are established on models in whichthe motion of the overriding plate is constant and the subductionsteady-state which may not correspond to the general case:How does the subduction adapt to the changes of overriding plateabsolute velocity? What is the impact on slab dip and upperplate deformation? Could it explain why correlations in Naturebetween overriding plate absolute motion and slab dip are poorerthan expected? We herein present a study coupling 3D analoguemodels of subduction and observations of present-day subductionparameters. The models show that instantaneous changes in theoverriding plate absolute velocity produce a transient stage duringwhich both upper plate deformation and slab dip adjust, followedby a period of stabilization. Specifically, deformation velocity adjustsrapidly while slab dip requires a larger time. For deep slabdip this time delay is on average 17 ± 4 Myr and is independent ofboth the absolute velocity and changes in absolute velocity. Thereis a systematic trend toward lower than expected slab dip whenupper plate slowdowns and higher when it accelerates. In Nature,there is the same general trends but correlations are poorer thanin laboratory models. The distribution of slab dip as a functionof overriding plate accelerations or decelerations above specificperiods of time supports the idea of a global adjustment time of10 Ma, close to what found in analogue models. Over this timeperiod, deep slab dips tend to be lower than expected when theoverriding plate constantly decelerated and higher when it accelerated.The same analysis for upper plate deformation gives abest result for an adjustment time of 5 Ma. Thus, there is a delayin the response of the system to the changes of overriding plateabsolute velocity. The poor correlation in Nature between presentdaysubduction kinematics, slab dip and upper plate deformationmay explain because the present-day velocities correspond to asnapshot in the evolving history of subductio