Abstract: Recent high–resolution reconstructions of plate motions reveal a complex history of alternating slowdowns and speedups, often over short timescales ( < 5 Myr). These rapid changes offer an opportunity to reassess the geodynamic processes driving tectonic plates, which we explore using an analytical inverse framework. This approach, however, inevitably yields non–unique solutions when inferring the forces behind a motion change. We partly address this issue by focusing on forces capable of varying at rates consistent with rapid kinematic shifts, though the specific driver behind any change may remain ambiguous. We adopt a two–step methodology, using torque changes as intermediaries linking force variations to reconstructed absolute plate motion changes. First, we employ an established method that combines rheological constraints with torque–balance principles to estimate the torque variation required for a given kinematic change. Second, we estimate torque–change vectors arising from a broad range of geodynamic scenarios — acting at plate boundaries (e.g., slab pull, interplate friction) and at the base of plates (e.g., asthenospheric flow). We then apply directional statistics to quantify the similarity between the motion–based torque–change distribution and each simulated vector. This comparison allows us to identify the location and direction of the force–change vectors most likely to produce the motion change of study. We apply this method to the Neogene Nazca–South America convergence. Our kinematic analysis reveals rapid slowdowns in the absolute motion of both plates and a pronounced Nazca speedup at ∼ 10–12 Myr. Our geodynamic analysis indicates that the force variations driving the slowdowns are likely concentrated along the central segments of the shared convergent boundary. This result aligns with established hypotheses linking reduced convergence to Central Andes orogeny, thereby supporting our approach. Key advantages of this novel method include fast computation, explicit treatment of kinematic uncertainties, and broad applicability across tectonic settings. • Forces driving motion is a non–unique inverse problem, fit for costly numerical models. • We present an alternative analytical–statistic approach that is fast yet thorough. • Likelihood of force changes is assessed based on their orientation and location. • We test isolated boundary and base acting forces against fast ( < 5 Myr) motion changes. • Results applied to the Andean convergence are well aligned with GPE– induced changes.

Abstract: This study presents three sets of scaled ‘sandbox’ experiments designed to understand the formation of mélanges during seamount subduction, focusing, in particular, on the role of talus aprons and moat basins, i.e., loose block accumulations around seamounts formed on the seafloor prior to subduction. In the first set of experiments, subduction of a single seamount increases wedge taper, inducing slumping of previously accreted material and producing sedimentary mélanges. Increasing the slab dip enhances off-scraping of material from the seamount's leading side, deposition on the rear side, and apron incorporation into the wedge interior. The second set simulates the subduction of a seamount chain, leading to wedge thickening, increased mass wasting, and the formation of a layered unit between the seamounts. The third set models oceanic plateau-wedge interactions, resulting in increased mass wasting on the wedge's surface and the formation of sedimentary mélanges. Loose blocks are deposited along the plateau surface and incorporated into the wedge along shear zones, creating tectonic mélanges. In conclusion, our results indicate that sedimentary and tectonic mélanges form during seamount subduction, especially from the surrounding talus. Sedimentary mélanges are deposited at the wedge surface through sliding and mixing of blocks with a terrigenous matrix, while the tectonic mélanges result from intense shear deformation inside the wedge, along the seamount/wedge interface. When compared with seismic profiles across modern accretionary wedges, our findings also suggest that the often-observed chaotic reflections may represent previously unrecognized seamount-derived mélanges, however, their type is difficult to establish just from geophysical data. • Thirteen analog experiments simulate subduction of seamounts surrounded by aprons. • Seamount aprons are a key prerequisite for mélange formation. • Sedimentary, tectonic and polygenetic mélanges form variably above seamounts. • Mélange formation is highly dependent on the geometry of the subduction zone. • Analog modeling may contribute to detection of mélanges in seismic profiles.