Can subduction initiation at a transform fault be spontaneous?
We present an extensive parametric exploration of the feasibility of “spontaneous” subduction initiation, i.e., lithospheric gravitational collapse without any external forcing, at a transform fault (TF). We first seek candidates from recent subduction initiation events at an oceanic TF that could fulfill the criteria of spontaneous subduction and retain three natural cases: Izu–Bonin–Mariana, Yap, and Matthew and Hunter. We next perform an extensive exploration of conditions allowing for the spontaneous gravitational sinking of the older oceanic plate at a TF using 2-D thermomechanical simulations. Our parametric study aims at better delimiting the ranges of mechanical properties necessary to achieve the old plate sinking (OPS). The explored parameter set includes the following: crust and TF densities, brittle and ductile rheologies, and the width of the weakened region around the TF. We focus on characterizing the OPS conditions in terms of (1) the reasonable vs. unrealistic values of the mechanical parameters and (2) a comparison to modern cases of subduction initiation in a TF setting. When modeled, OPS initiates following one of two distinct modes, depending mainly on the thickness of the overlying younger plate. The asthenosphere may rise up to the surface above the sinking old plate, provided that the younger plate remains motionless (verified for ages ≥5 Myr, mode 1). For lower younger plate ages (typically ≤2 Myr), the younger plate is dragged toward the older plate, resulting in a double-sided subduction (mode 2). When triggered, spontaneous OPS is extremely fast. The parameters that exert the strongest control over whether OPS can occur or not are the brittle properties of the shallow part of the lithosphere, which affect the plate resistance to bending, the distance away from the TF over which weakening is expected, and the crust density. We find that at least one mechanical parameter has to be assigned an unrealistic value and at least two other ones must be set to extreme ranges to achieve OPS, which we do not consider realistic. Furthermore, we point out inconsistencies between the processes and consequences of lithospheric instability, as modeled in our experiments and geological observations of subduction infancy, for the three natural candidates of subduction initiation by spontaneous OPS. We conclude that spontaneous instability of the thick older plate at a TF evolving into mature subduction is an unlikely process of subduction initiation in modern Earth conditions.