Subduction initiation at an oceanic transform fault experiencing compression: Role of the fault structure and of the brittle-ductile transition depth

It has recently been shown that among the subduction zones initiated in the Cenozoic era, the subducting oceanic lithosphere could have any age, and was as often older than the overriding plate as younger than it. To try to explain this observation, we consider the simple setup of an oceanic transform fault and perform 2D thermomechanical experiments by applying the same convergence velocity on the two adjacent plates. We combine a non-Newtonian viscous rheology and a pseudo-brittle behavior. We focus on the influence of the fault structure at convergence onset, by testing the effects of the fault gouge depth, zwz, and of the width of the thermal transition between the lithospheres, which allows for mimicking the settings of a transform fault as well as of an old fracture zone. We investigate large ranges of plate age and of plate age offset. In most of our experiments, the mode of convergence accommodation is basically a function of the gouge depth, zwz, that we compare to the mean of the 2 plate thicknesses. We observe three main behaviors: (1) failure of subduction initiation if zwz is shallow, i.e., ≲15% of the mean depth of the 2 plate bottoms if the age offset scaled by the younger age is very low (≲1) or ≲30% otherwise, (2) old plate subduction, that is best favored for ratios zwz/zTT∼0.25 and for low offsets in relative plate ages, and (3) young plate subduction if zwz is quite deep (≳40%), whatever the age offset. The thermal transition width only has a second-order effect, that facilitates subduction initiation when the transition is wide. Compressive forces are overshooting reasonable thresholds (≳1013 N/m) if at least one of the two plates is older than ∼30 Myr. The success of a young plate subduction and, to a lesser extent, of an old plate subduction, can be predicted by comparing zwz at convergence onset to the depth of the brittle-ductile transition inside the future upper plate. The subduction success and polarity depend on the possibility to form a low shear stress plane. We predict that only the young plate subduction can initiate at an active transform fault, while the old plate subduction is restricted to the setup of an old fracture zone separating two lithospheres with a small age offset scaled by the younger age. The significant predominance in our modeling of the thin plate underthrusting results from the strong influence of the plate strength contrast when lithospheres are stiff. Our experimental results are consistent with the recent records of subduction initiation at the Mussau and Hjort Trenches, as well as at the Gagua ridge (W Pacific).