SUBDUCTION INITIATION IN NATURE AND MODELS

How new subduction zones form is an emerging field of scientific research with important implications for our understanding of lithospheric strength, the driving force of plate tectonics, and Earth's tectonic history. Subduction initiation (SI) is a difficult process to detect and observe directly because it occurs infrequently and deep underwater; approaches emphasizing the effects on the overlying plate coupled with the results of numerical experiments are needed to achieve progress. Our understanding of how new subduction zones form is advancing as a result of combining field studies of on-land ophiolites, diving on and dredging of nonaccretionary forearcs, and drilling into forearc crust (e.g., IODP Legs 351, 352, and 371) and mantle to identify examples and reconstruct SI timing and magmatic evolution. These results inform increasingly realistic numerical modeling and together geologic and experimental results can be used to test and refine SI hypotheses. This presentation summarized the recent review of Stern and Gerya (Tectonophysics 2018) on the state of the art of SI hypotheses. Two SI modes are identified, spontaneous and induced. Induced SI occurs when a collision destroys an existing subduction zone and continuing plate convergence causes a new subduction zone to form whereas spontaneous SI occurs without pre-existing convergence when large lateral density contrasts occur across the weak zone. We have good natural examples of 3 modes of subduction initiation, one type by induced nucleation of a subduction zone (polarity reversal) and two types of spontaneous nucleation of a subduction zone (oceanic transform collapse and collapse around the margins of the large head of a mantle plume; the latter likely also controlled plate tectonics initiation). In contrast, two proposed types of subduction initiation are not well supported by natural observations: (induced) transference and (spontaneous) passive margin collapse, although we may have a first example of passive margin collapse in the Late Cretaceous SI of SW Iran. Further work is needed to expand on and understand the implications of these observations. Future advances in understanding SI will come from better geologic insights, laboratory experiments, and numerical modeling, and with improving communications between these communities.