Subduction initiation process is yet partly enigmatic and controversial. At least fourteen hypotheses have been proposed to explain physical mechanisms of this process:
(1) Plate rupture within an oceanic plate or at a passive margin (e.g., McKenzie, 1977).
(2) Reversal of the polarity of an existing subduction zone (e.g., Mitchell, 1984).
(3) Change of transform faults into trenches (e.g., Uyeda and Ben-Avraham, 1972).
(4) Sediment or other topographic loading at continental/arc margins (e.g., Dewey, 1969).
(5) Forced convergence at oceanic fracture zones (e.g., Mueller and Phillips, 1991).
(6) Spontaneous initiation of retreating subduction due to a lateral thermal buoyancy contrast at oceanic fracture zones separating oceanic plates of contrasting ages (e.g., Nikolaeva et al., 2008).
(7) Tensile decoupling of the continental and oceanic lithosphere due to rifting (Kemp and Stevenson, 1996).
(8) Rayleigh-Taylor instability due to a lateral thermal/compositional buoyancy contrast within the lithosphere (Matsumoto and Tomoda, 1983).
(9) Addition of water into the lithosphere (Regenauer-Lieb et al., 2001).
(10) Spontaneous thrusting of the buoyant continental/arc crust over the oceanic plate (Mart et al., 2005).
(11) Small-scale convection in the sub-lithospheric mantle (Solomatov, 2004).
(12) Interaction of thermal-chemical plumes with the lithosphere (Ueda et al., 2008).
(13) Large asteroid impacts (Hansen, 2007).
(14) Shear-heating induced localization along spontaneously forming lithospheric-scale fracture zones (Crameri and Kaus, 2010).
The major obstacle for the possibility to directly investigate subduction initiation by observation is the absence of definite knowledge about localities where subduction possibly starts now. The main future challenge is, therefore, to identify possible subduction initiation localities by evaluating numerically probability of subduction initiation at various passive margins and oceanic plate boundaries based on local lithospheric structures and acting tectonic forces. Simulated dynamics of most unstable margins and plate boundaries can then be analyzed in order to understand critical observables to be monitored in nature.