Fig. 6

a Modified from Holt and Condit (2021, their Fig. 3a). Solid lines show the temperature evolution at the top of the slab in a fully dynamic model of subduction initiation below a young oceanic lithosphere (see their Fig. 2). As time progresses the slab top cools and adjusts to a near-steady-state thermal structure after \(\sim\)20 Myr that is reminiscent of the near-steady-state ones discussed in this paper. A18=(Agard et al. 2018) with t’ their estimate whether rocks were subducted during subduction initiation (\(t'=0\)), termination (\(t'=1\)), or some time in between. Symbols become larger with duration of subduction. SPD15p: prograde paths from Penniston-Dorland et al. (2015) as used in van Keken et al. (2018). HC21 indicates the slab paths from Holt and Condit (2021) with three paths highlighted with age. b–d Slab paths from D80 shown every 1 Myr for 15 Myr in red for three subduction zones. Black line is the near-steady-state top of slab temperature (after 20 Myr for ocean-ocean settings and 40 Myr for ocean-continent ones). b Top of slab thermal evolution in a subduction zone where the slab sinks below a young oceanic lithosphere, as assumed in Holt and Condit (2021) with a similar slab temperature evolution as in a. c Thermal evolution of the slab top below Colombia Ecuador characterized by moderately fast convergence in an ocean-continent setting with relatively large dip. d Thermal evolution predicted for Central Honshu characterized by fast ocean-continent subduction with an old incoming plate. Note that in all frames we only show the slab paths to the pressure corresponding to the depth that the slab tip has reached