UNCORRECTED PROOF Gondwana Research xxx (xxxx) xxx-xxx Contents lists available at ScienceDirect Gondwana Research journal homepage: www.elsevier.com GR focus review Asymmetric dynamics at subduction zones derived from plate kinematic constraints Eleonora Ficini a, ⁎ , Marco Cuffaro b , Carlo Doglioni a, c a Sapienza Università di Roma, Dipartimento di Scienze della Terra, Rome, Italy b Istituto di Geologia Ambientale e Geoingegneria, CNR, Rome, Italy c Istituto Nazionale di Geofisica e Vulcanologia, INGV, Rome, Italy ARTICLE INFO Article history: Received 29 October 2018 Received in revised form 1 July 2019 Accepted 21 July 2019 Available online xxx Handling Editor: M. Santosh Keywords: Subduction zones Subduction rate Recycling of the lithosphere into the mantle Mantle convection ABSTRACT The lithospheric sinking along subduction zones is part of the mantle convection. Therefore, computing the volume of lithosphere recycled within the mantle by subducting slabs quantifies the equivalent amount of mantle that should be displaced, for the mass conservation criterion. The rate of subduction is constrained by the convergence rate between upper and lower plates and the motion of the subduction hinge H that may either converge or diverge relative to the upper plate. Here, starting from the analysis of the slab hinge kinemat- ics, we evaluate the subduction rate at 31 subduction zones worldwide, useful to compute volumes of sinking lithosphere into the mantle. Our results show that ~190 km 3 /yr and ~88 km 3 /yr of lithospheric slabs are cur- rently subducting below H-divergent and H-convergent subduction zones, respectively. We also propose sup- porting numerical models providing asymmetric volumes of the subducted lithosphere, using the subduction rate instead of plate convergence, as boundary condition. Furthermore, H-divergent subduction zones appear to be coincident with subductions having “westward”-directed slabs, whereas H-convergent subduction zones are mostly compatible with those that have “eastward-to-northeastward”-directed slabs. On the basis of this geographical polarity, our lithospheric volume estimation gives ~214 km 3 /yr and ~88 km 3 /yr of subducting lithosphere, respectively. This entails that W-directed subduction zones contribute more than twice in lithos- pheric sinking into the mantle with respect to E-to-NE-directed ones. In accordance with the conservation of mass principle, this volumetric asymmetry in the mantle suggests a displacement of ~120 km 3 /yr of mantle material from west to east, providing a constraint for global asymmetric mantle convection. © 2019. 1. Introduction Subduction zones dynamics and the evolution of the lithos- phere-mantle system at convergent margins have been thoroughly in- vestigated in the last decades by the scientific community (Capitanio et al., 2010; Coltice et al., 2017; Conrad and Hager, 1999; Jarrard, 1986). Several parameters seem to influence subduction zones behav- ior and different models have been invoked to explain the initiation of the slab sinking, as well as the driving forces of plate tectonics and re- lated mantle convection (Carcaterra and Doglioni, 2018; Faccenna et al., 2001; Forsyth and Uyeda, 1975; Garfunkel et al., 1986). However, regardless the origins and driving mechanisms for plate tectonics, it is generally accepted that subduction zones dynamics is strongly af- fected by the motion of the trench (Doglioni et al., 2007; Heuret and Lallemand, 2005; Lallemand et al., 2008; Schellart and Rawlinson, 2013). In this work, we compute the volumes of subducting lithosphere at subduction zones worldwide (Fig. 1), taking into account the dif- ferent subduction hinge kinematics, also improving the study ⁎ Corresponding author. Email address: eleonora.ficini@uniroma1.it (E. Ficini) of Doglioni and Anderson (2015). Indeed, relative to the fixed up- per plate, the slab hinge can move either towards the lower plate (e.g., Marianas, Tonga-Kermadec or Sandwich subduction zones, etc.) or towards the upper plate (e.g., South America, Cascadia or Burma-Sumatra subduction zones, etc.) (Fig. 2). The volume was cal- culated to collect information about the dynamics observed at sub- duction zones starting from kinematic data. Then, speculation on how much could the observed dynamics affect mantle convection, since lithosphere re-entering into the mantle influences in some way man- tle circulation at depth. The volumes were calculated using the sub- duction rate (i.e., the effective velocity with which the slab enters the mantle, considering both the convergence velocity of the lower plate and the transient motion of the subducting slab hinge), the length of the trench of the considered subduction zone and the thickness of the lithosphere (Conrad and Lithgow-Bertelloni, 2006). Basically, the pa- rameters used were: V L = convergence velocity of the lower plate (DeMets et al., 2010) V H = velocity of the subducting slab hinge (GPS data) l = length of each subduction trench d = variable lithospheric thickness (Conrad and Lithgow-Bertelloni, 2006) https://doi.org/10.1016/j.gr.2019.07.013 1342-937/ © 2019.