Exploring the effect of anisotropy on body-wave tomography models: Rollback and subduction of the Alboran slab

Teleseismic P-wave tomography models often show low-velocity anomalies behind subducted slabs (i.e. opposite the direction of subduction). One such anomaly, behind the Alboran slab in the westernmost Mediterranean, requires partial melt in the mantle if taken at face-value. However, mantle anisotropy can cause low-velocity anomalies in tomographic models that assume isotropy. In fact, results from SKS splitting suggest rollback-induced anisotropy within the low-velocity region, and we investigate if this anisotropy can explain the sub-slab anomaly. We include anisotropy as an a priori constraint on the inversion and test different magnitudes, azimuths, and dips within the low-velocity region. We find that a range of anisotropic models can fit the travel time data as well as the isotropic models while significantly reducing or eliminating the low-velocity anomaly behind the slab. We conclude that this alternative interpretation (delays are caused by anisotropic structure) is as consistent with the travel time data as an isotropic low-velocity anomaly, and more consistent with SKS splitting observations and the known history of rollback. In addition, we find that models that include anisotropy with steeply dipping fast axes, meant to simulate the effect of downgoing entrained mantle, provide a poorer fit to the travel times than all the other models. This suggests that the slab may no longer be actively subducting.