Résumé
Although the number of teleseismic shear wave splitting measurements has considerably increased world-wide during the last decade, there is no consensus on a unique and universal interpretation of these data. Splitting might primarily originate in the lithospheric and/or asthenospheric mantle, but the technique has no vertical resolution and therefore leaves this question of anisotropy location unresolved. Generally, the consequences of the two possibilities are used to infer which is more likely. If splitting occurs in the lithosphere, the splitting parameters (the orientation of the fast split shear wave ø and the delay time
δt) should be related to frozen or active deformation. On the other hand, if the splitting occurs in the asthenosphere, the observed splitting parameters should be related to the present-day flow induced by the differential motion between the tectonic plates and the underlying mantle. The aim of this paper is to test these hypotheses against shear wave splitting observations from western Europe and the eastern U.S. Indeed, except in a few places, splitting on either side of the Atlantic, on Hercynian, Caledonian or Grenvillian structures do not appear to have been deeply and pervasively affected by the opening of this ocean. This provides us the opportunity to place these results in the Pangaean reference frame of the Hercynian-Caledonian belt before the opening of the Atlantic. In many places the close parallelism between lithospheric structure and the absolute plate motion (APM) impedes a reliable identification of the origin of the anisotropy. Especially, an asthenospheric flow channelled around the deeply rooted North American craton may be compatible with the large-scale anisotropy pattern in the eastern U.S. Despite this ambiguity, fast polarization directions (ø) observed on or near some major lithospheric structures correlate better with the trend of the local tectonic structures than with the expected asthenospheric flow. Therefore, we suggest that these lithospheric structures, developed before the opening of the Atlantic, remain frozen since Hercynian, Caledonian or Grenvillian times and are responsible for the observed splitting at these places.