The calculation of the lattice preferred orientation of crystals in mantle flow has become a major step in the interpretation of the upper mantle seismic anisotropy. Available models for such calculations are based on experimental data obtained on dry crystal aggregates. Recent simple shear experiments showed that the presence of water may radically change the fabrics produced, which casts doubt on the validity of the models in water rich environments like subduction zones. Water may change both the mechanisms of dynamic recrystallization and the mechanical properties of the crystals. This paper illustrates how a change of the critical resolved shear stress of the slip systems is sufficient to reproduce the new experimental results. In contrast, no changes are necessary in the theory of dynamic recrystallization. For subduction zones, first order calculations predict that trench parallel flow rather than water is more likely to be responsible for the anisotropic signature.