The inversion of the focal mechanism (FM) provides an estimate of the fault plane orientation and the direction of slip of an earthquake, giving us valuable insights into the mechanical processes involved in the occurrence of an earthquake. Given the recorded first motion polarities at a set of stations, there are always two possible planes that explain the observations equally well. This so-called fault plane ambiguity is often resolved based on expert judgment, considering knowledge about the local geology and the locations of fore- or aftershocks. With seismic networks and inversion algorithms continuously improving, we can obtain large numbers of inverted FMs, even for events of low magnitudes, which calls for an automated procedure to resolve the fault plane ambiguity. Using an enhanced epidemic-type aftershock sequence (ETAS) model, we jointly invert the plausibility of each of the two fault planes specified by the inverted FM and a magnitude-dependent shape of elliptic aftershock triggering oriented in the direction of strike, based on FMs of M≥2.5 earthquakes in Southern California since 1981. Results of this inversion do not only provide an approach to resolve fault plane ambiguity but also an ETAS model which goes beyond the common assumption of spatially isotropic triggering. Preliminary results suggest that aftershocks occur predominantly in strike direction relative to their triggering events and that the shape of the ellipse describing this behaviour is magnitude-independent. We conduct pseudo-prospective forecasting experiments to compare our novel anisotropic ETAS model based on fault plane plausibility estimates to the current state-of-the-art isotropic ETAS model to test the utility of understanding source anisotropy for earthquake forecasting.