Title: Scaling and precursor motifs in earthquake networks

Abstract: A measure of the correlation between two earthquakes is used to link events to their aftershocks, generating a growing network structure. In this framework one can quantify whether an aftershock is close or far, from main shocks of all magnitudes. We find that simple network motifs involving links to far aftershocks appear frequently before the three biggest earthquakes of the last 16 years in Southern California. Hence, networks could be useful to detect symptoms typically preceding major events.