• Earthquake interaction;
  • forecasting;
  • and prediction


The Northeastern Caribbean region accommodates ∼20 mm  yr−1 of oblique convergence between the North American and Caribbean plates, which is distributed between the subduction interface and major strike-slip faults within the overriding plate. As a result, this heavily populated region has experienced eleven large (M ≥ 7.0) earthquakes over the past 250 yr. In an effort to improve our understanding of the location and timing of these earthquakes, with an eye to understand where current seismic hazards may be greatest, we calculate the evolution of Coulomb stress on the major faults since 1751 due to coseismic, postseismic, and interseismic deformation. Our results quantify how earthquakes serve to relieve stress accumulated due to interseismic loading and how fault systems communicate with each other, serving both to advance or retard subsequent events. We find that the observed progressive westwards propagation of earthquakes on the Septentrional and Enriquillo strike-slip faults and along the megathrust was encouraged by coseismic stress changes associated with prior earthquakes. For the strike-slip faults, the loading of adjacent segments was further amplified by postseismic relaxation of a viscoelastic mantle in the decades following each event. Furthermore, earthquakes on the Septentrional fault relieve a small level of Coulomb stress on the parallel Enriquillo fault to the south (and vice versa), perhaps explaining anticorrelated timing of events on these respective fault systems. The greatest net build-up of Coulomb stress changes over the past 250 yr occurs along the central and eastern segment of the Septentrional and the Bowin strike-slip faults (65°–71°W), as no recent earthquake has relieved stress in these regions. For oblique thrust faults, net stress build-up over the past 250 yr is the largest on the North American/Caribbean megathrust west of 70.5°W. High Coulomb stress has also developed east of 65.5°W, where no historic events have been inferred to have relieved stress, though uncertainties in fault slip rates from our block model associated with a lack of GPS observations in this region may have led to an over-estimation of stress changes.