The origin and prevalence of triggered seismicity and remote aftershocks are under debate. As a result, they have been excluded from probabilistic seismic hazard assessment and aftershock hazard notices. The 2004 M = 9.2 Sumatra earthquake altered seismicity in the Andaman backarc rift-transform system. Here we show that over a 300-km-long largely transform section of the backarc, M≥4.5 earthquakes stopped for five years, and over a 750-km-long backarc section, the rate of transform events dropped by two-thirds, while the rate of rift events increased eightfold. We compute the propagating dynamic stress wavefield and find the peak dynamic Coulomb stress is similar on the rifts and transforms. Long-period dynamic stress amplitudes, which are thought to promote dynamic failure, are higher on the transforms than on the rifts, opposite to the observations. In contrast to the dynamic stress, we calculate that the mainshock brought the transform segments approximately 0.2 bar (0.02 MPa) farther from static Coulomb failure and the rift segments approximately 0.2 bar closer to static failure, consistent with the seismic observations. This accord means that changes in seismicity rate are sufficiently predictable to be included in post-mainshock hazard evaluations.
