We investigate southern California seismicity in order to characterize its temporal evolution during the last decades. We analyze the time series composed of the number of events per year and the focal mechanisms of earthquakes since 1933. The results show a statistically significant nonstationarity, with a change that occurred in the 1960s in both time series. The seismicity before the change point is mostly characterized by a strike-slip focal mechanism of San Andreas type; after the 1960s the seismicity appears to show more scattered focal mechanisms and a lower seismicity rate. We provide a possible physical explanation of the significant nonstationarity by modeling the postseismic stress perturbation field induced by the two strongest earthquakes of the last century, the Chile (1960) and Alaska (1964) earthquakes, which both occurred in the 1960s. To first order, the postseismic stress rate seems to be in agreement with the observed changes in seismicity, supporting a causality hypothesis. The model also foretells the future behavior of the trend of southern California seismicity; this forward prediction provides an important opportunity to validate the causal hypothesis of a remote (and long-term) coupling between earthquakes.