Resonances as a record of planetary migration
While hundreds of extrasolar planetary systems have been discovered, only a small fraction of them contain more than one known planet. An even smaller fraction of known multi-planet systems are relatively compact, containing giant planet pairs with period ratios below three or four (Saturn-to-Jupiter period ratio is about 2.5). Most multi-planet systems contain planet pairs on eccentric orbits, with orbital periods varying by orders of magnitude. This is almost certainly a consequence of planet-planet scattering, which would have erased much of information about the prior dynamical evolution of the system.
Current theory of planet formation suggests that there are generally two epochs of giant planet migration: one caused by interactions with the gas disk within the first few Myr of the system's history, and a later one (possibly lasting for tens to hundreds of Myr), driven by gravitational interaction between the planets and solid planetesimals. Planets with smaller masses are expected to be affected more by interactions with planetesimals, due to the limited amount of mass available as solids (cf. Raymond and Armitage 2009).
We note that very massive known compact exoplanet pairs tend to occupy resonances more often than those with masses similar to Jupiter and Saturn. Using numerical simulations, we show that the non-resonant compact pairs can be plausibly derived from an initial resonant or near-resonant configuration, much like the Nice model (Tsiganis et al. 2005) proposes for our solar system. We conclude that the gas-driven migration might be often ending with many of the planets in resonances, despite the effects of turbulence (Adams et al. 2008). It is possible that after the gas has dissipated, the scattering of sizable solid protoplanets enables resonance-breaking in less massive systems but not in more massive ones. We propose observational tests of these hypotheses.