A Bayesian modeling approach for determining productivity regimes and their characteristics.

Oscillations in the environment result in substantial alterations to population dynamics as evidenced by time series of abundance and recruitment. Depending on the reference timescale, these oscillations are referred to as regime shifts. Regime shifts may occur on very short time scales and are often undetected for several years. Consequently, tools that allow the estimation of regime-specific population dynamic parameters may be of great value. Using a hidden Markov model to describe the unobserved regime state, we develop methods to infer regime-specific parameters for a commonly used model of density-dependent recruitment in addition to identifying the unobserved regime state. We apply the method to recruitment data for Japanese sardine (Sardinops melanostictus).