An Evaluation of Natural Mortality Assumptions in Stock Assessments

Although some stock assessments now include age varying―or age and time varying ―natural mortality (M), many stock assessments still assume M is constant across ages and years, which is likely not valid.  We modeled two generic fish populations, long-lived and short-lived, with age and time varying M, and simulated the collection of data commonly used in stock assessments.  Statistical catch at age models were then fit to the simulated data incorrectly assuming M was constant across ages and years. The bias in estimates of spawning biomass (SSB) and fishing mortality (F) varied among years for both long- and short-lived populations; the median bias was often +/- 100% in magnitude.  The trend, magnitude, and direction of the bias were partially defined by how the true M varied among years.  Bias variation was greater for the short-lived population implying that mis-specification of M for short-lived species may have different consequences than for long-lived species.