Consequences of error in natural mortality and its estimation in stock assessment models

Abstract

Natural mortality (M) is often considered to be one of the most important parameters in a fish stock assessment and affects productivity estimates for the population. However, it is also among the most difficult parameters to estimate using commonly available data. The magnitude of error (both bias and variance) when estimating this parameter can be substantial and can be affected by ignoring its variation over time, space, age, and length. In this study we explore the implications of errors in M on estimation and management performance using simulations and illustrative examples. The error in management reference points such as F35% and FMSY is related directly to the error associated with M. Estimates of biomass are expected to be positively biased when M is over-estimated and vice versa. Use of feedback management strategies reduces the impact of errors in M, but performance in meeting management objectives is compromised when M is in error. Estimating M was found to perform better than pre-specifying M in closed-loop simulations. Also, we found that the consequences of setting M to an incorrect value were reduced if stock-recruitment steepness was estimated. Based on our study and a review of related work, we advocate estimating M within an assessment, ideally with a prior for M tailored to the stock concerned.