Effects of recruitment variability and fishing history on estimation of stock-recruitment relationships: Two case studies from US West Coast fisheries

The Beverton-Holt stock-recruitment (SR) relationship has typically been reparameterized such that a single parameter, steepness, can be used to define recruitment compensation along a common scale. However, steepness is notoriously difficult to estimate in stock assessment models, particularly for the integrated analysis models used to assess groundfish stocks along the U.S. West Coast, due in part to the high levels of recruitment variability observed in the California Current Ecosystem. Additional complicating factors include the effects of exploitation history and subsequent level of contrast in stock abundance, which also influence the extent to which the SR function can be adequately estimated. We conducted a simulation study to explicitly evaluate the effects of recruitment variability on the ability to estimate the steepness parameter. We used two case study models to represent relatively short- and long-lived species of rockfish (Sebastes spp.), and for each we also evaluated effects of two fishing histories, reflecting high and moderate levels of exploitation and historical stock depletion. Not surprisingly, we found that even in data rich scenarios, using appropriately parameterized models, increasing recruitment variability alone can lead to high uncertainty in the estimation of steepness. Perhaps more importantly, with high recruitment variability steepness is frequently estimated to be near or at the upper boundary of the potential range of this value, and there is a bias towards overestimation of steepness in the face of increasing recruitment variability. This over-estimate of steepness occurred less frequently in high fishing scenario, suggesting that recruitment variability has less effects on estimation of steepness if the population has had greater contrast in relative abundance levels. The same results were found in both the relatively short and the long-lived species models. Inclusion of informative priors, either correctly or incorrectly specified, had a greater influence on estimated steepness values at higher levels of recruitment variability, further demonstrating the challenges and risks associated with the application of informative priors when recruitment variability is high.