Evaluation of the precision of parameters of the Schaefer production model applied to South Atlantic swordfish (Xiphias gladius)

• Published in: Fisheries research 2019-03, Vol.211, p.21-26
• Main Authors: Xavier, Érika Fialho Morais, de Andrade, Humber Agrelli
• Format: Article
• Language: English
• Subjects: Bayesian inference; MCMC; Poor data; State-space models; Surplus production model
• ISSN: 0165-7836; 1872-6763
• DOI: 10.1016/j.fishres.2018.11.001

•Information of datasets was assessed by state space production models as new data were added to the time series.•The approach was applied to South Atlantic swordfish stock.•There was no improvements in precision of the estimations of carrying capacity as new data were added to the time series.•The approach proved to be useful to assess how poor the data is, and how much we are learning from it across the years. Standardized catch rates calculated from commercial fishery data have been used for many years to inform production models, which are usually the basis of stock assessments whenever data are limited. Statistical theory implies that precision should increase as we gather more informative data about the process. It is therefore expected that the precision of population parameter estimates should increase as catch rate series increase in length. Does this actually occur? We explore whether new catch rate data, added each year improve the ability to assess stocks using a Bayesian state-space version of the production model. One can assess how much we have “learned” about the parameters of interest by comparing estimates calculated from shorter time series with those from longer time series. That line of reasoning is applied to datasets for swordfish of South Atlantic. Most of the results suggest that standardized catch rates added during the last two or three decades did not result in a significant gain of knowledge regarding model parameters.