2018 updates of stock assessment for Pacific saury in the North Pacific Ocean by using Bayesian state-space production models (Japan)
Stock assessment was conducted for the North Pacific saury using most updated dataset of catch and abundance indices. Times series of the total catch record is available up to 2017. Time series of the abundance indices (standardized CPUE series for commercial fisheries by China, Japan, Korea, Russia and Chinese Taipei and biomass series from fishery-independent surveys by Japan) were submitted by Members and then discussed intersessionally by correspondence in Pacific saury stock assessment group via a collaboration site in the NPFC. There were some positive and negative comments to the submitted abundance indices (see summary compiled in NPFC-2018- SSC-PS-WP11), but a default option is to use all the indices available in the stock assessment. In addition, each member could select a set of indices as its preferred option (e.g. elimination of some indices in its stock assessment).
The model employed in the analysis is the state-space surplus production models as agreed in the SSC-PS01 as an interim stock assessment model. The model can account for process and observation errors in the abundance indices. The three agreed assumptions for relative bias in the Japanese biomass survey (Base cases 1-3) still exist, and therefore they were used as they are in this stock assessment. Given that all the abundance indices observed are not necessarily synchronized possibly because of difference in spatial use of fishing and survey grounds, a jackknife approach was used for Base case 1 to evaluate the sensitivity of analysis to the use of indices. Also, as a preferred option selected by Japanese CPUE scientists, stock assessment without use of Chinese and Korean CPUE series was also conducted. Parameters in the models were estimated based on Bayesian framework with a Markov chain Mote Carlo method.
The models were diagnosed with respect to shapes of posterior distributions, diagnosis plots, residual plots, retrospective pattern and predictability of the future population status. In terms of parameter estimation, shapes of posterior distributions were generally good except for Sensitivity test (Japanese biomass survey is removed from the analysis). The results of fitting tended to show that the estimated population dynamics fitted well to the Japanese and Russian CPUE series and the biomass indices by Japanese survey. Since there are somewhat different behaviors among abundance indices, standardized residual plots showed that the residuals are almost perfectly within the 95% range and the variance is homogeneous across years, but there observed directional residuals for Chinese and Chinese Taipei’s CPUE series. Model selection was not conducted formally in this paper partly because the models are not based on IID observations and therefore use of Bayesian types of information criteria such as WAIC may not be suitable, and partly because Bayes factor may have Monte Carlo errors. Instead, not yet comprehensively conducted, as a way of model checking, a kind of hindcasting approach (in terms of predictability; e.g. Kell et al. 2016) was used to identify retrospective patterns and predictability in the models.
As for the stock assessment, the current median depletion level is around 17-20% of the carrying capacity, and B-ratio (=B/Bmsy) and F-ratio (=F/Fmsy) are both around 90%, which indicated that the population status is assessed as overfished and subject to overfishing. When dropping the absolute biomass index (as an agreed Sensitivity test), the B-ratio tended to further go down.
These results are dramatically different from the previous conclusion made in SSC-PS in 2017, wWoverfishing”. Now estimated population dynamics strongly indicated that the population status is no longer healthy. The specification of models is not different from those used in the previous stock assessment. A big change occurred in the abundance indices of Members by their revision and addition of recent two years’ fishery/survey information. Elimination of earlier period of Japanese CPUE slightly drove the depletion level down, and the recent decline of abundance indices drove the current population status further down. This latter kind of change was not able to predict by the models.
For considering management implications, population dynamics were projected for some scenarios with respect to several levels of reduction/increase of catch as well as status quo. The results showed that continuation of the current level of catch cause further decline in the population size in the next decade. A safer option is of course reduction of catch to keep the population as the current level or make the population recovery to the size above the MSY level.
Given the results stated above, the authors concluded that the North Pacific saury population is “overfished” and the fishing pressure is in a state of “overfishing”, which means that the current catch level is harmful to the North Pacific saury population, and reduced catch are recommended. In addition, continued works/efforts for improving indices and models would be required toward better stock assessment and development of management procedures based on the assessment.