We made five sensitive runs of single-sex age-structured CASAL models for stock status assessments of Dissostichus eleginoides in the research block 58.4.4b_1 following the recommendation of the WG-FSA-2014 meeting. IUU catch was estimated within the models using disease mortality functionality. Firstly four sensitive runs were made as follows: 1) inclusion of all observations and estimate of same double normal selectivity for disease mortality and fishery (OB_1a model), 2) removal of tag and CPUE observations in 2008 from OB_1a model (OB_1b model), and 3) modified OB_1b model as fix of fishing selectivity at the estimates of OB_1b model, use of knife edge selectivity at fixed 8 year for disease mortality assuming IUU operations using gillnets, and weighting for tag and CPUE observations as per OB_1b model (OB_2b model). Next, further two sensitive runs based on OB_1b were made as follows: 4) removal of all tagging data (OB_1b_notag model) and 5) removal of age observations, and fix of both selectivitties at the estimates of OB_1b model (OB_1b_noage model).
The fits to CPUE in 2008 and MPD profiles for tags in the same year in OB_1a model were spurious. This suggests that Shinsei-maru No. 3 was learning operation in its first research in 2008. The removal of 2008 CPUE data improved the CPUE fits in OB_1b model. The fix of knife edge selectivity at 8 year for disease mortality in OB_2b model did not reduce the total objective function values compared to the original OB_1b model. The selectivity for IUU fishery likely represents the size of fish available for catches in this area.
The OB_1b_notag model did not improve any fits of observations. Although the CPUE fitted better than other models, the fits to tag data did not improve in OB_1b_noage model. The IUU estimate in 2013 for this model was unrealistically high.
From the results on MPD fits to observations, OB_1b model was considered as the most plausible model among five models. This model generally showed well convergence for each MCMC estimate.
The median MCMC estimates of the initial and current biomass were 1060 and 570 tonnes in OB_1b model. The CCAMLR yield was calculated at 51 tonnes. We recommend this model be used as management advice.