I revised the CASAL catch-at-length models for stock status assessments of Dissostichus eleginoides in SSRUs C in Division 58.4.4 a & b (WG-SAM-13/21) following several advices during the last WG-SAM meeting as follows: 1)I included 2012 tag data (which was not used for model runs in WG-SAM-13/21), and a combined value of over-dispersions were used for all tag-release years (2008 and 2010‒12) instead of separated value for each year, 2)I calculated effective sample size for the proportions-at-length using regression line between coefficient variances estimated by a bootstrap procedure and expected proportions, 3)I changed the tagging mortality rate estimate of 0.1 instead of 0.2 because Japanese trotline system showed the captures of an adequate number of single-hooked fish in a state suitable for tagging. I made following four sensitive runs; 1)base, 2)25%_IUU, 3)100%_IUU and YCS models. The base model assumed the constant year class strength (YCS) and no IUU fishing occurrence in SSRU C. I changed the base model from that used in WG-SAM-13/21 which assumed estimated variability of YCS and a quarter of observed IUU fishing occurrence in SSRU C because estimated variability of YCS was inconsistence with the estimated vulnerable biomasses, and stock size estimate with an assumption of no IUU occurrence was conservative.
Fits of tag data were a little improved compared to those in the WG-SAM-13/21 in the base model. However, there was some evidence of lack of convergence in MCMC posterior trace and in the median MCMC jump statistics for the shape of the right hand limb of the double normal selectivity in each model. Hence, the MCMC results are not yet robust.
Median MCMC estimates (95 % credible intervals) of initial and current biomasses in SSRU C were 1100 tonnes (530-3900 tonnes) and 1040 tonnes (470-3840 tonnes), respectively in the base model. The CCAMLR yield was 28 tonnes in the model.