An assessment of toothfish in Subarea 48.3 using CASAL

The CASAL software has been successfully applied to an assessment of the stock of toothfish around South Georgia in Subarea 48.3. Both point estimate (“MPD”) and MCMC runs were carried out, with an extensive range of sensitivity tests and examination of diagnostics.
A baseline assessment model was selected to best represent the fishery and the available data sources. This incorporated two fleets with different selectivities, and fitted to catch length frequencies, tag recapture data, CPUE data, and the survey recruitment series (treated as a relative index). This produced excellent fits to the catch length frequencies, but a poor fit to the early CPUE data, for which a substantial process error CV was estimated (0.39). A possible alternative assessment model was also identified using a single fleet, as it is possible that overall trends in the CPUE series were not fully taken into account in the two fleet assessment. This alternative assessment produced worse fits to the catch length frequencies, and a poor fit to the CPUE data, for which an even higher process error CV was estimated (0.46).
For the baseline assessment incorporating two fleets, the 2005 vulnerable biomass was estimated to be 45,893 t, and the spawning stock biomass was estimated to be 120,360 t, around 67% of its unexploited level. For the alternative single fleet assessment, the 2005 vulnerable biomass was estimated to be 49,943 t, and the spawning stock biomass was estimated to be 100657 t, some 64% of its unexploited level. These estimates of current biomass are somewhat lower than the modified Petersen tagging estimates alone, which were 53,600 t and 55,600 t respectively (Agnew and Payne 2005: WG-FSA-05/17) but the confidence intervals for the estimates overlap.
The current implementation of CASAL is unable to reproduce all the features of GYM projections to estimate long term yields according to the CCAMLR decision rules. Using GY, the estimated long term yield is 4910 t for the baseline two-fleet assessment, while for the one-fleet assessment, the estimated long term yield is 4560 t.