Accounting for tag loss rate is an important step in reducing bias in tag-based stock size estimates in CASAL. Here, we used 10,968 records of double-tagged Patagonian toothfish (Dissostichus eleginoides) released in Subarea 48.3 between 2004 and 2020, that were recaptured with at least one tag, to estimate tag loss rates over time at liberty. We explored different tag loss models to update the tag loss parameter for toothfish in Subarea 48.3 and test whether there was evidence of initial tag loss as well as ongoing constant tag loss; whether tag loss rates differed by size-class of tagged fish; and whether tag loss rates differed by season of tag release.
We found evidence for initial single tag loss, estimated as 2.8% (95% C.Is 2.0% - 3.6%), as well as ongoing constant single tag loss over time, estimated as 0.037 y-1 (95% C.Is 0.035-0.041 y-1) in the best fitting model. Our results also showed that there was no clear trend in initial tag loss or ongoing constant tag loss by season of tagged fish release, suggesting that initial tag retention has remained consistent over time and that there was no evidence for a tagging cohort effect on ongoing tag loss. When comparing tag loss rates between size-classes of toothfish at tagging, we showed no trend in initial tag loss with size-class, however there was an increase in ongoing constant tag loss from 0.023 y-1 (95% C.Is 0 - 0.039 y-1) at 50 cm to 0.046 y-1 (95% C.Is 0.037 - 0.057y -1) at 90 cm. Furthermore, we demonstrated that the single tag loss parameters, λ = 0.0061 y-1 (for recaptures at one to four years at liberty) and λ = 0.0074 y-1 (for recaptures at one to six years at liberty) best approximated the best fitting double tag loss model, and were comparable to λ = 0.0064 y-1 (one to four years at liberty) currently used in the stock assessment for Subarea 48.3.