Evaluation of the impacts of using a double tag loss rate function and changing the time at liberty in the assessment of Ross Sea region Antarctic toothfish (Dissostichus mawsoni)

We investigated alternative parameterisations of the tag loss rate function and the effect of including tag recapture observations with greater time at liberty in the Ross Sea region (RSR) Antarctic toothfish (Dissostichus mawsoni) assessment using Casal2. The analyses used the Casal2 implementation of the 2021 CASAL assessment model that was used by the Scientific Committee to provide management advice for the RSR in 2021.

Antarctic toothfish in the RSR have been double tagged since 2004, with single tags used in 2001 and 2002 and a mix of double and single tagging used in 2003. The 2021 assessment model approximated the tag loss rate of double-tagged fish using a single tag loss rate function and only included tag recapture observations for a time at liberty of up to six years from the year of release. The single tag approximation was used for double-tagged fish because CASAL only implements the single tag loss rate function, and this approximation was assumed to be inadequate for times of liberty greater than six years.

To evaluate the consequences of changing tag loss rates in the assessment, we implemented models that use the estimate of tag loss for double-tagged fish rather than the single tag approximation. Further, we evaluated the implications of increasing the observations of tag recaptures with more than six years at liberty by including all the tag recapture data available for the 2005–2020 tag release cohorts.

Using the double tag loss rate function, instead of the single tag approximation, resulted in slightly higher estimates of initial biomass (B₀), current biomass (B₂₀₂₁) and current status (B₂₀₂₁/B₀). Diagnostics and residuals using the single tag approximation or the double tag loss rate function were almost identical and did not significantly favour either model. However, given that all fish since 2004 have been double tagged, we recommend the use of the double tag loss rate function for years since 2004 in future models.

Including tag recapture observations with greater times at liberty resulted in higher estimates of B₀ and B₂₀₂₁, however, model diagnostics suggested a pattern of trend in residuals and likelihood profiles with increased time at liberty. Four hypotheses for the pattern were considered: (i) that the tag loss rate for tagged fish was higher than was estimated outside the model, (ii) that tagged fish suffered ongoing additional mortality associated with being tagged, (iii) that migration and dispersal of the tagged fish resulted in a lower probability of recapture with increased time at liberty, and (iv) that the effect of tagging on fish resulted in a lower probability of their recapture in the year (or years) immediately following release.

Models were implemented that included processes to account for each of these effects. Diagnostics indicated that either (ii) or (iii) were more consistent with the tag recapture observations, with migration and tag dispersal the most plausible hypothesis, but the pattern was not fully resolved especially in the early years of time at liberty. The lower probability of recapture in the year following release was not supported by the data, with the trend best explained by a decreasing (not increasing) probability of capture with greater years at liberty (i.e., equivalent to the effect in (ii) or (iii)).

Based on these results, we recommend that sensitivity models include all times at liberty for the tag recapture observations and also investigate effects for ongoing tag-related mortality or tagged fish dispersal. We also recommend that further work be carried out to investigate these hypotheses further, including approaches to account for tag related mortality, dispersal and migration, and the relative availability of tagged fish following area-based management changes in the Ross Sea region.

In addition, while tagging of toothfish in the RSR has been undertaken since 2001, CCAMLR only required vessels to tag toothfish as a part of fishing operations from 2003, with double tagging required since 2004 and the requirement to tag fish in proportion to the length frequency of catch introduced in 2005. Hence, while previous analyses used tag release data from 2001, the requirements for tagging toothfish in the RSR have only been broadly consistent since 2005. For the 2023 assessment, eighteen years of release and recapture data since 2005 will be available and the inclusion of tag recaptures with all years at liberty will significantly increase the amount of tag recapture observations over the current model. Hence, to avoid potential biases from the use of the early data for single-tagged fish, and noting the increased number of observations that will be available by including tag recaptures from all years at liberty, we recommend that tag release data from 2001–2004 be excluded from future assessments of RSR Antarctic toothfish.