CCAMLR

An accurate history of catch is an important input into the assessment of toothfish stocks, and uncertainty or bias in the amount of catch from an area must be taken into account in the assessment of a stock.

In 2013, Korea held a workshop on the anomalous CPUE data reported from some Korean vessels in data-poor exploratory fisheries (Delegation of Korea 2013). That paper requested that the data be analysed for the development and testing of plausible hypotheses to account for its inherent patterns.

One plausible hypothesis is that the catches were not reported proportionally from the areas where they were taken. For the purposes of developing an alternative catch history for these SSRUs, we use these data regarded as anomalous, along with that from other vessels fishing in the same area and season to illustrate a method for re-adjusting catches that may be area-misreported. We do this by using a standardised CPUE from these vessels to develop an estimate of the amount of catch that could plausibly be area-misreported on these vessel trips across the areas fished.

This allows the calculation of possible alternative catches in areas where reports may not reflect catches. In illustrating this approach to the vessels with anomalous CPUE, we can develop alternative catch histories that can be used to assess the sensitivity for assessments of toothfish.

We detail the procedure followed to check the Ross Sea CCAMLR fisheries data prior to using them in a stock assessment of Antarctic toothfish (Dissostichus mawsoni). The R codes used in this procedure have been made available to the Secretariat.

This paper proposes extending the trial of daytime setting of longlines between 15 and 30 April in the Patagonian toothfish longline fishery in CCAMLR Statistical Division 58.5.2 to encompass the 2013/14 and 2014/15 fishing seasons.  Extending the trial for a further two years will enable sufficient daytime fishing effort to occur in the trial period to ensure there is a proper basis for assessing whether to allow daytime fishing between 15 and 30 April on a permanent basis.

This paper proposes extending the fishing season in the Patagonian toothfish longline fishery in CCAMLR statistical division 58.5.2 on a trial basis during the 2013/14 and 2014/15 fishing seasons by a further two weeks. This would increase the fishing season to include the period from 1 to 14 November as part of the post-season extension with day/night setting allowed.

We have identified key uncertainties in the stock assessment of Antarctic toothfish in Subareas 88.1 ad 88.2 which could be addressed using specific research surveys and other dedicated voyages. Given the continued uncertainty arising from the relative paucity of abundance data, age and length composition data and gonad data for toothfish in some geographic areas and seasons, and from lack of sufficient data to understand toothfish movements and stock structure, we believe that research of this kind should be given a high priority by the Scientific Committee. We recommend the Scientific Committee invite members to submit proposals to address these uncertainties, and consider mechanisms by which coordinated research plans involving multiple Members could be generated either by the Scientific Committee or in its working groups.

We review tagging protocol changes made in 2012/13 and provide further recommendations to improve the quality of collected tagging programme data, and conduct research to improve effective tag releases and recoveries. We also provide a tagging training module to inform vessel observers and crew on tagging procedures.

We present a spatially explicit age-structured population dynamics operating model for Antarctic toothfish in the Ross Sea region, for a medium scale spatial resolution (189 spatial cells) covering the Ross Sea region. In this model run we spatially restrict the stock to cells where at least 5% of the depth is deemed suitable as habitat for toothfish (120 cells – semi-restricted model), and compare with previous models where the stock was either restricted to the fished cells only (65 cells – restricted model) or allowed to occupy the entire Ross Sea region including depths outside of those normally considered suitable habitat for toothfish (unrestricted model). The semi-restricted model provides a plausible distribution hypothesis in-between the two extreme bounding hypotheses represented by the previously presented restricted and unrestricted models (Mormede et al. 2013).

The most plausible model is the one that restricts toothfish distribution to depths which are known to provide suitable toothfish habitat. This is unsurprising because the model utilises fishery-dependent data and has no information about the distribution of toothfish in areas where no fishing has occurred, such that the unrestricted model estimates fish movements into cells outside of the fished area including those with implausible depths for toothfish.
 
Further data collection would be useful to improve the parameterisation of the model, in particular making collection of gonad weight measurements routine on all fishing vessels, surveying likely spawning grounds during winter, and obtaining fishery or survey data from areas not fished to date.

We propose that the modelling platform presented here can be used to test the likelihood and examine the consequences of alternative fish movement hypotheses given the data available. For a given movement scenario it can also be used to examine the likely effects on the toothfish stock assessment of alternate spatial management options affecting the distribution of fishing effort, data collection, and/or tagged fish releases.

This paper presents the Bayesian sex and age structured population stock assessment of Antarctic toothfish (Dissostichus mawsoni) in Subarea 88.2 covering SSRUs 88.2C, D, E, F, G, and H (SSRUs 88.2C–H). We examined several models including the 2013 implementation of the 2011 base case (R1), using the revised data selection method (R2), using a logistic selectivity for the northern fishery (R3), down-weighting the age data (R4), and having annual age length keys in the north (R5).

The stock assessment has strongly conflicting data: age frequency data push the biomass high, data from tags released in 2010, 2011 and 2012 suggest a much lower biomass, and data from tags released from 2004 to 2009 suggest an intermediate biomass.

The two models (R1 and R2) which estimate a higher biomass are those where the age data were more influential. However, these age data are not reliable because a single age length key has been applied. When annual age-length keys are applied there was a strong shift towards younger ages (R5) and lower biomass. The remaining two models (R4 and R5) suggest a substantially lower biomass than previously estimated, mainly due to the signal in the last three years of tag data. These are the data in which we have the highest degree of confidence for this fishery, due to improvements in tagging protocols and data collection.

We believe that there is sufficient conflicting information, and uncertainty in the age data, to warrant a decision to adopt the model which down-weights the age data in favour of the tag data as the base case. We therefore recommend that model R4 is used for providing management advice. In this model, the initial stock biomass is estimated at 6590 t (95% confidence interval 4800 – 9190 t), stock status is at 65%Bo (CI 52%–75%) and the yield based on the CCAMLR Decision Rules is 266 t. We also recommend that further ageing is carried out by fishing members involved in the fishery to allow annual full age length keys for all fisheries for future assessments.

We provide an update of the Bayesian sex and age structured population stock assessment model for Antarctic toothfish (Dissostichus mawsoni) in the Ross Sea region (Subareas 88.1 and SSRUs 88.2A–B), using revised catch, catch-at-age, and tag-recapture data for the 2012–2013 seasons.  Several models were examined including an update of the 2011 base case (R1), a model using the revised data selection method, new maturity ogive, and new data weighting method (R2), and a model which used a logistic selectivity for the fishery in the North. We recommend model R2 be used to provide catch limits for the stock because this model reflects the best science and methodologies available.
 
Retrospective analysis and MPD profiles suggest that the tag recapture data from the last few years tend to push the model toward a higher biomass, in particular the data from the 2012 tagged fish. We expect this is due to higher catches in SSRU88.1K which has a comparably high biomass but a shorter catch history (therefore fewer tags available for recapture) than the other slope SSRUs 88.1H and I.  Fish released or recaptured in SSRU K have also consistently shown higher rates of movement between SSRUs and lower rates of recapture in the location of release, perhaps indicative of a more mobile population in this location.

In contrast the updated data selection algorithm resulted in less tag data being selected for input into the stock assessment, which tended to push the model toward lower biomass.  The updated maturity ogive a slightly positive effect, and the new data weighting had no effect on model biomass. 

Sensitivity runs showed that cryptic biomass is not an issue for this stock assessment, and that the data from the sub-adult survey series is expected to be helpful in estimating year class strengths in the future.

Overall, model fits to the data were adequate, and, as in previous assessments, the tag-release and recapture data provided the most information on stock size. Monte-Carlo Markov Chain (MCMC) diagnostics suggested little evidence of non-convergence in the key biomass parameters, although there was some evidence of non-convergence in the annual shift parameters for the shelf fishery. MCMC estimates of initial (equilibrium) spawning stock abundance (B₀) for the 2013 reference model were estimated as 68 790 t (95% credible intervals 59 540–78 470); and current (B₂₀₁₃) biomass was estimated as 74.8% B₀ (95% C.I.s 71–78). The estimated yield, using the CCAMLR decision rules, was 3044 t.

Vessel-specific indices of tag detection rates and tagging mortality rates were developed in 2012 and further developed in 2013, using a methodology which controls for the spatial and temporal variability of fishing operations by pairing each individual tag release or recapture event with all other fishing events which occurred in the same time and place (i.e. within a specific distance and in the same fishing season). The method showed that when the confounding effects of variable fishing location and time were controlled there were nonetheless significant differences between vessels with respect to the tag detection index in the Ross Sea region.

Here we present the results of simulations which indicate that the index of tagging mortality rate obtained using this method is not as useful as is the index of tag detection rate. We recommend that in the absence of a better index of tagging mortality, the tag detection index be used to select data for use in the 2013 toothfish stock assessments for the Ross Sea and Subarea 88.2 fisheries, with respect to both tag releases and tag recaptures. This recommendation is in accordance with that of most of the participants at WG-SAM 2013 (paragraph 4.7).