CCAMLR

At its 2011 meeting, the Scientific Committee agreed that a time series of relative recruitments from a well-designed survey could be a useful input into the Ross Sea stock assessment model and endorsed a proposal to carry out this work once the fishery had closed at the end of the 2011/12 season. The first survey was successfully completed in February 2012 and the results presented to the 2012 meeting of WG-SAM (Hanchet al. 2012). In this paper we provide a proposal to continue this time series.

The objectives of this second survey of the time series are as follows:

(i) To carry out the second of a time series of longline surveys to monitor pre-recruit (<110 cm TL) toothfish in the southern Ross Sea (strata A12–C12) using standardised gear in a standardised manner; and

(ii) To define additional experimental stations in adjacent areas to identify areas of high pre-recruit abundance which could be included as additional strata in future annual surveys.

CCAMLR Conservation measures currently require vessels to tag only toothfish with a high probability of survival, but no objective criteria are provided for this evaluation. Achieving this requires an objective evaluation of the probability of survival of each fish to be tagged, which in turn requires objective criteria that link observable characteristics to survival. This paper proposes criteria based on assessments of injuries and vitality used on other species and includes several options for implementation in CCAMLR tagging programmes.

Mark-recapture methodologies are commonly used to inform the management of fisheries. In addition to answering targeted research questions, tagging programmes are typically used to describe movement patterns relative to stock unit identification or to support spatial disaggregation of stock assessment areas, and in some instances to derive population abundance estimates. This paper is intended as a discussion tool to identify and prioritize tagging-related issues for further research drawing on the experience tagging programmes internationally. To the extent that CCAMLR tagging programmes are unique, an evaluation of potential programme biases and corresponding information needs is presented. The paper focusses on the design and implementation of toothfish tagging programmes, but many of the potential issues apply equally to other tagged species such as skates.

With the recognition that CCAMLR tagging programmes are used for abundance estimation of toothfish and skates, tagging operations need to be appropriately supported through documentation, training, and supplies. This paper reviews how tagging programme information could be effectively organised, provides a template training module for observers and vessel crew, and makes recommendations to update tagging protocols and to create a tagging checklist for use at the tagging station.

A reliable commercial tagging program is critical to the successful assessment and management by CCAMLR of a number of toothfish fisheries in Antarctica. The evaluation of the tagging programme relies on two key aspects (i) that the tagged fish have a good chance of survival and (ii) that the tagged fish are reliably detected when recaptured. Previous attempts to evaluate individual vessel tagging performance in relation to these two aspects has been thus far inconclusive because of confounding factors such as time and location of tagging and subsequent fishing effort, as well as size of fish tagged.

We propose that, by controlling for the spatial and temporal confounding factors using a case-control study design, we can derive meaningful indices of relative tagging performance of groups of fishing effort (e.g. vessel, vessel-trip). We developed indices to compare (i) the tag detection rate of recaptured fish and (ii) the tagging survival of released fish.

This method was applied to the toothfish tagging data in CCAMLR Subareas 88.1 and 88.2. Initial results show that the indices developed can provide evidence of significant differences in tagging performance between different groups of fishing effort. Preliminary investigation showed these indices are robust to the choice of the control group and the area included in the analysis. Subject to further checks, this method could be used to investigate the relative tagging performance of different components of the CCAMLR tagging program across all fisheries, and more generally the relative performance of spatially and temporally heterogeneous data sets.

At its 2011 meeting, the Scientific Committee agreed that a time series of relative recruitments from a well-designed survey could be a useful input into the Ross Sea stock assessment model and endorsed a proposal to carry out this work once the fishery had closed at the end of the 2011/12 season. The survey had two main objectives: (i) To establish the feasibility of developing a time series of longline surveys to monitor pre-recruit (<100 cm TL) toothfish in the south of SSRUs 881.J and 881.L in the southern Ross Sea using standardised gear in a standardised manner; and (ii) To carry out experimental depth-stratified fishing in 400–600 m depth adjacent to the survey boundaries to establish the most appropriate depth strata for future annual surveys.

The survey demonstrated it is possible to carry out a longline survey of pre-recruit toothfish from commercial fishing vessel in Ross Sea using standardised gear in a standardised manner. Based on a total of 45 sets in the three core strata (A12-C12) the survey had an overall c.v. of 9%, which met the target c.v. of 10%. The survey caught mainly 70–110 cm TL Antarctic toothfish. The survey provided new data on the depth distribution of pre-recruit fish in this area. Catch rates of pre-recruit toothfish were highest between 500 and 900 m and very low shallower than 450 m. The size distribution of toothfish was very similar between the four main strata suggesting no depth preference between 400 and 900 m. It is recommended that future surveys in this area should focus on depths greater than 500 m. The survey also demonstrated the feasibility of collecting samples for wider ecosystem monitoring.

Although the main aim of the 2012 survey was to monitor pre-recruit toothfish, the survey provided the opportunity to compare catch rates in 2012 with historical catch rates by San Aotea II and its sister ship Janas in earlier years of the fishery. The results of a standardized CPUE analysis suggest that there has been no significant change in catch rates in the southern Ross Sea over the past decade.

Currently the target tagging rate to be achieved by vessels fishing in the CCAMLR area is set in terms of the number of fish tagged per mass of fish caught thereby mixing “units” of measure. This note questions whether it is appropriate to set the tagging targets these terms and suggests that defining the target in terms of a percentage of the number caught may be more appropriate.

Using a modeling framework for toothfish population dynamics, fishing and data collection, this study investigated how the bias and precision of a CASAL assessment is influenced by various aspects of a tagging program, in particular the effects of the numbers of fish tagged, the duration of a tagging program, the size of tagged fish, and the type of auxiliary data used in the assessment.

The simulation study indicated that:

(1) A high tag size-overlap (at least 60%, better 100%) and high tagging numbers were important particularly in the early stages of an exploratory fishery to ensure some recaptures and maximise the likelihood of a robust assessment.

(2) Using catch-at-age data instead of catch-at-length data substantially improved all assessment estimates and was required for adequately estimate annual recruitment and year class strengths.

(3) For a simulated lightly fished stock, CASAL assessments with good quality tagging data had a low bias and reasonable precision for SSB₀ and annual SSB estimates, however the precision of SSB status estimates in the final assessment year was relatively low. Bias and precision of CASAL assessments were better for the lightly fished stock than for a heavily fished stock.

(4) This modeling framework appears to be useful for evaluating data sampling strategies, assessment approaches and management strategies.

Model used by WG-SAM-12/24

Using the simplest form of an abundance estimator using tag-recapture data, the Lincoln-Petersen equation, we summarise the processes that may occur during a tagging program that lead to difficulty in accurately estimating key parameters, and consequently introduce bias into an abundance estimate. We also review the methods that have been used in CCAMLR tag recapture to program to mitigate against this bias, or to account for their effects in assessment models. In nearly all cases, examples of at sea or model-based approaches are available from established toothfish tagging programs to reduce these biases. Estimates of post-capture mortality, tag detection rates and where it occurs, post-release depredation rates are a priority for new tagging programs and tag-based assessments. Due to the complexity of toothfish movements throughout their life cycle, as well as the spatial structure of release and recapture efforts by fishing and research vessels, development of spatially explicit modelling approaches is an important next step for toothfish assessments.

In order to reflect the level of confidence when linking a tag recapture to a tagging event, as well as an indication of the time and effort taken to make the link, the Secretariat introduced in 2011 a tag recapture link status code.  For the 2011 season, a total of 1433 tags have been returned; 50% provide an exact match with tag numbers held in the database, 22 % provide an exact match with tag numbers but had inconsistent ancillary data. For those tags that did not provide exact or unique matches only 109 (8%) could not be successfully linked using additional information. For the current season (2012), a total of 164 tags have been returned; 64% provide an exact match; 34% provided an exact match with inconsistent ancillary data; and less than 2% could not be successfully linked. An approach to developing thresholds for ancillary data to be used in allocating tag link status are provided for comment.