This paper examines fishery data collected by Russian scientific observers on longline vessels that fished toothfish using Spanish longline and trotlines in CCAMLR and adjacent Atlantic waters during the 2002-2017 fishing seasons. The factors influencing the longline fishing impact zone    are discussed in this paper.  It is shown that the existing approach to the definition of "fishing impact zone" is aimed at determining the impact of fishing on the bottom. This is primarily important for assessing the risk areas where fishing can impact on VMEs.  However, the assessment of the CPUE, the understanding of the fish size composition in the catches requires an expanded understanding of the term "fishing impact zone". It should include both the interaction of the longline with the bottom, and the presence of a bait smell field that attracts fish to the longline.

Stock Annex for the 2021 assessment of the Antarctic toothfish (Dissostichus mawsoni) population of the Ross Sea region, including stock structure and definition, fishery information, catch data, biological information, abundance information, and stock assessments.

Here, we provide diagnostic plots for the 2021 assessment model for Antarctic toothfish (Dissostichus mawsoni) in the Ross Sea region presented in Grüss et al. (2021a), following the recommendations of WG-SAM-2015 (SC-CAMLR-XXXIV 2015 Annex 5). The stock assessment model is described in Grüss et al. (2021a), and a detailed description of the stock area, stock assessment methods and the stock assessment parameters are given in Grüss et al. (2021b).

We update the Bayesian sex- and age-structured integrated stock assessment model for Antarctic toothfish (Dissostichus mawsoni) in the Ross Sea region (Subareas 88.1 and Small-Scale Research Units (SSRUs) 88.2A-B) using the most recent available data for the Antarctic toothfish fishery. The assessment model employs reported catch for 1998–2021, tag-release data for 2001–2019 and associated tag-recapture observations for 2002–2020, commercial fishery age frequencies for 1998–2020, abundance observations from the Ross Sea Shelf Survey (RSSS) for 2012–2021, and age observations from the RSSS for 2012–2020.

The assessment model used slightly revised catch data and observations for 1998–2019, and new data and observations for 2020 and 2021. Observational data (tag releases, tag recaptures, and age/length data) from vessel trips that had been quarantined since the previous assessment were also excluded. The model structure was the same as that employed in 2019. Additional data that should be available to enable an update of this assessment to be reviewed by the 2021 meeting of the Working Group on Fish Stock Assessment (WG-FSA) will include age data from the 2021 RSSS, tags released in 2020, and tags recaptured in 2021.

The 2021 model (R1.1) maximum posterior density (MPD) estimated the equilibrium pre-exploitation mature (spawning) stock biomass (B₀) as 79 140 t, and the current stock status (B₂₀₂₁) as 61.9%. Markov chain Monte Carlo (MCMC) estimated B₀ as 78 530 t (95% CIs 72 090–86 470 t) and the current stock status (B₂₀₂₁) as 62.0% B₀ (95% CIs 58.9–65.2% B₀). The estimated status in 2019 from the updated model was almost the same as that estimated in the 2019 assessment (B₂₀₁₉ was 66.0% in the 2019 model and was 65.7% in the updated 2021 model).

The key outcome of the sensitivity analyses was that exclusion of the initial three years of tag-release data (2001–2003) and associated tag-recapture data resulted in almost no differences in estimated B₀ or current biomass and improved the overall model fit. Although exclusion of the initial three years of tag-release data and associated tag-recapture data did not improve the Pearson residuals of the catch-at-age data (and decreased the negative log likelihood for the remaining tag-recapture observations), results suggested that a model without the early tag data could be used as the base case model for the 2021 assessment of Ross Sea Antarctic toothfish. The main advantage of doing this would be to reduce the number of partitions within the assessment model, thereby decreasing the computer memory allocation required and processing time to evaluate the model. We recommend that the sensitivity excluding the 2001–2003 tag data (R1.2) be employed as the base case for the 2021 stock assessment.

We recommend that future development should include, inter alia, aggregation of the very old fish (> 35 years) in the observations as a plus group, recruitment of fish into the model at age four or five, and division of fisheries into discrete temporal periods.

Ukrainian vessel CALIPSO (Fig. 1) (shipowner FC NEPTUNO LLC, Ukraine) performed the research survey in the statistical subarea 48.1 according to the research plan SC-CAMLR-39/BG/08 and management advices of the WG-FSA-2020 and SC-CAMLR-39 in the second part of February 2021. The survey was carried out partly and stopped in advance due to the exhaustion of the limit (a total of 6.56 tons were caught from the allocated 7 tons) for fish species Macrourus spp.

The problem of differences more than 10% between vessel catch reporting form (C2) and landed Dissostichus catch (DCD) on Ukrainian fishing vessels Simeiz, Koreiz and Calipso was raised  by the CCAMLR Secretariat in 2018, and the data from these Ukrainian fishing vessels for the period 2014-2018 has been quarantined. There is proposed to include these data, revised by Ukraine, into the CCAMLR database (with special mark), leaving the uncorrected  data with the appropriate mark also available for use.

In RB5, the SICs in Feb. 2021 were the highest and the SSTs were the lowest for the years 2016-202.  In Mar.  2021, the highest SICs decreased to nearly the long-term average while the SST increased accordingly. In the same year, the SICs and SSTs had two peaks in Feb. and Mar. respectively.  In RB4, the SICs during Jan.- Feb (Austral summer) in 2021 were also the highest since 2016. The sharp spikes of SST (rapid increasing SST) had become smaller year by year from 2017 to 2021, which indicates that the SSTs had a cooling phase in 5-6 year periodical cycles corresponding to an increasing trend in SICs.

Spatial dynamics of SICs with SSTs contour of -1.8°C and -0.8°C were analyzed. It was found that the ice edges are at approximately -1.8°C and partially broken ices exist between -1.8°C and -0.8°C when comparing imagery by GIBS and SICs distribution by AMSRs with SSTs by NOAA.

Daily wind stick plots indicate that the eastward winds could encourage the off-shore Ekman transport at the end of Feb. and the beginning of Mar. which resulted in late (slow) ice retrieval in 2021.

Recommendations

When modelling multiple fleets in a GYM assessment, we recommend the use of the presented extension to the Grym package based on its shown capacity to model more complex fisheries and fishing practices with a high degree of consistency in results.

Abstract

Grym is an implementation of the Generalized Yield Model in R that provides greater transparency and extensibility (Maschette et al., 2020, Wotherspoon & Maschette, 2020). This paper describes an extension of the Grym to permit multiple fleets within a season, allowing the Grym to model more complex fishery behavior and evolving fisheries practices. Results from the expansion are compared with existing analyses from the Grym. The model can include vessels using different gear types (e.g. trawl and longline) or identical selectivity. An example is presented for Patagonian toothfish.

Recommendations

That WG-SAM note these preliminary explorations of alternative decision rules for managing toothfish fisheries, which indicate that an approach based on a harvest rate H:

• Would be consistent with the CCAMLR decision rule and its objective, and;
• Could provide a higher level of predictability and certainty about the likely changes in future catch limits, and reduce fluctuations in stock size around the target level in the long term.

Abstract

In this paper, we conducted simple simulations to outline alternative decision rules with the potential to manage toothfish fisheries according to the current CCAMLR decision rule. In contrast to the current decision rule, the rules in this paper were based on a harvest rate H which was stochastically estimated from the stock productivity and fishery selectivity to result in the long-term 50% SSB depletion with a probability of 50%. Catch limits were then calculated from the harvest rate, the fishery selectivity and the current vulnerable stock biomass, bound by minimum and maximum levels of change to reduce short-term variability in the catch limit. We evaluated two different types of decision rules (constant H and hockey-stick response function for H). All evaluated scenarios demonstrated that:

• The SSB target level was reached in around 35 years in all simulations, indicating that an approach based on a harvest rate would be consistent with the CCAMLR decision rule and its objective;
• The expected catch limit variability between assessments would be around 5–10% on average and rarely hitting the 20% maximum level used here.

An alternative rule based on a harvest rate could provide a higher level of predictability and certainty about the likely changes in future catch limits, and reduce fluctuations in stock size around the target level in the long term.

Recommendations

For GYM assessments with proportional recruitment, we recommend to use:

• Formula-based methods for low variance in proportional recruitment;

• Simulation modelling for high variance in proportional recruitment to accurately reproduce mean and standard deviation;

as described in this paper and implemented in the Grym (open source package).

Abstract

Krill are a keystone species in the Southern Ocean food-web, and, as such, it is crucial to effectively manage the krill fishery to ensure its long-term sustainability. To assess the impacts of current harvesting pressures, evaluations rely on sampling and population modelling. Krill stock projections are developed with the Generalised Yield Model (GYM), which provides an assessment for stock status under current harvesting scenarios and various levels of uncertainties. One of the fundamental components of the GYM is the simulation of recruitment. De la Mare (1994) presents a proportional recruitment function for estimating krill recruitment based on estimates from field surveys. The De la Mare (1994) function uses estimates of the mean and variance of recruitment from survey data to determine the scaling of natural mortality and the distribution of random recruits that reproduce the observed mean and variance estimates. We evaluated De la Mare’s (1994) proportional recruitment function and found that for large variations in recruitment the function does not reproduce the observed mean proportion of recruits and its variance accurately. We review the deficiencies within the model and provide two alternative methods, which can support a wider range of values and possible extreme scenarios, such as years of low recruitment.