Benchmark workshop on Northern Shelf cod stocks (WKBCOD)

The North Sea cod stock was last benchmarked in 2021 including a workshop on Stock Identification of North Sea Cod (WKNSCodID) reviewing the population structure information on the cod in the North Sea and adjacent waters. The workshop concluded that the North Sea includes different stock components of Viking and Dogger cod, and that the Dogger cod population extends to the West of Scotland (6.a.N). However, it was not possible to develop spatial approaches in time for the benchmark in 2021 because of (1) unexplained discrepancies between spatially disaggregated data and the data as used in the current North Sea assessment; and (2) the constraint that the 6.a stock was not included in the benchmark process. Nevertheless, the North Sea cod assessment was improved by revising the survey indices, biological data, and SAM assessment model configuration, thus lessening the data and assessment issues that had triggered the benchmark process. ICES recommended that further work be conducted on a data call to consider the different stock components as well as inclusion of the West of Scotland cod stock in the evaluation. This stock was last benchmarked in 2020 at WKDEM (ICES, 2020).

Several pre-data meetings were conducted with the data providers to ensure that the data asked for in the data call met the requirement of the stock assessors, and were realistic to provide for the relevant countries. The data set on landings for the substocks components covering 1995–2021 were evaluated at the data evaluation meeting and was found appropriate to use for the assessment. Presently the North Sea cod time series starts in 1963 and it would therefore be beneficial if the landings by substock could be extended further back in time. For the landings data covering the time period 1963–1995, ICES has a historic database with information on annual cod landings by area. It was decided to investigate if these data could be converted into landings by substock with some country specific assumptions. Also, recreational data were requested in the data call. However, very different quality levels for these data, with many missing years, were submitted by countries; therefore, it was decided that it is currently not possible to incorporate recreational data in the analytic stock assessment. It was concluded that catch level could, as in previous years, be given as a supplementary information in the WGNSSK report. A workshop on including recreational catches in stock assessment has been planned for in April 2023.

As the substocks are considered mixed in Q3-4, a combined index for the whole area was evaluated during the data evaluation meeting. It was decided to combine the survey data for Q3 and Q4 to ensure a full stock area coverage. As the substocks are considered separate in Q1, it was decided to split the Q1 indices based on the assumption that all fish observed during the Q1 surveys can be allocated to substocks based on where they were found. Given mixing is assumed to occur during Q3 and Q4, the decision was made to let the Q3+4 index remain aggregated and representative of the total stock. To allow testing with as many ages as possible in SAM, it was agreed to prepare indices with a 7+ group for the benchmark meeting.

A multi-stock (SAM) model was developed to take into account the substock structure in Northern Self cod stocks. This new model is estimated using substock Q1 survey indices and information about substock fishery catch compositions, as well as catch and survey indices that are only available as a sum for the substocks. For example, yearly catch-at-age data from the North Sea is only available as a sum of the catch of the Southern, Northwestern, and Viking stocks. Furthermore, the model can include genotype data to estimate stock- or catch-compositions.  

The benchmark meeting decided to use the BioPar option in the multi-stock (SAM) model for maturity, stock weights, and natural mortality. Three options were explored to address data deficiencies for maturities derived from Q1 surveys prior to 1990. New procedures were used to calculate catch and stock weights-at-age. The stochastic multi-species model SMS estimates of natural mortality rates were used for the three substocks assessed.

Many configurations and sensitivity analyses of the multi-stock model were reviewed during the benchmark meeting. In the final preferred model formulation, stock dynamics were modelled from 1983 onwards. The substock model could not be reliably extended prior to 1983 because of a lack of substock information about landings. Another motivation for truncating the assessment time-series was that there were no maturity estimates prior to 1983 to derive SSB. These values had to be assumed. Hence, the meeting concluded that the assessment model time-period should be constrained to 1983-present. However, the full landings time-series from 1963 onwards still provides some historic perspective on the size of the total stock during 1963–1982.

Estimates from the preferred model indicated that the Southern stock SSB decline steadily from around 21 600 tonnes in 1983 to 2700 in 2020, followed by a small increase in both 2021 and 2022. The northwestern stock was the largest component; during 1983 to 1997 its SSB fluctuated around 49 500 tonnes, followed by a large decline to 14 700 in 2005 and a generally increasing trend since then, except for 2017–2020. Throughout the period, the Viking stock SSB fluctuated around 14 000 tonnes[1].

Single combined-stock SAM models were also investigated during the benchmark meeting. Their results were very consistent with the multi-stock model results when combined for the three substocks, but the residual diagnostics were problematic for the single-stock SAM, and the retrospective pattern was at best borderline acceptable. The benchmark concluded that the multi-stock model was better suited to support the intended management options for northern shelf cod.

There is evidence from the literature of three recruitment periods or productivity regimes for cod in the North Sea: before 1988, between 1988 and 1998, and after 1998. The benchmark meeting consensus from the multi-stock model results was that all three substocks had higher recruitment rates (i.e., recruits per spawner) prior to 1997; that is, there were two recruitment regimes since 1983. Hence, the benchmark meeting consensus was that reference points should be based on the stock-recruit times-series since 1997, the last three years for selectivity, and the last 5 or 10 years for other biological parameters. However, agreement to split the stock-recruit time-series was not unanimous, and an alternative perspective is documented in this report.  

The meeting report documents a minority statement from Swedish participants. They did not agree with the setting of the Btrigger reference points, which they suggested were too low and should have been set at 50% of BMSY, at a minimum.  

The benchmark process provided many research recommendations dealing with these topics: 1) catch sampling programs that take the new cod substock structure into account, 2) improved genetic sampling information for the substocks, 3) M information for each substock, 4) substock specific landings fractions and catch weights-at-age, 5) further simulation testing of the multi-stock model, 6) fecundity information for the substocks.

[1] The values in this paragraph were corrected in version 2 of the WKBCOD report.