​Benchmark workshop on horse mackerel and boarfish (WKBHMB)

The benchmark workshop on horse mackerel and boarfish (WKBHMB) evaluated the assessment (input data and methodology), short-term forecast procedures, and reference points for three horse mackerel stocks and one boarfish stock. A key focus has been on evaluating stock structure for horse mackerel making new data available for catches, surveys and CPUE, and applying new modelling approaches to the assessments of the stocks.

General introduction to horse mackerel in the Northeast Atlantic (Section 2)

The data call for WKBHMB led to a revision of the landings and discards data, and some additional survey and CPUE data, for the three horse mackerel stocks. Catch data were supplied by 14 countries and length sampling data and age–length keys were supplied by 7 countries. The majority of countries delivered data for the 2003–2022 period. Based on the periods covered by the data submissions to the benchmark, revised dataseries were compiled for the period 2003–2022. No revisions were made to the existing data for earlier years. Because the data were mostly supplied at métier level 6, the InterCatch platform could not be used (impractical to raise all the unsampled catches). Therefore, R scripts were developed to check and clean individual length and age-based InterCatch national data submissions and to raise sampled catches to the total catch at the appropriate level.

The stock structure of horse mackerel In the Northeast Atlantic has been reviewed, based on several genetic research projects that have been carried out. The widescale application of the genetic stock identification approach has concluded that:

• The North Sea stock is primarily restricted to the southern part of 4.b, 4.c, and 7.d. Therefore, catches in divisions 3.a and 4.a are now considered to belong to the Western stock.
• There is significant mixing of the Western stock and the North Sea stock in 7.d and 7.e resulting in the need to carry out sensitivity analysis on the consequences for stock assessment (which unfortunately was not possible to carry out during WKBHMB).
• Although the small number of baseline samples from the Southern biological unit prevented the development of an assignment model between the Southern and Western biological units, there was enough information to conclude that there was mixing of the Western and Southern units along the Portuguese coast. The majority of the Southern individuals were caught south of Lisbon.

Further development of the genetic assignment models is required to establish a robust basis for realignment of the stock assessment areas and/or for splitting catch and survey data by biological unit. A horse mackerel research plan has been outlined in Section 2.4.

The approach to estimating natural mortality and maturity has been aligned between all three horse mackerel stocks.

Western horse mackerel (Section 3)

The western horse mackerel stock is now defined as horse mackerel which occur in ICES Subarea 8 and divisions 2.a, 3.a, 4.a, 5.b, 6.a, 7.a–c, 7.e–k. Based on the stock structure analyses, WKBHMB has reallocated catches from divisions 3.a (west) and 4.a in quarters one and two from the North Sea stock to the Western stock.

The input data for western horse mackerel was adapted as follows:

• Catch data for the period 2003–2022 was updated with the revisions submitted to the catch data call and now includes both the landing and discard components.
• The biomass trend derived from the acoustic survey PELACUS was replaced with a combined biomass index derived from three acoustic surveys that cover most of the stock area during spring.
• A standardized commercial CPUE time-series (2017–2022) from the Pelagic Freezer Trawler Association (PFA) self-sampling program has been added as an additional biomass index.
• The methodology to estimate the recruitment index was improved. The index is now calculated with a species distribution model in TMB, which includes new bottom-trawl surveys to cover most of the stock area during the spawning season.
• The maturity ogive was updated using data from the acoustic survey WESPAS and from the fishery.
• The natural mortality is now estimated by age following the Gislason model.

Stock assessments were performed with both SAM and Stock Synthesis (SS3). Both SAM and SS3 yielded acceptable and comparable results. In the absence of a compelling reason to choose either SS3 or SAM, the choice was made to continue using SS3 because the stock assessor was more familiar with that model framework. The final model proposed for Western horse mackerel is an annual age-based model with one single area and one single fishing fleet performed in SS3. Overall, the model fits reasonably well the index and catch data, although, similar to the previous model, there is a tendency to overestimate SSB. The retrospective pattern has been investigated and WKBHMB concluded that it is likely driven by the data. The retrospective pattern was corrected as much as possible by reducing the temporal variability of the fishing selectivity.

The reference points were updated following the ICES EqSim approach. B_lim was quantified as the minimum biomass observed since 1998 and B_pa was derived based on the uncertainty in the model estimates of SSB. MSY reference points were estimated based on simulated populations with an SR segmented regression model with a fixed breakpoint at B_lim and autocorrelation in recruitment fixed at 0.42.

There was no time during WKBHMB to set up the short-term forecast. A stochastic forecast within SS3 will be set up during WGWIDE 2024 to provide catch advice for 2025.

North Sea horse mackerel (Section 4)

The North Sea horse mackerel stock is now defined as horse mackerel which occur in divisions 4.b, 4.c, and 7.d. There appears to be significant mixing of the Western stock and the North Sea stock in 7.d and 7.e but unfortunately a sensitivity analysis on the consequences for stock assessment could not be carried out during WKBHMB.

The survey data from the FR-CGFS and NS-IBTS has been revisited and biomass information has been re-evaluated. Additional information on maturity at length (taken from the western horse mackerel) allowed for a transformation into an SSB index to be used in a SAM assessment. A standardized commercial CPUE time-series (2017–2023) from the Pelagic Freezer Trawler Association (PFA) self-sampling program has been added as an additional biomass index.

The re-evaluated catch data in combination with biological assumptions on maturity and mortality-at-age allowed for a category 1 assessment to be run within the FLSAM framework. The North Sea horse mackerel assessment has been upgraded from a category 3 to a category 1 assessment.

Biological reference points were estimated using the EqSim framework (ICES, 2021) but are considered to be very uncertain. The appropriateness of the estimated biomass reference points was explored using 200-year projections with a dynamic stock–recruitment relationship based on a segmented regression which mimicked the loess-smoothed SR relation as closely as possible. The exploration showed that it would take approximately until 2060 until the stock would reach the estimated B_lim. Although the appropriateness of the specific biomass reference points is questionable, stock status is far from any candidate reference points and trying to estimate the reference points with more accuracy would not make any difference to the stock status in the short term.

A stochastic short-term forecast method has been implemented for the stock. However, no catch scenarios will be presented because the uncertainties about the biological reference points should be addressed first.

Southern horse mackerel (Section 5)

A large-scale genetic study identified issues with the delineation of the Western and Southern horse mackerel stocks. The genetic analyses indicate that the Western biological unit extends significantly into Division 9.a. Further baseline samples are required from Division 9.a throughout the spawning season to allow a genetic assignment model to be developed for potentially splitting mixed survey and commercial catches by biological unit. It is suggested that the results of the genetic analyses need to be analysed in a multidisciplinary context that also includes comparative assessments of life-history traits such as growth, otolith shape, reproduction, and distribution. In the absence of sufficient information on the stock delineation of the Western and Southern horse mackerel stocks, no changes to the southern horse mackerel stock definition were made during the benchmark meeting.

Data and assessment

All the available index information has been reviewed and evaluated. Two new tuning indices have been added to the assessment: a commercial CPUE index and a DEPM survey index. The commercial CPUE index was developed using an extensive dataset of revised and compiled data from the Portuguese trawl logbooks. This index uses a combination of multiple métiers/areas to capture the abundance across several stock areas and a broader coverage of the stock components. A revision of the SSB estimates from the DEPM surveys between 2010 and 2022 is now included as additional information in the assessment of southern horse mackerel.

The assessment method has been changed from AMISH to SS3. The challenge with the AMISH model has been that making changes to the model was difficult because of the limited documentation and limited availability of the developer.

The major change in the SS3 model is an updated selectivity pattern for the commercial fleet. This provides more flexibility by using time-varying selectivity through random walks. Other relevant updates to the model include a revised maturity ogive and natural mortality values. The SS3 assessment provides a substantially different perspective on stock trends compared to the previous AMISH assessment.

Reference points were estimated based on the ICES (2021) guidelines using the standard EQSIM software (ICES, 2014). The short-term forecast approach developed during WKBHMB was deterministic but with a clear understanding that during the upcoming WGHANSA meeting a stochastic forecast should be set up.

Northeast Atlantic boarfish (Section 6)

No changes to the boarfish stock definition were made during the benchmark process.

The benchmark data call resulted in additional catch and sampling information being available. Revisions to the existing catch time-series were relatively minor but the additional sampling information was particularly useful for the stock assessment. A time-series of the catch length composition is now also available.

A complete re-analysis of the first 6 years of the WESPAS acoustic survey (2011–2016) was undertaken for the benchmark. All acoustic survey data are now available from the ICES Acoustic data portal. The WESPAS survey estimates have been supplemented by those from the neighbouring PELGAS survey from the Bay of Biscay. Estimates of boarfish TSB and size structure are available and, given the southward expansion in stock distribution in recent years form an important component of the total stock estimate.

The 6 individual IBTS groundfish survey indices developed for the previous assessment have been replaced by a single biomass index from a VAST model fit to a combined groundfish survey dataset of catch rates. Accompanying the biomass index from the spatio-temporal VAST model fit is the survey length composition, compiled by aggregating all hauls.

The previous, surplus production (category 3) assessment model has been replaced by an implementation of SS3 which includes data on catch, an index from 2 acoustic surveys and a VAST index from a number of groundfish surveys. The updated assessment offers a revised view of the development of the boarfish stock. Although both the previous and new assessments indicate an increasing stock size in the most recent period, the new assessment estimates the stock to be at a historic high with relatively low fishing pressure.

Reference points were estimated using the ICES EqSim approach. A B_lim reference point was identified based on the mean stock size that gave rise to the 4 observed above-average recruitments.

The SS3 assessment has been configured to include a projection period during which it is assumed that the biology and selection is the mean of the final 3 years for the assessment data period. Recruitment is drawn from the assessment Beverton–Holt spawner recruitment relationship for the final years of the assessment data and projection periods.

General findings

The benchmark workshop on horse mackerel and boarfish (WKBHMB) was carried out in two meetings and several intersessional activities. The data workshop was held 22–25 January 2024 and the benchmark workshop was held 8–12 April 2024. Following the benchmark meeting in April 2024, not all assessments, reference points and short-term forecast methodologies had been finalized. This was mainly due to large challenges that presented themselves for the assessments of the horse mackerel stocks. This led to a series of six online meetings aimed at finalizing the results for each of the stocks. Unfortunately, completing the report was then further delayed by the ransomware attack on the ICES systems which led to a temporary inability to access the files needed to complete the work and several experts not being allowed to communicate directly with ICES systems or staff. Ultimately, this resulted in the WKBHMB report to be finalized only by late summer 2024.