General information
Greenland halibut in ICES Subareas 5, 6 12 and 14 (East-Greenland, Iceland, Faroe-islands) are assessed as one stock. In Icelandic waters, it is found on the continental shelf around Iceland with the highest abundance west, north and east off the coast in deeper and colder waters. It is mainly found on a muddy substrate at depths ranging from 200-1500 m. The main spawning grounds are located west off the coast at around 1000 m depth and eggs and larvae drift between Iceland and the east coast of Greenland until juveniles seek bottom post metamorphosis. After spawning, Greenland halibut migrates further north and east to their main feeding grounds. No juvenile grounds are known within the assessment area, and migration is known to occur from adjacent management units.
In the water East of Greenland it mainly found at depths greater than 600 m on the steep continental slope where as in the Faroe Islands it is mainly found North and East of the islands at 200 to 600 m.
Fishery
Spatial distribution of the 2023 fishery and historic catch and effort in the trawl fishery in Subareas 5, 6, 12 and 14 is provided in Figure 1 and Figure 2. Fishery in the entire area did in the past occur in a seemingly continuous belt on the continental slope from the slope of the Faroe plateau to southeast of Iceland extending north and west of Iceland and further south to southeast Greenland. Fishing depth ranges from 350-500 m southeast, east and north of Iceland to about 1500 m at East Greenland.
In 2001–2008 a directed and a by-catch fishery by Spain, France, Lithuania, UK and Norway developed in the Hatton Bank area of Division 6.b, however, most of these fisheries ceased after 2008. Presently UK and France have a small fishery in the area. All catches in Subareas 6 and 12 are assumed to derive from the fishing on Hatton Bank area.
Landing trends
In 1980–1990, about 75–90% of catches were caught by Iceland ( Figure 3). Since 1990, the Icelandic proportion has decreased, and has in recent years been 50–60%. Highest catches were recorded in 1986, about 60 thous. tonnes. Landings in Icelandic waters (usually allocated to Division 5a) have historically been predominated by the total landings in areas 5+14 (Icelandic waters), but since the mid-1990s fisheries in Subarea 14 and Division 5b have developed. Landings have since 1997 been between 20-31 thous. tonnes ( Figure 4).
Demersal trawl has been the main fishing gear for Greenland halibut in Icelandic waters, followed by gillnets, while a small proportion of the catch is taken on longlines and in shrimp trawls. Since 2015, landings by gillnets have, however, increased, reaching 62% of total catch in 2019 ( Figure 5). The Greenland halibut trawl fishery is considered clean with respect to by-catches. The mandatory use of sorting grids in the shrimp fishery in Icelandic and Greenland waters since 2002 is observed to have reduced by-catches of Greenland halibut considerably. Greenland halibut is caught in relatively deep waters, with most of the catch (70%) taken between 400-800 meters depth. In 2003, most of Greenland halibut was caught at 800 meters or deeper (73%), but since then, catch has increased steadily in more shallow waters ( Figure 6). Changes in depth range where Greenland halibut was caught seem to be reasonably synchronized with changes in fleet and therefore gear structure that target Greenland halibut in most recent years ( Figure 5 and Figure 6).
The number of vessels accounting for 95% of the catch of Greenland halibut in Icelandic waters changed from about 75 vessels in 1994-1998 to little less than 20 ( Figure 7). This change coincided with reduced catches. Since 1998,the number of vessels accounting for 95% of the catch has been relatively constant despite variable annual catches, with the lowest number of vessels observed in 2018
Catch per unit effort
Estimates of catch per unit effort (CPUE) for the Icelandic trawl fleet directed at Greenland halibut for the period 1985–onwards is provided in Figure 8. The overall CPUE index for the Icelandic fishery is compiled as the average of the standardized indices from the whole area. Catch rates of Icelandic bottom trawlers decreased for all fishing grounds during 1990–1996 but peaked again in 2001. Since 2003, CPUE has been relatively stable.
An analysis of the CPUE by area is shown in Figure 9. The CPUE west of Iceland showed a substantial drop in the period but after that the CPUE in the western area followed similar trends as other areas around Iceland.
Sampling from Greenland halibut landings
Area 5a
In general sampling is considered good from commercial catches in Icelandic waters from the main gears (gillnets, longlines and trawls). The sampling does seem to cover the spatial and seasonal distribution of catches (see Figure 11 and Figure 10). In 2020 sampling effort was reduced substantially, on-board sampling in particular, due to the COVID-19 pandemic. This reduction in sampling is, however, considered not to substantially affect the assessment of the stock in the short term. Sampling effort has now started to increase.
The bulk of the length measurements in Icelandic waters are from the three main fleet segments, i.e. trawls, longlines and gillnets. The number of available length measurements by gear has fluctuated in recent years in relation to the changes in the fleet composition.
Length distributions from the main fleet segments are shown in Figure 12. The sizes caught by the main gear types (bottom trawl and gillnets) appear to be fairly stable, primarily catching halibut in the size range between 40 and 80 cm while gillnets tend to catch slightly larger fish.
There has been a gradual shift towards larger fish in the length distribution of landed catch (Figure 12) following periods of poorer recruitment.
Areas 14 (East Greenland) and 5b (Faroese waters)
Samples collected from the fishing grounds in Faroe islands and east of Greenland are shown in Figure 13.
Other areas
No samples are available and reported catches have been negligible in recent years.
By-catch and discard
The Greenland halibut trawl fishery is mostly a clean fishery with little by-catches. Eventual bycatches are mainly beaked redfish and cod. Southeast of Iceland the cod fishery and a minor Greenland halibut fishery coincide spatially. In East Greenland where fishery is located on the steep slope, fishing grounds for cod and redfish are close to the Greenland halibut fishing grounds, but nevertheless the catches from single hauls are clean catches of Greenland halibut.
The mandatory use of sorting grids in the shrimp fishery in Iceland since the late 1980s and in Greenland since 2002 was observed to have reduced by-catches considerably. Based on few samples in 2006–2007, scientific staff observed by-catches of Greenland halibut to be less than 1% compared to about 50% by weight observed before the implementation of sorting grids (Sünksen, 2007). No information has since been available but the fishery in Division 14b generally report discard rates less than 1% by weight in logbooks.
Survey information
Three surveys are conducted in the distribution area of the Greenland halibut stock; in East Greenland (14.b), in Iceland waters (5.a) and in Faroese waters (5.b). The total surveyed area is provided in Figure 14. The two surveys in 5.a and 14.b are combined to one index and used as biomass index input for the assessment model. In the years between 2017 and 2021 no survey data were available from East Greenland (area 14.b), for those years the 2016 estimate was used to fill in the combined survey estimate. A relative comparison of the two surveys is provided in Figure 15.
The Icelandic autumn groundfish survey (hereafter autumn survey) was commenced in 1996. Spatial distribution and abundance in recent years are shown in Figure 14 and Figure 15 while Figure 16 shows trends in various biomass indices, and a recruitment index based on abundance of Greenland halibut \(\leq\) 40 cm. Survey length distributions are shown in Figure 17. In the recent years, Greenland halibut were mainly caught on the continental slope south east, north, and north-west of the country (Figure 15).
Since the survey was commenced in 1996, the distributional pattern has remained quite stable, with the greatest biomass index in the northeast and northwest. Since 1996, biomass index in the west has been steadily decreasing, while increasing in the southeast (Figure 15).
Biomass indices for the total stock of Greenland halibut and Greenland halibut larger than 40 cm (harvestable part of the stock), that are based on the combined Icelandic and Greenlandic autumn surveys, showed an increase from 1996-2001. After peaking in 2001, indices dropped but increased steadily from 2004 till 2017 when the stock started to decrease (Figure 16). The same holds for the index of Greenland halibut larger than 60 cm. The index of juvenile abundance (<40 cm) has fluctuated between years, peaking in 2002 but remained low in the past six years ( Figure 16). Since 2016 the East Greenland area has not been surveyed, and for the indices the values from 2016 are used for the years after that.
Length distributions from the Icelandic autumn survey shows more dynamics in observed sizes compared to catch samples (Figure 17), and average size has be decreasing in recent years.
Age distribution of the sexes of Greenland halibut from the autumn survey 2015-onwards show that the greatest proportion males are between 9 and 10 years old and range between 4-16 years. The greatest proportion of females are 11-13 years old and range from 3 to 22 years (Figure 18).
It is worth noting that aging recently resumed after a long period where otoliths were sampled but not age read. Recent advances in age reading techniques suggested that older age reading methods used previously were biased and thus older age-readings are not considered representative of the age structure in the population. Further, otoliths sampled prior to 2015 were not stored in a manner compatible with the newer age-reading method. It is therefore uncertain whether data on the historic age structure will ever be available.
According to the length distribution by age of Greenland halibut, it reaches 60 cm at the roughly the age of 12 on the average (Figure 19). The growth of Greenland halibut appears to be similar between the sexes, while female exhibit larger variability in size. It is noteworthy that males tend to be on average smaller in the catches than females, even though both sexes seem to have similar mean length at age. This may suggest differences in behavior of the sexes, such as catchability with respect to gear and/or natural mortality.
Faroese survey
The annual Greenland halibut survey in Faroese waters was started in 1995. The samples taken using a commercial trawl and the survey design varies between years. The average tow time has increased steadily from an average of 3 hours in 1995 to nearly 7.5 hours in 2020.
Aging resumed in 2015 and information is available from four years (2015 to 2017 and 2021). Preliminary results from an aging workshop on Greenland halibut ototliths suggest that further calibration between labs is needed to ensure that they are appropriate (Windsland pers. comm).
Maturity data
Information on maturity for Greenland halibut is sparse, and the maturity scale used in the surveys is considered to be imprecise. A gonadsomatic index (GSI) value above 1% is considered be a good indicator of maturity (Kennedy pers comm). Information on gonad size is available from the Icelandic autumn survey (Figure 21). Work has started to update the maturity scale used in the survey.
Updates to reference points
This year all reference point were estimated and revised. This re-estimate was needed after an error was discovered during the update assessment of Greenland halibut and beaked redfish. The routine that collated SSB from the model output incorrectly calculated mean weight at length resulting in a downward bias in total and spawning stock biomass estimates (Figure 23). As the estimate of B\(_{lim}\) for GHL was based on B\(_{loss}\) it was affected by this error and thus the biomass and fishing pressure reference points needed to be revised. Figure 22 illustrates the difference between last years estimates and the corrected last years assessment.
The revision of reference points followed the same procedure as used at WKBNORTH and resulted in higher estimated of B\(_{lim}\) and B\(_{pa}\) and lowered estimates of F\(_{lim}\) and F\(_{pa}\) (see Table 1. F\(_{msy}\) was also lowered from 0.24 to 0.22. This is not related to precautionarity, as the estimate F\(_{pa}\) is higher but only minor pertubations in yield as the range of fishing mortalities that have expected yields within 5% of the MSY is wide (see Figure 24).
Updated Value | 2023 benchmark | Basis | |
---|---|---|---|
MSY approach | |||
Fmsy | 0.22 | 0.24 | F leading to MSY |
Btrigger | 24895 | 21402 | Bpa |
Precautionary approach | |||
Blim | 18213 | 15657 | Lowest observed stock biomass |
Bpa | 24895 | 21402 | Blim x exp(1.645 sigma_SSB) |
Flim | 0.41 | 0.5 | F leading to P(SSB < Blim) = 0.5 |
Fpa | 0.29 | 0.38 | F, when ICES AR is applied, leading to P(SSB > Blim) = 0.05 |
MSY | 25567 | 26554 | Avg. MSY |
Due to this update of the reference point last years advice needed also to be corrected based on the updated F\(_{msy}\) and B\(_{trigger}\) valuees.
Stock assessment
The stock was benchmarked in 2023 (WKBNORTH) where the basis for advice was changed and reference points were updated. The approved assessment is an age–length based assessment model (GADGET) incorporating available data on the population dynamics. An overview of the settings are listed below:
- Start year 1985
- Two timesteps, equal in length, within the year
- Age range: 1 to 20\(^+\)
- Size range: 4 – 100 cm, 1 cm length groups
- Growth:
- Length based Von Bertalanffy size update (\(k\), \(L_{\infty}\))
- Beta-binomial size dispersal with a maximum length group growth set as 15 cm (\(\beta\))
- Length – weight relationship estimated externally
- Natural mortality set as 0.15
- Initial population and recruitment
- Annual recruitment occurs in the first timestep, one parameter per year \(R_y\).
- Mean length and standard deviation at recruiment is estimated
- Initial population at age is set as \(S \times \mathfrak{n}_a \times e^{-a (M_a + \hat{F})}\)
- Initial mean length at age is defined using the Von B growth curve, and initial numbers at length are dispersed assuming a normal distribution around the mean length with a fixed CV.
- Fishing split by fleet:
- 6 fleets, 1 survey, 3 bottom trawl (Greenland, Iceland and Faroese), gillnet and longlines in Iceland
- Logit selectivity for each fleet (\(\alpha_f\), \(l_{50,f}\))
- Maturity at length estimated externally based on autumn survey samples
- Likelihood functions:
- Survey indices are fit assuming that \(\log(I) = \alpha + \beta \log(\hat{I})\), where \(I\) and \(\hat{I}\) are observations and model predictions respectively. \(\alpha\) and \(\beta\) are estimated using linear regression.
- Composition data are assumed randomly sampled and fit using sums of squares of proportions
- Uncertainties are estimated using a spatial bootstrap for the composition data and simulated survey indices based on estimated survey CV.
Input data
The Gadget assessment of Greenland halibut relies on a number of disparate datasets, ranging from survey indices from the autumn survey, landings by gear and area, and catch composition data from the various fleets that target Greenland halibut. An overview is shown in Figure 25.
The model fit to the observed biomass index from the combined Iceland and Greenland autumn survey is shown in Figure 26. The model appears to capture the main trends in the index, although in order to do that a non-linear relationship between the model biomass and the observed survey biomass is assumed.
The model estimated catch composition is illustrated in Figure 27 to Figure 34, with residual plot shown in Figure 35). In general the fit is best to the autumn survey data. Other datasets that have had fairly consistent sampling through the years, such as the bottom trawl samples, show no discernible patterns in the residuals, with the Icelandic bottom trawl and gillnet samples exhibit the lowest deviation in residuals. Observed longline size distributions, however, are fairly inconsistent from year to year and the model seems therefore to have higher propensity to ignore that dataset.
Model results
The results from the model are shown in Figure 36. The total and spawning stock biomass are estimated to have decreased since its highest value at the start of the model period and reached its lowest point in SSB around 2005. The stock biomass increased to 2015 due to incoming recruitment but has since then deacreased and the assessment suggests that the spawning stock biomass at the start of 2024 is below B\(_{trigger}\). Fishing mortality appears to fluctuate without trend. Analytical retrospective analysis is shown in Figure 37. These results indicate a potential for bias, as indicated by the Mohn’s \(\rho\). This can be explained by the exclusion of age data when a year of data is removed, as the model only has age observations for the last nine years. The recruitment is estimated to fall outside the uncertainty bounds in the current assessment, suggesting that little observations are available on the recruitment at age 5.
Estimated selection by fleet is shown in Figure 38. The estimated selectivities range considerably, with the Faroese bottom trawl fleet catching the smallest fish while longline and gillnet boats in Iceland the largest. The Greenlandic and the autumn survey catch similar sizes.
Conclusions
Overall the gadget model presented here captures the overall trends in the data, and in spite of minor mis-fits the model is usable for assessing the stock and to base advice to managers.
In a complicated such as the gadget model that has many parameters and many data-sets of varying quality it is to be expected that there may be problems with some parameters and fit to some data-sets.
The main problem encountered when building the model during the benchmark were strong year factors in the autumn survey. Although fitting to a single survey seems improve the retrospective estimates it does cause some concern. However as more age data becomes available in the coming years it is expected that this issue will be easier to reconcile within the model.
Short term prognosis
Short-term forecasts for Greenland halibut are done in Gadget using the settings described below.
- F and M before spawning: NA
- Weight-at-age in the stock: GADGET uses a weight–length relationship and von Bertalanffy growth (no weights-at-age are supplied to GADGET)
- Weight-at-age in the catch: GADGET uses a weight–length relationship and von Bertalanffy growth (no weights-at-age are supplied to GADGET)
- Exploitation pattern:
- Landings: logistic selection-at-length by fleet, with parameters estimated within GADGET. Catch proportions by fleet are assumed fixed based on last three years.
- Intermediate year assumptions: Advice constraint (catch of 21 590 t)
- Stock–recruitment model used: Constant, based on last years age 1 recruitment estimate.
- Catch scenarios: F=Fmsy, F=0 and F=Fsq
The results of the prognosis are shown in Table 2.
Recruitment | Catch | SSB | Fbar | approach | |
---|---|---|---|---|---|
Status Quo | |||||
2021 | 29750 | 22635 | 26648 | 0.28 | SQ |
2022 | 34959 | 20899 | 25569 | 0.27 | SQ |
2023 | 46769 | 25425 | 25204 | 0.33 | SQ |
2024 | 50180 | 21590 | 23871 | 0.28 | SQ |
2025 | 49516 | 23164 | 23971 | 0.29 | SQ |
2026 | 47986 | 25246 | 25142 | 0.29 | SQ |
Zero catch | |||||
2021 | 29750 | 22635 | 26648 | 0.28 | Zero catch |
2022 | 34959 | 20899 | 25569 | 0.27 | Zero catch |
2023 | 46769 | 25425 | 25204 | 0.33 | Zero catch |
2024 | 50180 | 21590 | 23871 | 0.28 | Zero catch |
2025 | 49516 | 0 | 23971 | 0.00 | Zero catch |
2026 | 48004 | 0 | 29826 | 0.00 | Zero catch |
Fmsy | |||||
2021 | 29750 | 22635 | 26648 | 0.28 | Fmsy |
2022 | 34959 | 20899 | 25569 | 0.27 | Fmsy |
2023 | 46769 | 25425 | 25204 | 0.33 | Fmsy |
2024 | 50180 | 21590 | 23871 | 0.28 | Fmsy |
2025 | 49516 | 17890 | 23971 | 0.21 | Fmsy |
2026 | 47990 | 20966 | 26287 | 0.22 | Fmsy |
Management
Available biological information and information on distribution of the fisheries suggest that Greenland halibut in East Greenland, Iceland and Faroe Islands might be separated into subpopulations but that they do mix between these. Recent information of tagging experiments in the Barents Sea suggests high mixing between the Barents Sea and Iceland and also connectivity to West Greenland. This connectivity is not accommodated for in the present assessment.
Figure 39 shows the Icelandic national TAC, and catches since the 1991/1992 fishing year. In 2014, the Greenland and Iceland entered a five-year bilateral agreement to limit the fishing pressure of the Greenland halibut stock in East-Greenland, Iceland and Faroes to F\(_{msy}\). According to this agreement 56.4% of the TAC was allocated to Iceland and 37.6% to Greenland. This agreement was renewed in 2023. Other countries, notably the Faroe Islands were not party to this agreement.
In recent fishing years, landings have been similar to the advised TAC. Figure 40) shows the net transfers in the Icelandic ITQ-system since 1991 to Greenland halibut. In this period, transfers to Greenland halibut from other species (positive values) and transfers from Greenland halibut to other species (negative values) have fluctuated, but have mostly been from Greenland halibut to other species.
Data consideration and Assessment quality
The fishery for Greenland halibut in the vast stock area from East Greenland to west of the British Isles is conducted by an international fleet and catch recordings are therefore dependent on reporting from many nations. Even though it is believed that reporting is reliable the many data sources do not always agrees. In example logbook information, reporting’s to national authorities, data submissions to ICES, Eurostat and FAO often deviate and there are difficulties associated with choosing what is believed to be the correct number. Even data within ICES do not agree even though the source is the same, namely EuroStat. Thus ICES Catch data set 2006-2020 has huge deviations from its database 1950-2010 in its 5 overlapping years. An effort has been made to correct obvious deviations back in time, but this work is expected to continue and revisions of historic catch data are therefore foreseen. For the forthcoming years logbook data that agrees with reporting on catch from quota will superimpose other official reporting.
With the change to an age and length-based assessment more requirements will be put on biological sampling and sampling from the fisheries. This is especially the case for SA 14 (East Greenland) where sampling have been inadequate so far. Ageing of Greenland halibut ceased for many of the marine institutes in Greenland, Iceland, Faroe Island and Norway around 2000 due to reading difficulties and lack of inter-calibration. A new method has been agreed upon and cooperation between institutes has been initiated on age calibration. With respect to this stock Iceland has now progressed so far that an ALK is available for the 9 previous years. The Greenland institute of Natural Resources has also initiated age reading.
References
ICES. 2013. Report of the Benchmark Workshop on Greenland Halibut Stocks (WKBUT), 26–29 November 2013, Copenhagen, Denmark. ICES CM 2013/ ACOM:44. 367 pp.
ICES. 2017. Report of the Workshop on age reading of Greenland halibut 2 (WKARGH2), 22-26 August 2016, Reykjavik, Iceland. ICES CM 2016/SSGIEOM:16. 40 pp.
Sünksen, K. 2007. Bycatch in the fishery for Greenland halibut. WD 17, NWWG 2007. ICES 2023. WKBNORTH Report