COD

Gadus morhua


Assessment report
Published by

Marine and Freshwater Research Institute, Iceland

Published

6 June 2025

Key signals

  • Survey biomass increased steadily from 2008 to 2012, after which high but variable levels were observed in both the spring and autumn surveys. Since 2012, discrepancies have emerged between the two surveys: the spring survey shows a stronger upward trend, while the autumn survey indicates more stable or modest increases. The most recent spring survey, however, shows a marked decline in biomass.

  • Recruitment (age 3) was substantially higher prior to 1985 but has remained relatively stable over the past decade. The year classes 2020–2022, recruiting in 2023–2025, are at or below the long-term average, suggesting a likely decline in reference biomass as these cohorts reach age 4. Several of these cohorts also exhibit low stock and catch weights at ages 4–7, which make up the bulk of the reference biomass.

  • Length distributions have been stable over the past decade, reflecting periodic recruitment and a broad range of sizes in the population. However, the abundance of fish in the 30–50 cm range is lower than in most recent years, supporting expectations of reduced biomass in the near term.

  • Spawning stock biomass (SSB) peaked in 2018–2019, reaching its highest level since before 1970. Although SSB is now declining, it remains relatively high compared to most levels observed since 1980.

  • Harvest rates exceeded the management target of 0.2 prior to 2022 but have since fluctuated around the target level.

General information

Cod (Gadus morhua) is widely distributed in Icelandic waters, with the highest abundance found on the northwestern, northern, and northeastern parts of the continental shelf. As a demersal species, cod occupies a wide depth range, typically from a few meters down to 600 m, and occasionally deeper. Adult cod are not strongly selective regarding bottom substrate and can be found on a variety of seabed types. In contrast, juveniles prefer moderately sheltered, shallow habitats such as kelp forests and seagrass beds. Cod thrive in sea temperatures around 4–7°C, although they are frequently caught in waters below 2°C.

Cod spawn around the Icelandic coastline in distinct regional spawning components. The primary spawning areas are located in the south, southwest, and west, where spawning begins early in spring (March–April) in warmer waters. In recent years, spawning activity has shifted westward. Historically, spawning occurred later in the colder northern waters, but the spawning period there has advanced considerably. Pelagic eggs and larvae drift predominantly northward and eastward, following a clockwise pattern to nursery areas in the north and northeast. Adult cod undertake feeding migrations to deeper waters in the northwest and southeast, although some remain in shallow areas to feed. Cod is Iceland’s most significant groundfish species in terms of commercial importance.

Fishery

Landing trends

Cod landings from Iceland waters have been consistently high since records began in 1903, typically ranging from 300 to 450 thousand tonnes prior to World War II and again between 1950 and 1990 (Figure 1). During the World Wars, foreign fleets ceased operations in Icelandic waters, while Icelandic catches remained stable. Throughout much of the 20th century, foreign vessels accounted for a significant share of the catch until Iceland extended its Exclusive Economic Zone (EEZ) to 200 miles in 1976.

Catch levels have largely been shaped by stock productivity, influenced by both recruitment and immigration from Greenland. Significant migration events from Greenland to Iceland occurred in the 1930s and in 1953. A reduction in fishing effort during World War II also contributed to stock rebuilding.

Between 1955 and 1990, landings gradually declined. From 1992 onward, average catches have been around 218 thousand tonnes (range: 148–270 thousand tonnes). In 2024, landings totaled 220402 tonnes. A small share of this (2314 tonnes) was landed by Norwegian and Faroese vessels under bilateral agreements (Figure 1).

Historically, bottom trawl and gillnets were the primary gear types (Figure 2). Bottom trawl has consistently made up at least 40% of catches, rising to about 55% in recent years, while gillnet use declined from 20% before 2001 to around 7% more recently. Most bottom trawl catches occur in the northwest, while gillnet catches have been concentrated in the south and west during spawning. The share of longline catches has increased over time, with large longline vessels with automatic baiting extending the fishery into deeper waters. Before 2003, longlines accounted for about 20% of catches; this rose to around 35% during 2005–2016. Demersal seines have contributed a steady 5–8% of total catches. In some regions, especially the northwest and southeast, cod aggregate in dense schools, allowing captains to quickly fill quotas, particularly prior to landing. Cod condition and size, along with proximity to landing ports, also influence fishing locations.

In 2024, bottom trawls accounted for more than half of total catches (57%), followed by longlines (23%), gillnets (7%), jiggers (7%), and demersal seines (7%). The largest proportion of the catches were taken from the western and northwestern regions, followed by the northeast and southwest. Catches occurred at similar depths as in previous years, although slightly shallower overall (Figure 2).

The depth distribution of the fishery has shown little change (Figure 3).

Due to its wide distribution, cod is fished across the Icelandic shelf using various gear types (Figure 4, Figure 5, Figure 6). Demersal trawl is the main fishing gear (Figure 4), with primary fishing areas located in deeper, colder waters offshore in the northwest, northeast, and east. Over recent years, trawl effort has increasingly concentrated in these hotspots.

Longlines account for the second-largest share of cod catches and are widely used around the shelf, with the lowest effort on the southern and southeastern coasts (Figure 5). The spatial pattern of longline catches has remained generally consistent, with periodic hotspots. Gillnets, demersal seines, and jiggers contribute smaller shares and primarily operate in shallower waters (Figure 6).

Since 2000, the share of cod catches from the northwestern area increased from about 35% to 40–50% in 2011–2024 (Figure 7 and Figure 8 ). In contrast, catches west of Iceland declined from 25–30% to just under 20%, and from the southeast, from 5–10% to below 5%. Other areas have remained stable, with the northeast contributing around 20% and the southwest about 10%.

The number of vessels accounting for 95% of annual cod landings declined from around 1000 to 750 during 1994–1999, despite a notable increase in total catch (Figure 9). Between 1999 and 2008, vessel numbers declined further to below 400, in line with reduced catches. After a temporary increase during 2009–2014, the number of vessels has since decreased to around 300. This trend is evident across all fleet segments (Table 1).

Figure 1: Cod in Division 5.a. Recorded landings since 1903.
Figure 2: Cod in Division 5.a. Landings in tons and percent of total by gear and year
Figure 3: Cod in Division 5.a. Depth distribution of cod catches from bottom trawls, longlines, trawls and demersal seine from Icelandic logbooks
Figure 4: Cod in Division 5.a. Spatial distribution of catches by demersal trawls.
Figure 5: Cod in Division 5.a. Spatial distribution of catches by longlines.
Figure 6: Cod in Division 5.a. Spatial distribution of catches by gillnets, demersal seine, and jiggers.
Figure 7: Cod in Division 5.a. Changes in spatial distribution of cod catches as recorded in Icelandic logbooks.
Figure 8: Cod in Division 5.a. Spatial distribution of catches by all gears for selected years.
Figure 9: Cod in Division 5.a. Number of vessels (all gear types) accounting for 95% of the total catch annually since 1994. Left: Plotted against year. Right: Plotted against total catch. Data from the Directorate of Fisheries.
Table 1: Cod in Division 5.a. Number of Icelandic vessels landing cod, and all landed catch divided by gear type.
Year
Bottom trawl
Long lines
Other gears
Total
Vessel (number) Catch (tonnes) Vessel (number) Catch (tonnes) Vessel (number) Catch (tonnes) Catch (tonnes)
2000 183 103 558 530 49 946 827 17 623 186 057
2001 160 99 071 515 47 172 766 17 002 180 260
2002 155 87 885 450 42 405 724 19 305 163 179
2003 147 88 422 461 44 654 721 16 026 162 477
2004 135 95 769 470 57 397 722 14 840 182 234
2005 134 84 018 463 69 444 604 8 106 174 338
2006 126 82 417 447 71 037 509 5 859 169 671
2007 123 71 499 425 58 943 473 4 397 143 550
2008 113 58 172 370 53 843 427 4 151 124 607
2009 113 79 667 336 61 005 798 8 190 159 232
2010 111 75 609 286 57 491 1 008 9 372 150 768
2011 110 73 538 290 57 711 1 061 12 665 153 020
2012 118 85 265 305 67 777 1 099 13 417 176 448
2013 110 101 453 297 74 835 1 054 15 237 201 617
2014 110 95 830 292 77 807 1 012 16 355 200 399
2015 103 103 530 266 79 244 943 13 957 208 669
2016 99 111 016 246 84 509 956 15 299 226 754
2017 94 117 891 221 75 244 832 14 945 223 478
2018 83 135 030 201 78 316 813 16 221 245 385
2019 81 135 661 190 78 326 794 13 592 241 760
2020 81 146 788 158 68 103 839 15 884 246 973
2021 87 140 773 146 69 632 841 16 014 244 114
2022 83 124 950 121 62 795 850 15 172 219 413
2023 81 117 580 99 54 527 866 13 389 201 511
2024 79 123 832 90 49 139 849 15 238 203 602

Data available

Sampling coverage from commercial catches is generally good for the main gears (demersal seines, longlines, gillnets, bottom trawls), with broad spatial and seasonal representation (Figure 10, Figure 11).

Figure 10: Cod in Division 5.a. Ratio of samples by month (bars) compared with proportion landings by month (solid black line) split by year and main gear types. Numbers of above the bars indicate number of samples by year, month and gear.
Figure 11: Cod in Division 5.a. Fishing grounds last year as reported in logbooks (contours) and positions of samples taken from landings (crosses) by main gear types.

Landings and discards

Landings data before 1982 (and foreign landings until 2005) are sourced from the ICES STATLANT database. Icelandic landings between 1982–1993 were collected by the Fisheries Association of Iceland, and from 1994 onward by the Directorate of Fisheries. Foreign landings were recorded by the Coast Guard before 2014 and by the Directorate thereafter.

Discarding is prohibited by law. Management measures, such as inter-species quota transfers and landing flexibility, discourage discarding. To help prevent high grading and quota mismatches, fisheries are permitted to land catches that are exempt from quota deductions, provided that the proceeds from their sale are transferred to a government-managed fund. More information is available at https://www.responsiblefisheries.is/seafood-industry/fisheries-management.

Figure 12: Cod in Division 5.a. Estimates of annual discards by gear (point estimates and 95% confidence interval). No estimates are available after 2017.

Length compositions

The length distribution of landed cod has shifted toward larger fish over the past decade (Figure 13). Most length data come from bottom trawl, longline, gillnet, and demersal seine fleets (Table 2). Gear-specific sample sizes have varied with fleet composition.

Figure 13: Cod in Division 5.a. Commercial length distributions by gear and year
Table 2: Cod in Division 5.a. Number of samples and length measurements from landed catch.
Year
Bottom Trawl
Danish Seine
Gillnets
Long Line
Num. samples Num. lengths Num. samples Num. lengths Num. samples Num. lengths Num. samples Num. lengths
2000 766 172 132 23 3 265 27 4 517 124 29 780
2001 1 131 170 398 79 13 660 541 39 836 281 39 915
2002 1 233 162 365 122 19 057 443 45 520 338 46 560
2003 1 131 114 366 73 13 283 512 32 729 444 67 124
2004 1 239 107 977 97 17 943 310 28 422 467 85 952
2005 1 092 101 166 298 18 050 267 37 554 549 108 597
2006 859 79 264 159 15 812 349 34 329 776 115 401
2007 946 75 259 416 16 252 304 24 024 498 96 662
2008 849 67 630 297 14 065 370 23 676 530 105 864
2009 884 76 100 216 15 306 205 25 436 420 86 909
2010 806 77 979 139 10 621 152 26 303 401 81 373
2011 596 64 643 86 6 604 162 29 203 280 55 614
2012 604 54 037 104 8 282 131 22 678 438 97 729
2013 659 73 853 50 2 196 25 3 532 393 83 001
2014 531 46 615 40 5 118 141 27 103 453 96 534
2015 554 65 641 47 5 880 56 5 860 376 83 342
2016 493 57 116 51 6 201 150 25 919 421 96 525
2017 518 67 512 61 7 073 47 7 149 368 77 691
2018 264 48 111 46 5 222 114 15 746 395 74 874
2019 451 81 165 49 4 679 43 5 754 292 56 710
2020 191 35 494 27 3 038 89 12 469 84 13 242
2021 325 53 645 48 5 014 13 1 453 38 4 333
2022 228 38 180 49 5 336 9 961 52 11 348
2023 150 23 230 44 3 854 81 9 659 34 4 187
2024 246 43 591 36 4 161 21 1 987 45 6 816

Age compositions

Table 3 presents the number of otolith samples and age readings by gear type, while Figure 11 shows the sampling locations.

Over the past few decades, the age composition of the catch has shifted toward older fish, likely due to reduced fishing pressure (Figure 14). In recent years, the number of year classes represented in the catch has increased, reflecting sustained low fishing mortality (Figure 15).

Table 3: Cod in Division 5.a. Number of samples and otoliths collected from landed catch.
Year
Bottom Trawl
Danish Seine
Gillnets
Long Line
Num. samples Num. otoliths Num. samples Num. otoliths Num. samples Num. otoliths Num. samples Num. otoliths
2000 205 10 034 18 885 21 1 051 44 2 223
2001 515 9 193 10 500 288 2 239 136 2 830
2002 587 8 612 30 693 226 2 055 162 2 815
2003 684 8 144 9 450 308 1 217 170 3 409
2004 824 9 052 8 400 90 802 100 2 297
2005 654 6 485 208 788 12 598 103 3 548
2006 529 5 720 71 931 102 1 471 271 3 507
2007 625 6 094 328 1 244 125 1 203 82 3 650
2008 543 5 024 217 816 196 1 029 65 2 771
2009 569 5 417 136 561 23 1 150 33 1 650
2010 493 5 880 83 501 16 799 59 2 931
2011 340 5 339 49 578 14 700 46 2 294
2012 403 5 757 63 688 21 1 031 73 3 209
2013 375 6 192 39 397 21 1 050 62 3 065
2014 378 5 102 20 525 29 850 65 1 856
2015 312 4 937 29 708 45 917 54 1 467
2016 320 5 015 41 1 025 41 1 025 60 1 544
2017 236 3 817 39 975 26 644 46 1 119
2018 92 2 369 30 750 16 400 39 945
2019 113 2 828 27 675 12 300 50 1 237
2020 73 1 847 21 520 12 300 31 775
2021 85 2 171 34 850 8 200 30 750
2022 57 1 264 16 320 2 40 18 362
2023 79 1 511 28 560 6 118 22 440
2024 91 1 871 31 620 5 100 22 440
Figure 14: Cod in Division 5.a. Catch at age from the commercial fishery in Iceland waters. Bar size is indicative of the catch in numbers and bars are colored by cohort.
Figure 15: Cod in Division 5.a. Catch at age from the commercial fishery in Iceland waters. Proportion biomass caught by year and age, bars are colored by cohort.

Weight at age in the catch

Mean weight at age in the catch (Figure 16) declined from 2001 to 2007, reaching historical lows in many age groups. Although mean weights increased in subsequent years, nearly all age classes were at least slightly below the 1985–present average in both this year and the previous year. For 2025, mean weights for ages 3–9 are estimated using the relationship between spring survey weights and commercial weights from the preceding year. Weights for older fish are assumed to remain unchanged from the previous year (see section on short-term projections).

Figure 16: Cod in Division 5.a. Mean weight at age in the catch from the commercial fishery in Icelandic waters. Bars are coloured by cohort.

Catch per unit of effort from commercial fisheries

Catch per unit of effort (CPUE) data (Figure 17) indicate a steady increase since 1990 for the main gear types in hauls where cod accounts for more than 50% of the catch. Overall CPUE across all gears is among the highest on record. CPUE could not be estimated after 2021 due to missing effort data for several years.

Figure 17: Cod in Division 5.a. Catch per unit of effort in the most important gear types. The dashed lines are based on locations where more than 50% of the catch is cod and solid lines on all records where cod is caught. A change occurred in the longline fleet starting September 1999. Earlier only vessels larger than 10 BRT were required to return logbooks but later all vessels were required to return logbooks. Data are excluded for years after 2021 for gillnets and longlines due to effort data inconsistencies.

Survey data

The Icelandic spring groundfish survey (hereafter spring survey) has been conducted annually in March since 1985. An additional survey, the Icelandic autumn groundfish survey (autumn survey), was initiated in 1996. However, a full autumn survey was not carried out in 2011.

Figure 18 presents a recruitment index based on cod smaller than 55 cm, alongside trends in various biomass indices. Survey abundance by tow location is shown in Figure 19, while changes in spatial distribution are illustrated in Figure 20.

The total biomass index from the spring survey has remained relatively high since 2012 but has shown an overall declining trend over the past decade. Both spring and autumn survey indices declined markedly from their 2017 peak to 2020, followed by a slight increase from 2023 to the present (Figure 18). While spring survey measurements in 2021 and 2022 were similar to those in 2018 and 2019, the autumn survey index in 2021 declined further, reaching its lowest level since 2004. The 2020 survey indices were considerably below expectations for the size classes making up the majority of the fishable biomass, a trend that continued in the 2021 autumn survey but not in the 2021 spring survey.

Overall, both surveys have shown similar long-term trends (Figure 18), although fluctuations since the late 2000s have been more pronounced in the autumn survey. The discrepancy between spring and autumn biomass measurements reached a record high in 2021 and remains the largest observed since 2000. However, the recent decline in the spring survey index, combined with a stabilization in the autumn index, has narrowed this gap in the most recent survey year.

In the spring survey of 2025, cod were caught throughout Icelandic waters, with hotspots observed offshore in the north and southwest, and in shallow southern waters (Figure 19). Biomass along the western continental slope was similar to the previous year. The spatial distribution of the total biomass index consistently shows the northwest (NW) and northeast (NE) areas dominating across years (Figure 20). However, recent years have seen some shifts, with declining biomass in the NE and increases in the west and southeast (SE). In 2025, biomass increased in all areas except the northern regions.

The spatial distribution of cod in the autumn survey of 2024 was similar to previous years, with the majority of biomass measured on traditional fishing grounds in the northwest and northeast regions (Figure 20).

Length distributions from both surveys clearly show distinct modes for the youngest age groups (Figure 21). In older age groups, separation becomes less distinct due to individual variability in growth and maturity, though modes corresponding to 2- and 3-year-old fish can sometimes be identified.

Survey age-based indices for older fish have remained relatively high over the past decade, despite the fact that many of these cohorts showed only low to moderate indices at younger ages (Figure 22). The 2020 spring survey anomaly is particularly evident — e.g., the 2014 and 2015 year classes were around the long-term average in 2019 (as 4- and 5-year-olds), but dropped to roughly half that in 2020 (as 5- and 6-year-olds). In 2021, the 2015 cohort appeared well above average in both survey and catch-at-age data, a trend that continued in subsequent years and suggests likely immigration from Greenland waters.

Figure 18: Cod in Division 5.a. Indices (total biomass, biomass > 55 cm, biomass > 80 cm and abundance ≤55 cm) in the Spring Survey (March) 1985 and onwards (line shaded area) and the autumn survey (point ranges).
Figure 19: Cod in Division 5.a. Location of cod in the 2024 March (SMB) and the 2023 Autumn (SMH) surveys, bubble sizes are relative to catch sizes, and crosses indicate stations where no cod was observed.
Figure 20: Cod in Division 5.a. Changes in geographical distribution of the survey biomass.
Figure 21: Cod in Division 5.a. Length disaggregated abundance indices from the March and Autumn surveys.
Figure 22: Cod in Division 5.a. Age disaggregated indices in the Spring Survey (left) and the autumn survey (rights). Note different scales on y-axes.

Stock weight at age

Mean weights in the spring survey for all cod age groups were below average during approximately 2000–2010 (Figure 23). Following this period, mean weights for ages 5–10 were generally above average for several years. However, in the past two years, mean weights have declined across both younger and older age groups.

Figure 23: Cod in Division 5.a. Stock weights from the March survey in Icelandic waters. Bars are coloured by cohort.

Stock maturity at age

Maturity-at-age data from the spring survey are presented in Figure 24. In recent decades, maturity at younger ages has generally been below average, indicating a trend toward later maturation.

Figure 24: Cod in Division 5.a. Maturity at age in the survey. Bars are coloured by cohort. The values are used to calculate the spawning stock.

Data analyses

Analytical assessment

The assessment is based on a separable statistical catch-at-age model (commonly referred to as MUPPET, see H. Björnsson, Hjörleifsson, and Elvarsson (2019)), which assumes constant selectivity within four time periods. The most recent period spans from 2007 to the present. Survey residuals are modeled using a multivariate normal distribution to account for potential year effects — an approach implemented since 2002. The model is a statistical cohort framework in which fishing mortality can change gradually over time, constrained by a random walk. This same framework is used to project stock dynamics for estimating reference points and evaluating the harvest control rule (HCR). It was benchmarked in 2021 as part of the HCR evaluation (ICES (2021b)).

In the current assessment, the model configuration was updated to account for a likely migration event. The 2015 year class appeared in 2021 at age 6 in higher numbers than would be expected based on earlier observations. To address this, an additional parameter was included to estimate recruitment at age 6 in 2021 — representing the 13th such migration event accounted for since the model’s starting year of 1955.

Data used by the assessment

The assessment relies on four primary data sources: two groundfish surveys (spring and autumn), commercial samples, and landings. Commercial data are used to construct catch-at-age estimates, which enter the likelihood function alongside age-specific survey indices from both surveys. Stock weights and catch weights at age are derived from the spring survey and commercial catches, respectively, while maturity-at-age data are obtained from the spring survey. Prior to the start of the spring survey in 1985, stock weights and maturity were assumed constant at their 1985 values.

A detailed description of data preparation for both tuning and input is provided in the stock annex (ICES (2021a)), and all input data are available on the MFRI website (www.hafogvatn.is).

The input to the analytical age-based assessment consists of catch-at-age data from 1955 to 2024 (ages 3–14), spring survey data (SMB) from 1985 to 2025 (ages 1–14), and autumn survey data (SMH) from 1996 to 2024 (ages 3–13). Catch-at-age data are derived based on 20 metiers, combining two areas (north and south), two seasons (January–May and June–December), and five fleets (bottom trawl, longline, handline/jiggers, gillnet, and Danish seine).

The reference biomass (age 4+) used to set the TAC for the upcoming fishing year is calculated as the product of population numbers at the beginning of the assessment year and catch weights for that year. Since catch weights for 2025 are not yet known, they must be predicted from stock weights measured in the spring survey, using a relationship established from the previous year’s data.

Mean weight at age in the catch (Figure 16) declined between 2001 and 2007, reaching historical lows in many age groups. The observed variation in catch weights at age partly reflects underlying variability in stock weights, as indicated by survey measurements (Figure 23).

No direct information is available on natural mortality. For assessment and advisory purposes, a fixed natural mortality rate of 0.2 is assumed for all age groups.

Diagnostics

The diagnostics (Figure 25) reveal large negative residuals in the 2020 spring survey for the key age groups (ages 4 to 8), along with smaller negative residuals in the surrounding the years, 2019 and 2021, particularly in the autumn survey. A summary diagnostic comparing observed and predicted survey biomass (Figure 26) highlights discrepancies between model estimates and survey point estimates. There are indications that interannual variability in survey measurements has increased in recent years in both surveys, relative to earlier periods.

Figure 25: Cod in Division 5.a. Residuals from the model fit to survey and catch data based on the both the surveys. Red circles indicate negative residuals (observed < modelled), while blue postive. Residuals are proportional to the area of the circles.
Figure 26: Cod in Division 5.a. Aggregated model fit to the total biomass indices. Note that residual correlation is estimated (see text for further details).

Model results

The results of this year’s assessment estimate the spawning stock biomass (SSB) for the assessment year at 379 thousand tonnes. Weight and maturity-at-age data used in the SSB calculation are presented in Table 4. The estimated SSB in recent years is higher than any values observed over the past five decades. The reference biomass for 2025 is estimated at 972146 t, and the harvest rate for 2024 is estimated at 0.197.

Year classes since the mid-1980s are estimated to be relatively stable, but with a mean approximately 35% lower than that observed during 1955–1985. Assessment results are summarized in Table 4 and Figure 27, and are also available on the MFRI website (www.hafogvatn.is).

Reference biomass has remained stable in recent years. However, the estimates for the 2021 and 2022 year classes have not been revised upward from the relatively low values reported last year (close to 100 million). These cohorts are expected to form a substantial part of the fishable stock in 2026–2027, suggesting a potential decline in fishable biomass during that period. Moreover, the most recent recruitment estimate (2023 year class) is only slightly higher and remains below the past decade’s average, indicating that the decline in reference biomass could continue into 2028, unless future recruitment estimates are revised upward.

Figure 27: Cod in Division 5.a. Summary from assessment. Dashed vertical line indicates the assessment year and yellow shaded region the uncertainty as estimated by the model.

Estimated spawning stock biomass (SSB), although variable, increased to a peak in 2017, the highest level observed in nearly 60 years. Since the most recent revision of the harvest control rule, the harvest rate has declined. Given that recruitment since 1988 has been substantially lower than the average observed during 1955–1985, the increase in SSB was primarily driven by lower harvest rate.

In 2023, the harvest rate reached its lowest level in the assessment period. However, due to the decline in stock and catch weights during the past year, a greater number of cod will be required to fill the current TAC compared to previous years. As a result, the harvest rate has increased over the last year.

This rise in fishing mortality, combined with the potential for elevated natural mortality due to lower energy reserves and increased predation on recruiting age groups (B. Björnsson, Sólmundsson, and Woods 2022), could increase the risk of overestimating stock size in projections. This potential overestimation, along with recent recruitment levels falling below the past decade’s average, suggests a continued decline in biomass is likely in the short term.

Biomass estimates have been revised upwards in each of the past three years, suggesting a slight tendency for the model to underestimate spawning stock biomass (SSB). In the current assessment, the notable upward revisions in SSB for 2021 and subsequent years result from a change in the model configuration to account for a likely migration event. Specifically, an additional parameter was introduced to estimate recruitment to the population at age 6 in 2021.

This migration event was first identified through unexpectedly high numbers of the 2015 year class appearing at age 6 in 2021, a signal that persisted through consecutive years and was consistently detected in all three data sources used in the model. The same 2015 cohort also appears as a large year class in East Greenland catch-at-age and survey data for the East Greenland–Iceland offshore spawning cod stock (ICES stock cod.21.27.1.14), indicating a likely shared origin.

This represents the 13th migration event accounted for in the model since 1958, all of which involved recruitment at ages 6–9 and were most likely associated with immigration from Greenland waters.

In addition to this 13th migration event, fishing activity and catches have increased over the past decade on both the Icelandic and Greenland sides of the EEZ near Dohrn Bank. Currently, the cod stock located on the Greenland side of the EEZ boundary is assessed using landings data only, as attempts to implement a category 1 assessment have not successfully captured trends across most fishing activities in the area. As a result, catch-at-age data are generated as part of the Greenland stock assessment. These data were included in the Icelandic stock assessment as a sensitivity run, both with and without estimation of the 13th migration event.

Omitting the 13th migration from the assessment resulted in a slight upward revision of the 2020 biomass estimate and a marginally poorer model fit. Including catch data from Dohrn Bank generally led to upward revisions of SSB estimates across the period following 2015. When both the migration event and Dohrn Bank data were included, these effects were combined; however, the inclusion of Dohrn Bank data also caused a decline in model fit. As a result, only the estimation of the 13th migration was retained in the final model configuration used for providing advice. The sharp decline in SSB projected for 2025 was consistent across all model runs.

Figure 28: Cod in Division 5.a. Comparison of assessment results (Advice, both surveys) with sensitivity runs, where the Alternative prognosis was used to produce catch advice (only affects terminal year in B4+), additional catches at age were included from Dohrn Bank in Greenland waters, the 13th migration was left out, or both of the latter two conditions were used.

The retrospective pattern of the assessment is shown in Figure 29, along with the Mohn’s rho values. The default 5-year peels resulted in the following values: 0.025 for recruitment, 0.026 for harvest rate, and for spawning stock biomass.

Leave-one-out analyses indicate that the assessment aligns most closely with trends in the spring survey. Removing the autumn survey series results in a slight increase in biomass estimates, while removing the spring survey leads to a more substantial decrease — similar to the effect of relying on catch data alone (Figure 30).

Figure 29: Cod in Division 5.a. Analytical retrospective analysis of the assessment of cod with a 5 year peel.
Figure 30: Cod in Division 5.a. Comparison of assessment results where either the spring survey, the autumn survey, or both are omitted from the estimation.

Short term projections

Landings of Icelandic cod in 2024 are estimated at 220402 tonnes, the majority of which were taken by the Icelandic fleet.

To perform short-term projections, an estimate of the catch for the current calendar year is required. This year’s projection assumes that the remainder of the TAC for the current fishing year (ending August 31) will be fully taken, along with an expected 3 thousand tonnes catch by the foreign fleet. The projected stock status for the interim year, along with the projection output, is provided in Table 4.

Mean annual discards of cod during 2001–2012 were approximately 1% of landings by weight. More recent data suggest that discards may have increased slightly (MRI 2016). These discard estimates are based on the assumption that discarding primarily occurs as high-grading.

In line with the management plan, advice for the upcoming fishing year (2025/2026) is based on the biomass of age 4+ cod at the beginning of the calendar year (2025). Because reference biomass is calculated using catch weights, a deterministic projection of catch weight growth is required. In recent years, catch weights at age for age groups 3–9 have been predicted using spring survey weights from the same year, applying a linear relationship (slope and intercept) derived from the previous year’s data. For age groups 10 and older, catch weights from the previous year are carried forward.

The same method was used to estimate weights-at-age for the following year, i.e., \(\alpha\) and \(\beta\) were estimated according to the following equation:

\[ \text{cW}_{a,y-1}=\alpha+\beta\text{sW}_{a,y-1} \]

and catch weights for the assessment year were then estimated according to:

\[\text{cW}_{a,y}=\alpha+\beta\text{sW}_{a,y}\]

where cWa,y denotes the catch weight and sWa,y the stock weight for age a and year y.

Based on this approach, the predicted mean catch weights-at-age in 2025 are below average for most age groups, as both the mean weights in the spring survey and in the catch were below average in 2024, and stock weights were below average this year (Figure 23).

An alternative method, labeled ‘alt’, was also explored. This model predicts catch weight at age using age-specific spring survey weights within each year, based on data from 1990 onwards. It has previously yielded more plausible estimates of catch weights, including in the current assessment year (2025), where the deviation from the standard model was -5%.

Retrospective analyses indicate that this alternative model has improved predictive performance, with a lower coefficient of variation (0.035 vs. 0.050) and reduced bias (-0.0020 vs. -0.0049), as shown in Figure 31 and Figure 32.

The alternative model has been discussed by the NWWG in 2022 and subsequent years as a potential improvement over the current SPALY weight prediction model. However, it was agreed that a more extensive evaluation should be conducted before implementation, especially in light of recent changes in cod mean weights. External review, either through a working document or during the next benchmark (anticipated for 2026 or 2027), would also be beneficial.

Figure 31: Cod in Division 5.a. Comparison of the short term prediction of catch weights from stock weights (blue), used in the calculation of reference biomass, using a regression across ages of catch weight on stock weight data observed in 2024 (red). Historical values and a regression based on all years of data represent the long-term average relationship in grey.
Figure 32: Cod in Division 5.a. Comparison of the short term prediction of reference biomass using the standard procedure (spaly) versus an alternate method (alt) to the realised value a year later (black line).
Table 4: Cod in division 5.a. Summary of the assessment and results of the short-term projections when the harvest control rule is applied. In 2025, the assessment includes estimates for SSB, B4+, calendar year catch, HR, and recruitment. Projections begin with 2026. Fishing year catches are only given for intermediate years and projections.
Year Calendar Year Catch Fishing Year Catch SSB B4+ HR Recruit. (age 3)
1955 539 551 — 727 229 2 091 810 0.242 151 046
1956 461 938 — 588 478 1 819 490 0.256 143 788
1957 454 387 — 575 328 1 640 750 0.303 161 493
1958 508 487 — 690 670 1 651 320 0.290 215 095
1959 436 910 — 639 893 1 581 200 0.295 303 665
1960 474 708 — 583 890 1 657 470 0.246 153 856
1961 387 789 — 465 596 1 430 420 0.269 195 930
1962 393 640 — 505 902 1 464 320 0.275 125 429
1963 408 347 — 460 574 1 299 160 0.329 173 331
1964 417 437 — 420 245 1 211 390 0.333 197 643
1965 376 228 — 323 089 1 053 240 0.346 219 546
1966 346 195 — 295 925 1 063 580 0.321 232 974
1967 343 449 — 280 828 1 139 710 0.322 319 777
1968 398 926 — 248 645 1 242 220 0.319 171 377
1969 388 868 — 354 462 1 335 840 0.338 239 738
1970 459 022 — 355 015 1 332 950 0.341 179 800
1971 436 750 — 253 195 1 084 060 0.380 193 147
1972 393 201 — 225 844 979 285 0.386 142 197
1973 361 931 — 245 558 831 519 0.443 277 908
1974 362 100 — 189 180 910 210 0.402 187 247
1975 361 637 — 175 453 892 060 0.395 259 266
1976 331 327 — 146 418 949 482 0.360 368 544
1977 337 654 — 199 740 1 298 080 0.257 144 349
1978 329 654 — 213 162 1 308 870 0.271 224 367
1979 357 425 — 308 382 1 410 570 0.291 237 572
1980 423 455 — 371 134 1 512 700 0.300 141 665
1981 451 111 — 276 608 1 255 600 0.325 145 417
1982 386 460 — 183 990 988 908 0.329 141 178
1983 293 425 — 145 132 804 440 0.357 227 951
1984 282 924 — 155 476 916 263 0.338 143 689
1985 325 622 — 170 645 943 360 0.372 140 271
1986 366 575 — 196 292 869 332 0.439 297 710
1987 379 876 — 146 534 988 391 0.386 249 467
1988 370 744 — 160 003 975 808 0.377 176 556
1989 328 389 — 161 704 947 977 0.363 97 360
1990 320 288 — 198 041 816 985 0.388 130 447
1991 305 544 — 157 352 699 833 0.399 113 492
1992 267 222 — 143 686 567 136 0.450 159 789
1993 243 551 — 115 008 586 369 0.345 129 738
1994 171 459 — 151 372 568 867 0.302 80 865
1995 158 512 — 173 075 565 685 0.312 142 044
1996 178 116 — 157 067 679 342 0.288 166 005
1997 204 918 — 191 485 796 733 0.289 92 171
1998 250 081 — 201 810 738 696 0.345 155 981
1999 261 671 — 178 491 728 421 0.334 76 148
2000 230 132 — 164 166 590 749 0.400 167 753
2001 213 708 — 160 483 664 563 0.329 155 450
2002 195 315 — 193 080 714 217 0.291 158 023
2003 203 600 — 189 797 742 425 0.298 180 785
2004 232 268 — 197 000 812 378 0.269 85 052
2005 225 077 — 226 783 729 491 0.278 155 009
2006 199 109 — 218 417 693 166 0.260 132 378
2007 179 391 — 203 010 669 255 0.233 95 294
2008 151 528 — 254 558 673 325 0.255 131 368
2009 179 076 — 234 809 748 357 0.233 117 333
2010 168 054 — 264 579 797 363 0.215 125 927
2011 171 584 — 323 629 843 979 0.223 165 532
2012 195 447 — 359 333 964 190 0.222 175 731
2013 228 092 — 381 977 1 087 150 0.205 125 253
2014 219 763 — 349 421 1 093 690 0.208 174 520
2015 228 172 — 456 456 1 175 280 0.208 148 570
2016 246 996 — 398 755 1 230 290 0.200 100 220
2017 247 357 — 526 734 1 159 450 0.224 155 079
2018 270 747 — 518 258 1 205 370 0.219 154 937
2019 271 787 — 462 885 1 136 040 0.236 119 708
2020 269 538 — 409 439 1 028 880 0.260 147 741
2021 263 365 — 418 669 1 150 440 0.217 133 391
2022 228 463 — 436 454 1 136 600 0.199 171 381
2023 207 694 — 419 378 1 181 400 0.186 136 037
2024 206 049 — 392 372 1 069 360 0.197 107 477
2025 205 911 203 901 379 529 972 941 0.208 117 968
2026 201 286 196 221 382 916 942 700 0.209 136 658
2027 194 667 191 724 377 265 936 135 0.205 140 595
2028 191 118 190 066 364 547 942 045 0.202 131 979

Management

History

The Ministry of Industries is responsible for the management of Icelandic fisheries and implementation of relevant legislation. Cod was incorporated into the Individual Transferable Quota (ITQ) system in 1984. During the early years of the Total Allowable Catch (TAC) system, effort management was also implemented, partly to address concerns from stakeholders who felt they had received an unfair share of the quota. This “additional effort” management system resulted in catches exceeding the TAC by 20–30% in the initial years of the ITQ system.

In 1990, legislation was amended to eliminate effort management, except for the smallest coastal fleet, which remained under a fishing days regime. At the same time, many restrictions on quota transferability were lifted, and the fishing year was redefined to run from 1 September to 31 August. These changes took effect on 1 September 1991. In the early 1990s, scientific advice from the Marine Research Institute (MRI) was based on reducing fishing mortality (F) by 40%. However, during this period, TACs exceeded scientific advice, and catches exceeded TACs.

The cod stock declined sharply in the early 1990s due to a combination of low recruitment and high fishing mortality. Recognizing the need for stricter fisheries control, a group of fisheries scientists developed a Harvest Control Rule (HCR), which was adopted for the 1995/96 fishing year. This led to a significant reduction in fishing mortality.

Since the introduction of the HCR, TACs have been set according to the rule. However, catches have exceeded TACs by an average of approximately 5% in recent years (Figure 33). The primary reason for this overage is that catches from vessels operating under the effort control system have exceeded predictions. These predicted catches are deducted from the TAC calculated under the HCR before allocating the remaining quota to Icelandic vessels.

The TAC system does not account for catches taken by Norway and the Faroe Islands under bilateral agreements. Although the size of these catches is known in advance, they were not historically considered in the Ministry’s TAC allocation to Icelandic vessels.

There is no minimum landing size for cod in ICES Division 5.a. However, economic penalties are imposed on landings valued below a minimum size threshold to discourage such practices. Iceland, Norway, and the Faroe Islands also maintain bilateral agreements governing fishing access to restricted areas within Iceland’s EEZ. Under these agreements, Faroese vessels are permitted to catch up to 5,600 tonnes of demersal fish species in Icelandic waters, including a maximum of 1,200 tonnes of cod.

Catches from Marine and Freshwater Research Institute surveys are outside vessel quotas; this catch has typically accounted for approximately 0.5 to 1.5% of the total cod catch. Other catches not included in vessel quotas (so-called VS-catch) have increased in recent years, rising from less than 0.5% to nearly 1.5% in 2024 (Figure 35).

Figure 33: Cod in Division 5.a. Comparison of the realised catches and the set TAC for the fishing operations in Icelandic waters.
Figure 34: Cod in Division 5.a. An overview of the net transfers of quota between years and species transformations in the fishery in 5a.
Figure 35: Cod. Overview of landings outside vessel quotas. The upper figure shows total landings classified as VS-catch and landings from surveys conducted by the Marine and Freshwater Research Institute. The lower figure illustrates the proportion of these catches relative to the total cod catch.

Harvest control rule

The TAC for the next fishing year (starting 1 September in the assessment year and ending 31 August of the following year) is based on a multiplier of 0.20 applied to the reference biomass of fish aged four years and older in the assessment year (\(B_{4+,y}\)). A stabiliser is then applied by averaging this result with the previous year’s TAC to determine the TAC for the next fishing year:

\[ \mathrm{TAC}_{y+1} = \frac{0.20 \cdot B_{4+,y} + \mathrm{TAC}_y}{2} \]

If the spawning stock biomass (SSB) is below the trigger value of 220,000 tonnes, the stabiliser is not applied. Instead, the harvest rate is scaled linearly according to the ratio of SSB to the trigger value:

\[ \mathrm{TAC}_{y/y+1} = 0.20 \cdot B_{4+,y} \cdot \frac{\mathrm{SSB}_y}{220{,}000} \]

A system of catch-quota balancing allowances is in place that permits quota transfers between years and some species transformations, as illustrated in Figure 34. Transfers into cod from other species are not allowed. Net quota transfers out of cod to other species have remained relatively low in recent years (Figure 34, upper panel). Transfers of unused cod quota between fishing years typically range from 0–7%.

The harvest control rule (HCR) has undergone several amendments over time. The most recent significant change occurred in 2007, when the harvest rate multiplier used to determine TAC was reduced from 0.25 to 0.20. The current HCR also includes a catch stabiliser: when SSB in the assessment year is estimated to exceed 220 thousand tonnes, the rule applies the averaging procedure as shown above.

The TAC set for the current fishing year (2023/2024), based on last year’s assessment, was 213214.

Following the 2021 benchmark, the reference biomass used in advice was estimated to be approximately 20% lower in recent years than under the previous framework. This has contributed to slightly higher realized harvest rates than intended, although these remain within the expected range of the HCR simulations. During the benchmark, reference points and the definition of harvest rate were also revised.

Management considerations

The assessment indicates that the number of cod contributing to the reference biomass (B{4+) has decreased by 8% compared to last year, with a corresponding 9% reduction in biomass. The decline in numbers follows recent decreases in spring survey indices, while the biomass reduction also reflects the progression of younger year classes, with relatively low mean weights, into the fishable stock. For most age groups, stock and catch weights have been below the long-term average over the past two years.

As capelin is a major prey species for cod in Icelandic waters (Pálsson and Björnsson 2011), and capelin stock levels have been linked to cod growth (Frater et al. 2019), the recent decline in cod weights is likely associated with the low capelin biomass observed over the past two years in Icelandic and Faroese waters, East Greenland, and the Jan Mayen area (cap.27.2a514).

Unexpectedly low weights at age in the catch will result in higher fishing mortality under the current TAC. In addition, reduced energy reserves and lower prey availability could lead to increased natural mortality. Together, these factors increase the risk of stock size declining in the near future.

Despite this, commercial catch data and surveys indicate that cod in Division 5.a is currently in good condition. This is supported by the stock assessment and the most recent HCR evaluation (ICES 2021b). While the stock is presently in a strong state, the highly variable recruitment observed over the past decade suggests future fluctuations in stock size are likely.

Current recruitment estimates for the 2021–2023 year classes remain below the decade average. As these low recruitment years occur consecutively, a decline in reference biomass can be expected in the coming years unless stronger recruitment is observed. However, because the HCR includes a cap on changes in TAC, the impact of this decline will be moderated in future advice.

References

Björnsson, Björn, Jón Sólmundsson, and Pamela J Woods. 2022. “Natural Mortality in Exploited Fish Stocks: Annual Variation Estimated with Data from Trawl Surveys.” ICES Journal of Marine Science 79 (5): 1569–82.
Björnsson, Höskuldur, Einar Hjörleifsson, and Bjarki ór Elvarsson. 2019. “Muppet: Program for Simulating Harvest Control Rules.” Reykjavik: Marine; Freshwater Researh Institute. http://www.github.com/hoski/Muppet-HCR.
Frater, Paul N, Birgir Hrafnkelsson, Bjarki Th Elvarsson, and Gunnar Stefansson. 2019. “Drivers of Growth for Atlantic Cod (Gadus Morhua l.) in Icelandic Waters–a Bayesian Approach to Determine Spatiotemporal Variation and Its Causes.” Journal of Fish Biology 95 (2): 401–10.
ICES. 2021a. “Stock Annex: Cod (Gadus morhua) in Division 5.a (Iceland grounds).” International Council for the Exploration of the Seas; ICES publishing. https://doi.org/https://doi.org/10.17895/ices.pub.18622199.
———. 2021b. “Workshop on the Re-Evaluation of Management Plan for the Icelandic Cod Stock (WKICECOD), ICES Scientific Reports. 3:30.” International Council for the Exploration of the Seas; ICES publishing. https://doi.org/https://doi.org/10.17895/ices.pub.7987.
MRI. 2016. “Mælingar á brottkasti þorsks og ýsu (e. Measurments of discards of Cod and Haddock), 2014–2016, Reykjavik, Iceland.” Vol. 3. Marine; Freshwater Research Institute, Iceland; Marine Research Institute, Iceland. https://www.hafogvatn.is/static/research/files/fjolrit-183pdf.
Pálsson, Ólafur K, and Höskuldur Björnsson. 2011. “Long-Term Changes in Trophic Patterns of Iceland Cod and Linkages to Main Prey Stock Sizes.” ICES Journal of Marine Science: Journal Du Conseil 68 (7): 1488–99.