Report of the Interbenchmark Protocol on Greenland Cod (Ibpgcod)

ICES IBPGCOD REPORT 2018

ICES ADVISORY COMMITTEE

ICES CM 2018/ACOM:30

Report of the InterBenchmark Protocol on Greenland Cod (IBPGCod)

8–9 January 2018

Copenhagen, Denmark

International Council for the Exploration of the Sea Conseil International pour l’Exploration de la Mer H. C. Andersens Boulevard 44–46 DK-1553 Copenhagen V Denmark Telephone (+45) 33 38 67 00 Telefax (+45) 33 93 42 15 www.ices.dk [email protected]

Recommended format for purposes of citation:

ICES. 2018. Report of the InterBenchmark Protocol on Greenland Cod (IBPGCod). ICES IBPGCod Report 2018 8–9 January 2018. Copenhagen, Denmark. ICES CM 2018/ ACOM:30. 205 pp. https://doi.org/10.17895/ices.pub.5266

The material in this report may be reused using the recommended citation. ICES may only grant usage rights of information, data, images, graphs, etc. of which it has owner- ship. For other third-party material cited in this report, you must contact the original copyright holder for permission. For citation of datasets or use of data to be included in other databases, please refer to the latest ICES data policy on the ICES website. All ex- tracts must be acknowledged. For other reproduction requests please contact the Gen- eral Secretary.

This document is the product of an Expert Group under the auspices of the International Council for the Exploration of the Sea and does not necessarily represent the view of the Council.

Contents

Executive summary ...... 3

1 Introduction ...... 1

2 Description of the benchmark process ...... 2

3 Cod (Gadus morhua) in East Greenland ...... 3 3.1 Stock ID and sub-stock structure (ToRs a.i) ...... 3 Tagging studies ...... 3 3.2 Life-history data (ToRs a.ii) ...... 3 Natural Mortality ...... 3 Maturity ogive ...... 3 Weight (stock and catch)...... 3 3.3 Fisheries-dependent and Fisheries-independent data (ToRs a.iii) ...... 4 3.3.1 Description of the surveys...... 4 3.3.2 Description of fisheries ...... 5 3.4 Assessment method and issues ...... 6 3.4.1 SAM ...... 6 3.5 Reference points (ToRs c) ...... 7 3.5.1 Recruitment ...... 7

3.5.2 Blim estimation ...... 7 3.5.3 EqSim settings ...... 7 3.5.4 Stock recruitment relation, derived PA points and MSY simulations ...... 8

MSY reference points (MSY Btrigger and FMSY) ...... 8 3.6 Short term projections (ToRs b) ...... 9 3.7 Future research and data requirements (ToRs d) ...... 9

4 Cod (Gadus morhua) inshore Greenland ...... 10 4.1 Stock ID and sub-stock structure (ToRs a.i) ...... 10 4.2 Life-history data (ToRs a.ii) ...... 10 Natural Mortality ...... 10 Maturity ogive ...... 10 Weight (stock and catch)...... 11 4.3 Fisheries-dependent and Fisheries-independent data (ToRs a.iii) ...... 11 4.3.1 Description of the surveys...... 11 Age range in survey data ...... 11 Description of fisheries ...... 12 Age range in catch data...... 12 4.4 Assessment method and issues ...... 12 4.4.1 SAM ...... 12 4.5 Reference points (ToRs c) ...... 13

4.5.1 Recruitment ...... 13

4.5.2 Blim estimation ...... 13 4.5.3 EqSim settings ...... 13 4.5.4 Stock recruitment relation, derived PA points and MSY simulations ...... 14 MSY reference points (MSY Btrigger and FMSY) ...... 14 4.6 Short term projections (ToRs b) ...... 15 4.7 Future research and data requirements (ToRs d) ...... 15

5 Conclusions ...... 16

6 External Reviewers Report ...... 17 6.1 Summary ...... 17 6.2 Analytical assessment and associated forecast of the West and East Greenland cod stocks ...... 17 6.3 Assessment and forecast considerations specific to the East Greenland cod stock ...... 17 6.4 Assessment and forecast considerations specific to the West Greenland cod stock ...... 18 6.5 Assessment and forecast considerations common for both stocks ...... 19 6.5.1 Description of fisheries in recent years ...... 19 6.5.2 Stock components and stock delineation ...... 19 6.5.3 Survey and catch data ...... 20

7 References ...... 22

Annex 1: List of participants ...... 34

Annex 2: Agenda ...... 35

Annex 3: List of stock annexes ...... 36

Annex 4: Working documents ...... 37

Report of the InterBenchmark Protocol on Greenland Cod (IBPGCod)

Executive summary

The ICES InterBenchmark protocol for East Greenland and the inshore Greenland cod (IBPGreenlandCod) was held at ICES headquarters on 8 and 9 of January 2018. The meeting was chaired by Marie Storr-Paulsen, Denmark, with 10 participants from 5 nations (Greenland, Norway, Germany, Iceland and Denmark). External chair was Arved Staby, Norway and external experts Rasmus Nielsen and Bjarki Elvarsson, Ice- land. The InterBenchmark was mainly conducted to, if possible, upgrade the two Green- landic cod stock assessments from DLS assessment to a full analytic stock assessment (category 1 stocks). The meeting addressed issues regarding survey data, stock structure and stock dynamics, weight and catch at age data from the commercial fishery, maturity ogive data, changes in fisheries and selection curves, as well as reference points and short term forecast settings. Specifically, issues around natural mortality from fish migrating at first spawning to Iceland and differences in mean weight at age between the Greenlandic survey on Paamiut and the German survey on Walter Herwig were discussed. This report is structured around the terms of reference covering these points. The group decided that for East Greenland cod, where tagging data clearly documented a spawning migration with increasing age, migration should be reflected in increased natural mortality with age. The SSB estimated by the SAM model was robust for different mortality rates, but as expected the increase in natural mortality had an effect on F. As the spawning migration starts at age 5 this was the first age with an increased natural mortality of 0.3. For age 6, the natural mortality was set to 0.4 and for the older ages to 0.5. For the inshore Greenland cod stock the group decided not to include any further migration mortality, as the migration pattern was not as strong as seen in East Greenland. There was a discrepancy between the two East Greenland ground fish surveys with respect to weight at age, but not length at weight. This was interpreted as a potential age reading problem between Greenland and Germany. Presently, an age reading exchange is conducted between the two nations to solve this issue. However, for this benchmark is was decided to only use the weight at age from the Greenlandic survey on Paamiut. The commercial catches are only read by Greenland. During the time period of the assessment the selection in the fishery has changed in both East Greenland due to periodic closures and in west Greenland due to changes in the main types of gears used. Several SAM models with different settings were run to test the robustness of recruitment, SSB and F to changes in survey tuning series, the inclusion of older ages, and different levels of natural mortality. The meeting recommended to use age 1–6 from the inshore gill net survey for the inshore Greenland cod stock and did not included any further migration mortality. For the east Greenland cod stock age 1–9 was used from both the Greenlandic Paamiut survey the German vessel Walter Herwig. Reference points were estimated using EqSim with some discussions regarding the type of the stock – recruitment plot. This was particularly the case for the eastern Greenland cod stock, which had some very high peaks of recruits and large inflow of cod of Icelandic origin. For both the East Greenland and the inshore Greenland cod stocks, a shortened 5 year period for the selection pattern was decided on due to recent changes in fishing pattern.

4 | ICES IBPGCod Report 2018

Short-term forecasts were estimated in SAM and the agreed settings (time-period) for the short term are included in this report as well as in the relevant stock annex. The meeting concluded that although some issues should be further investigated in future full benchmarks, such as a more robust calculated migration rate, the group concluded that both stocks could be upgraded to a full analytic assessment including new reference points and a short term forecast.

Report of the InterBenchmark Protocol on Greenland Cod (IBPGCod) | 1

1 Introduction

An Inter-benchmark process (IBP) on Greenland cod (IBPGCod), chaired by External Chair: Arved Staby, Norway and ICES Chair Marie Storr-Paulsen*, Denmark, and at- tended by two invited external experts, Rasmus Nielsen, Denmark and Bjarki El- varsson, Iceland, will be established and meet at ICES Headquarters for an Inter- Benchmark meeting, 8–9 January 2018 to:

a) Evaluate the appropriateness of data and methods to determine stock status and investigate methods for short-term outlook taking agreed or proposed management plans into account for the stocks listed in the text table below. The evaluation shall include consideration of:

i.Stock identity and migration issues; ii. Life-history data; iii. Fishery-dependent and fishery-independent data; iv. Further inclusion of environmental drivers, multi-species information, and ecosystem impacts for stock dynamics in the assessments and outlook b) Agree and document the preferred method for evaluating stock status and (where applicable) short term forecast and update the stock annex as appropriate. Knowledge about environmental drivers, including multispecies interac- tions, and ecosystem impacts should be integrated in the methodology a ) If no analytical assessment method can be agreed, then an alternative method (the former method, or following the ICES data-limited stock approach) should be put forward; c) Re-examine and update if appropriate necessary) MSY and PA reference points according to ICES guidelines (see Technical document on reference points); Develop recommendations for future work to improve the assessment and data collection and processing.

2 | ICES IBPGCod Report 2018

2 Description of the benchmark process

A physical meeting was held on 8 and 9 January 2018, and was preceded by a WebEx meeting on 1 November 2017 to ensure that work and data availability was on track for the benchmark. The main part of the work was focussed around Stock ID and the migration of mature fish to Iceland to be able to conduct an analytic assessment (category 1) for both stocks, presently assessed as DLS stocks. The stock assessors provided several working documents prior to the meeting and during the meeting, it was possible to update both stocks to category 1 stocks. Other issues, such as the weight at age data from different surveys, commercial catches and reference points were also discussed and agreed upon. The well prepared and well documented (with working documents) stock assessor was much appreciated by the group.

Report of the InterBenchmark Protocol on Greenland Cod (IBPGCod) | 3

3 Cod (Gadus morhua) in East Greenland

3.1 Stock ID and sub-stock structure (ToRs a.i)

Tagging studies Extensive tagging data shows that there is a significant emigration of East Greenland cod to Iceland. This emigration starts at approximately age 5 when fish become mature, and the proportion of migrating fish increases with increasing age. The tagging data also suggests that the number of fish returning to East Greenland waters is low. There is every reason to assume that these conclusions are also valid for the period prior to that presented here, given that a high proportion of tagged cod is recorded in Icelandic annually in the period 2001–2010 (Table 3.1.1, i.e. Hansen and Hermann, 1953; Storr-Paulsen et al., 2004). The exact level of migration in terms of the proportion of a year class that migrates to Iceland from East Greenland each year is difficult to quantify. Hence, the fishing mortality and effort is unknown in East Greenland and there might be differences in the return percentages from the vessels. However, based on the numbers shown here, at least 50% of a cohort appears to go to Iceland (Table 3.1.2). Conclusion Natural mortality was increased to 0.3 at age 5, to 0.4 at age 6 and kept constant at 0.5 for ages 7+. Issue list Future work regarding stock structure and stock dynamics should include a detailed analysis by age and season (year), which would give an indication of the variation of the proportion of migrating fish at age between years.

3.2 Life-history data (ToRs a.ii)

Natural Mortality See Stock ID and sub-stock structure (3.1)

Maturity ogive Based on maturity data for 1557 cod, collected in ICES Division 14.b and NAFO 1F during the spawning season April to May in the period 2007–2016 on commercial vessels, L50 was estimated at 5.00 years (SE = 0.06). The average proportion of mature fish at age for this period was applied to the entire time series 1973–2016. The maturity ogive was estimated by a general linear model (GLM) with binomial errors (Figure 5, WD01). Conclusion It was agreed to use a constant maturity ogive sampled in the time-period 2007–2016 for the entire dataset 1973–2016.

Weight (stock and catch) Mean weights-at-age estimated for the two east Greenland surveys differ considerably with a doubling of weight at age in the German survey (2008–2015) compared to the Greenland survey (2006–2016; Figure 3.2.1). Length-weight relationship between the surveys indicate that there might be an age reading issue as the length-weight relationship are similar in the two surveys (Figure 3.2.2).

4 | ICES IBPGCod Report 2018

Conclusion Stock weight: It was decided to exclude the German mean weight at age data and only use the Green- land survey weight data from 2008. Stock weights used for the period prior to 2008 are the mean of weights at age calculated from this data for the period 2008–2016. Weight in catch: From 2005, the catch at age data is derived from annual on board sampling conducted by the fishermen. Gaps in the time series of weight in catch (1996–2004) were filled with a constant average weight at age based on weights from the periods 1973–1995 and 2005–2016.

3.3 Fisheries-dependent and Fisheries-independent data (ToRs a.iii)

3.3.1 Description of the surveys Abundance at age data are available for two annually conducted offshore trawl surveys - the Greenland Shrimp and Fish survey (GRL-GFS) and the German groundfish survey (Ger(GRL)-GFS-Q4). Both are stratified random bottom-trawl survey. The Greenland survey started in 1992 in West Greenland covering the area south of 72000’N and depth from 0–600 m. Since 2008, East Greenland south of 67000’ N was included in the survey (ICES, 2017a). The Greenland survey covering East Greenland and NAFO Division 1F since 2008 was used as a tuning series for this assessment. Ap- proximately 125 hauls are taken each year. An internal consistency plot was produced and while for most age groups the relationship was reasonable it was considered poor between age 4 and 5 and 5 and 6 (Figure 3.3.1). The German survey has been conducted since 1982 (ICES, 2017a). The survey covers both East (area south of 67°00’ N) and West Greenland (area south of 67°00’ N). The German survey covering East Greenland and NAFO Division 1F provides number at age for age groups 1 to 9 and was used as a tuning series for this assessment. Numbers of hauls in East Greenland and NAFO Division 1F were initially ca. 110 per year but were reduced from the early 1990s to 50–60 per year. An internal consistency plot was produced and for all age groups the relationship was considered high (Figure 3.3.2).

Age range in survey data The survey provides number at age for the age groups 1 to 9.

Comparison between Greenland and German survey Given the potential age reading discrepancies between Germany and Greenland inclusion of the German survey data as a tuning fleet may influence the assessment results. The leave-one out plots generated in SAM which show the consequences for the SSB and the fishing mortality when omitting one or the other survey, depict the same perception of the stock over time, but with differences in both SSB and fishing mortality. Including only the Greenland survey estimates would for example increase the SSB and decrease the fishing mortality – this is probably because the German survey catches much fewer older fish than the Greenland survey.

Report of the InterBenchmark Protocol on Greenland Cod (IBPGCod) | 5

Conclusion In most years, the differences between the two surveys lie within the confidence limits and it was thus decided, that until the age reading issue has been resolved, to keep both surveys in as tuning fleets.

3.3.2 Description of fisheries During the last two decades the bottom trawl cod fishery off East Greenland has seen changes in fishing patterns, partially because of the closure of northern areas for commercial fishing from 2007–2009. Vessels in the offshore fisheries are vessels above 75BT/120BT and are restricted to the area outside 3 nm off the baseline. Trawlers and longliners exclusively conduct the fishery and longliners have in recent years increased the amount taken to 40% of the total catch in 2016 (Retzel, 2017). Mesh size in the trawl fishery is 140 mm. The large mesh size is one of the main reasons to assume negligible discarding taking place, the other reason being that quota does not seem to have been limiting the fisheries. The East Greenland area has been subject to several area closures in recent years. In 2008, fishing north of N63°00’ was not allowed in order to protect the potential spawning segments, especially on Kleine Banke. In 2009–2010 the delimitation line was at N62°00’ and additionally NAFO 1F was closed in 2010, primarily to protect the relatively strong incoming year classes. The high effort in 2008 and 2009 was caused by increased catches in NAFO 1F combined with decreasing CPUE. With the closure of 1F in 2010, effort decreased. Since 2011, the area north of 630N has been closed in spawning season (April and May).

Commercial catch at age data Since 2005, a directed cod fishery started in both East and South Greenland. Annual catch-at-age are obtained from a self-sampling program in this recent period Information on historic catch in numbers 1973–1995 for West and East Greenland is from a compilation in Hovgård and Wieland (2008). From this data, combined catch- at-age for East Greenland and South Greenland (NAFO Division 1F) was calculated (Retzel, 2015). In the period 1996–2004 no sampling was conducted due to the very low catches. The model setup used total catch weight per year as a “survey” for this period and the model was set to utilize this aggregated information.

6 | ICES IBPGCod Report 2018

Information on the sampling intensity on the East Greenland cod

Aged cod Length measured cod

Year Count Stations Count Stations 2005 306* 26* 4150 26 2006 70 / 732* 8 / 53* 3554 14 2007 559 / 216* 9 / 6* 18531 109 2008 146 / 882* 4 / 69* 1763 11 2009 886* 81* 8777 43 2010 430 / 689* 20 / 62* 13073 63 2011 1947 123 21781 159 2012 1665 132 12250 83 2013 2082 147 19565 158 2014 1828 138 13966 171 2015 1140 191 12059 153 2016 885 122 14469 168

3.4 Assessment method and issues

3.4.1 SAM All the settings in the SAM configuration can be seen in Table 3.4.1. Some of the key settings are shortly explained below.  The landing fraction was assumed to be 1 (i.e. no discarding). This is consistent with the overall impression of the fishery and the gears typically allow for the release of small cod (see Section 3.3.2).  The Greenland (2008–2016) and German (1982–2016) surveys in East Green- land and NAFO Division 1F age 1 to 9 were applied for tuning.  The catchability in both surveys was assumed different for age 1 to 8 and age 8 and 9 coupled. The reason for coupling age 8 and 9 was that number of age 9 caught in the survey very rather few, especially in former years.  There is no catch of ages 1 and 2 in any years and the fishing mortality coefficients were not estimated for these ages. The coefficients were estimated independently for ages 3 to 8, and set to a common coefficient for ages 9, and 10+. The selectivity curve shows a gradually increase up to age 8 followed by a mi- nor decrease (Figure 3.5.3).  The Fbar range of 5 to 10 years old was applied as these age groups constitute the main part of the commercial catches.  The process variance parameters for the log(N)-process were separated for age 1 and for age 2–10 as age 1 may be estimated with more uncertainty.  The variance parameters for catches and surveys were set similar for all ages (catches 3–10 years old, and surveys 1–9 years old) but separately for the catch and each of the surveys.  An independent covariance structure for each fleet was assumed.  The covariance structure for each fleet was set to be independent as no knowledge was available to suggest another structure. Similarly, it was assumed that there was no coupling of correlation parameters in the surveys. The correlation across ages was set to be auto-correlated with a lag of 1. The

Report of the InterBenchmark Protocol on Greenland Cod (IBPGCod) | 7

fishery is mainly a trawl fishery and some auto-correlation across ages is expected. Furthermore, applying the auto-correlation option lead to a decrease in AIC compared to the independent option.  We assumed a random-walk recruitment for the short-term forecast. For the full settings used, see Table 3.4.1

3.5 Reference points (ToRs c) The cod stock in East Greenland was previously assessed qualitatively as a category 3 stock. With the acceptance of the present analytical assessment (SAM), reference points are therefore appropriate to define. The group took basis in the ICES advice technical guidelines as published 20 January 2017 (ICES, 2017b) for the estimation of the reference points.

3.5.1 Recruitment Whether age 1 is appropriate to describe recruitment in relation to spawning stock i.e. the cohort tracking from age 1 to age 5, was analysed by sensitivity analyses of con- ducting SR with age 3 as recruits instead. This did not change the perception of the SR relation with respect to year-class strength and age 1 was therefore kept as recruiting age to the stock. Exclusion of years with a potential high Icelandic input was discussed (1973, 1984 and 2003 year classes). However, as previous tagging studies showed that a certain proportion of Icelandic cod contributed to not only the big year-classes but all year classes (Table 3.1.1) it was decided to keep all year classes in the time series for the reference point estimation.

3.5.2 Blim estimation

For estimating Blim a categorization of the stock-recruitment relationship into type is required (ICES, 2017b). The group agreed that the Type 1 S-R relationship corre- sponded best to the stock- recruitment relationship with some very high recruitment peaks, and low recruitment even at high SSB values (Figure 3.5.1). Type 1 is described as “Spasmodic stocks – stocks with occasional large year classes”. According to this SR type, Blim is based on the lowest SSB where large recruitment is observed. The first large yc is the 2003 yc which is record high for the time series. Because of the proportion of the Icelandic stock in East Greenland waters, it was decided by the group to make the Blim estimate more robust by calculating an average of three years SSB where recruitment seems not to be impaired (year-class 2002, 2003 and 2004). This gave a Blim of 10 354 tonnes. At the sandeel benchmark in 2016, a similar decision was taken (ICES, 2017c).

3.5.3 EqSim settings Data from the accepted SAM assessment (run: EastCod_2017_final) were used for the following simulations. The Eqsim software (ICES, 2015) was used to explore SR rela- tions and define both PA and MSY reference points. The present EqSim version (edited by D.C.M. Miller, ICES) reads input data and assessment results directly from the required SAM run at assessment.org. The text table below provides the simulation settings and the justification. The number of simulations was set to 2000.

8 | ICES IBPGCod Report 2018

Settings for the EqSim for East Greenland cod

Data and parameters Setting Comments

Full range The group did not recognize any specific re- SSB-recruitment data of data gime shift in productivity (1973–2014) Mean weights and pro- 10 years average is recommended in ICES 2007–2016 portion mature guidelines (ICES 2017) 5 years period to reflect recent fishery. Selec- Exploitation pattern 2012–2016 tivity has changed over time (Figure 3.5.1). Assessment error in the F: 0.343 Based on final year in assessment advisory year. CV of F SSB: 0.217 Based on final year in assessment and SSB Autocorrelation in as- Since no previous analytical assessments have sessment error of the 0.423 be conducted for this stock, default values advisory year, Fphi was set according to ICES guidelines

3.5.4 Stock recruitment relation, derived PA points and MSY simulations Three different models, Ricker, Beverton&Holt and segmented regression, were weighted by the default “Buckland” method, and applied to the time series, 1973–2014 (Figure 3.5.2). The two last years (2015 and 2016) were also removed as SAM estimated numbers of recruits for these most recent years are considered uncertain. Segmented regression was weighted far higher (0.83) than the two other methods and was thus decided for the further simulations (Figure 3.5.2).

Blim was estimated to 10 354 tonnes as described in 3.5.2.

Bpa was calculated from Blim by: Blim * exp(1.645 *σ), where σ is SD of ln(SSB) in 2016 - here estimated by SAM to 0.217. Bpa is then estimated at 14 802 tonnes.

Flim was estimated by simulation using the above values of Blim and Bpa, setting Fcv, Fphi and SSBcv = 0 (no assessment and advice noise) and with no MSY Btrigger. The range of years are the same as used for the SSB-R relationship (1973–2014) F50 is median Flim, here estimated to 2.34 (Tables 3.5.1 and 3.5.2).

Fpa is calculated from the formula Fpa = Flim * exp(-1.645 * σ), where σ is SD of ln(F) in 2016 here estimated by SAM to 0.34. Fpa is then 1.33 (Tables 3.5.1 and 3.5.2). The simulations were done with 200 runs, scanning F from 0 to 3 divided into 100 intervals.

MSY reference points (MSY Btrigger and FMSY)

FMSY is initially estimated as the F that maximize median long-term yield in the simulation under constant F exploitation. The recommended default values of cvF = 0.212, phiF = 0.423 and cvSSB = 0 were applied to the simulation since no assessment/advice history is available for this stock. The initial FMSY was estimated at 0.65, which is far below the above estimated Fpa (1.33).

The final FMSY, capped by the option SSB ≥ Blim with a 95% probability (Fp05), is estimated by a simulation using the default Fcv, Fphi, the estimated Blim, Bpa and Btrigger here set equal to Bpa. Fp05 was estimated at 0.46. The precautionary principle states that if FMSY > Fp05, which is the case here, then FMSY should be reduced to Fp05. The final FMSY therefore equals Fp05 = 0.46 (Tables 3.5.1 and 3.5.2).

Report of the InterBenchmark Protocol on Greenland Cod (IBPGCod) | 9

3.6 Short term projections (ToRs b) Settings: A random-walk recruitment was assumed for the short-term forecast as no clear relationship has been seen for this stock. The time-period from where the recruitment was drawn was from the full time-series. The biological parameters is as default set as the 5 last years’ average and the selectivity is taken from the final assessment years. Three short-term forecast scenarios were simulated. The results of the forecast are not described as short-term forecasts is not a part of the ToR.

3.7 Future research and data requirements (ToRs d)  Future work regarding stock structure and stock dynamics should include a detailed analysis by age and season (year), which would give an indication of the variation of the proportion of migrating fish at age between years.  Age reading exchange: during autumn 2017, 300 otolith are being exchanged. Results from this will be available at NWWG 2018. o Otolith exchange program/workshop (following ICES guidelines under the ICES group WGBIOP) is recommended. o If there is an age reading discrepancy between Greenland and Ger- many, this will have an effect on the tuning series presently used in the stock assessment.  Several hauls were made in end September by the Greenland survey vessel in 2017 in the same area as the German survey in order to investigate the weight at age in collections made in the same time period and area between the two surveys. Results from this will be available at NWWG 2018.  Thorough investigation of variability in the weight at age in the surveys especially to investigate if there is an increasing trend in weight at age through the time period in the German survey and the catches.  Investigate growth rate in the tagging data.

10 | ICES IBPGCod Report 2018

4 Cod (Gadus morhua) inshore Greenland

4.1 Stock ID and sub-stock structure (ToRs a.i) There has been extensive tagging in the inshore region for decades. Jointly, these studies show that cod tagged in the deeper parts of the West Greenland fjords remain there for years, suggesting a distinct inshore population. This is confirmed by recent genetic studies. Cod in the coastal region tends to be a mix of the different cod stocks present in Greenland waters, and from this area, cod migrates both further inshore and to offshore banks. The stocks are not morphologically distinguishable and any stock assign- ment hinges on genetic analyses. The survey for the inshore stock is conducted with gill nets, which are set throughout the fjords. Recent genetic studies suggest that the main component in these surveys is the inshore stock component. A similar study on the catches suggests different proportions, with the offshore stocks constituting a significant part of the landings. The 1984 year class was highly abundant in both Green- land inshore and offshore waters. The general perception is that this year class “disap- peared over night”, and following its disappearance the inshore area had very few cod. Consequently, the conclusion is that this year class was most likely of offshore origin and to mimic extensive migration, the natural mortality for this year class was increased. Conclusion For the 1984 year class, the natural mortality was set at 0.4 from age 6 and older as the migration is associated with maturity and is considered a spawning migration. For all other year classes and for all ages natural mortality was set at 0.2. Issue list Future work should focus on estimating the level of stock mixing in the survey and in the fishery. This can also be extended back in time by analyzing otoliths available in collections.

4.2 Life-history data (ToRs a.ii)

Natural Mortality See Stock ID and sub-stock structure (Section 4.1)

Maturity ogive Due to large fluctuations between years in maturity sampling, with years with little or no sampling, it is not possible to generate an annual maturity ogive. Maturity information prior to 2000 is only available for November in 1987 (N = 484). Recent information is available since 2007, especially for 2010 and 2011, where dedicated spawning collections where made (April and May). There appears to be a shift in maturity between the early pre-2000 period and recently. The estimated length at 50% maturity decreased from 5.07 years (SE = 0.18) in 1987 to 4.32 years (SE = 0.04) in 2007–2016. Ma- turity ogives are available for earlier periods (1930s) and these are similar to the 1976– 1999 estimates, supporting that a shift has occurred. Conclusion Because of the shift in maturation, the conclusion from the group was to use two different fixed maturity ogives: 1976–2006 and 2007–present.

Report of the InterBenchmark Protocol on Greenland Cod (IBPGCod) | 11

Weight (stock and catch) Because cod from the inshore fishery are landed primarily as decapitated and gutted fish, there is a lack of age (and weight) data from the commercial fishery. Due to the lack of mean weight at catch from the commercial fishery, the age-length key and length at weight relationship for the Greenland inshore survey was used to derive a mean weight at age for the commercial fishery, using the length samples from the commercial fishery. This may introduce an auto-correlation error and most likely underestimate the mean weight at age, since jigging and long liners will target the larger fish from corresponding year classes. This is especially in later years were the fraction of jigging and long liners has increased. However, the main fishery remains the pound net fishery, and this takes place at similar depths and locations as the survey, meaning that the effect should be small. Issue list It is recommended that in future otoliths and corresponding length/ weight data be obtained from the different commercial fisheries to get a better understanding and independent estimate of mean weight at age of the catch.

4.3 Fisheries-dependent and Fisheries-independent data (ToRs a.iii)

4.3.1 Description of the surveys A multi meshed gill net survey designed to target juvenile cod age 2 and 3 years old in the inshore area in West Greenland has been conducted annually since 1987. The objective of the survey is to assess the abundance and distribution of recruiting cod. How- ever, given the different ways of being caught in a gill net other than being gilled the selectivity is not entirely dome shaped but elongated towards larger fish. The survey uses gangs of gill nets with different mesh sizes (16.5, 18, 24, 28 and 33 mm, ½ mesh size). 100–150 nets are set annually and are set parallel to the coast in order to keep the depth constant. The survey effort is allocated evenly between the depth zones of 0–5 m, 5–10 m, 10–15 m and 15–20 m. The abundance index used in the survey is defined as 100*(caught/net*hour). Historically three areas were covered: north west (Sisimiut, NAFO Division 1B), mid west (Nuuk, NAFO Division 1D) and south west (Qaqortoq, NAFO Division 1F). South Greenland has only been covered in the period 1987–1995, 1998, 2000 and 2007–2009 and due to very scarce data from this survey this area is not included as a tuning fleet. The 1B survey covers the period 1987–1998, 2002–2007 and 2010–2016. The 1D survey covers the period 1987–2016 except in 2002 and 2007 where no survey was conducted.

Age range in survey data The surveys target primarily 2- and 3-year-old cod. Initially the assessment was conducted using ages 1–5, but since data are also for older age classes and because these data contained information especially in later years, it was decided to include age 6 data in the tuning series. Conclusions It was suggested to combine the two surveys to increase the overall weighting in the model. However, due to the lack of coverage in certain (but different) years in both surveys, data needed to be tabulated to fill in the data holes. For this reason, it was

12 | ICES IBPGCod Report 2018

thought to be more transparent to keep the two surveys separate, including the data gaps for the ages 1–6.

Description of fisheries The inshore Greenland commercial cod fishery in West Greenland started in 1911 by opening cod trading facilities where cod seemed to occur regularly. The fishery expanded over the next decades with the addition of new trading facilities. Annual catches above 20 000 tonnes have been taken inshore during the period 1955–1969 but declined to around 5000 tonnes in the 1970s. There are no data on catch-at-age for this yearly period. In the 1980s catches fluctuated between 5000 and 35 000 tonnes, partly driven by a few strong year classes (1979 and 1984) entering from the offshore stocks. From 1993 to 2001, the inshore catches were low; in the range 500–2000 tonnes. In the 2000s, catches have gradually increased and currently exceeds 30 000 tonnes. The most important gear has been pound-net (between 60% and 80% of the annual catches). These are anchored to the shore and fishing the upper 20 m. Due to ice, pound nets are replaced by jigs, longlines and gill nets during November–April. Trawling is not allowed within 3 nm off the baseline. The fishery is carried out along the entire coastline of West Greenland from Disko Bay to Cap Farewell, with the majority of the catches being taken in mid-Greenland (NAFO areas 1B–1D). In recent years, the fishery has expanded north, and catches in this area are to a large extent caught as bycatch in the Greenland halibut fishery. This has shifted the selectivity slightly, increasing the catchability of larger cod. Discarding is considered negligible for the fisheries, as undersize cod can be released alive from the pound nets and as the TAC is not limiting for the fisheries.

Age range in catch data The age range in the catch data is 1–10, with 10 being a plus group. The weight-at-age for the commercial catch-at-age data are derived from the survey age-length-key as the main parts of the commercial catches are landed gutted and without head.

4.4 Assessment method and issues

4.4.1 SAM All the settings in the SAM configuration are shown in Table 4.4.1. Some of the key settings are explained in short below.  The landing fraction was assumed to be 1 (i.e. no discarding). This is consistent with the overall impression of the fishery. There has not been an enforced TAC constraint in any years and the TAC is therefore not seen as limiting for the catches and the pound nets allow for the release of small cod alive.  The two gill net surveys in NAFO 1B (Sisimiut) and NAFO 1D (Nuuk) were used for tuning separately. Because of the survey gear catchability, only ages 1–6 were used in the model.  We assumed a random-walk recruitment for the short-term forecast.  The catchability in both surveys was assumed different for ages 1 to 6. The surveys target relatively small cod, which at these life stages maintain distinct length frequencies. Therefore, we assume that they are not caught equally well in the survey and let the model estimate the parameters.

Report of the InterBenchmark Protocol on Greenland Cod (IBPGCod) | 13

 There is no catch of ages 1 and 2 in any years and the fishing mortality coefficients were not estimated for these ages. The coefficients were estimated independently for ages 3 to 8, and set to a common coefficient for ages 9, and 10+. There is a relatively steep selectivity curve for the fishery, but a plateau is not reached. Instead, the older age groups are not caught as well as younger age groups, and this is particularly true for the main fishing gear; pound net.  The Fbar range of 4 to 8 years old was applied as these age groups constitute the main part of the catches. On average, the 9 and 10+ old cod constitute less than 1% of the total catch, and no more than 5% in any single year.  The covariance structure for the fishery was set to be independent as no knowledge was available to suggest another structure. Similarly, it was assumed that there was no coupling of correlation parameters in the surveys. The correlation across ages was set to be independent. Using the auto-correlation options provides an improved model fit (∆AIC of 18), but this comes at a cost of a more variable F estimate. We do not think F changes this much from year-to-year and opted for the independent option despite a worse model fit.

4.5 Reference points (ToRs c) The inshore cod stock in West Greenland was previously assessed qualitatively as a category 3 stock. With the acceptance of the present analytical assessment (SAM), reference points can be defined. The group took basis in the ICES advice technical guidelines as published 20 January 2017 (ICES, 2017b) for the estimation of the reference points.

4.5.1 Recruitment Whether age 1 is appropriate to describe recruitment in relation to spawning stock i.e. the cohort tracking from age 1 to age 5, was analysed by sensitivity analyses of con- ducting SR with age 2 as recruits instead. This did not change the perception of the SR relation with respect to year-class strength and age 1 was therefore kept as recruiting age to the stock (Figure 4.5.1). No historic trends were observed in the stock recruitment relation and therefore the entire time series from 1976 was utilized in the estimation of reference points.

4.5.2 Blim estimation

For estimating Blim, a categorization of the stock-recruitment relationship into type is required (ICES, 2017b). The group agreed that the Type 2 S-R relationship corre- sponded best to the stock- recruitment relationship with a wide dynamic range of SSB and evidence that recruitment is or has been impaired (Figure 4.5.1). According to this SR type Blim, is based on the breakpoint in a segmented regression. This gave a Blim of 4346 tonnes. (Figure 4.5.3).

4.5.3 EqSim settings Data from the accepted SAM assessment (run: WestIns_2017_final) were used for the following simulations. The Eqsim software (ICES, 2015) was used to explore SR rela- tions and define both PA and MSY reference points. The present EqSim version (edited by D.C.M. Miller, ICES) reads input data and assessment results directly from the required SAM run at assessment.org. The text table below provides the simulation settings and the justification. The number of simulations was set to 2000.

14 | ICES IBPGCod Report 2018

Settings for the EqSim inshore West Greenland cod

Data and parameters Setting Comments

Full range of data The group did not recognize any specific SSB-recruitment data (1976–2016) regime shift in productivity.

20 yrs average since no change is seen for this time span. 10 yrs average is recom- Mean weights and propor- 1997–2016 mended in ICES guidelines (ICES, 2017b) tion mature however, EqSim crashed with this setting.

5 yrs period to reflect recent fishery pat- Exploitation pattern 2012–2016 tern. Selectivity has changed over time (Figure 4.5.2).

Assessment error in the advi- F: 0.252 Based on final yr in assessment. sory year. CV of F and SSB SSB: 0.194

Since no previous analytical assessments Autocorrelation in assess- have be conducted for this stock default ment error of the advisory 0.423 values was set according to ICES guide- year, Fphi lines.

4.5.4 Stock recruitment relation, derived PA points and MSY simulations Three different models, Ricker, Beverton&Holt and segmented regression were run and applied to the time series, 1976–2016 (Figure 4.5.3). Beverton&Holt was clearly weighted highest (0.47) by the three methods and was thus decided for the further simulations.

Bpa was calculated from Blim by: Blim * exp(1.645 *σ), where σ is SD of ln(SSB) in 2016 - here estimated by SAM to 0.194. Bpa is then estimated at 5983 tonnes.

Flim was estimated by simulation using the above values of Blim and Bpa, setting Fcv, Fphi and SSBcv = 0 (no assessment and advice noise) and with no MSY Btrigger. F50 is median Flim, here estimated to 3.61 (Table 4.5.1).

Fpa is calculated from the formula Fpa = Flim * exp(-1.645 * σ), where σ is SD of ln(F) in 2016 here estimated by SAM to 0.252. Fpa is estimated to 2.39 (Table 4.5.1). The simulations was done with 200 runs, scanning F from 0 to 5 divided into 101 intervals.

MSY reference points (MSY Btrigger and FMSY)

FMSY is initially estimated as the F that maximize median long-term yield in the simulation under constant F exploitation. The recommended default values of cvF = 0.212, phiF = 0.423 and cvSSB = 0 were applied to the simulation since no assessment/advice history is available for this stock. The initial FMSY was estimated at 0.268, which is below the above estimated Fpa (2.39).

Report of the InterBenchmark Protocol on Greenland Cod (IBPGCod) | 15

then FMSY should be reduced to Fp05. This is not the case here and the final FMSY therefore equals initial FMSY = 0.268 (Table 4.5.1).

4.6 Short term projections (ToRs b) A random-walk recruitment was assumed for the short-term forecast as no clear relationship has been seen for this stock. The time period from where the recruitment was drawn was from the full time-series. The biological parameters are as a default set as the last 5 years average and the selectivity is taken from the final assessment years. Three short-term forecast scenarios were simulated. The results of the forecasts are not described as short-term forecasts is not a part of the ToR.

4.7 Future research and data requirements (ToRs d)  It is recommended that in future otoliths and corresponding length/ weight data be obtained from the different commercial fisheries to get a better understanding and independent estimate of mean weight at age of the catch. As the fish are mainly landed decapitated and gutted a self-sampling program or sampling directly from the fishery will be needed. An alternative could be a reference fleet.  It should also be looked into if it is sound to use ages up to 10 years in the catch matrix as the biology to this data comes from the survey were the numbers from age groups above 6 are very limited.  Future work should focus on estimating the level of stock mixing in the survey and in the fishery. This can also be extended back in time by analyzing otoliths available in collections.

16 | ICES IBPGCod Report 2018

5 Conclusions

The conclusion of the group was to recommend that ICES use the final SAM assessments as the official category 1 analytical assessment for both the inshore West Green- land and the East Greenland cod stock that was output from this InterBenchmark meeting. Further, the group reached consensus on the reference points as well as the settings used for the short-term forecast for both stocks.

Report of the InterBenchmark Protocol on Greenland Cod (IBPGCod) | 17

6 External Reviewers Report

IBPGCod Inter-Benchmark Meeting, ICES Secretariat 8–9 January 2018, Copenhagen External reviewers (experts): J. Rasmus Nielsen, Bjarki Elvarsson External chair: Arved Staby

6.1 Summary The reviewers have been through the working documents and acknowledge the comprehensive effort made in the preparation and presentation, as well as the timely de- livery of these working documents in advance of the InterBenchmark working group meeting. The prepared material provided a good basis for further discussions and eval- uations of the assessments brought forward during the InterBenchmark working group meeting. The reviewers note that this was an InterBenchmark assessment with a 2 day meeting duration. Comprehensive ToRs were formulated for this InterBenchmark assessment. The ToRs comprised full benchmark considerations, including: i) evaluation of stock entity, delineations and mixing (including migrations between stocks), ii) implemen- tation of new assessment models and upgrading from data poor stock assessments to analytical assessments (type 3/4 assessments to type 1 assessments), and iii) definition of revised MSY or PA reference points for the stocks, based on evaluation of appropriate stock-recruitment relationships according to the guidelines set out by ICES. The reviewers also note that the ToRs included the development of recommendations for future work to improve the assessments, including data collection and processing.

6.2 Analytical assessment and associated forecast of the West and East Green- land cod stocks The assessments were performed using current information and aggregation levels of presently available data. The reviewers recommend that the final SAM baseline assessments resulting from the InterBenchmark meeting for each of the stocks are adopted by ICES as the official category 1 analytical assessment for the inshore West Greenland cod stock and for the East Greenland cod stock. Overall, the baseline assessments performed relatively well without strong trends in the residuals and without strong retrospective patterns. The explored scenarios did in general not deviate from the baseline and were within the certainty levels of the baseline assessments. The assessments generally seem to perceive the development and dynamics in the stocks well and within reasonable certainty limits for both stocks over time.

6.3 Assessment and forecast considerations specific to the East Greenland cod stock Regarding the overall conclusion mentioned above, it is worthy to note that the assessment of the East Greenland cod stock seems to slightly underestimate F and slightly overestimate SSB. With respect to the East Greenland cod stock, Figure 1 in WD01 indicates temporal trends in MWA in the landings with an increasing tendency in the more recent years compared to historical levels. This trend was confirmed during the meeting. It was questioned whether there are similar trends over time in the survey MWA data time

18 | ICES IBPGCod Report 2018

series (German survey) which should be further explored. Furthermore, there is inconsistency in MWA between the Greenland and German surveys, with an increasing tendency by age. Impacts of these age-related weight differences were discussed in relation to use of survey data as proxies for MWA in the stock, and agreement was reached on how to address this issue for the time being. For future full benchmark assessment, it is recommended that ANOVA of temporal trends (including confidence limits of MWA by age by year) for those different data time series are made. Accordingly, historical biological raw data from commercial fisheries should be scrutinized. Preliminary analyses indicate age reading inconsistency between the two survey series. The reviewers acknowledge that an initiative under an otolith exchange program is addressing this issue. The results of these analyses should be considered in relation to the MWA used in the stock in different time periods.

Regarding the East Greenland stock, the reviewers questioned the choice of Blim in the assessment presented in the working documents. The reasoning for the choice of Blim given in the working documents needed to be elaborated further. The reviewers perceived the SSB-R-relationship as a Type 1, and it could be considered whether the average of 3 years where low biomasses were able to produce strong year classes should be used. It is further suggested that sensitivity analysis be performed using age 1 or 3 as recruitment, taking into consideration exploitation patterns and gear selectivity in the data time series, as the estimated number at age 3 may provide a better indication of the recruitment to the fisheries. With respect to the forecast for the East Greenland stock, the reviewers questioned the exploitation pattern over time. The selectivity presented reflects the more recent northerly fishery, which is expected to continue in the future. The estimated selection and exploitation pattern for the last 3 or 5 years was therefore recommended to be used in forecasts.

6.4 Assessment and forecast considerations specific to the West Greenland cod stock Regarding the use of MWA from surveys in the catches, the reviewers recommended that a future full benchmark assessment should further investigate gear selectivity. Im- pact of the use of aggregated or disaggregated data time series, from various sources, on the MWA estimates used in the assessment should be investigated. These sources include surveys with gill nets and the commercial fishery, consisting mainly of pound nets but with significant catches from jigs and longlines. Accordingly, selectivity differences between the gill net surveys and the pound nets should be explored, and biological data should be sampled from the jig and longline fisheries, i.e. gears that tend to select for larger fish. Autocorrelation in the assessment due to the application of the age-length key from surveys to the commercial fishery catch at length data should also be explored further in future during a full benchmark. Issues related to stock identity and mixed fisheries of different stock components and potential local depletions were discussed but could not be addressed at this benchmark due to the unavailability of data on a finer spatial scale. It is therefore recommended that the analysis of time series of catches, effort and survey data should be done in a future full benchmark, considering several inshore components and a West Greenland offshore component.

Report of the InterBenchmark Protocol on Greenland Cod (IBPGCod) | 19

Regarding survey data the question was raised whether the two gill net surveys (1D and 1B) should be kept separate or combined in the assessment. This was discussed according to scenario analyses, as these survey series show different trends through time. As it is necessary to make assumptions based on regression analysis to fill in data gaps, and the effect of combining them was not perceived to have a noticeable effect on the assessment, the reviewers recommend to keep them separate. However, this issue should be revisited in a future benchmark in light of the results of the analysis of catch data on a finer scale, as suggested above.

For the West Greenland forecast, a type 2 SSB-R relationship should be used. Here Blim was be set at the inflection point for a segmented regression. Upon the reviewer’s recommendation a sensitivity analyses using age 2 as recruitment age was performed.

6.5 Assessment and forecast considerations common for both stocks

6.5.1 Description of fisheries in recent years Selection patterns in the fishery have changed in later years with the closure of southern areas and more northerly fishery in East Greenland, and with the introduction of new gears - among other longlines and jigs - besides the pound net fishery in inshore West Greenland areas. Accordingly, the selection and exploitation patterns for recent years should be used in the forecasts, based on a detailed description of changes in fishing patterns and gears used. As noted above, regarding the mixed fishery off West Greenland, the exploitation of different stock components and / or potential local depletion should be explored further by analysing time series of catches, effort and survey data. These analyses should be done in preparation for a future full benchmark considering several inshore components and West Greenland offshore components. In this context, catch at age matrices should also, to the extent possible, be disentangled and disaggregated on area level (stock component) in the catch data.

6.5.2 Stock components and stock delineation With respect to assessment of different stock components and stock delineation, the migration issues in relation to results from available tagging studies were discussed for both cod stocks. For both the East and West Greenland cod stocks, there seems to be a clear pattern with increasing emigration to Iceland by age which impacts both cod stocks and their assessments, though it appears that the degree of emigration from West Greenland to Iceland is lower than from East Greenland to Iceland. For the east Greenland cod stock, tagging studies indicate that a considerable proportion of cod emigrate from East Greenland to Iceland from age 5. The available tagging data indicate an increasing tendency by age of the proportion emigrating. This emigration has for the East Greenland cod stock been dealt with by increasing the natural mortality (M) by age in the stock assessment, assuming a certain proportion of the fish for the older age groups emigrate to Iceland. The reviewers notice that the assessment is sensitive to the settings of the M parameter for the older age groups in the assessment and the increasing tendency herein. The current approach accounts for the emigration by estimating the fishing mortality in the stock component remaining in Greenland waters, given a certain East Greenland spawning component. The aim is to protect the East Greenland spawning component in the stock. However, the precise values of the increased M for the older age groups

20 | ICES IBPGCod Report 2018

are not fully documented and have not been investigated exhaustively in full population dynamics, feed-back analyses. Given the sensitivity of the assessment to this emigration and the associated levels and values of different M by age, these analyses are central in improving our understanding of varying Ms on the assessment. Also, the sensitivity of dealing with this emigration as partial fishing mortality instead of natural mortality in either Iceland or Greenland waters (see section below) could be investigated and evaluated further. The part of the stock migrating to Iceland is still vulnerable to commercial fisheries off Iceland, and will be subject to the fishing mortality in the Iceland cod stock and as a result will be included in Icelandic catches. Here the question can be raised whether this should be considered as a partial fishing mortality in Iceland, i.e. as a partial F, instead of being accounted for in the natural mortality? Furthermore, the extent of a certain proportion of the stock emigrating to Iceland is perceived as available for Greenland fishery, is not fully investigated. In this context sensitivity analysis of chang- ing / excluding partial F when calculating FMSY are necessary. Also, the sensitivity of the assessment outcome and results can be tested according to the inclusion of actual catches in Iceland (and accordingly partial fishing mortality) on the cod emigrating to Iceland both in Greenland and Iceland waters, i.e. consider those catches and fishing mortality directly in the stock assessment given available information on this instead of integrating this emigration (only) as increased M by age for the East Greenland stock component. Overall, many of the stock component aspects pertaining to the East Greenland stock assessment are not fully evaluated, especially with regard to robustness and sensitivity to different scenarios of partial Fs instead of increased Ms. The reviewers recommend that further investigations should include sensitivity analyses of different scenarios according to the above mentioned considerations and that these should be dealt with in a full future benchmark assessment of the East Greenland cod stock, with adequate time set aside to peruse the outcome of these analyses. To ensure continued evaluation of the cod stock regarding mixing and migration between stocks (areas), the reviewers recommend that tagging studies are continued and that these should preferably be extended into Icelandic waters. Furthermore, the reviewers recommend that ongoing genetic studies of the West Greenland stock components in relation to the offshore and (potential different) inshore stock components should be included in future analyses of stock units, delineation, mixing and migration. The reviewers note that the outcome of the two cod assessments give a nearly identical perception of stock development over time. With respect to SSB, this could indicate similar stock dynamics and a possibly strong connection between the stocks.

6.5.3 Survey and catch data In addition to the evaluation of historical tagging studies, it is recommended that further analyses be conducted focusing on the development of different cohort strengths of specific age groups in the different areas (East, West Inshore – potentially several inshore, West Offshore, Iceland) and for the different cod stock components. These analyses should respectively cover different time-series of catches and different surveys, and consider effort and gear selectivity. This is recommended for a future full benchmark assessment to further comparatively analyse stock entity, stock mixing and stock integration between the different Greenland stock components and the Iceland cod stock. The analysis should be based on both survey data and catch at age data to

Report of the InterBenchmark Protocol on Greenland Cod (IBPGCod) | 21

evaluate the consistency between strong and weak year classes in the different stock components. In the above context, it is also noted and strongly recommended that the age reading issues in relation to potential consistent differences between Greenland and German age readers are thoroughly investigated in a full scale otolith exchange program. The reviewers note that initiatives on the latter are ongoing. If historical age readings cannot be standardized the sensitivity of the assessment to these potential differences needs to be analysed in depth in a future full scale benchmark assessment. Also, this needs to be taken into consideration when trying to follow cohort strengths using different types of data time series (numbers in catch at age and age-based survey indices).

22 | ICES IBPGCod Report 2018

7 References

Hansen P.M. and Hermann F. 1953. Fisken og havet ved Grønland. Danmarks Fiskeri og Havundersøgelser. Nr. 15.

ICES. 2014. Report of the North-Western Working Group (NWWG), 24 April-1 May 2014, ICES HQ, Copenhagen, Denmark. ICES CM 2014/ACOM:07.902 pp. WD22

ICES. 2015. Report of the Joint ICES-MYFISH Workshop to consider the basis for FMSY ranges for all stocks (WKMSYREF3), 17–21 November 2014, Charlottenlund, Denmark. ICES CM 2014/ACOM:64. 156 pp.

ICES. 2016. Report of the North-Western Working Group (NWWG), 27 April- 4 May 2016, ICES HQ, Copenhagen, Denmark. ICES CM 2016/ACOM:08. 703pp. WD02

ICES. 2017a. Report of the North Western Working Group (NWWG), 27 April – 4 May 2017, Copenhagen, Denmark. ICES CM 2017/ACOM:08. 642 pp.

ICES 2017b. ICES fisheries management reference points for category 1 and 2 stocks. DOI: 10.17895/ices.pub.3036

ICES. 2017c. Report of the Benchmark on Sandeel (WKSand 2016), 31 October - 4 November 2016, Bergen, Norway. ICES CM 2016/ACOM:33. 319pp.

Storr-Paulsen, M., Wieland, K., Hovgård, H., and Rätz, H. J. 2004. Stock structure of Atlantic cod (Gadus morhua) in West Greenland waters: Implications of transport and migration. ICES Journal of Marine Science, 61: 972–982.

Report of the InterBenchmark Protocol on Greenland Cod (IBPGCod) | 23

Table 3.1.1: Recapture percentages of fish tagged in East Greenland waters by year class.

Year class West Greenland (%) East Greenland (%) Iceland (%) 2001 (N=10) 0% 80% 20% 2002 (N=31) 0% 13% 87% 2003 (N=43) 0% 33% 67% 2004 (N=23) 0% 43% 57% 2005 (N=3) 0% 33% 67% 2006 (N=6) 0% 50% 50% 2007 (N=54) 4% 48% 48% 2008 (N=25) 8% 28% 64% 2009 (N=11) 0% 9% 91% 2010 (N=4) 0% 50% 50%

Table 3.1.2: Distribution of recaptures of East Greenland marked cod at specific recapture ages

Age when recaptured West Greenland East Greenland Iceland

4 (N=6) 0% 83% 17% 5 (N=34) 3% 68% 30% 6 (N=46) 7% 28% 61% 7 (N=48) 0% 31% 69% 8 (N=51) 0% 22% 78% 9 (N=15) 0% 7% 93% 10+ (N=8) 0% 38% 63%

Table 3.4.1 SAM configuration for the East Greenland cod stock.

# Where a matrix is specified rows corresponds to fleets and columns to ages. # Same number indicates same parameter used # Numbers (integers) starts from zero and must be consecutive # $minAge # The minimium age class in the assessment 1

$maxAge # The maximum age class in the assessment 10

$maxAgePlusGroup # Is last age group considered a plus group (1 yes, or 0 no). 1

$keyLogFsta # Coupling of the fishing mortality states (nomally only first row is used). -1 -1 0 1 2 3 4 5 6 6 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1

$corFlag # Correlation of fishing mortality across ages (0 independent, 1 compound symmetry, or 2 AR(1) 2

$keyLogFpar

24 | ICES IBPGCod Report 2018

# Coupling of the survey catchability parameters (nomally first row is not used, as that is covered by fishing mortality). -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 0 1 2 3 4 5 6 7 7 -1 8 9 10 11 12 13 14 15 15 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1

$keyQpow # Density dependent catchability power parameters (if any). -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1

$keyVarF # Coupling of process variance parameters for log(F)-process (nomally only first row is used) 0 0 0 0 0 0 0 0 0 0 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1

$keyVarLogN # Coupling of process variance parameters for log(N)-process 0 1 1 1 1 1 1 1 1 1

$keyVarObs # Coupling of the variance parameters for the observations. -1 -1 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 -1 2 2 2 2 2 2 2 2 2 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1

$obsCorStruct # Covariance structure for each fleet ("ID" independent, "AR" AR(1), or "US" for unstructured). | Pos- sible values are: "ID" "AR" "US" "ID" "ID" "ID" "ID"

$keyCorObs # Coupling of correlation parameters must be specified if the AR(1) structure is chosen above. NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA -1 NA NA NA NA NA NA NA NA -1 -1 -1 -1 -1 -1 -1 -1 -1 -1

$stockRecruitmentModelCode # Stock recruitment code (0 for plain random walk, 1 for Ricker, and 2 for Beverton-Holt). 0

$noScaledYears # Number of years where catch scaling is applied. 0

$keyScaledYears # A vector of the years where catch scaling is applied.

$keyParScaledYA # A matrix specifying the couplings of scale parameters (nrow = no scaled years, ncols = no ages).

$fbarRange

Report of the InterBenchmark Protocol on Greenland Cod (IBPGCod) | 25

# lowest and higest age included in Fbar 5 10

$keyBiomassTreat # To be defined only if a biomass survey is used (0 SSB index, 1 catch index, and 2 FSB index). -1 -1 -1 3

$obsLikelihoodFlag # Option for observational likelihood | Possible values are: "LN" "ALN" "LN" "LN" "LN" "LN"

$fixVarToWeight # If weight attribute is supplied for observations this option sets the treatment (0 relative weight, 1 fix variance to weight). 0

Table 3.5.1. Estimated by eqSim reference points for the East Greenland cod stock

Btrigger Bpa Blim Fpa Flim Fp05 Fmsy unconstr Fmsy

14803 14802 10354 1.33 2.34 0.46 0.65 0.46

Table 3.5.2 Reference points for cod in East Greenland

Reference Value Technical basis point

Equilibrium scenarios using segmented regression and capped FMSY 0.46 by Fp05

Equilibrium scenarios prob (SSB < Blim) < 50% with stochastic FLIM 2.34 recruitment

1.645σ FPA 1.33 Flim / e , σ = 0.34

BLIM 10354 t. Average of SSB 2002, 2003 and 2004

1.645σ BPA 14803 t Blim×e , σ = 0.217

MSY Btrigger 14803 t. BPA

Table 4.5.1 Reference points for inshore West Greenland cod

Reference point Value Technical basis

FMSY 0.268 Equilibrium scenarios using Beverton&Holt SR Equilibrium scenarios prob (SSB < Blim) < 50% with FLIM 3.61 stochastic recruitment

1.645σ FPA 2.39 Flim / e , σ = 0.252

BLIM 4346 t. Breakpoint in segmented regression

1.645σ BPA 5983 t Blim×e , σ = 0.194

MSY Btrigger 5983 t. BPA

26 | ICES IBPGCod Report 2018

Table 4.4.1: SAM configuration for the West Greenland inshore stock.

$maxAge # The maximum age class in the assessment 10

$maxAgePlusGroup # Is last age group considered a plus group (1 yes, or 0 no). 1

$keyLogFsta # Coupling of the fishing mortality states (nomally only first row is used). -1 -1 0 1 2 3 4 5 6 6 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1

$corFlag # Correlation of fishing mortality across ages (0 independent, 1 compound symmetry, or 2 AR(1) 0

$keyLogFpar # Coupling of the survey catchability parameters (nomally first row is not used, as that is covered by fishing mortality). -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 0 1 2 3 4 5 -1 -1 -1 -1 6 7 8 9 10 11 -1 -1 -1 -1

$keyQpow # Density dependent catchability power parameters (if any). -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1

$keyVarF # Coupling of process variance parameters for log(F)-process (nomally only first row is used) 0 0 0 0 0 0 0 0 0 0 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1

$keyVarLogN # Coupling of process variance parameters for log(N)-process 0 1 1 1 1 1 1 1 1 1

$keyVarObs # Coupling of the variance parameters for the observations. 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 -1 -1 -1 -1 2 2 2 2 2 2 -1 -1 -1 -1

$obsCorStruct # Covariance structure for each fleet ("ID" independent, "AR" AR(1), or "US" for unstructured). | Pos- sible values are: "ID" "AR" "US"

Report of the InterBenchmark Protocol on Greenland Cod (IBPGCod) | 27

"ID" "ID" "ID"

$keyCorObs # Coupling of correlation parameters must be specified if the AR(1) structure is chosen above. NA NA NA NA NA NA NA NA NA NA NA NA NA NA -1 -1 -1 -1 NA NA NA NA NA -1 -1 -1 -1

$stockRecruitmentModelCode # Stock recruitment code (0 for plain random walk, 1 for Ricker, and 2 for Beverton-Holt). 0

$noScaledYears # Number of years where catch scaling is applied. 0

$keyScaledYears # A vector of the years where catch scaling is applied.

$keyParScaledYA # A matrix specifying the couplings of scale parameters (nrow = no scaled years, ncols = no ages).

$fbarRange # lowest and higest age included in Fbar 4 8

$keyBiomassTreat # To be defined only if a biomass survey is used (0 SSB index, 1 catch index, and 2 FSB index). -1 -1 -1

$obsLikelihoodFlag # Option for observational likelihood | Possible values are: "LN" "ALN" "LN" "LN" "LN"

$fixVarToWeight # If weight attribute is supplied for observations this option sets the treatment (0 relative weight, 1 fix variance to weight). 0

28 | ICES IBPGCod Report 2018

Figure 3.2.1. Weight at age by cohorts from the German survey (black line, data from 2008–2015), Greenland survey (red line, data from 2008–2016) and the commercial fishery (green line, data from 2006–2016). Uncertainties are SD.

SouthWest Greenland (1F): 2010-2015, age 4 2500 German Survey Greenland survey 2000

1500

1000 Total weight (g) weight Total

500

0 0 10 20 30 40 50 60 70 Length (cm)

Figure 3.2.2: Length-weight relationship in the Greenland and German survey in the south western part of the management area (NAFO 1F) 2010–2015 at age 4.

Report of the InterBenchmark Protocol on Greenland Cod (IBPGCod) | 29

Figure 3.3.1. Internal consistency in cohort tracking: GRL-GFS (Pamiut)

Figure 3.3.2. Internal consistency in cohort tracking: Ger(GRL)-GFS-Q4 (Walter Herwig).

30 | ICES IBPGCod Report 2018

Figure 3.5.1. Stock-recruitment relationship for the East Greenland cod. Year of recruits at age 1 are shown for each data point.

Report of the InterBenchmark Protocol on Greenland Cod (IBPGCod) | 31

Predictive distribution of recruitment for cod14

Segreg 0.83

300000 Ricker 0.12

250000 Bevholt 0.06

200000

Recruits

150000

100000

50000

0 20000 40000 60000 80000 100000 120000

SSB ('000 t)

Figure 3.5.2. East Greenland cod. Ricker, B&H and segmented regression fitted to the SR data for the East Greenland cod. (black line is segmented regression, dashed line is Ricker and dotted line is BH. Yellow line is average fit weighted by Buckland method.

Figure 3.5.3. East Greenland cod. Selectivity (relative F at age) for 4 recent periods.

32 | ICES IBPGCod Report 2018

Figure 4.5.1. Stock-recruitment relationship for the inshore West Greenland cod. Labels correspond to year (i.e. 1 year should deducted to get year class).

Figure 4.5.2. Inshore West Greenland cod. Selectivity (relative F at age) for 4 recent periods.

Report of the InterBenchmark Protocol on Greenland Cod (IBPGCod) | 33

Figure 4.5.3. Inshore West Greenland cod. Ricker, B&H and segmented regression fitted to the SR data for the East Greenaldn cod. (black line is segmented regression, dashed line is Ricker and dotted line is Beverton&Holt. Yellow line is average fit weighted by Buckland method.

34 | ICES IBPGCod Report 2018

Annex 1: List of participants

Name Responsibility Institute Country Email

Inshore Rasmus Greenland cod GNI Greenland [email protected] Hedeholm assessor East Greenland Anja Retzel GNI Greenland [email protected] cod coordinator Frank Farsø East Greenland GNI Greenland [email protected] Riget cod assessor Bjarki Invited Expert Haf og vatn Iceland [email protected] Elvarsson Rasmus Invited Expert DTU Aqua Denmark [email protected] Nielsen

Arved Staby External Chair IMR Norway [email protected]

GNI / DTU Greenland Jesper Boje SAM [email protected] Aqua /Denmark Marie Storr- ICES Chair DTU Aqua Denmark [email protected] Paulsen ICES Sarah Louise Professional ICES Denmark [email protected] Millar Secretary Lisbeth Royal Schönemann Industry Denmark [email protected] Greenland -Paul

Report of the InterBenchmark Protocol on Greenland Cod (IBPGCod) | 35

Annex 2: Agenda

Monday the 8th of January

9.00 Presentations of participants and housekeeping, role of the different participants. 9.15 Input data – East Greenland cod East Greenland cod – we have several WD available on the share point as the data was discussed at the skype meeting 1/11 – 2017 (minutes available on the share point) I would prefer to focus on the new information (ToR a) 11.00 Input data – West Greenland cod Again we have several WD available on the share point and as the data was discussed at the skype meeting 1/11 – 2017 (minutes available on the share point) I would prefer to focus on the new information (ToR a). 13.00 Lunch – 1 hour 14.00 Models We would like to have a presentation on the different possible assessment runs. Sensi- tivity runs can be needed if new or changed data has been applied (ToR b). 18.00 End of day

Tuesday the 9th of January

9.00 Short review of final decisions from yesterday 10.00 Short term forecast Discus the settings in the short term forecast. Do we have reference points? (ToR b and c) 13.00 Lunch 14.00 SA Write stock annex with all final settings. 17.30 Recommendation on future work 18.00 Closing of meeting

36 | ICES IBPGCod Report 2018

Annex 3: List of stock annexes

Last Stock ID Stock name Link updated

Cod (Gadus morhua) in NAFO Sub- January cod.21.1 area 1, inshore (West Greenland cod.21.1 2018 cod) Cod (Gadus morhua) in ICES Subarea January cod.2127.1f14 14 and NAFO Division 1.F (East cod.2127.1f14 2018 Greenland, South Greenland)

Report of the InterBenchmark Protocol on Greenland Cod (IBPGCod) | 37

Annex 4: Working documents

WD01 Description of input data for SAM for East Greenland cod WD02 Description of input data for SAM for Inshore cod WD03 Analysis of 2003-2016 tagging data from Greenland waters WD04 Procedure for estimating the proportional WD05 Inshore West Greenland Cod SAM assessment WD06 East Greenland Cod SAM assessment

ICES Benchmark, 8.-9. January 2018, Working doc 01:

Assessment input for Cod (Gadus morhua) in ICES Subarea 14 and NAFO Division 1F (East Greenland, South Greenland)

*Anja Retzel, Jesper Boje and Frank Riget Greenland Institute of Natural Resources, Nuuk, Greenland *[email protected]

Catch in numbers Information on historic catch in numbers 1973-1995 for West and East Greenland is from a compilation in Hovgård and Wieland, 2008. From this data combined catch-at-age for East Greenland and South Greenland (Nafo division 1F) was calculated (Retzel, 2015).

In the period 1996-2004 no sampling was conducted due to the very low catches. The model setup used total catch weight per year as a “survey” for this period and the model was set to utilize this aggregated information.

Since 2005 a directed cod fishery started in both East and South Greenland. Annual catch-at-age are obtained from sampling on board fishing vessels in this recent period.

Tabel 1 outlines the sampling since 2005.

Landing mean weight Information on historic weight in catches 1973-1995 for West and East Greenland is from a compilation in Hovgård and Wieland, 2008. From this data combined weight-at-age for East Greenland and South Greenland (Nafo division 1F) was calculated (Retzel, 2015).

1996-2004: There was no sampling in this low fishery period. The weight-at-age applied in these years is an average of weight-at-age in the period 1973-1995 and 2005-2016 (Figure 1).

Since 2005 a directed cod fishery started in both East and South Greenland. Annual weight-at-age are obtained from sampling on board fishing vessels in this recent period (table 1).

For the 1 and 2 years old an assumed constant mean weight at age of 0.1 kg and 0.3 kg was applied for the entire time series.

Discard mean weight Since no discarding is assumed to take place same values as landing mean weight was applied.

Stock mean weight There are two offshore bottom trawl surveys covering the cod stock in East and South Greenland (see Survey section). The Greenland survey is in July and August, whereas the German survey is in October and November. The weight-at-age pattern in the two surveys are different with higher weight-at-age in the German survey than the Greenland survey especially for older age groups (figure 2). The catch-at-age pattern in the two surveys does not indicate an age reading issue (figure 3). The weight-at-age in the fishery corresponds overall to the weight-at-age in the Greenland survey (figure 2). Further 80% of the fishery takes place from January-August (ICES 2017).

We decided to use the weight-at-age from the Greenland survey as stock mean weight in the assessment because: - The German weight-at-age differ significantly from the weight-at-age in the Greenlandic survey and the fishery. - The German survey takes place outside the main part of the fishery season and is therefore likely not representative of the fishable biomass

As the Greenlandic survey started in 2008 we used an average of the weight-at-age from 2008-2016 for the period 1973-2007 (Figure 4)

Catch mean weight Since discarding is observed to be insignificant landings mean weight and catch mean weight are set equal.

Maturity ogive Based on maturity determination of 1557 cod from ICES Div 14b and NAFO 1F caught in spawning period April to May during the period 2007-2016, the average for this period was applied for the entire time series 1973-2016 (Figure 5). The maturity ogive was estimated by a general linear model (GLM) with binomial errors. L50 was estimated to 5.00 years (SE = 0.06).

Surveys Data from two annual offshore trawl surveys were used. The Greenland Shrimp and Fish survey (GRL-GFS) and the German groundfish survey (Ger(GRL)-GFS-Q4). Both are stratified random bottom-trawl survey.

The Greenland survey started in 1992 in West Greenland covering the area south of 72o00’N and depth from 0-600 m. Since 2008 East Greenland south of 67°00’ N was included in the survey (ICES, 2017). The Greenland survey covering East Greenland and NAFO Division 1F since 2008 was applied as tuning for this assessment. Approximately 125 hauls are taken each year. The survey provides number at age for the age groups 1 to 9. SAM input do not allow + groups. Indices and cohort tracking is provided in Fig. 6 and 8.

The German survey has been conducted since 1982 (ICES, 2017). The survey covers both East (area south of 67°00’ N) and West Greenland (area south of 67°00’ N). The German survey covering East Greenland and NAFO division 1F were applied as tuning for this assessment. The survey provides number at age for age groups 1 to 9 from East Greenland and NAFO division 1F. Numbers of hauls in East Greenland and NAFO division 1F were initially ca. 110 per year but were reduced from the early 1990s to 50–60 per year. Indices and cohort tracking is provided in Fig. 7 and 9.

Figure 10 illustrates the consistency between the two surveys.

More detailed information on the surveys are available in the stock annex for Cod in 14 and 1F.

Natural mortality Natural mortality is set to 0.2 for all ages for all years for this stock as for other cod stocks in the North Atlantic since no accurate estimates presently are available.

Emigration Analysis of tagging experiments 2003-2016 in East Greenland suggest an emigration of fish into Icelandic waters (Hedeholm, WD to this meeting, IBPCOD17). This trend is increasing with age such that 17% of tagged fish age 4 are recaptured in Iceland while this proportion is about 80% for ages 8-10. This emigration rate is sought implemented in the assessment as an additional mortality at 0.1 for age 6 and 0.2 for older age groups (7-10). In the SAM input data this value is found combined with M as an addition.

F before spawning In the assessment it is assumed that spawning takes place 1 January and fishing mortality before spawning is therefore set to 0. M before spawning In the assessment it is assumed that spawning takes place 1 January and natural mortality before spawning is therefore set to 0.

Landing fraction Discarding is assumed insignificant in the fishery and landing fraction is therefore set to 1.

Addition During a webex on the 1st of November 2017, certain points were raised regarding this working document. This addition has been added to address some of these issues.

 Include uncertainties in the weight estimates from the German and Greenland survey and the commercial fishery (figure 2): Figure 1a is based on raw data of weight at age and are not weighted with abundance by age and area as figure 2. The data file starts in 2008 for both the German and Greenland survey.

Figure 1a: Weight at age by cohorts from the German survey (black line, data from 2008-2015), Greenland survey (red line, data from 2008-2016) and the commercial fishery (green line, data from 2006-2016). Uncertainties are SD.

 Compare recruitment index for the Icelandic cod with the index for recruitment in the east Greenland cod: The recruitment index of age 3 cod in East Greenland was compared with the recruitment index of age 3 cod in Iceland (figure 2a). The relationship is significant but with a low R-value.

Figure 2a: Consistency of age 3 cod between East Greenland and Iceland.

 Description of the fishery to address a better understanding of the assumption of no discard. Further information about effort over time: Vessels in the offshore fisheries are vessels above 75BT/120BT and restricted to the area more than 3 nm off the baseline. Trawlers and longliners exclusively conduct the fishery and longliners have in recent years increased the amount taken to 40% of the total catch in 2016 (Retzel 2017). Mesh size in the trawl fishery is 140 mm.

The East Greenland area has been subject to several area closures in recent years. In 2008 fishing north of N63°00’ was not allowed in order to protect the potential spawning segments, especially on Kleine Banke. In 2009–2010 the delimitation was at N62°00’ and additionally NAFO 1F was closed in 2010 primarily to protect the relatively strong incoming year classes. The high effort in 2008 and 2009 was caused by increased catches in NAFO 1F combined with decreasing CPUE. With the closure of 1F in 2010 effort decreased (Figure 3a). Since 2011 the area north of 63oN has been closed in spawning season (April and May).

Figure 3a: Effort expressed as catch/standardized CPUE in East and South Greenland (ICES 14 and NAFO 1F).

References Hedeholm, R. 2017 Analysis of 2003-2016 tagging data from Greenland waters as it relates to assessment of the East Greenland offshore stock and the West Greenland inshpre stock. WD 03, IBPCOD. Hovgård, H., Wieland, K. 2008. Fishery and Environmental Aspects Relevant for the Emergence and Decline of Atlantic Cod (Gadus morhua) in West Greenland Waters. Resiliency of Gadid Stocks to Fishing and Climate Change Alaska Sea Grant College Program. AK-SG-08-01. ICES (2015). Report of the Benchmark Workshop on Icelandic Stocks (WKICE). ICES CM 2015/ACOM: 31. ICES (2017). Cod (Gadus morhua) in ICES subarea 14 and NAFO Division 1.F (East Greenland, South Greenland). In Report of the North Western Working Group (NWWG). ICES CM 2017/ACOM:08. Pp. 404-442. Retzel, A. (2015). Combined VPA data for Cod in East Greenland and NAFO div 1F. WD01 in Report of the Benchmark Workshop on Icelandic Stocks (WKICE). ICES CM 2015/ACOM: 31. Retzel, A. (2017). Greenland commercial data for Atlantic cod in East Greenland offshore waters for 2016. ICES North Western Working Group (NWWG) April 27- May 4, 2017, WD 02.

Table 1: Sampling from South and East Greenland offshore cod fisheries from 2005-2016. Count (total number) and stations of aged and length measured cod by year. * from survey.

Aged cod Length measured cod Year Count Stations Count Stations 2005 306* 26* 4150 26 2006 70 / 732* 8 / 53* 3554 14 2007 559 / 216* 9 / 6* 18531 109 2008 146 / 882* 4 / 69* 1763 11 2009 886* 81* 8777 43 2010 430 / 689* 20 / 62* 13073 63 2011 1947 123 21781 159 2012 1665 132 12250 83 2013 2082 147 19565 158 2014 1828 138 13966 171 2015 1140 191 12059 153 2016 885 122 14469 168

Figure 1. Landing weight at age.

2002 cohort 2006 cohort 16 14

14 12

12 10 10 8 8 6

Weight (kg) Weight (kg) 4 4 2 2 0 0 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 Age Age 2003 cohort 2007 cohort 12 9 8 10 7 8 6 5 6

4 Weight (kg) Weight (kg) 4 3 2 2 1 0 0 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 Age Age 2004 cohort 2008 cohort 12 8 7 10 6 8 5 6 4

3 Weight (kg) Weight (kg) 4 2 2 1 0 0 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 Age Age 2005 cohort 2009 cohort 12 6

10 5

8 4

6 3 Weight (kg) Weight (kg) 4 2

2 1

0 0 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 Age Age German survey Greenland survey Fishery German survey Greenland survey Fishery

Figure 2: Mean weight at age for yearclasses 2002-2009 in the German (blue line), Greenland (red line) survey and the fishery (green line). 2008 German survey 2012 30000 14000 Greenland survey 25000 12000 10000 20000 8000 15000 6000 10000

4000

Totalabundance Totalabundance

5000 2000

0 0 1 2 3 4 5 6 7 8 9 1 2 3 4 5 6 7 8 9 2009Age 2013Age 12000 50000 45000 10000 40000 35000 8000 30000 6000 25000 20000 4000

15000

Totalabundance Totalabundance 10000 2000 5000 0 0 1 2 3 4 5 6 7 8 9 1 2 3 4 5 6 7 8 9 2010Age Age 14000 70000 2014

12000 60000

10000 50000

8000 40000

6000 30000

20000

4000 Totalabundance Totalabundance

2000 10000

0 0 1 2 3 4 5 6 7 8 9 1 2 3 4 5 6 7 8 9 2011Age Age 18000 35000 2015

16000 30000 14000 25000 12000 10000 20000 8000 15000 6000

10000 Totalabundance Totalabundance 4000 5000 2000 0 0 1 2 3 4 5 6 7 8 9 1 2 3 4 5 6 7 8 9 Age Age Figure 3: Abundance index (‘000) by age in the German (blue line) and Greenland (red line) survey from 2008-2015.

Figure 4. Stock weight at age.

Figure 5: Maturity at age for all years.

Figure 6. The Greenland survey, GRL-GFS (Pamiut) indices

Figure 7. The German survey Ger(GRL)-GFS-Q4 (Walter Herwig) survey indices.

Figure 8. Internal consistency in cohort tracking: GRL-GFS (Pamiut)

Figure 9. Internal consistency in cohort tracking: Ger(GRL)-GFS-Q4 (Walter Herwig)

Figure 10. Between fleet consistency; Paamiut versus Walter Herwig survey by age.

ICES Benchmark, 8.-9. January 2018, Working doc 02:

Assesment data for Cod (Gadus morhua) in NAFO Subarea 1, inshore (West Greenland cod)

*Anja Retzel, Jesper Boje, Rasmus Hedeholm and Frank Riget Greenland Institute of Natural Resources, Nuuk, Greenland *[email protected]

Catch in numbers Catch at age are obtained from port sampling representing all divisions 1A to 1F along West Greenland from 1976 to 2016 (table 1). In the period with very low catches (1990’ies) sampling was restricted to Nuuk (1D). Further in 1998 and 2001 no sampling was conducted. Length measurements are conducted as random samples while age/weight/maturity sampling is length stratified.

Tagging experiments indicate that the offshore and inshore cod are generally separated but that they mix in the coastal area, especially at ages 4-6 before the offshore cod starts their spawning migration (ICES 2017a, Storr-Paulsen et al. 2004). The fishery in the fjords and especially in the coastal areas therefore catch a mix of offshore West Greenland cod, EastGreenland/Icelandic cod and inshore cod. The proportion is not yet possible to quantify but ongoing genetic analyses will reveal this.

Landing mean weight A well-balanced sampling of the Greenland coastal fleets catches has always been impeded by the geographical conditions, i.e. the existence of many small landing sites separated along the over 1,000 km coast. Cod has also been landed without head, which hinder otolith sampling. Limited age information therefore exists from the commercial fishery. Mean-weight-at-age in the landings are therefore primarily calculated survey sampling and set equal to stock mean-weight-at-age.

A more comprehensive description of the fishery and sampling procedures are provided in stock annex for inshore cod in NAFO subarea 1 (ICES 2017b).

Discard mean weight Since no discarding is assumed to take place same values as landing mean weight are used.

Stock mean weight Mean-weight-at-age in the landings are primarily calculated survey sampling and set equal to stock mean- weight-at-age.

Catch mean weight Due to insufficient sampling from the fishery catch mean weight is set equal to stock mean weight derived from surveys.

Maturity ogive Maturity information from the early period of the assessment is only available for November in 1987 (n=484 cod). This sampling is disqualified with respect to spawning time since November is outside the observed spawning period (April-May). However, the sample is taken in the bottom of the fjord where there is minimal mixing with the offshore stock (Storr-Poulsen et al. 2004). From a more recent period (2007-2015) maturity information is available from the spawning season (n=3326 cod).

The maturity ogive for the two periods was estimated by a general linear model (GLM) with binomial errors. The ogive for the two periods are estimated to be different: L50 from 1987 was 5.07 years (SE=0.18), and for 2007-2015 L50 was estimated to 4.32 years (SE =0.04).

It was decided to use the years with very low catches (600-800 tons) as transition years between the two maturity ogives. The maturity ogive for the period 1976-1999 was set constant to the estimated 1987 ogive. For the remaining period (2000-2016) the maturity ogive was set constant based on the maturity ogive for the period 2007-2015. The reason for not applying different maturity ogive for each year is due to high variation in number of samples between years that results in noisy data.

Even though the maturity ogive for the period 1976-1999 is based on few fish caught outside spawning season it was decided to use this as this maturity ogive is supported by earlier maturity ogives from the 1930’ies that have a similar L50 (Hansen, 1949).

Surveys A multi meshed gillnet survey designed to target juvenile cod age 2 and 3 yrs old in the inshore area in West Greenland has been conducted annually since 1987. The objective of the survey is to assess the abundance and distribution of pre recruiting cod (ICES 2017b) since cod recruits to the fishery at age 4. However, given the different ways of being caught in a gillnet other than being gilled the selectivity is not entirely dome shaped but elongated towards larger fish. Therefore, gillnet catches of older fish ages 1-5 were included in the data set.

Historically three areas were covered: north west (Sisimiut, NAFO division 1B), mid west (Nuuk, NAFO division 1D) and south west (Qaqortoq, NAFO division 1F). South Greenland has only been covered in the period 1987-1995, 1998, 2000 and 2007-2009 and due to very scarce data from this survey this area is not included as a tuning fleet.

Due to local stock dynamics for each fjord complex the survey is split into two survey indices as follows: - NAFO division 1B, survey index for the period 1987-1998, 2002-2007 and 2010-2016. In 1999-2001 and 2008-2009 no survey was conducted. - NAFO division 1D, survey index for the period 1987-2016 except in 2002 and 2007 where no survey was conducted.

Indices and cohort tracking in the two areas are provided in Figures 1 - 4. Figure 5 illustrates the consistency in the survey between the two areas; there is generally little agreement in cohort tracking between the two areas. This might be due to different population dynamics in NAFO division 1B (Sisimiut area) and 1D (Nuuk area).

More detailed information on the survey is available in the stock annex for Cod in NAFO subarea 1 (ICES 2017b).

Genetic investigations of the cod caught in the survey in NAFO division 1B in 2016 showed a clear dominance of fish from the West Greenland inshore stock component (72%). The West Greenland offshore component (20%) is the only other stock with a contribution above 10% (figure 6). The survey indices are therefore primarily an expression of the status of the inshore cod stock than the fishable stock which is more a mix of the different cod stocks. A master thesis, investigating the mix of the different stocks in the fishery, found that the West Greenland inshore stock component constituted 50% of the commercial catch in NAFO division 1B in 2014 (Henriksen 2015). In 2017 an extensive collection of the fishery with samples taken over the entire year at different locations will investigate the mix of the different stocks in the fishery on a regional and temporal scale.

Natural mortality Natural mortality is set to 0.2 for all ages for all years for this stock as for other cod stocks in the North Atlantic since no accurate estimates presently are available .

Emigration To account for the observed emigration of the big 1984 year class, emigration for age 6 and older has been set to 0.2 for this year-class. In the SAM input data this value is found as an addition to M.

F before spawning In the assessment it is assumed that spawning takes place 1 January and fishing mortality before spawning is therefore set to 0.

M before spawning In the assessment it is assumed that spawning takes place 1 January and natural mortality before spawning is therefore set to 0.

Landing fraction No discarding is observed to take place in the inshore fishery and landing fraction is therefore set to 1.

Addition During a webex on the 1st of November 2017, certain points were raised regarding this working document. This addition has been added to address some of these issues.

 Table 1 gives a fine overview of numbers of age and length sampled. Is it possible to have numbers of harbor visits or a description of the commercial sampling design?

Numbers of sampling days at the factories in each NAFO division has been added in table 1. Cod in other NAFO divisions than 1D (the capital Nuuk) are length measured on designated sampling trips lasting between 2-5 days, usually in late summer where the fishery peaks. In Nuuk (NAFO 1D) cod are measured 1- 3 days each month.

Length and weight at age are primarily based on cod collected on surveys as cod, especially in recent years, have been landed without head and always gutted. A conversion factor (of 1.23) of length measured cod without head is used to obtain total length. In addition some factories do not allow the head of the cod to be cut open.

A more comprehensive description of the fishery and sampling procedures are provided in stock annex for inshore cod in NAFO subarea 1 (ICES 2017b).

 Investigate if the recruitment in the inshore component is matching the recruitment in the east Greenland and or the Icelandic cod stock.

The recruitment index of age 3 cod in inshore West Greenland was compared with the recruitment index of age 3 cod in Iceland (figure 1a). The relationship is not significant.

Figure 1a: Consistency of age 3 cod between Inshore West Greenland and Iceland.

The index of age 1-9 cod in the inshore area in West Greenland was compared with the index of age 1-9 cod in East Greenland (figure 2a). The relationship for all ages are significant.

Figure 2a: Consistency of age 1-9 cod between Inshore West Greenland and East Greenland.

 Partial F has a strange pattern since 2010 were older fish has a large increase in F and age 4 and 5 are decreasing. Can this be explained? Mean length in the catches has increased from 47 and 48 cm in 2008 and 2009 to 58 cm in 2014 (figure 3a). In 2015 and 2016 mean length in the catches were 57 and 56 cm respectively. Poundnet has been the dominating gear and believed to constitute around 80% of the catch. Poundnets are shorebased and usually catches cod between 4-6 yrs as older and larger cod tend to stay deeper in the watercolumn. Since 2012 it has been possible to split the catch into gear based on landing saleslips (table 1a) and the proportion taken by poundnets dropped to 47% in 2014 where especially jigs took a higher proportion of the catch. Jigs are better able to catch larger cod than poundnets.

Table 1a: Proportion (%) of total catch by gear. Year Poundnet Jig Gillnet Longline 2012 74% 7% 12% 7% 2013 62% 16% 14% 8% 2014 47% 25% 18% 10% 2015 50% 22% 15% 13% 2016 62% 14% 13% 11% 2008, Mean Length 47 cm 800

600

400

200

Raised numbers ('000) numbers Raised 0 30 35 40 2009, Mean45 Length50 48 cm55 60 65 70 Length (cm) 800

600

400

200

Raised numbers ('000) numbers Raised 0 30 35 40 2010, Mean45 Length50 53 cm55 60 65 70 Length (cm) 800

600

400

200

Raised numbers ('000) numbers Raised 0 30 35 40 2011, Mean45 Length50 52 cm55 60 65 70 Length (cm) 800

600

400

200

Raised numbers ('000) numbers Raised 0 30 35 40 2012, Mean45 Length50 53 cm55 60 65 70 Length (cm) 800

600

400

200

Raised numbers ('000) numbers Raised 0 30 35 40 2013, Mean45 Length50 53 cm55 60 65 70 Length (cm) 800

600

400

200

Raised numbers ('000) numbers Raised 0 30 35 40 2014, Mean45 Length50 58 cm55 60 65 70 Length (cm) 800

600

400

200

Raised numbers ('000) numbers Raised 0 30 35 40 45 50 55 60 65 70 Length (cm)

Figure 3a: Length distribution in the inshore fishery in the period 2008-2014.

References

Hansen, P.M. (1949). Studies on the biology of the cod in Greenland waters. Reprint from Rapports et Procés-Verbaux des Réunions vol. CXXIII.

Henriksen, O (2015). Genetic insight into the population composition of two regional inshore mixed stocks of Atlantic cod (Gadus morhua) in West Greenland. Master thesis, Technical University of Denmark, National Institute of Aquatic Resources (DTU Aqua), Section for Marine Living Resources.

ICES (2017a). Cod (Gadus morhua) in NAFO subarea 1, inshore (West Greenland cod). In Report of the North Western Working Group (NWWG). ICES CM 2017/ACOM:08. Pp. 369-403.

ICES (2017b). Stock Annex: Cod (Gadus morhua) in NAFO Subarea 1, inshore (West Greenland cod). http://www.ices.dk/sites/pub/Publication%20Reports/Stock%20Annexes/2017/cod.21.1_SA.pdf

Storr-Paulsen, M., Wieland K., Hovgård H. and Rätz H-J. 2004. Stock structure of Atlantic cod (Gadus morhua) in West Greenland waters: implications of transport and migration. ICES Journal of Marine Science. 61: 972–982.

Table 1: Sampling from West Greenland inshore cod fisheries from 1976-2016. Number (count) of aged and length measured cod by year and NAFO divisions combined. (X) indicate number of habour visits (days) in each NAFO division.

Aged cod Length measured cod Year NAFO div Count NAFO div Count 1976 1D 1182 1D (8) 2860 1977 1B,D 1783 1B (5),D (9) 3361 1978 1D,E 1172 1D (15),E (1) 4829 1979 1B,C,D,E 3669 1B (4),C (2),D (15) ,E (1) 15019 1980 1B,D,E,F 4420 1B (1),D (16) ,E (6),F (1) 14768 1981 1B,C,D,E,F 2181 1B (3),C (4),D (11),E (5),F (5) 21452 1982 1B,D,E 1173 1B (1),D (4),E (3) 9308 1983 1D,E,F 3616 1D (13),E (6),F (2) 15292 1984 1B,C,D,E,F 2682 1B (3),D (6),E (1) 13057 1985 1C,D,E,F 840 1D (3),E (1),F (3) 3347 1986 1A,B,C,D,E,F 2101 1B (1),D (7),E (5),F (2) 5940 1987 1B,C,D,E,F 1983 1B (1),C (3),D (4),E (4),F (3) 6553 1988 1B,C,D,E,F 3035 1B (6),C (4),D (11),E (5),F (6) 20002 1989 1B,C,D,E,F 3328 1B (5),C (2),D (10),E (7),F (5) 20048 1990 1B,D,E,F 2884 1B (14),C (2),D (11),E (3),F (4) 9990 1991 1B,D,E,F 1777 1B (3),D (6),E (1),F (4) 7525 1992 1B,D 487 1B (4),C (1),D (10),E (3) 5610 1993 1A,B,D 1119 1B (1),C (1),D (4),F (1) 4397 1994 1B,D,F 696 1C (1),D (4),F (1) 2585 1995 1B,C,D 799 1B (1),D (3) 1183 1996 1B,C,D 737 1C (7),D (2) 1908 1997 1B,C,D 478 1C (1),D (2) 1013 1998 No sampling 1999 1D 201 1D (1) 1137 2000 1A,D,F 145 1D (1) 375 2001 No sampling 2002 1B,D,F 427 1F (9) 10157 2003 1B,D 617 1B (5),D (8) 4402 2004 1B,D 655 1C (1),D (3) 1585 2005 1B,D 600 1B (2),D (1) 1820 2006 1B,D 349 1D (1),E (17),F (4) 9496 2007 1A,B,C,D,E,F 967 1A (6),B (9),C (8),D (9),E (6),F (8) 19297 2008 1A,B,D,F 2189 1A (6),B (11),D (5),F (4) 5355 2009 1A,B,C,D,F 1849 1D (12),F (7) 11506 2010 1A,B,D,E,F 2709 1A (4),B (3),D (6),E (4),F (9) 11590 2011 1A,B,D,E,F 1685 1A (12),B (4),C (7),D (13),E (1),F (3) 9565 2012 1A,B,D,F 1027 1A (3),B (1),C (24),D (20),E (2),F (6) 13335 2013 1A,B,C,D,E,F 1200 1A (5),B (2),C (13),D (7),E (4),F (1) 11255 2014 1A,B,D,E,F 1137 1A (7),C (6),D (10),E (5),F (2) 6446 2015 1A,B,D 912 1A (11),B (14),C (9),D (8),E (4) 21854 2016 1A,B,D 1070 1A (14),B (14),C (7),D (11) 21816

Figure 1. The survey indices for area 1B.

Figure 2. The survey indices for area 1D.

Figure 3. Internal consistency in cohort tracking for survey in NAFO div 1B.

Figure 4. Internal consistency in cohort tracking for survey in NAFO div 1D.

Figure 5. Between area consistency; NAFO 1D versus NAFO 1B survey by age.

Iceland inshore

East Greenland/Iceland Offshore

West Greenland inshore

West Greenland offshore

Figure 6: The stock specific survey index for NAFO division 1B (Sisimiut area) in 2016.

Working document 03

Analysis of 2003-2016 tagging data from Greenland waters as it relates to assessment of the East Greenland offshore stock and the West Greenland inshore stock

Rasmus Hedeholm, Greenland Institute of Natural Resources, Kivioq 2, 3900 Nuuk [email protected]

Introduction

There are three offshore surveys in Greenland starting in 1982, 1991 and 2008 (ICES 2017). For all years, data are available disaggregated by age. Catch data go back more than 50 years, also disaggregated by age. In spite of this, there is no quantitative analytical assessment for the offshore cod in either West or East Greenland. One major issue hampering the development of any models is the variable and unquantifiable migration from west to east (Bonanomi et al., 2016). This means that in East Greenland there is an inflow of fish from West Greenland. These fish either stay or continue to Iceland. This means that the natural mortality is “inflated” by migration. Natural mortality for is often set at 0.2 as a default in cod assessment models (e.g. ICES 2017). In this document we argue that this is not valid in East Greenland and we use recent tagging experiments to support this view which has also been shown previously (Hansen and Hermann 1953) (Storr-Paulsen et al., 2004). We also illustrate the extent of immigration from West Greenland to the East Greenland area.

For the West Greenland inshore area, the challenge in developing any useful models is similar. The area is supposedly a nursery area for both the West Greenland offshore stock and the East Greenland/Iceland offshore stock. Upon reaching maturity, these fish leave the area (Bonanomi et al., 2016) and the natural mortality appears higher based on catch at age data. If this is unaccounted for in assessment models, the models will interpret this as a result of fishing mortality, thereby overestimating the fishing pressure which in turn directly affects the scientific advice. In this document we will examine if recent tagging data suggest that there is sound scientific basis for setting the natural mortality higher than the usual default of 0.2 at least for some yearclasses.

The questions addressed are in summary:

1. Is there substantial migration from East Greenland to Iceland across yearclasses? 2. Is there substantial migration from West Greenland to East Greenland across yearclasses? 3. Is there emigration from the West Greenland inshore area to offshore areas? 4. Is there migration from the West Greenland offshore area to the inshore area?

Materials and methods

The tagging procedure is described in detail in (Hansen, 1949; Storr-Paulsen et al., 2004) and Hedeholm et al. 2013 and will not be repeated here. The 2003-2016 data analyzed in this document have been supplemented when needed. Hence, all the presented recaptures have an associated marked length, recapture length and each individual fish is assigned to a year class. Naturally, not all fish were aged upon recapture, and recapture lengths are often highly uncertain. Hence, both recapture length and missing ages were estimated based on a typical growth pattern of Greenland cod and age-length keys routinely developed as part of surveys.

In this document we use the ICES delineations of ‘East’, ‘West’ and ‘Inshore’. Hence, ‘Inshore’ is NAFO regions 1A-1F in West Greenland, ‘West’ is NAFO divisions 1A-1E in West Greenland and ‘East’ is offshore East Greenland and NAFO 1F in West Greenland.

Results

In total there were 1558 recaptured cod. The overall recapture pattern in seen in Fig. 1.

The vast majority of fish were tagged during summer. To allow for possible migration to take place, I only consider fish that had at least 100 days at liberty. This would allow them to migrate during winter before being recaptured. This reduced the recaptures to 1096. These are the only ones considered in further analysis.

Question 1 Is there substantial migration from East Greenland to Iceland across yearclasses?

Of the 208 recaptured cod that were tagged offshore in East Greenland waters, 4 (2%) were recaptured in West Greenland, 71 (34%) were caught in East Greenland and 133 (64%) were caught within the Iceland EEZ (Fig. 2). Doing this exercise by yearclass shows that all yearclasses have similar substantial migration to Iceland (Table 1, Fig. 3)

Table 1: Recapture percentages of fish tagged in East Greenland waters by yearclass.

If only recaptures of cod smaller than 60 cm (N=28) are considered, only 18% were recaptured in Iceland, while 71% of the fish recaptured when larger than 60 cm were recaptured in Iceland waters. If the same approach is applied to age groups, then percentage recaptured in Iceland increases with age (Table 2)

Table 2: Distribution of recaptures of East Greenland marked cod at specific recapture ages.

Age when recaptured West Greenland East Greenland Iceland 4 (N=6) 0% 83% 17% 5 (N=34) 3% 68% 30% 6 (N=46) 7% 28% 61% 7 (N=48) 0% 31% 69% 8 (N=51) 0% 22% 78% 9 (N=15) 0% 7% 93% 10+ (N=8) 0% 38% 63%

If the time at liberty is set at 500 days rather than 100, the number of recaptures of fish tagged in East Greenland is reduced to 120. Of these fish, 1 (1%) were recaptured in West Greenland, 29 (24%) were recaptured in East Greenland and 90 (75%) were recaptured in Iceland.

Conclusions regarding question 1:

There is substantial migration from East Greenland to Iceland. This holds true for all yearclasses analyzed here, and the migration appears to start at age 5, with the proportion of the recaptures at Iceland increasing with age. There is every reason to assume that these conclusions are also valid for the period prior to that presented here (i.e. Hansen and Hermann 1953; Storr-Paulsen et al. 2004). The exact level of migration in terms of the proportion of a yearclass that migrates to Iceland from East Greenland is difficult to quantify. Hence the fishing mortality and effort is unknown in East Greenland and there might be differences in the return percentages from the vessels. However, based on the numbers shown here, at least 50% of a cohort appears to go to Iceland.

Question 2: Is there substantial migration from West Greenland to East Greenland across yearclasses?

I still only consider fish with more than 100 days at liberty.

In total, 266 cod tagged between 2003 and 2016 in West Greenland were recaptured (Fig. 4). Of these, 217 (82%) were recaptured in West Greenland, 21 (8%) in East Greenland and 28 (11%) were recaptured in Iceland waters. The vast majority of recaptures in West Greenland were from the inshore region (67%). If I only use cod marked in the offshore region, there are 82 recaptured cod and of these 48% were recaptured in West Greenland, 18% in East Greenland and 34% in Iceland. Of the recaptures in East Greenland, only five are taken outside NAFO area 1F. When the 82 recaptures are assessed based on age at recapture (Table 3), the proportion recaptured in Iceland increase with age, suggesting that East Greenland is a transit area, which is also suggested by the overall pattern across ages. The transit time is however difficult to determine from the present data set.

Table 3: Distribution of recaptures of West Greenland offshore marked cod at specific recapture ages.

Age when recaptured West Greenland East Greenland Iceland 4 (N=7) 86% 14% 0% 5 (N=12) 58% 25% 17% 6 (N=22) 50% 23% 27% 7 (N=18) 39% 17% 44% 8 (N=15) 53% 13% 33% 9 (N=6) 0% 17% 83% 10+ (N=2) 0% 0% 100%

The pattern is similar across yearclasses (Table 4, Fig. 5). All yearclasses except for 2002 had substantial migration to Iceland, and of the 2002 yearclass, only one cod had been at liberty for more than a year and none of the fish were more than six years old suggesting that a spawning migration could still occur at a later time. The recapture proportions in East Greenland were generally low, suggesting that East Greenland is a transit area for the majority of the migrating cod. The 2009 has not yet been recaptured at ages older than 7, but it seems to have just started its spawning migration; i.e. there a substantial recaptures in East Greenland and a lower proportion from Iceland.

Table 4: Recapture percentages of fish tagged in West Greenland offshore waters by yearclass.

Year class West Greenland East Greenland Iceland 2002 (N=6) 100% 0% 0% 2003 (N=5) 60% 20% 20% 2006 (N=10) 20% 10% 70% 2007 (N=20) 40% 5% 55% 2008 (N=16) 50% 13% 38% 2009 (N=19) 47% 42% 11% 2010 (N=4) 50% 25% 25%

Conclusions regarding question 2:

Similar to cod tagged in East Greenland, cod tagged in West Greenland offshore waters are predominantly recaptured in Iceland waters, especially when reaching ages older than 6 years. This is consistent across yearclasses. Based on the recaptures by age (Table 4) it seems that the East Greenland area is a transit area and this conclusion is supported by survey data with the cod becoming progressively larger moving northwards along in East Greenland (ICES 2017). The present data does not allow for an evaluation of the East Greenland residence time, and therefore not on the inflow of fish form West to East. But that fish migrate to East Greenland is irrefutable, and should be taken into account when setting natural mortality values in modelling of yearclass dynamics. Question 3: Is there emigration from the West Greenland inshore area to offshore areas?

I still only consider fish with more than 100 days at liberty.

In total 250 cod were tagged in the West Greenland inshore area and recaptured (Fig. 6). The recaptures were predominantly from the inshore area (98%) with limited migration to East Greenland (1%) and Iceland (1%). The five cod that left the inshore area were all tagged in the southern part of West Greenland (NAFO 1F) and were from the 2002 and 2003 yearclasses.

Within the inshore area, there were generally three tagging areas: Qaqortoq, Nuuk and Sisimiut (Fig. 6). Within the individual areas there was very little migration outside the same fjord system. For instance, in the Nuuk fjord 108 cod were tagged and recaptured. Of these only four cod migrated to another part of the inshore area.

Conclusions regarding question 3:

In the period of the present data, there has been limited fishery in Greenland offshore waters, especially in West Greenland. This could skew the picture with lower than expected recaptures in offshore waters. However, the distribution of times at liberty, suggest that the cod remain for years inshore (Fig. 7). However, 76% of the recaptures were made at ages 5 and younger and only 10% at age 7 or older. The lack of mature recaptures could be related to the inshore fleet selectivity. The main gear is pound nets, and these are thought to have a relatively low selectivity at older ages. For instance, age 5 dominates in catches in consecutive years although the yearclass strength varies (ICES 2017). The very low number of offshore/Iceland recaptures also suggest that the larger mature fish stay inshore but are unavailable to the fishery. Storr-Paulsen et al. (2004) show that cod tagged in the coastal region are a mix of inshore and offshore cod, and migrate accordingly. This is not the case here. In the present study there is less coastal tagging than in Storr-Paulsen et al. (2004), but the limited West Greenland offshore fishery also limits the recapture effort. The tentative conclusion is, that the inshore fish tend to stay inshore, or they do at least not to any great extent migrate to East Greenland/Iceland. The emigration to the West Greenland offshore area is unknown, but based on the recent resurgence of the stock, it seems unlikely that this is driven solely by the inshore stock, but rather involves an input from the offshore area. When modelling stock dynamics, this should be considered for instance by increasing M for older age groups coincident with the onset of maturation. Question 4: Is there migration from the West Greenland offshore area to the inshore area?

Of the 89 recaptures of cod tagged in offshore West Greenland waters, 8 were recaptured in the West Greenland inshore region (Fig. 8). These recaptures are in the southwestern part of Greenland, and were from the coastal zone, i.e. not in the deeper parts of the fjords. A similar conclusion was reached by Storr-Paulsen et al. (2004). No recent tagging has been done on the northern part of the West Greenland offshore distributional area, and any possible migration from the West Greenland offshore banks to the inshore area cannot be addressed. Based on historical tagging data, some migration from these areas to the coastal part of the inshore region should be expected.

Conclusions regarding question 4:

The present data does not allow for any in depth analyses of this question, as to few cod were tagged in the appropriate areas. The only conclusion that can be made is that based on the available data, some migration to the coastal zone is likely, and this would also be in line with previous similar conclusions (Storr-Paulsen et al. 2004; Hovgård and Wieland 2008). Addition

During a webex on the 1st of November 2017, certain points were raised regarding this working document. This addition has been added after the fact to address some of these issues.

Between 2003 and 2016 6809 cod were tagged in East Greenland and 210 were recaught. Two-thirds of the recaptures were in Iceland and one-third in Greenland. These are only cod that had more than 100 days at liberty. In an attempt to use this information to quantify the level of migration we did the following:

- According to the SAM baserun for the East Greenland cod available at the time of writing this F from 2003 to 2016 averaged 0.15 in Greenland. From the currently available assessment for Iceland cod, F averaged 0.41 in Iceland. F reflects the proportion of the stock that is caught, and integrats both stock size and effort. - We weighted the number of recaptures (72 in Greenland and 133 in Iceland) by these F

values; i.e. 72*(Ficeland/FGreenland) = 197. - The proportion of a yearclass that stays in Greenland waters is therefore estimated as 197/(197+133) = 0.6. Accordingly, the share migrating to Iceland waters is 0.4.

To try an alternative approach to estimate migration, we used effort data.

- The Greenland effort data was derived from CPUE and trawl catch data (Effort = catch/CPUE). - For Iceland, Icelandic scientist provided the trawl effort in NW and SW in hours. - For both areas we did the following: o Summed the effort from 2003-2016. o Averaged the SSB from 2003-2016 (SAM base run and current Iceland assessment). o Calculated an effort pr. t. (2.12 hours in Greenland and 3.77 in Iceland). o Calculated the number of hours it takes to catch a tagged fish: #recaptures/effort pr. t. - This procedure uses direct estimates to account for the difference in population size and exact data on effort. - The result is that 51% of a yearclass stays in Greenland, and 49% migrate to Iceland.

Conclusion

The current base run for the East Greenland cod is configured in such a way, that approximately 50% of each cohort migrates from East Greenland to Iceland. Based on tagging data this seems a valid assumption.

References

Bonanomi, S., Overgaard Therkildsen, N., Retzel, A., Berg Hedeholm, R., Pedersen, M. W., Meldrup, D., Pampoulie, C., et al. 2016. Historical DNA documents long-distance natal homing in marine fish. Molecular Ecology, 25: 2727–2734.

Hansen, P. M. 1949. Studies on the biology of the Cod in Greenland waters. ICES Rapports et procès- verbaux, 123: 3–85.

Hansen P.M. and Hermann F. 1953. Fisken og havet ved Grønland. Danmarks Fiskeri og Havundersøgelser. Nr. 15.

Hedeholm R., Retzel A. and Boje J. 2013. A short review of available mark-recapture studies on Atlantic cod (Gadus morhua) in Greenland with the objective to separate in- and offshore components. ICES, North Western Working Group 2013. WD no. 12.

Hovgård H. and Wieland K. 2008. Fishery and environmental aspects relevant for the emergence and decline of Atlantic cod (Gadus morhua) in West Greenland waters. In: Resiliency of gadid stocks to fishing and climate change. Ed: Kruse G., Drinkwater K, Ianelli J. et al. Pages 89-110.

ICES. 2017. Report of the North Western Working Group (NWWG), 27 April – 4 May 2017, Copenhagen, Denmark. ICES CM 2017/ACOM:08. 642 pp.

Figure 1: All mark-recapture positions from 2003-2016. Lines show the shortest distance between mark and recapture events. The division between East and West Greenland is shown. The green line is the EEZ delineation.

Figure 2: Mark and recapture positions of all recaptured cod tagged in East Greenland

2001 YC 2002 YC 2003 YC

2004 YC 2005 YC 2006 YC

2007 YC 2008 YC 2009 YC

2010 YC

Figure 3: Mark and recapture positions for the YC’s 2001-2010. Only cod tagged in East Greenland are shown.

Figure 4: Mark and recapture positions of all recaptured cod tagged in West Greenland

2002 YC 2003 YC

2006 YC 2007 YC

2008 YC 2009 YC

2010 YC

Figure 5: Mark and recapture positions for YC’s between 2001 and 2010. Only cod tagged in offshore West Greenland are shown.

Figure 6: Mark and recapture positions of all recaptured cod tagged in West Greenland inshore waters.

160

140

120

100

Count

0 0 500 1000 1500 2000 2500

Time at liberty Figure 7: Distribution of times at liberty for fish tagged in West Greenland inshore waters.

Figure 8: Mark and recapture positions of all recaptured cod tagged in West Greenland offshore waters and recaptured in West Greenland inshore waters.

Working document 04

Procedure for estimating the proportional contribution of the different cod stocks to the West Greenland inshore gill net survey in Sisimiut in 2016.

Rasmus Hedeholm, Greenland Institute of Natural Resources, Kivioq 2, 3900 Nuuk [email protected]

In 2016 cod were sampled from the survey and in total 117 cod were genetically assigned to one of four stocks: West Greenland inshore, West Greenland offshore, East Greenland/Iceland offshore or Iceland inshore. Not all stations in the survey (N=58) were sampled (N=30). Hence, it was not possible to estimate the proportional contribution of the four stocks by station. Instead nearby stations were grouped together (Fig. 1) and the genetic composition from all those stations combined were assumed to reflect the overall catches in the given group.

Fig. 1. All stations from the 2016 inshore survey in Sisimiut. The colors and circles indicate which stations were grouped together for genetic similarity. The genetic composition and sample sizes within each group can be seen in table 1.

Table 1. Sample size (N) within each group and the share (%) of each stock found in the genetic analysis within each group.

Group N West Greenland, West Greenland East Greenland/Iceland Iceland inshore offshore offshore inshore 1 37 57% 27% 11% 5% 2 32 31% 31% 9% 0% 3 40 78% 15% 8% 0% 4 5 80% 20% 0% 0% 5 3 100% 0% 0% 0%

The index value for each station in each group was then multiplied by the stock specific proportion for that group. The stock specific survey indices were then averaged across all stations as this is the way things are tabulated for the index that is currently used in the assessment for this stock. This produced an index for each stock, and we estimated the share of the total index each stock contributed.

Using this procedure, the survey index by stock is shown in figure 2.

Iceland inshore

East Greenland/Iceland Offshore

West Greenland inshore

West Greenland offshore

Figure 2: The stock specific survey index for the Sisimiut area in 2016.

There is a clear dominance of fish from the West Greenland inshore stock component (72%). The West Greenland offshore component (20%) is the only other stock with a contribution above 10%.

Because migration is an issue in the area, size specific stock proportions may change. The most noticeable change is, that no East Greenland/Iceland offshore fish are found in the samples in cod smaller than 40cm (Fig. 3). The legal limit for the commercial fishery is 40cm, and hence the patterns demonstrated here may not apply to the commercial catches. 1,0

0,8

0,6

0,4 Proportion of Proportion sample

0,2

0,0 15-20 20-25 25-30 30-35 35-40 40-45 45-50 50-55 Size category (cm)

Iceland Inshore East Greenland/Iceland Offshore West Greenland inshore West Greenland offshore

Figure 3: Sample composition in 5 cm length categories for the whole Sisimiut area combined. IBPGCod, January 2018, Working document 05

A SAM assessment of the West Greenland Inshore cod stock (cod.21.1) Rasmus Hedeholm*, Jesper Boje and Frank Riget Greenland Institute of Natural Resources *[email protected]

Introduction The Atlantic cod (Gadus morhua) in Greenland waters is comprised primarily of three spawning stocks:

1 ) West Greenland inshore; 2 ) West Greenland offshore; and 3 ) East Greenland offshore.

Until 2011, ICES provided advice for all three stock components combined. Since 2012 a separate advice has been given for the inshore stock component. This separation was based on extensive tagging data (Storr-Paulsen et al. 2004) as well as the general understanding of stock dynamics in Greenland. Recent genetic studies supports this division (Therkildsen et al. 2013). This document presents the results of the state-space model (SAM) (Nielsen and Berg, 2014) for the West Greenland Inshore cod stock (ICES stock cod.21.1).

Input The input data for the SAM assessment model and consistency analyses are described in details in Working document #2 for the IBPGCod 2017 benchmark and are shown in table 1. Landing mean weights were set equal to Catch mean weight, no discarding was assumed. Fishery (F) and natural maturity (M) before spawning were set to zero assuming spawning takes place 1 January.

Assessment Main settings/assumptions Table 2 shows the SAM configuration of the base run. Several SAM runs with different input or other configurations were compared with the base run. In all runs, the landing fraction was assumed to be one (i.e. no discarding). This is consistent with the overall impression of the fishery. There has not been an enforced TAC constraint in any years and the gears typically allow for the release of small cod. The two gill net surveys in NAFO 1B (Sisimiut) and NAFO 1D (Nuuk) were used for tuning, and combined to one index in another run. Because of the survey gear catchability, only ages 1-5 were used in the model. The survey is designed to target 2-3 year old cod (20-30 cm).

Recruitment We assumed a random-walk recruitment for the short-term forecast. Because of the survey design, the recruitment is tracked in both surveys for at least two years prior before a yearclass enters the fishery. The stock-recruitment relationship was best estimated using a Beverton-Holt (0.48) rather than Ricker (0.2). Assuming a random walk recruitment produced a slightly better model fit (∆AIC of 2) and was implemented in all runs. IBPGCod, January 2018, Working document 05

Catchabilities The catchability in both surveys was assumed different for ages 1 to 5. The surveys target relatively small cod, which at these life stages maintain distinct length frequencies. Therefore, we assume that they are not caught equally well in the survey and let the model estimate the parameters.

Selectivity There is no catch of ages 1 and 2 in any years and the fishing mortality coefficients were not estimated for these ages. The coefficients were estimated independently for ages 3 to 8, and set to a common coefficient for ages 9, and 10+. There is a relatively steep selectivity curve for the fishery, but a plateau is not reached. Instead the older age groups are not caught as well as younger age groups, and this is particularly true for the main fishing gear; pound net.

Fbar The Fbar range of 4 to 8 years old was applied as these age groups constitute the main part of the catches. On average, the 9 and 10+ old cod constitute less than 1% of the total catch, and no more than 5% in any single year.

Variance The fishing mortality variance was assumed equal for all ages (this was only relevant for the catches). To allow for some uncertainty in the recruitment independently of other ages, the variance for the population numbers was assumed to be the same for ages 2 to 10+ but different for age 1. The survey variance-at-age was assumed to be the same for all included ages (1-5). The surveys targets smaller cod, and there is no reason to assume the variance is different for any of these age group.

Covariance and correlation The covariance structure for the fishery was set to be independent as no knowledge was available to suggest another structure. Similarly, it was assumed that there was no coupling of correlation parameters in the surveys. The correlation across ages was set to be independent. Using the auto-correlation options provides an improved model fit (∆AIC of 18), but this comes at a cost of a more variable F estimate. We do not think F changes this much from year-to-year and opted for the independent option despite a worse model fit.

Model diagnostics Parameter estimates The model fits the catch data relatively poorly with high uncertainty around both F and recruitment (Figure 1). In addition, the fit to the catches is poor, being most pronounced in years with high catches, including the recent period (Figure 2). The lack of an overall fit and the specific case with the catches are most likely both linked to the mixing of stocks in the area and the noise it adds to the data. Genetic data from the survey in 2016 shows that the inshore cod stock was the main driver of the index (>75%). The important stations in this survey are located relatively deep in the fjords, which is in some contrast to the fishery, which is more coastal. The limited available genetic data from the fishery documents a large input from other stocks to the catches (Henriksen, 2015). We interpret this, as cod that IBPGCod, January 2018, Working document 05

enter the area for feeding, but do not show up in the survey – i.e. they are most abundant in the coastal zone and not in the survey area, which also targets juveniles. The models relies on both surveys, and hence the model has no way of explaining where the fish in the catches come from (i.e. few recruits), and “solves” this by not fitting to the catches. Hence, the model should be considered an inshore stock model, and not a model representing the inshore area. Unfortunately, data on the proportional contribution of the different stocks are not available, and no stock split can be applied to the time series. The catches are known with a high degree of certainty, and we did a model run, where we reduced the variance allowed on the catches from 2006-present. This forces the model to track the catches very closely. This produced a worse model fit than the base run (∆AIC of 5) but produced the expected outcome – a higher current SSB. Our interpretation is that since ICES is not asked to produce an area based assessment, the run does not provide a meaningful alternative. Although we cannot quantify the contribution from other stocks, the fact that mixing occurs should be reflected in the assessment as best possible.

Residuals No apparent pattern are observed in the catch residuals. The survey residuals tended to have blocks of positive and negative residuals, particularly the NAFO 1D survey (Figure 3). Relative high process residuals are found both in case of N and F but appear to be random without any clear pattern (Figure 4).

Stock summary Fishing mortality The historical F trajectories are shown in Figure 5. From 1976 to the early 1990’s F increased from 0.6, peaking at 1.33 in 1992. F then decreased and has been at fluctuating around 0.7 since 2000 and is estimated at 0.77 in 2016.

SSB The SSB has a peak in 1979, but this is followed by decade of decline and a period of almost zero SSB (Figure 6). Around 2003 the biomass started to increase, and has increased since, except in 2016. The 2016 estimate of SSB is 28103 t.

Recruitment In the early part of the time series, there are three distinct recruitment peaks (Figure 7). These are associated with the 1977, 1979 and 1984 yearclasses. Yearclasses of a similar size have not been observed since, although the recent 2003 and 2009 yearclasses are relatively large. The most recent six year period have seen decreasingly smaller yearclasses and the 2016 value is approximately half of the time series mean (6481 and 14530, respectively).

Leave one out In order to assess the impact of the different surveys on the stock trajectories, the assessment was re-run leaving out the survey time series one at a time (Figure 8). F was insensitive to leaving out both the NAFO 1B or 1D survey. The recent estimate of SSB was in between the two surveys. Hence, the NAFO 1B survey reduces the current SSB while the reverse is true for the NAFO 1D survey. This is also seen for the recruitment, indicating that the surveys are currently showing opposing trends, which is also reflected in IBPGCod, January 2018, Working document 05

the models ability to track the catches. Without the NAFO 1B survey, which is indicating a low recent recruitment, the model estimates a higher catch.

Retrospective analysis The robustness of the assessment is validated by a 5 year retrospective analysis (Figure 9). In all years, F lies below the current estimate (Mohn’s rho =-0.142) but all estimates are within the confidence limits. Recruitment is also underestimated compared to the recent estimate (Mohn’s rho =-0.342), but again the estimates are within confidence limits. The SSB estimates are more robust, and there is only a slight tendency to underestimate (Mohn’s rho =-0.083). Accepted range for Mohn’s rho according to Hurtado- Ferro et al. (2015) is for long-lived species 0.20 to -0.15, and the present patterns are within these limits. However, ICES have not yet defined acceptable ranges for retrospective patterns.

Exploratory runs using other configurations The SAM model was run with different configurations in order to assess the effects on the outcome (Table 3, Figure 10). Table 3 shows the main statistics of some of the most relevant options together with the name of the stock as named on stockassessment.org. The run combining the survey indices was the only run that provided a significant improvement in model fit (∆AIC of 314). This does not only reflect a better fit, but also that less “penalty” is included as fewer parameters are estimated (10 vs 16). Combining the indices required filling out missing values in the numbers-at-age matrices or rejecting several years. We opted for keeping all data and estimating the missing values. In NAFO 1B, 1999-2001 and 2008-2009 were missing and in NAFO 1D, 2001 and 2007 were missing. We estimated the missing values by linear regressions on agey vs. agey+1. The missing values were estimated by projecting a known survey index-at- age either forward or backward. In preparation for the IBPGCod, there was a webex. Here it was suggested to try a model run separately for the two survey areas (NAFO 1B and 1D). This was not possible as the catch matrix is not area specific.

Short term forecast Based on the base run, we made three forecast scenarios. The first run was a status quo run. Here, F was fixed in all years at the same level as in 2016 (0.774). This resulted in a reduced SSB (40% in 2019) and reduced catches. If F is reduced to 0 after the first year, the SSB increased by 71% in 2019. Fishing at FMSY (0.28, see later) from 2017 onwards maintained the SSB at the same level in all years, but reduced the catches from ~18000 t to ~8000 t.

Reference points The estimation of reference points follows the ICES Reference Points Guidance, January 2016. The estimation was done using the simulation R-programme EqSim developed by D.C.M. Miller, which works directly on a specified SAM fit. The simulation settings for the Stock-Recruitment relationship were as follows. The number of simulations was set to 2000. No years were omitted as all years fall within a relatively narrow SSB/R range (Figure 11). Certain yearclasses are heavily influenced by an inflow from offshore stocks (e.g. 1984) but the contribution is unknown. We have kept them in the analysis, as they do not appear to be clear outliers. IBPGCod, January 2018, Working document 05

The segmented regression was the preferred SSB-R relationship, although the Beverton-Holt fit was better. However, when implemented in the model, the performance was best with the segmented regression. There was a significant one-year lagged autocorrelation in recruitment, and this was included in the simulation (Figure 13). Based on the segmented regression, Blim was set to 5000 t (point estimate of 4049 t), which appears reasonable when considering the trajectory of the SSB- R relationship. The simulations were done with 200 runs, scanning an F range from 0 to 3 divided into 100 intervals. Bpa was calculated from the formula Bpa = Blim * exp(1.645 *σ), where σ is SD of ln(SSB) in 2016 - here estimated by SAM to be 0.221. Accordingly, Bpa is estimated at 7192 t. Flim was estimated by simulation using the above values of Blim and Bpa, setting Fcv=0, Fphi=0 and SSBcv = 0 (no assessment and advice noise) and with no Btrigger. The range of years was from 1991 to 2015. F50 is then Flim, here estimated at 1.14 (Table 5). Fpa is calculated from the formula Fpa = Flim * exp(-1.645 * σ), where σ is SD of ln(F) in 2015 here estimated by SAM to be 0.259. Fpa is then 0.74 (Table 5).

MSY reference points (MSY Btrigger and FMSY).

FMSY was initially estimated as the F that maximizes the median long-term yield in the simulation under constant F exploitation. The default values of cvF = 0.212, phiF = 0.423 and cvSSB = 0 were applied to the simulation. The initial FMSY was estimated at 0.27, which is below the estimated Fpa. The final FMSY is

estimated by a simulation using the default Fcv, Fphi, the estimated Blim, Bpa and Btrigger which are equal to Bpa. The final FMSY estimate was 0.28. The precautionary principle states that if FMSY > F05, FMSY should be reduced to F05, but this is not the case here (Table 5).

Conclusion

This assessment is based on a model that fitted the observations reasonably well, and the estimated trends in stock size and fishing mortality seem valid. The model parameters are associated with high uncertainty. This is to some extent caused by the periodic occurrence of other cod stocks in the area. Based on genetics, the stock-mixing issue seems to be the least pronounced in the survey data, and the model relies relatively more on surveys than catches. As a result, the model fits the catch data poorly, especially in years when catches are large. In years when the inflow of cod from other areas is believed to have been particularly large, the modelled catches constitute between 45% and 80% of the total landings. The limited genetic data suggests that these are realistic values (Henriksen, 2015).

The assessment model utilizes all available data, which is a great improvement compared to the previous advisory procedure for this stock. Hence, the model provides a useful tool for assessing the inshore stock.

IBPGCod, January 2018, Working document 05

References

Henriksen O. 2015. Genetic insights into the population composition of two regional inshore mixed stocks of Atlantic cod (Gadus morhua) in West Greenland. Master Thesis. Technical University of Denmark. 82 pp.

Nielsen A, Berg CW. 2014. Estimation of time-varying selectivity in stock assessments using state-space models. Fisheries Research. 158: 96-101.

Storr-Paulsen M., Wieland K., Hovgård H. and Rätz H.-J. 2004. Stock structure of Atlantic cod (Gadus morhua) in West Greenland waters: implications of transport and migration. ICES Journal of Marine Science 61: 972-982.

Therkildsen N.O., Hemmer-Hansen J., Hedeholm R.B., Wisz M.S., Pampoulie C., Meldrup D., Bonanomi S., Retzel A. Olsen S.M. & Nielsen E.E. 2013. Spatiotemporal SNP analysis reveals pronounced biocomplexity at the northern range margin of Atlantic cod Gadus morhua. Evolutionary Applications 6: 690–705. IBPGCod, January 2018, Working document 05

Table 1. Input files to SAM runs Catch in Numbers (thousands) 1 2 1976 2016 1 10 1 0 0 2508 924 556 287 38 31 11 7 0 0 467 5437 1100 883 179 7 142 46 0 0 97 1262 9904 132 68 7 3 0 0 0 323 2297 2380 8281 170 96 4 14 0 0 4343 4334 1646 806 6492 106 29 37 0 0 87 15793 5225 725 499 2906 61 17 0 0 3013 1587 6309 1545 798 152 610 154 0 0 229 16877 1381 4352 368 139 65 75 0 0 520 4451 9269 346 634 18 42 12 0 0 5 2400 1028 2229 196 363 14 78 0 0 286 178 896 460 721 16 102 38 0 0 5503 1334 228 710 340 1084 46 265 0 0 419 15588 150 51 39 90 161 12 0 0 15 5962 23956 271 46 2 93 176 0 0 212 2997 15403 6732 33 11 7 16 0 0 124 6022 4910 5695 330 0 0 0 0 0 8 2408 2344 452 139 46 13 5 0 0 28 661 575 206 34 41 10 7 0 0 22 1468 342 62 45 8 11 1 0 0 1 834 773 37 5 0 0 0 0 0 2 165 362 130 25 3 1 0 0 0 1 397 311 179 31 0 0 0 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 0 0 87 465 105 1 0 0 0 0 0 0 4 228 336 7 0 0 0 0 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 0 0 532 2243 657 29 9 1 0 0 0 0 152 581 1547 258 51 16 15 11 0 0 530 1669 1095 228 37 3 0 0 0 0 1392 2408 944 186 36 10 4 0 0 0 4256 3363 680 22 0 0 0 0 0 0 1944 7910 1010 116 38 13 8 4 0 0 1176 5012 2793 319 36 6 2 0 0 0 487 3540 2372 194 13 3 0 4 0 0 301 1091 2475 1524 141 32 21 27 0 0 129 2929 2567 1480 255 90 12 7 0 0 735 1725 2681 850 182 21 13 13 0 0 143 3806 2477 1083 361 115 67 9 0 0 40 1389 4024 2292 328 168 103 52 0 0 20 2007 5683 3010 1338 133 9 8 0 0 32 2146 9701 5732 1179 239 57 7 IBPGCod, January 2018, Working document 05

Table 1 continued Mean Weight in Catch (kilograms) 1 3 1976 2016 1 10 1 0.042 0.100 0.811 1.114 1.662 2.738 3.226 4.062 5.831 12.747 0.042 0.200 0.674 1.382 2.201 2.649 3.322 6.363 3.920 4.616 0.042 0.250 0.668 0.965 1.801 2.472 2.845 3.649 4.733 7.313 0.042 0.350 0.800 1.309 2.111 3.153 3.696 4.371 6.861 8.007 0.042 0.500 0.753 1.017 1.884 2.580 3.823 4.107 5.715 7.902 0.042 0.190 0.308 1.045 1.576 2.190 2.590 4.029 3.529 7.831 0.042 0.138 0.844 1.118 1.604 2.605 3.875 5.495 5.425 6.278 0.042 0.200 0.552 0.937 1.337 2.039 2.795 3.378 4.218 4.109 0.042 0.175 0.624 0.967 1.385 1.869 2.469 3.286 3.985 4.433 0.100 0.100 0.420 0.754 1.134 1.662 2.065 2.669 3.486 4.337 0.027 0.262 0.582 1.248 1.414 2.043 2.689 3.188 3.893 8.401 0.040 0.109 0.872 1.187 2.043 2.302 2.963 3.294 4.114 5.107 0.040 0.088 0.659 1.106 1.251 1.691 2.677 3.046 3.478 5.111 0.032 0.116 0.558 0.855 1.308 1.821 3.161 4.252 4.397 5.862 0.042 0.090 0.649 0.889 1.031 1.452 2.614 3.765 5.846 10.868 0.052 0.107 0.802 0.966 1.088 1.146 1.595 3.964 4.682 7.313 0.020 0.115 0.567 0.869 1.028 1.697 1.849 2.845 3.253 4.402 0.042 0.107 0.585 0.820 1.239 1.830 1.802 2.873 3.976 8.777 0.037 0.158 0.430 0.883 1.359 1.706 3.103 3.900 4.976 16.271 0.048 0.156 0.768 0.930 1.093 1.799 2.493 4.130 6.490 7.313 0.042 0.149 0.501 0.814 1.201 2.176 2.955 4.151 5.507 6.577 0.034 0.115 0.560 0.956 1.397 1.767 1.830 3.239 4.682 7.313 0.036 0.182 0.368 0.833 1.459 1.585 1.714 3.485 3.140 7.313 0.040 0.180 0.739 0.895 1.240 2.254 3.387 4.556 4.682 7.313 0.040 0.171 0.642 1.121 1.453 2.378 2.621 2.409 4.682 7.313 0.042 0.138 0.606 0.945 1.296 1.832 2.470 3.485 4.455 7.474 0.044 0.191 0.708 0.999 1.397 2.318 1.884 2.853 3.560 3.356 0.032 0.203 1.046 1.391 2.069 2.565 3.300 3.988 5.095 6.958 0.044 0.232 0.988 1.236 1.584 2.158 3.149 6.132 6.443 9.737 0.038 0.217 0.811 1.106 1.728 2.415 2.810 6.955 6.443 9.737 0.032 0.199 0.724 0.944 1.560 3.102 4.522 9.931 9.931 9.737 0.044 0.145 0.703 0.950 1.543 2.574 4.003 5.136 6.541 10.250 0.037 0.150 0.615 0.884 1.406 2.332 3.709 5.463 7.263 9.737 0.062 0.155 0.641 0.898 1.461 2.348 4.055 5.132 5.869 14.181 0.034 0.209 0.659 0.976 1.517 2.120 3.204 4.872 6.929 9.796 0.032 0.178 0.657 0.918 1.466 2.013 3.305 5.396 7.527 10.366 0.031 0.176 0.764 1.109 1.810 2.700 3.554 5.964 6.910 14.345 0.032 0.121 0.766 1.258 1.623 2.235 3.059 3.636 4.114 7.430 0.032 0.121 0.690 1.226 1.935 2.534 3.408 5.327 5.746 7.766 0.028 0.121 0.783 1.131 1.754 2.548 3.378 4.924 7.829 12.922 0.030 0.120 0.505 1.042 1.474 2.233 3.120 3.716 3.914 9.304 IBPGCod, January 2018, Working document 05

Table 1 continued Mean Weight in Stock (kilograms) 1 4 1976 2016 1 10 1 0.042 0.100 0.811 1.114 1.662 2.738 3.226 4.062 5.831 12.747 0.042 0.200 0.674 1.382 2.201 2.649 3.322 6.363 3.920 4.616 0.042 0.250 0.668 0.965 1.801 2.472 2.845 3.649 4.733 7.313 0.042 0.350 0.800 1.309 2.111 3.153 3.696 4.371 6.861 8.007 0.042 0.500 0.753 1.017 1.884 2.580 3.823 4.107 5.715 7.902 0.042 0.190 0.308 1.045 1.576 2.190 2.590 4.029 3.529 7.831 0.042 0.138 0.844 1.118 1.604 2.605 3.875 5.495 5.425 6.278 0.042 0.200 0.552 0.937 1.337 2.039 2.795 3.378 4.218 4.109 0.042 0.175 0.624 0.967 1.385 1.869 2.469 3.286 3.985 4.433 0.100 0.100 0.420 0.754 1.134 1.662 2.065 2.669 3.486 4.337 0.027 0.262 0.582 1.248 1.414 2.043 2.689 3.188 3.893 8.401 0.040 0.109 0.872 1.187 2.043 2.302 2.963 3.294 4.114 5.107 0.040 0.088 0.659 1.106 1.251 1.691 2.677 3.046 3.478 5.111 0.032 0.116 0.558 0.855 1.308 1.821 3.161 4.252 4.397 5.862 0.042 0.090 0.649 0.889 1.031 1.452 2.614 3.765 5.846 10.868 0.052 0.107 0.802 0.966 1.088 1.146 1.595 3.964 4.682 7.313 0.020 0.115 0.567 0.869 1.028 1.697 1.849 2.845 3.253 4.402 0.042 0.107 0.585 0.820 1.239 1.830 1.802 2.873 3.976 8.777 0.037 0.158 0.430 0.883 1.359 1.706 3.103 3.900 4.976 16.271 0.048 0.156 0.768 0.930 1.093 1.799 2.493 4.130 6.490 7.313 0.042 0.149 0.501 0.814 1.201 2.176 2.955 4.151 5.507 6.577 0.034 0.115 0.560 0.956 1.397 1.767 1.830 3.239 4.682 7.313 0.036 0.182 0.368 0.833 1.459 1.585 1.714 3.485 3.140 7.313 0.040 0.180 0.739 0.895 1.240 2.254 3.387 4.556 4.682 7.313 0.040 0.171 0.642 1.121 1.453 2.378 2.621 2.409 4.682 7.313 0.042 0.138 0.606 0.945 1.296 1.832 2.470 3.485 4.455 7.474 0.044 0.191 0.708 0.999 1.397 2.318 1.884 2.853 3.560 3.356 0.032 0.203 1.046 1.391 2.069 2.565 3.300 3.988 5.095 6.958 0.044 0.232 0.988 1.236 1.584 2.158 3.149 6.132 6.443 9.737 0.038 0.217 0.811 1.106 1.728 2.415 2.810 6.955 6.443 9.737 0.032 0.199 0.724 0.944 1.560 3.102 4.522 9.931 9.931 9.737 0.044 0.145 0.703 0.950 1.543 2.574 4.003 5.136 6.541 10.250 0.037 0.150 0.615 0.884 1.406 2.332 3.709 5.463 7.263 9.737 0.062 0.155 0.641 0.898 1.461 2.348 4.055 5.132 5.869 14.181 0.034 0.209 0.659 0.976 1.517 2.120 3.204 4.872 6.929 9.796 0.032 0.178 0.657 0.918 1.466 2.013 3.305 5.396 7.527 10.366 0.031 0.176 0.764 1.109 1.810 2.700 3.554 5.964 6.910 14.345 0.032 0.121 0.766 1.258 1.623 2.235 3.059 3.636 4.114 7.430 0.032 0.121 0.690 1.226 1.935 2.534 3.408 5.327 5.746 7.766 0.028 0.121 0.783 1.131 1.754 2.548 3.378 4.924 7.829 12.922 0.030 0.120 0.505 1.042 1.474 2.233 3.120 3.716 3.914 9.304 IBPGCod, January 2018, Working document 05

Table 1 continued Proportion Mature at Year Start 1 6 1976 2016 1 10 1 0 0.011 0.045 0.171 0.475 0.799 0.946 0.987 0.997 0.999 0 0.011 0.045 0.171 0.475 0.799 0.946 0.987 0.997 0.999 0 0.011 0.045 0.171 0.475 0.799 0.946 0.987 0.997 0.999 0 0.011 0.045 0.171 0.475 0.799 0.946 0.987 0.997 0.999 0 0.011 0.045 0.171 0.475 0.799 0.946 0.987 0.997 0.999 0 0.011 0.045 0.171 0.475 0.799 0.946 0.987 0.997 0.999 0 0.011 0.045 0.171 0.475 0.799 0.946 0.987 0.997 0.999 0 0.011 0.045 0.171 0.475 0.799 0.946 0.987 0.997 0.999 0 0.011 0.045 0.171 0.475 0.799 0.946 0.987 0.997 0.999 0 0.011 0.045 0.171 0.475 0.799 0.946 0.987 0.997 0.999 0 0.011 0.045 0.171 0.475 0.799 0.946 0.987 0.997 0.999 0 0.011 0.045 0.171 0.475 0.799 0.946 0.987 0.997 0.999 0 0.011 0.045 0.171 0.475 0.799 0.946 0.987 0.997 0.999 0 0.011 0.045 0.171 0.475 0.799 0.946 0.987 0.997 0.999 0 0.011 0.045 0.171 0.475 0.799 0.946 0.987 0.997 0.999 0 0.011 0.045 0.171 0.475 0.799 0.946 0.987 0.997 0.999 0 0.011 0.045 0.171 0.475 0.799 0.946 0.987 0.997 0.999 0 0.011 0.045 0.171 0.475 0.799 0.946 0.987 0.997 0.999 0 0.011 0.045 0.171 0.475 0.799 0.946 0.987 0.997 0.999 0 0.011 0.045 0.171 0.475 0.799 0.946 0.987 0.997 0.999 0 0.011 0.045 0.171 0.475 0.799 0.946 0.987 0.997 0.999 0 0.011 0.045 0.171 0.475 0.799 0.946 0.987 0.997 0.999 0 0.011 0.045 0.171 0.475 0.799 0.946 0.987 0.997 0.999 0 0.011 0.045 0.171 0.475 0.799 0.946 0.987 0.997 0.999 0 0.011 0.045 0.171 0.475 0.799 0.946 0.987 0.997 0.999 0 0.011 0.045 0.171 0.475 0.799 0.946 0.987 0.997 0.999 0 0.011 0.045 0.171 0.475 0.799 0.946 0.987 0.997 0.999 0 0.011 0.045 0.171 0.475 0.799 0.946 0.987 0.997 0.999 0 0.011 0.045 0.171 0.475 0.799 0.946 0.987 0.997 0.999 0 0.011 0.045 0.171 0.475 0.799 0.946 0.987 0.997 0.999 0 0.011 0.045 0.171 0.475 0.799 0.946 0.987 0.997 0.999 0 0.077 0.196 0.415 0.674 0.858 0.946 0.981 0.993 0.998 0 0.077 0.196 0.415 0.674 0.858 0.946 0.981 0.993 0.998 0 0.077 0.196 0.415 0.674 0.858 0.946 0.981 0.993 0.998 0 0.077 0.196 0.415 0.674 0.858 0.946 0.981 0.993 0.998 0 0.077 0.196 0.415 0.674 0.858 0.946 0.981 0.993 0.998 0 0.077 0.196 0.415 0.674 0.858 0.946 0.981 0.993 0.998 0 0.077 0.196 0.415 0.674 0.858 0.946 0.981 0.993 0.998 0 0.077 0.196 0.415 0.674 0.858 0.946 0.981 0.993 0.998 0 0.077 0.196 0.415 0.674 0.858 0.946 0.981 0.993 0.998 0 0.077 0.196 0.415 0.674 0.858 0.946 0.981 0.993 0.998

IBPGCod, January 2018, Working document 05

Table 1 continued Natural Mortality 1 5 1976 2016 1 10 1 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.4 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.4 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.4 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.4 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.4 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 IBPGCod, January 2018, Working document 05

Table 1 continued "Tuning Data; Cod inshore Greenland" 102 ungtorsk (1B) 1987 2016 1 1 0.5 0.6 1 5 1 0 122 233 25 1 1 0 33 130 111 2 1 1 110 83 57 32 1 0 109 108 62 53 1 0 3 131 53 11 1 0 43 10 18 3 1 0 22 22 2 1 1 4 8 19 12 0 1 2 115 19 7 1 1 0 28 40 7 1 1 0 14 8 3 1 1 2 7 4 6 3 1 -1 -1 -1 -1 -1 1 -1 -1 -1 -1 -1 1 -1 -1 -1 -1 -1 1 31 207 72 21 9 1 1 68 69 21 3 1 32 28 29 9 5 1 47 123 35 7 5 1 32 148 60 24 1 1 7 170 82 15 1 1 -1 -1 -1 -1 -1 1 -1 -1 -1 -1 -1 1 138 155 120 58 12 1 20 526 106 44 19 1 7 184 304 30 8 1 4 158 105 104 27 1 7 46 45 25 19 1 2 39 44 59 49 1 6 31 98 42 36 ungtorsk (1D) 1987 2016 1 1 0.5 0.6 1 5 1 1 16 68 5 0 1 0 20 48 30 1 1 0 78 47 13 13 1 0 14 35 4 4 1 124 3 17 6 2 1 0 61 22 10 7 1 0 4 57 20 2 IBPGCod, January 2018, Working document 05

1 0 0 6 5 1 1 0 3 2 4 4 1 0 1 1 0 2 1 3 3 1 0 0 1 0 10 17 1 0 1 0 0 1 3 0 1 0 2 2 1 1 1 -1 -1 -1 -1 -1 1 0 7 4 3 0 1 0 6 4 2 1 1 3 43 6 3 1 1 9 27 7 2 0 1 2 114 37 13 4 1 -1 -1 -1 -1 -1 1 4 4 47 63 7 1 4 52 14 72 23 1 1 33 107 18 27 1 10 45 3 18 6 1 2 52 46 21 28 1 0 91 61 77 25 1 0 41 74 46 27 1 2 42 79 68 30 1 1 59 92 34 47

IBPGCod, January 2018, Working document 05

Table 2. SAM configuration of the base run in the assessment of the West Greenland inshore cod stock

# Configuration saved: Mon Sep 4 10:38:33 2017 # # Where a matrix is specified rows corresponds to fleets and columns to ages. # Same number indicates same parameter used # Numbers (integers) starts from zero and must be consecutive # $minAge # The minimium age class in the assessment 1

$maxAge # The maximum age class in the assessment 10

$maxAgePlusGroup # Is last age group considered a plus group (1 yes, or 0 no). 1

$keyLogFsta # Coupling of the fishing mortality states (nomally only first row is used). -1 -1 0 1 2 3 4 5 6 6 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1

$corFlag # Correlation of fishing mortality across ages (0 independent, 1 compound symmetry, or 2 AR(1) 0

$keyLogFpar # Coupling of the survey catchability parameters (nomally first row is not used, as that is covered by fishing mortality). -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 0 1 2 3 4 -1 -1 -1 -1 -1 5 6 7 8 9 -1 -1 -1 -1 -1

$keyQpow # Density dependent catchability power parameters (if any). -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1

$keyVarF # Coupling of process variance parameters for log(F)-process (nomally only first row is used) 0 0 0 0 0 0 0 0 0 0 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 IBPGCod, January 2018, Working document 05

-1 -1 -1 -1 -1 -1 -1 -1 -1 -1

$keyVarLogN # Coupling of process variance parameters for log(N)-process 0 1 1 1 1 1 1 1 1 1

$keyVarObs # Coupling of the variance parameters for the observations. 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 -1 -1 -1 -1 -1 2 2 2 2 2 -1 -1 -1 -1 -1

$obsCorStruct # Covariance structure for each fleet ("ID" independent, "AR" AR(1), or "US" for unstructured). | Possible values are: "ID" "AR" "US" "ID" "ID" "ID"

$keyCorObs # Coupling of correlation parameters must be specified if the AR(1) structure is chosen above. NA NA NA NA NA NA NA NA NA NA NA NA NA NA -1 -1 -1 -1 NA NA NA NA NA -1 -1 -1 -1

$stockRecruitmentModelCode # Stock recruitment code (0 for plain random walk, 1 for Ricker, and 2 for Beverton-Holt). 0

$noScaledYears # Number of years where catch scaling is applied. 0

$keyScaledYears # A vector of the years where catch scaling is applied.

$keyParScaledYA # A matrix specifying the couplings of scale parameters (nrow = no scaled years, ncols = no ages).

$fbarRange # lowest and higest age included in Fbar 4 8

$keyBiomassTreat # To be defined only if a biomass survey is used (0 SSB index, 1 catch index, and 2 FSB index). -1 -1 -1

$obsLikelihoodFlag # Option for observational likelihood | Possible values are: "LN" "ALN" IBPGCod, January 2018, Working document 05

"LN" "LN" "LN"

$fixVarToWeight # If weight attribute is supplied for observations this option sets the treatment (0 relative weight, 1 fix variance to weight). 0

IBPGCod, January 2018, Working document 05

Table 3. SAM runs with different configurations

Name of stock Changes to log(L) AIC No of SSB rho F rho Recruit baserun parameter rho codINBen_Baserun -751 1535 16 -0.083 -0.142 -0.342 Variance on codINBen_restriVar catch restricted. -766 1564 16 0.008 0.039 0.287 CorFlag set to codINBen_CorFind AR, not ID. -741 1517 17 0.107 -0.343 -0.318 Two survey codINBen_comb_surveys indices joined. -600 1221 10 -0.044 -0.161 -0.279 Includes emigration (M=0.3) from age codINBen_migration 5 -752 1537 16 -0.058 -0.184 -0.335

IBPGCod, January 2018, Working document 05

Table 4: Forecast runs. Top: same F as in final assessment year. Middle: F reduced to 0. Bottom: F set to MSY.

IBPGCod, January 2018, Working document 05

Table 5. Reference points.

Framework Reference Value Technical basis Point

MSY approach MSY Btrigger 7196 MSY Btrigger = Bpa

FMSY 0.28 FMSY < simulated F05

Precautionary Blim 5000 Segemented regression, judgement

approach Bpa 7196 Blim * exp(1.645 * σ), σ = 0.221

Flim 1.14 F50 deterministic simulated

Fpa 0.74 Flim * exp(-1.645 * σ), σ = 0.259

Y/R approach F0.1 0.15 SAM estimated

Fmax 0.25 SAM estimated

F0.35SPR 0.18 SAM estimated

IBPGCod, January 2018, Working document 05

Figure 1. Parameter estimates from the SAM model and associated confidence intervals..Left: Std dev of estimated terminal Fbar, recruitment and process error. Right: Std dev of observations of catch, survey 1B and survey.1D

IBPGCod, January 2018, Working document 05

Figure 3. Normalized residuals derived from the SAM base run. Blue circles indicate positive residuals (observation larger than predicted) and filled red circles indicate negative residuals.

IBPGCod, January 2018, Working document 05

Figure 4. Normalized residuals derived from the SAM base run. Blue circles indicate positive residuals (observation larger than predicted) and filled red circles indicate negative residuals.

IBPGCod, January 2018, Working document 05

Figure 5. Estimated historical patterns of fishing mortality (F)

IBPGCod, January 2018, Working document 05

Figure 6. Estimated historical patterns of spawning stock biomass (SSB) IBPGCod, January 2018, Working document 05

Figure 7. Estimated historical patterns of age 1 recruitment.

IBPGCod, January 2018, Working document 05

Figure 8. Leave out plots of Fbar, SSB and Recruits.

IBPGCod, January 2018, Working document 05

Figure 9. Retrospective plots of Fbar, SSB and Recruits.

IBPGCod, January 2018, Working document 05

Figure 10. Plot of SSB and F from exploratory SAM runs.

IBPGCod, January 2018, Working document 05

Figure 11. SSB – Recruitment relationship estimated by simulation using EqSim. Labels indicate yearclasses.

IBPGCod, January 2018, Working document 05

Figure 12: SSB – Recruitment relationship estimated by simulation using EqSim with fitted SSB- Recruitment relationships shown. Dashed: Ricker curve. Dotted: Beverton-Holt curve. Black line: Segmented regression. The curve fits are indicated.

IBPGCod, January 2018, Working document 05

Figure 13. Auto-correlation plots of numbers of Recruits. IBPGCod, January 2018, Working document 05

IBPGCod, January 2018, Working document 05

IBPGCod, January 2018, Working document 06

A SAM assessment of the East Greenland cod stock Frank Rigét1, Rasmus Hedeholm1, Jesper Boje1,2

1 Greenland Institute of Natural Resources, Box 570, DK-3900 Nuuk, Greenland 2Technical University of Denmark, National Institute of Fisheries Research, DK-2920 Charlottenlund [email protected]

Introduction The Atlantic cod (Gadus morhua) in Greenland waters is comprised primarily of three spawning stocks:

1 ) West Greenland inshore; 2 ) West Greenland offshore; and 3 ) East Greenland offshore.

Until 2011, ICES advice was provided for all three stock components combined. Since 2012 separate advice has been given for the inshore stock component. WKICE (ICES, 2015) evaluated the stock identity for the two offshore components based on the tagging (Storr-Poulsen et al., 2003) and genetic studies (Therkildsen et al., 2013) as well as advice from ICES SIMWG and concluded that they belong to different stock entities and should be assessed separately.

This document presents the input data and the assessment of the East Greenland stock (NAFO 1F and ICES 14b). The assessments are visible at stockassessment.org and Table 3 outlines the different runs/options from the basic run.

Input Input data to the assessment together with consistency analyses are described in details in WG_Doc1 for the IBPGCod 2017 benchmark and will not be discussed further here (Table 1).

Assessment Main settings/assumptions The state-space model (SAM, Nielsen and Berg, 2014) was applied to the stock. Table 2 show the SAM configuration of the base run. Several SAM runs with different data input or other configurations were made and compared with the base run. The landing fraction was assumed to be one, meaning no discard of the catch. This is consistent with the overall impression of the fishery. Fishery and natural maturity before spawning were set to zero assuming spawning takes place 1 January. The Greenland and German surveys in East Greenland and NAFO division 1F age 1 to 9 were applied for tuning. The plus group (10+) was not included. From 1996 to 2004 no age-aggregated catch data were available because of very low catches. Total catch weights for this period was integrated in the model as a third survey.

Catchabilities The catchabilities in the two surveys were estimated separately ages 1 to 7 and ages 8 and 9 the catchability was coupled, whereas age 10+ was not estimated as only were few has been caught.

IBPGCod, January 2018, Working document 06

Selectivity Fishing mortalities was set to be uniquely estimated for ages 3 to 8 and set to be estimated equal for ages 9 and 10.

Fbar The Fbar range of 5 to 10 years old was applied since these age groups constitute the main part of the catches (Table 1). Age 4 was not included as they constituted below 1.5% by number of the total catch during the period.

Variance The process variance parameters for the log(N)-process were separated for age 1 and for age 2-10 as age 1 may be more estimated with more uncertainty.

The variance parameters for catches and surveys were set similar for all ages (catches 3-10 years old, and surveys 1-9 years old) but separately for the catch and each of the surveys.

An independent covariance structure for each fleet was assumed.

Covariance and correlation The covariance structure for each fleet was set to be independent as no knowledge was available to suggest another structure. Similarly, it was assumed that there was no coupling of correlation parameters in the surveys. The correlation across ages was set to be auto-correlated with a lag of 1. The fishery is mainly a trawl fishery and some auto-correlation across ages is expected. Furthermore, applying the auto-correlation option lead to a decrease in AIC compared to the independent option.

Model diagnostics Parameter estimates The parameter estimates and their uncertainty are shown in Figure 1. The model give a relative good fit to the catch data (Figure 2). The observation standard deviations of the Greenland survey is lower than that of the German survey. This indicate that the assessment give relative higher weight to the Greenland survey than the German surveydespite it’s shorter timeseries

The process error standard deviation is reasonable and low compared to the standard deviation of the F random walk (Figure 1). The recruitment variance is high, which may be expected.

The sd(logF), sd(logSSB) and sd(logR) in 2016 were 0.343, 0.219 and 0.498, respectively and in all cases higher than in 2015.

Residuals No apparent pattern are observed in the catch and surveys residuals except for high positive residuals for several age-groups in the mid 1990ies in the German surveys (Figure 3 upper). Relative high process IBPGCod, January 2018, Working document 06

residuals are found both in case of numbers (N) and F but appear to be random without any clear patterns (figure 3 lower.

Stock summary F

The historical trajectories of F5-10 is shown in Figure 4. In the period from 1973 to late 1980ies, F5-10 varied

between ca 0.2 and ca 0.9. The period that followed had a marked peak of 1.7 in 1992. F5-10 then dropped

within a few years and was between ca 0.1 and 0.2 the mid 2000’s. Since 2013 F5-10 has increased and in 2016 it was 0.47.

SSB The SSB has marked peaks three times during the period (Figure 4). The first peak of ca 94,000 t in the late 1970s, the second peak of ca 66,000 t around 1990 and the third peak in 2014 also ca 66,000 t. The SSB was historical low in the mid 1990ies being less than 2,000 t. In 2016 the SSB has decreased to ca 47,000 t, but remains above the third quartile of the period.

Recruitment The number of age 1 recruits peaks one year after the well-known large year-classes of 1973, 1984, 2003 and 2009, with the latter being considerably lower than the three earlier peaks. Beside these years with marked peaks, the numbers of recruits is rather low. In 2016, the number of recruits has increased to a relative high level but the uncertainty is high on the terminal estimate.

Tuning fleet sensitivity In order to assess the impact of the different surveys on the stock trajectories, the assessment was re-run leaving out one survey a time (Figure 5). All runs showed relatively parallel trajectories in SSB, F5-10 and R meaning that no specific survey index drives the overall trend. The largest impact is from the German survey, which also cover a much longer period. Removing the German survey resulted in much higher F5-

10 in the period from 1993 until late 2000’s. This coincides with the period with historical low catches and where no age aggregated catch data were available in the period 1996 to 2004. Leaving out the Greenland survey resulted in a higher F5-10 in the two recent years.

Removing the German survey resulted in lower SSB in the period from the beginning of 2000s until recently, which indicate that the relative number of older cod are found in German survey compared to the fishery. Removing the Greenland survey resulted in lower SSB in the two most recent years.

Recruitment estimation is rather robust to all dataseries, except that the pronounced peak in 2004 (the large 2003 yc) almost disappear when removing the German survey. The reason is that 2003 yc were found in relatively high numbers as 1 year old in the German survey whereas only few was observed in the Greenland survey.

Retrospective analysis IBPGCod, January 2018, Working document 06

The robustness of the assessment is validated by a 5 year retrospective analysis (Figure 6). In no cases the

retrospective analyses derive at estimates of F5-10, SSB and R outside the confidence limits of the estimates

by the full time series. There is a tendency for the model to slightly underestimate F5-10 when related to the full time series (Mohns rho = -0.24) and to overestimate the SSB (Mohns rho = 0.275). The Recruitment appear to be minor underestimated (Mohn rho = -0.167). Accepted range for Mohn’s rho according to Hurtado-Ferro et al. (2015) is for long-lived species 0.20 to -0.15, and the present patterns are beyond these limits. However, ICES have not yet defined acceptable ranges for retrospective patterns.

Exploratory runs using other configurations The SAM model was run with different configurations in order to assess the robustness of the model. Table 3 show the main statistics of the runs as named on the stockassessment.org. None of the runs with other configurations showed a better fit in terms of AIC than the base run. The runs using Beverton-Holt or the Ricker stock-recruitment relationship instead of the random walk has a ΔAIC of approx. 9. Similarly, the run with M = 0.2 for all ages, assuming no emigration, instead of the increased M to 0.3 for age 6 and 0.4 for the older used by the base run, also has a ΔAIC of approx. 13. However, especially the run with M=0.2 resulted in a considerably lower SSB and higher F5-10 than the other runs (Figure 7). The run using independent fishing mortalities across ages instead of assuming an auto-correlated structure gave the highest AIC (ΔAIC = 98).

All retrospective runs showed clear tendencies to overestimate SSB (positive Mohns rho) and underestimate F (negative Mohns rho) (Table 3). The base run is somewhat in-between the other runs.

Short term forecast A random-walk recruitment from 2000-2016 was assumed for the short-term forecast as no clear relationship has been seen for this stock. An average of mean weight at age for the last 5 years was used in the forecast. Forecast scenarios for the years 2017 to 2019 were made using the SAM simulation procedure (Figure 8 and Table 4). The first scenario is with median F = 0.47 (mean F in SAM estimated to 0.47 for 2016) for all years resulted in a decrease of SSB from about 48,000 t in 2016 to below 18,000 t in 2019. Also, the catch decrease markedly from ca. 17,000 t to 4,000 t. The second scenario is with median F = 0.47 (SAM estimate for 2016) for the first year follow by no fishing in the next years resulted in a decrease in SSB from 2016 to 2017 and then minor increases to 36,000 t in 2019. The third scenario with a catch in

2017 similar to the TAC in 2016 of 14,800 t followed by fishing at FMSY = 0.28 in 2018 and 2019 would result in considerably decrease of SSB and catch from 2017 to 2018 and then rather similar levels in 2019.

Reference points The estimation of reference points follows the ICES Reference Points Guidance, January 2017. The estimation has been done using the simulation R-programme EqSim developed by D.C.M. Miller, which works directly on a specified SAM run.

The simulation settings for the Stock-Recruitment relationship were as follows:

IBPGCod, January 2018, Working document 06

Stock type was assumed type 1, ie. Stocks with occasional large year classesFor this type of stock Blim is based on the lowest SSB where large recruitment are observed….

The number of simulations was set to 2000. The year-classes 1974, 1985, 2004, 2010 were omitted as they were isolated peaks one year later than well-known very large year-classes (1973, 1984, 2003, 2009) and are believed to be of Icelandic origin (Figure 9). Hence, these recruitment peaks happen independently of this stock’s SSB and should therefore not be considered in the SSB-R relationship. Furthermore, the two last years were also removed as SAM estimated numbers of recruits for these most recent years are considered uncertain. The segmented regression was applied for the Stock-Recruitment relationship as simulation analysis showed much higher weight for this relationship than those of Ricker and Beverton-Holt relationships (Figure 9). No auto-correlation in the number of recruits was found (Figure 10). The estimated break point of the segmented regression was 5,613 t. However, it was considered to be too low. The trajectory of the SSB and the SSB-R relationship (Figure 4 and Figure 9) show that the SSB has been able to recover in a few years when the SSB has been above ca 20,000 t. However, during the 1990ies and early 2000ies, where the SSB were very low (between 400 and 5,000 t) it took 15 to 20 years stock for the stock to recover. Therefore, Blim was increased from the estimated break point of the segmented regression to 10,000 t. The simulation was done with 200 runs, scanning F from 0 to 3 divided into 100 intervals.

Bpa is calculated from the formula Bpa = Blim * exp(1.645 *σ), where σ is SD of ln(SSB) in 2016 - here estimated by SAM to 0.22. Bpa is then 14,347 t.

Flim is estimated by simulation using the above values of Blim and Bpa, setting Fcv, Fphi and SSBcv = 0 (no assessment and advice noise) and with no Btrigger. The range of years are from 1996 to 2015. F50 is then Blim, here estimated to 0.75 (Table 5).

Fpa is calculated from the formula Fpa = Flim * exp(-1.645 * σ), where σ is SD of ln(F) in 2015 here estimated by SAM to 0.34. Fpa is then 0.43 (Table 5).

MSY reference points (MSY Btrigger and FMSY)

FMSY is initially estimated as the F that maximize median long-term yield in the simulation under constant F exploitation. The default values of cvF = 0.212, phiF = 0.423 and cvSSB = 0 were applied to the simulation since no assessment/advice history is available for this stock. The initial FMSY was estimated at 0.38, which is below the above estimated Fpa.

The final FMSY is estimated by a simulation using the default Fcv, Fphi, the estimated Blim, Bpa and

Btrigger which is equal to Bpa. The final FMSY estimate was 0.45. The precautionary principle states that if

FMSY > F05, which is the case here, then FMSY should be reduced to F05 meaning FMSY = 0.28 (Table 5).

Conclusion This SAM assessment fitted the observations well and described the historical trajectories of F, SSB and recruits in accordance with the general view of the stock. The model parameters were estimated with IBPGCod, January 2018, Working document 06

reasonable size of uncertainties and the normalized residuals of both the process errors and the observations showed no crucial patterns. The model give highest weight to the catch observation and higher weight to the Greenland survey than to the German survey. Sensitivity to tuning fleets and catches were reasonable robust except for a period in the mid 1990ies where the catches were very low. The retrospective analyses also show reasonable consistency but a tendency a retrospective pattern with a overestimation of SSB and underestimation of F.

The increase of the natural mortality to 0.3 for the 6 years old and to 0.4 for age 7 and above in order to mimic emigration to Icelandic waters has large influence on the perception of this stock assessment. We believe that there is justification for this suggestion. The amount of emigration in the model corresponds to approximately 50% of each cohort migrating from East Greenland to Iceland, which seems valid when compared with results from tagging data (WD03 for the IBPGCod 2017 benchmark).

The assessment model utilizes all available data, which is a great improvement compared to the previous advisory procedure for this stock.

References.

Hurtado-Ferro, F., Szuwalski, C. S., Valero, J. L., Anderson, S. C., Cunningham, C. J., Johnson, K. F., Licandeo, R., McGilliard, C. R.,Monnahan, C. C., Muradian, M. L., Ono, K., Vert-Pre, K. A., Whitten, A. R., and Punt, A. E. 2015 Looking in the rear-view mirror: bias and retrospective patterns in integrated, age-structured stock assessment models. ICES J. Mar. Sci., 72: 99–110.

IBPGCod, January 2018, Working document 06

Table 1. Input files to SAM runs Catch in Numbers (thousands) 1 2 1973 2016 1 10 1 0 0 8 109 793 223 308 122 637 1024 0 0 12 79 76 579 176 200 63 584 0 0 75 293 517 205 1173 247 123 190 0 0 1279 564 384 750 274 1817 183 257 0 0 52 8836 1803 681 760 261 498 334 0 0 8 670 9164 2275 245 254 138 709 0 0 17 567 1961 6686 3546 1033 157 112 0 0 54 137 293 247 2411 689 77 23 0 0 0 87 104 267 388 2952 234 45 0 0 14 14 654 1326 1970 1562 2060 169 0 0 7 1153 788 3934 1137 419 150 287 0 0 70 297 1342 551 1999 334 138 123 0 0 84 263 339 1121 123 313 25 50 0 0 61 49 415 269 709 59 123 14 0 0 772 156 60 301 149 642 55 323 0 0 550 10794 327 110 498 123 347 172 0 0 22 2840 15527 219 43 252 87 277 0 0 22 702 7199 17371 140 9 112 70 0 0 9 812 724 4989 5232 49 21 49 0 0 2 127 266 163 2594 1225 29 5 0 0 1 15 18 59 10 121 78 2 0 0 41 53 27 16 27 0 5 0 0 0 24 8 37 9 3 9 1 4 0 0 -1 -1 -1 -1 -1 -1 -1 -1 0 0 -1 -1 -1 -1 -1 -1 -1 -1 0 0 -1 -1 -1 -1 -1 -1 -1 -1 0 0 -1 -1 -1 -1 -1 -1 -1 -1 0 0 -1 -1 -1 -1 -1 -1 -1 -1 0 0 -1 -1 -1 -1 -1 -1 -1 -1 0 0 -1 -1 -1 -1 -1 -1 -1 -1 0 0 -1 -1 -1 -1 -1 -1 -1 -1 0 0 -1 -1 -1 -1 -1 -1 -1 -1 0 0 5 33 57 103 94 57 16 6 0 0 232 376 135 175 115 14 1 0 0 0 49 1529 668 158 124 120 18 15 0 0 77 586 6015 2417 592 44 26 12 0 0 307 1287 1231 434 119 28 16 2 0 0 10 87 331 193 334 58 8 4 0 0 3 70 137 425 355 371 96 31 0 0 13 109 471 281 258 253 148 58 0 0 0 36 127 615 237 226 153 104 0 0 1 4 279 434 658 335 173 131 0 0 3 57 457 1554 1324 828 242 182 0 0 4 33 343 736 1130 766 427 257

IBPGCod, January 2018, Working document 06

Table 1 continued Mean Weight in Catch (kilograms) 1 3 1973 2016 1 10 1 0.1 0.3 0.508 1.251 1.652 2.392 3.311 4.504 5.487 7.423 0.1 0.3 0.578 1.084 1.609 2.407 3.308 4.454 5.465 7.698 0.1 0.3 0.474 1.267 1.546 2.497 3.466 4.398 5.368 8.04 0.1 0.3 0.752 1.183 1.738 2.328 3.232 4.396 5.447 7.693 0.1 0.3 0.74 1.176 1.478 2.234 3.237 4.406 5.468 7.507 0.1 0.3 0.65 1.134 1.458 2.267 3.236 4.382 5.516 7.899 0.1 0.3 0.597 1.197 1.534 2.306 3.214 4.387 5.536 7.709 0.1 0.3 0.749 1.217 1.529 2.339 3.257 4.384 5.529 8.335 0.1 0.3 0.83 1.09 1.528 1.846 2.894 4.246 5.949 10.199 0.1 0.3 0.83 1.11 1.401 1.978 2.878 3.992 5.332 7.352 0.1 0.3 0.78 0.954 1.296 2.129 3.057 3.74 4.699 6.145 0.1 0.3 0.693 0.878 1.35 2.149 3.05 3.717 4.705 5.743 0.1 0.3 0.78 0.96 1.421 2.128 3.102 3.9 4.704 5.665 0.1 0.3 0.253 0.8 1.595 2.582 3.636 4.894 5.803 5.733 0.1 0.3 0.341 0.944 1.79 2.737 3.671 4.567 5.359 6.127 0.1 0.3 0.367 1.012 1.578 2.298 3.682 4.153 5.504 7.192 0.1 0.3 0.288 0.758 1.458 2.593 3.276 4.87 4.868 6.358 0.1 0.3 0.785 0.917 1.226 2.038 3.151 4.27 5.365 7.89 0.1 0.3 0.78 1.034 1.167 1.55 2.558 3.3 5.41 9.37 0.1 0.3 1.326 1.77 1.807 2.071 2.217 3.586 4.143 8.168 0.1 0.3 0.79 1.47 1.16 2.38 2.77 3.87 5.66 8.08 0.1 0.3 0.518 1.98 2.962 4.791 4.738 4.21 5.742 7.444 0.1 0.3 0.426 1.427 2.949 4.176 5.233 5.926 9.645 7.442 0.1 0.3 0.618 1.097 1.603 2.465 3.467 4.616 6.044 8.186 0.1 0.3 0.618 1.097 1.603 2.465 3.467 4.616 6.044 8.186 0.1 0.3 0.618 1.097 1.603 2.465 3.467 4.616 6.044 8.186 0.1 0.3 0.618 1.097 1.603 2.465 3.467 4.616 6.044 8.186 0.1 0.3 0.618 1.097 1.603 2.465 3.467 4.616 6.044 8.186 0.1 0.3 0.618 1.097 1.603 2.465 3.467 4.616 6.044 8.186 0.1 0.3 0.618 1.097 1.603 2.465 3.467 4.616 6.044 8.186 0.1 0.3 0.618 1.097 1.603 2.465 3.467 4.616 6.044 8.186 0.1 0.3 0.618 1.097 1.603 2.465 3.467 4.616 6.044 8.186 0.1 0.3 0.354 0.718 1.073 1.963 2.738 3.699 5.271 7.366 0.1 0.3 1.323 1.602 2.349 3.608 4.42 5.44 7.191 8.127 0.1 0.3 0.387 0.917 1.597 3.294 6.092 8.524 11.114 14.435 0.1 0.3 0.359 0.644 1.266 1.799 3.025 4.936 5.84 8.29 0.1 0.3 0.489 0.776 1.396 2.797 4.634 6.453 7.804 9.993 0.1 0.3 0.699 1.124 1.636 2.494 3.354 5.334 8.06 10.475 0.1 0.3 0.553 1.026 1.541 2.297 3.377 4.685 6.285 10.022 0.1 0.3 0.501 0.891 1.434 2.37 3.559 5.137 7.167 11.417 0.1 0.3 0.48 0.998 1.698 2.272 3.408 4.745 6.827 9.024 0.1 0.3 0.564 1.163 1.853 2.603 3.636 4.732 6.4 8.841 0.1 0.3 0.484 0.833 1.435 2.097 3.46 4.699 6.846 9.115 0.1 0.3 0.406 0.845 1.42 2.135 3.267 4.693 6.693 10.071

IBPGCod, January 2018, Working document 06

Table 1 continued Mean Weight in Stock (kilograms) 1 4 1973 2016 1 10 1 0.025 0.147 0.372 0.788 1.452 2.385 3.737 5.382 7.22 10.033 0.025 0.147 0.372 0.788 1.452 2.385 3.737 5.382 7.22 10.033 0.025 0.147 0.372 0.788 1.452 2.385 3.737 5.382 7.22 10.033 0.025 0.147 0.372 0.788 1.452 2.385 3.737 5.382 7.22 10.033 0.025 0.147 0.372 0.788 1.452 2.385 3.737 5.382 7.22 10.033 0.025 0.147 0.372 0.788 1.452 2.385 3.737 5.382 7.22 10.033 0.025 0.147 0.372 0.788 1.452 2.385 3.737 5.382 7.22 10.033 0.025 0.147 0.372 0.788 1.452 2.385 3.737 5.382 7.22 10.033 0.025 0.147 0.372 0.788 1.452 2.385 3.737 5.382 7.22 10.033 0.025 0.147 0.372 0.788 1.452 2.385 3.737 5.382 7.22 10.033 0.025 0.147 0.372 0.788 1.452 2.385 3.737 5.382 7.22 10.033 0.025 0.147 0.372 0.788 1.452 2.385 3.737 5.382 7.22 10.033 0.025 0.147 0.372 0.788 1.452 2.385 3.737 5.382 7.22 10.033 0.025 0.147 0.372 0.788 1.452 2.385 3.737 5.382 7.22 10.033 0.025 0.147 0.372 0.788 1.452 2.385 3.737 5.382 7.22 10.033 0.025 0.147 0.372 0.788 1.452 2.385 3.737 5.382 7.22 10.033 0.025 0.147 0.372 0.788 1.452 2.385 3.737 5.382 7.22 10.033 0.025 0.147 0.372 0.788 1.452 2.385 3.737 5.382 7.22 10.033 0.025 0.147 0.372 0.788 1.452 2.385 3.737 5.382 7.22 10.033 0.025 0.147 0.372 0.788 1.452 2.385 3.737 5.382 7.22 10.033 0.025 0.147 0.372 0.788 1.452 2.385 3.737 5.382 7.22 10.033 0.025 0.147 0.372 0.788 1.452 2.385 3.737 5.382 7.22 10.033 0.025 0.147 0.372 0.788 1.452 2.385 3.737 5.382 7.22 10.033 0.025 0.147 0.372 0.788 1.452 2.385 3.737 5.382 7.22 10.033 0.025 0.147 0.372 0.788 1.452 2.385 3.737 5.382 7.22 10.033 0.025 0.147 0.372 0.788 1.452 2.385 3.737 5.382 7.22 10.033 0.025 0.147 0.372 0.788 1.452 2.385 3.737 5.382 7.22 10.033 0.025 0.147 0.372 0.788 1.452 2.385 3.737 5.382 7.22 10.033 0.025 0.147 0.372 0.788 1.452 2.385 3.737 5.382 7.22 10.033 0.025 0.147 0.372 0.788 1.452 2.385 3.737 5.382 7.22 10.033 0.025 0.147 0.372 0.788 1.452 2.385 3.737 5.382 7.22 10.033 0.025 0.147 0.372 0.788 1.452 2.385 3.737 5.382 7.22 10.033 0.025 0.147 0.372 0.788 1.452 2.385 3.737 5.382 7.22 10.033 0.025 0.147 0.372 0.788 1.452 2.385 3.737 5.382 7.22 10.033 0.025 0.147 0.372 0.788 1.452 2.385 3.737 5.382 7.22 10.033 0.019 0.088 0.262 0.52 1.067 1.982 3.385 5.699 8.447 8.564 0.059 0.14 0.452 0.976 1.73 2.977 4.186 5.447 7.423 10.8 0.041 0.206 0.406 0.823 1.728 2.499 3.496 5.48 7.363 10.686 0.017 0.152 0.366 0.783 1.408 2.209 3.891 5.711 7.218 10.859 0.025 0.201 0.367 0.916 1.519 2.634 4.068 5.658 7.565 10 0.02 0.194 0.45 0.771 1.396 2.353 3.663 5.14 7.062 10.354 0.003 0.129 0.36 0.773 1.402 2.758 4.145 5.173 6.217 9.06 0.017 0.1 0.357 0.697 1.194 1.808 3.241 4.835 6.809 10 0.025 0.116 0.327 0.831 1.623 2.245 3.557 5.299 6.879 9.973

IBPGCod, January 2018, Working document 06

Table 1 continued Proportion Mature at Year Start 1 6 1973 2016 1 10 1 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891 0.02 0.049 0.116 0.249 0.456 0.679 0.843 0.931 0.972 0.9891

IBPGCod, January 2018, Working document 06

Table 1 continued Natural Mortality 1 5 1973 2016 1 10 1 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.3 0.4 0.4 0.4 0.4

IBPGCod, January 2018, Working document 06

Table 1 continued Tuning "Data;" Cod East Greenland (XIVb+1F) 102 Paamiut oest(XIVb+1F) 2008 2016 1 1 0.58 0.65 1 9 1 372 1113 7968 6582 23794 5412 2235 736 1006 1 7642 8019 4504 5378 5664 6610 2537 225 554 1 2436 3959 5759 3253 12785 7969 11264 2958 450 1 162 5682 8288 16346 5409 4707 2226 3382 1834 1 258 1208 12748 7154 12041 4155 2428 1345 1849 1 157 1432 1954 44843 25373 26654 5209 3440 1852 1 15 207 1849 1558 21863 8805 12411 2875 3790 1 86 38 1259 4916 11445 29010 7407 4793 1954 1 3847 1818 998 555 2089 2399 6779 4874 3398 WH oest(XIVb+1F) 1982 2016 1 1 0.83 0.92 1 9 1 23 214 2500 1760 4451 1952 793 223 927 1 -1 -1 -1 -1 -1 -1 -1 -1 -1 1 8 54 1134 507 2434 582 1242 229 125 1 2521 242 160 1658 947 1439 344 831 96 1 3367 9255 1128 273 1631 603 1300 165 473 1 4 10193 24656 2689 720 1368 296 966 80 1 18 335 9769 23391 876 200 559 83 337 1 2 111 732 23945 49864 1007 44 756 70 1 36 58 715 706 11679 12101 139 15 74 1 73 150 171 539 102 2128 1762 31 11 1 10 196 103 61 53 67 67 51 0 1 4 15 869 152 95 97 31 83 34 1 71 5 16 84 39 22 38 8 0 1 1 621 347 260 1399 372 120 403 32 1 0 0 353 130 131 110 23 25 0 1 0 12 17 687 557 191 78 48 0 1 73 39 4 11 173 138 48 10 0 1 426 389 346 118 257 174 156 29 0 1 202 243 323 208 40 72 20 46 61 1 166 568 493 631 362 190 60 50 18 1 1 395 2119 601 477 454 217 61 21 1 629 53 553 1761 1026 1015 541 220 37 1 10687 1770 448 617 1667 921 620 228 39 1 1603 39549 8091 1250 2819 2549 727 189 40 1 439 3375 48140 9269 1328 2404 1309 193 30 1 154 2007 5149 65974 8166 713 658 634 70 1 265 513 8213 4401 22939 4201 516 220 199 1 322 1057 391 1620 2863 11241 1964 111 134 1 700 1425 1388 845 2887 2518 5707 1362 236 1 120 1246 3475 4874 2402 2949 1179 2324 310 1 50 1624 10093 10233 9846 2827 1778 1166 379 1 17 35 4312 27014 11146 7455 1314 517 291 1 7 55 602 20847 58174 9275 3284 1316 494 1 37 68 341 752 3688 3598 1881 644 187 1 442 88 115 724 1007 3031 1545 534 141 TOTALCW IBPGCod, January 2018, Working document 06

1996 2004 1 1 0 1 -1 -1 1 192 1 355 1 345 1 116 1 152 1 125 1 401 1 485 1 775

Landing mean weight was set equal to Catch mean weight

No discard is assumed

F and M before spawning set to zero allover

IBPGCod, January 2018, Working document 06

Table 2. SAM configuration of the basis run in the assessment of the East Greenland cod stock.

# Configuration saved: Fri Sep 8 15:00:00 2017 # # Where a matrix is specified rows corresponds to fleets and columns to ages. # Same number indicates same parameter used # Numbers (integers) starts from zero and must be consecutive # $minAge # The minimium age class in the assessment 1

$maxAge # The maximum age class in the assessment 10

$maxAgePlusGroup # Is last age group considered a plus group (1 yes, or 0 no). 1

$corFlag # Correlation of fishing mortality across ages (0 independent, 1 compound symmetry, or 2 AR(1) 2

$keyLogFpar # Coupling of the survey catchability parameters (nomally first row is not used, as that is covered by fishing mortality). -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 0 1 2 3 4 5 6 7 7 -1 8 9 10 11 12 13 14 15 15 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1

$keyQpow # Density dependent catchability power parameters (if any). -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1

$keyVarLogN # Coupling of process variance parameters for log(N)-process 0 1 1 1 1 1 1 1 1 1

IBPGCod, January 2018, Working document 06

$keyVarObs # Coupling of the variance parameters for the observations. -1 -1 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 -1 2 2 2 2 2 2 2 2 2 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1

$obsCorStruct # Covariance structure for each fleet ("ID" independent, "AR" AR(1), or "US" for unstructured). | Possible values are: "ID" "AR" "US" "ID" "ID" "ID" "ID"

$stockRecruitmentModelCode # Stock recruitment code (0 for plain random walk, 1 for Ricker, and 2 for Beverton-Holt). 0

$noScaledYears # Number of years where catch scaling is applied. 0

$keyScaledYears # A vector of the years where catch scaling is applied.

$keyParScaledYA # A matrix specifying the couplings of scale parameters (nrow = no scaled years, ncols = no ages).

$fbarRange # lowest and highest age included in Fbar 5 10

$keyBiomassTreat # To be defined only if a biomass survey is used (0 SSB index, 1 catch index, and 2 FSB index). -1 -1 -1 3

$obsLikelihoodFlag # Option for observational likelihood | Possible values are: "LN" "ALN" "LN" "LN" "LN" "LN"

$fixVarToWeight # If weight attribute is supplied for observations this option sets the treatment (0 relative weight, 1 fix variance to weight). 0

IBPGCod, January 2018, Working document 06

Table 3. SAM runs with different configurations

Name of Changes SAM to No of Recruit stock baserun log(L) AIC parameter SSB rho F rho rho codEGBen_baserun -899.1 1844.3 23 0.275 -0.240 -0.167 codEGBen_02M all M=0.2 -905.4 1856.8 23 0.391 -0.314 -0.147 codEGBen_corFind corFlag=0 -948.9 1941.8 22 0.246 -0.268 -0.196 none corFlag=1 -924.3 1894.6 23 4.378 -0.639 1.420 codEGBen_Rick Ricker -901.5 1853.0 25 0.281 -0.242 -0.135 none BH -901.3 1852.6 25 0.265 -0.232 -0.134

IBPGCod, January 2018, Working document 06

Forecast table 1. SQ F= .47 all years.

Year fbar:median fbar:low fbar:high rec:median rec:low rec:high ssb:median ssb:low ssb:high catch:median catch:low catch:high

2016 0.470 0.243 0.956 40376 14579 113101 48489 31075 76683 16845 11020 25226 2017 0.470 0.099 2.106 9175 510 176280 32709 15610 64297 10918 2931 29873 2018 0.470 0.052 3.829 9175 510 176280 20441 5365 51748 5416 984 17106 2019 0.470 0.036 4.870 9175 510 176280 18045 4936 50646 3905 661 16365

Forecast table 2. SQ then zero.

Year fbar:median fbar:low fbar:high rec:median rec:low rec:high ssb:median ssb:low ssb:high catch:median catch:low catch:high

2016 0.489 0.253 0.995 40376 14579 113101 48489 31075 76683 17386 11405 25939 2017 0.000 0.000 0.000 9175 510 39457 32127 15130 63560 0 0 0 2018 0.000 0.000 0.000 7668 510 39457 32975 16309 66948 0 0 0 2019 0.000 0.000 0.000 6162 510 39457 35864 18628 75370 0 0 0

Forecast table 3. catch 14.8K, then Fmsy=.28.

Year fbar:median fbar:low fbar:high rec:median rec:low rec:high ssb:median ssb:low ssb:high catch:median catch:low catch:high

2016 0.489 0.253 0.995 40376 14579 113101 48489 31075 76683 17386 11405 25939 2017 0.760 0.160 3.407 9175 510 39457 32127 15130 63560 14818 4435 36138 2018 0.280 0.031 2.281 7668 510 39457 15787 3361 44414 2622 510 9339 2019 0.280 0.021 2.901 6162 510 39457 16588 5056 48427 2601 375 11723

IBPGCod, January 2018, Working document 06

Table 5. Reference points. Cod in 14b and NAFO division 1F.

Framework Reference Value Technical basis Point

MSY approach MSY Btrigger 14347 MSY Btrigger = Bpa

FMSY 0.28 FMSY = simulated F05

Precautionary Blim 10000 Judgment

approach Bpa 14347 Blim * exp(1.645 * σ), σ = 0.22

Flim 0.75 F50 deterministic simulated

Fpa 0.43 Flim * exp(-1.645 * σ), σ = 0.34

Y/R approach F0.1 0.30 SAM estimated

Fmax 0.60 SAM estimated

F35spr 0.34 SAM estimated

IBPGCod, January 2018, Working document 06

Figure 1. Parameter estimates from the SAM model and associated confidence limits).Left: Std dev of observations of catch, Greenland survey and German survey. Right: Std dev of estimated terminal Fbar, recruitment and process error.

Figure 2. Estimated catch and with observed catch shown as crosses. Note the period 1996- 2004 with zero catches because no age disaggregated catch data were available. IBPGCod, January 2018, Working document 06

Figure 3. Upper: Normalized residuals derived from the SAM run in the data series. Lower: Residuals for the estimates of stock numbers and F. Blue circles indicate positive residuals (observation larger than predicted) and filled green circles indicate negative residuals. IBPGCod, January 2018, Working document 06

Figure 4. Estimated historical patterns of Fbar, SSB and Recruits. IBPGCod, January 2018, Working document 06

Figure 5. Leave out plots of Fbar, SSB and Recruits. IBPGCod, January 2018, Working document 06

Figure 6. Retrospective plots of Fbar, SSB and Recruits. IBPGCod, January 2018, Working document 06

Figure 7. Plot of Fbar vs SSB estimated by exploratory SAM runs.

IBPGCod, January 2018, Working document 06

Figure 8. Three forecast scenarios estimated by SAM. Above: Fbar in all years equal to 0.47 Middle: Fbar in 2016 then zero catch

Below: Fbar in 2017 based on a catch of 14,818 t (TAC in 2016) then FMSY = 0.28 in 2018 and 2019.

IBPGCod, January 2018, Working document 06

Figure 9. SSB – Recruits relationship estimated by simulation using EqSim Above: Each point labelled by year Below: With segmented regression (yellow) omitting the years 1974, 1985, 2004, 2010, 2015, 2016.

IBPGCod, January 2018, Working document 06

Figure 10. Auto-correlation plots of numbers of Recruits. IBPGCod, January 2018, Working document 07

A SAM assessment of the East Greenland cod stock Frank Rigét1, Rasmus Hedeholm1, Jesper Boje1,2

1 Greenland Institute of Natural Resources, Box 570, DK-3900 Nuuk, Greenland 2Technical University of Denmark, National Institute of Fisheries Research, DK-2920 Charlottenlund [email protected]

Introduction The Atlantic cod (Gadus morhua) in Greenland waters is comprised primarily of three spawning stocks:

1 ) West Greenland inshore; 2 ) West Greenland offshore; and 3 ) East Greenland offshore.

Input Input data to the assessment together with consistency analyses are described in details in WG_Doc1 for the IBPGCod 2017 benchmark and will not be discussed further here (Table 1).

IBPGCod, January 2018, Working document 07

Selectivity Fishing mortalities was set to be uniquely estimated for ages 3 to 8 and set to be estimated equal for ages 9 and 10.

Variance The process variance parameters for the log(N)-process were separated for age 1 and for age 2-10 as age 1 may be more estimated with more uncertainty.

The variance parameters for catches and surveys were set similar for all ages (catches 3-10 years old, and surveys 1-9 years old) but separately for the catch and each of the surveys.

An independent covariance structure for each fleet was assumed.

The process error standard deviation is reasonable and low compared to the standard deviation of the F random walk (Figure 1). The recruitment variance is high, which may be expected.

The sd(logF), sd(logSSB) and sd(logR) in 2016 were 0.343, 0.219 and 0.498, respectively and in all cases higher than in 2015.

residuals are found both in case of numbers (N) and F but appear to be random without any clear patterns (figure 3 lower.

Stock summary F

The historical trajectories of F5-10 is shown in Figure 4. In the period from 1973 to late 1980ies, F5-10 varied

between ca 0.2 and ca 0.9. The period that followed had a marked peak of 1.7 in 1992. F5-10 then dropped

within a few years and was between ca 0.1 and 0.2 the mid 2000’s. Since 2013 F5-10 has increased and in 2016 it was 0.47.

Retrospective analysis IBPGCod, January 2018, Working document 07

The robustness of the assessment is validated by a 5 year retrospective analysis (Figure 6). In no cases the

retrospective analyses derive at estimates of F5-10, SSB and R outside the confidence limits of the estimates

All retrospective runs showed clear tendencies to overestimate SSB (positive Mohns rho) and underestimate F (negative Mohns rho) (Table 3). The base run is somewhat in-between the other runs.

The simulation settings for the Stock-Recruitment relationship were as follows:

IBPGCod, January 2018, Working document 07

Stock type was assumed type 1, ie. Stocks with occasional large year classesFor this type of stock Blim is based on the lowest SSB where large recruitment are observed….

Bpa is calculated from the formula Bpa = Blim * exp(1.645 *σ), where σ is SD of ln(SSB) in 2016 - here estimated by SAM to 0.22. Bpa is then 14,347 t.

Fpa is calculated from the formula Fpa = Flim * exp(-1.645 * σ), where σ is SD of ln(F) in 2015 here estimated by SAM to 0.34. Fpa is then 0.43 (Table 5).

MSY reference points (MSY Btrigger and FMSY)

The final FMSY is estimated by a simulation using the default Fcv, Fphi, the estimated Blim, Bpa and

Btrigger which is equal to Bpa. The final FMSY estimate was 0.45. The precautionary principle states that if

FMSY > F05, which is the case here, then FMSY should be reduced to F05 meaning FMSY = 0.28 (Table 5).

The assessment model utilizes all available data, which is a great improvement compared to the previous advisory procedure for this stock.

References.

IBPGCod, January 2018, Working document 07

1996 2004 1 1 0 1 -1 -1 1 192 1 355 1 345 1 116 1 152 1 125 1 401 1 485 1 775

Landing mean weight was set equal to Catch mean weight

No discard is assumed

F and M before spawning set to zero allover

IBPGCod, January 2018, Working document 07

Table 2. SAM configuration of the basis run in the assessment of the East Greenland cod stock.

$maxAge # The maximum age class in the assessment 10

$maxAgePlusGroup # Is last age group considered a plus group (1 yes, or 0 no). 1

$corFlag # Correlation of fishing mortality across ages (0 independent, 1 compound symmetry, or 2 AR(1) 2

$keyQpow # Density dependent catchability power parameters (if any). -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1

$keyVarLogN # Coupling of process variance parameters for log(N)-process 0 1 1 1 1 1 1 1 1 1

IBPGCod, January 2018, Working document 07

$keyVarObs # Coupling of the variance parameters for the observations. -1 -1 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 -1 2 2 2 2 2 2 2 2 2 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1

$obsCorStruct # Covariance structure for each fleet ("ID" independent, "AR" AR(1), or "US" for unstructured). | Possible values are: "ID" "AR" "US" "ID" "ID" "ID" "ID"

$stockRecruitmentModelCode # Stock recruitment code (0 for plain random walk, 1 for Ricker, and 2 for Beverton-Holt). 0

$noScaledYears # Number of years where catch scaling is applied. 0

$keyScaledYears # A vector of the years where catch scaling is applied.

$keyParScaledYA # A matrix specifying the couplings of scale parameters (nrow = no scaled years, ncols = no ages).

$fbarRange # lowest and highest age included in Fbar 5 10

$keyBiomassTreat # To be defined only if a biomass survey is used (0 SSB index, 1 catch index, and 2 FSB index). -1 -1 -1 3

$obsLikelihoodFlag # Option for observational likelihood | Possible values are: "LN" "ALN" "LN" "LN" "LN" "LN"

$fixVarToWeight # If weight attribute is supplied for observations this option sets the treatment (0 relative weight, 1 fix variance to weight). 0

IBPGCod, January 2018, Working document 07

Table 3. SAM runs with different configurations

IBPGCod, January 2018, Working document 07

Forecast table 1. SQ F= .47 all years.

Year fbar:median fbar:low fbar:high rec:median rec:low rec:high ssb:median ssb:low ssb:high catch:median catch:low catch:high

Forecast table 2. SQ then zero.

Year fbar:median fbar:low fbar:high rec:median rec:low rec:high ssb:median ssb:low ssb:high catch:median catch:low catch:high

Forecast table 3. catch 14.8K, then Fmsy=.28.

Year fbar:median fbar:low fbar:high rec:median rec:low rec:high ssb:median ssb:low ssb:high catch:median catch:low catch:high

IBPGCod, January 2018, Working document 07

Table 5. Reference points. Cod in 14b and NAFO division 1F.

Framework Reference Value Technical basis Point

MSY approach MSY Btrigger 14347 MSY Btrigger = Bpa

FMSY 0.28 FMSY = simulated F05

Precautionary Blim 10000 Judgment

approach Bpa 14347 Blim * exp(1.645 * σ), σ = 0.22

Flim 0.75 F50 deterministic simulated

Fpa 0.43 Flim * exp(-1.645 * σ), σ = 0.34

Y/R approach F0.1 0.30 SAM estimated

Fmax 0.60 SAM estimated

F35spr 0.34 SAM estimated

IBPGCod, January 2018, Working document 07

Figure 4. Estimated historical patterns of Fbar, SSB and Recruits. IBPGCod, January 2018, Working document 07

Figure 5. Leave out plots of Fbar, SSB and Recruits. IBPGCod, January 2018, Working document 07

Figure 6. Retrospective plots of Fbar, SSB and Recruits. IBPGCod, January 2018, Working document 07

Figure 7. Plot of Fbar vs SSB estimated by exploratory SAM runs.

IBPGCod, January 2018, Working document 07

Figure 8. Three forecast scenarios estimated by SAM. Above: Fbar in all years equal to 0.47 Middle: Fbar in 2016 then zero catch

Below: Fbar in 2017 based on a catch of 14,818 t (TAC in 2016) then FMSY = 0.28 in 2018 and 2019.

IBPGCod, January 2018, Working document 07

Figure 10. Auto-correlation plots of numbers of Recruits. IBPGCod, January 2018, Working document 06