4 Reference points and assessment of salmon

Transcription

1 194 ICES WGBAST REPORT Reference points and assessment of salmon 4.1 Introduction In this chapter results of the assessment model and alternative future projections of salmon stocks in assessment units (AU) 1 4 are presented. Furthermore, the current status of salmon stocks in AUs 5 6 is evaluated against the reference points. The methodological basis and details of the assessment model and stock projections are given in the Stock Annex (Annex 2). Here we describe methodological updates, which are not reported yet in the recent benchmark (WKBALTSalmon, ICES, 2017d) or in the current version of the Stock Annex (to be updated later in 2018). Note that, as described below, the modelling results presented in this section originate from a run that had not fully converged at the time of the working group meeting. The approximations of posterior distributions of some parameters was therefore poor. Results originating from an extended (i.e. longer and more converged) model run were presented to the review group (RG) during the RG/ADG meeting, and the most important of these updated results are presented in Appendix 1. The advice draft document delivered by the Advice Drafting Group (ADG) is also based on these updated model results. Overall the differences between the results presented below and in the Appendix 1 are minor, and they do not affect conclusions or the perception of stock status and development. The input data used in the extended model run was also completed by correcting the piece of the model code, which reads in river specific smolt priors (derived mostly from the river model). The error in the code was located during the meeting and due to this error the last 2 years (2019 and 2020 in this year s assessment) of smolts priors became excluded from the model run. 4.2 Historical development of Baltic salmon stocks (assessment units 1 6) Changes in the assessment methods WinBUGS to JAGS During autumn 2016 and 2017, the FLHM was transferred from WinBUGS to JAGS (Just Another Gibbs Sampler; Plummer 2003). This work was initiated for the purpose of comparing alternative parameterisations of the stock recruitment function for the 2017 benchmark (ICES, 2017d) and continued thereafter to implement all stocks and observation models in the JAGS FLHM. This work has enabled continued use of the FLHM to assess stock status following technical failure to implement MCMC sampling of WinBUGS FLHM with updated time-series (ICES, 2017). The JAGS implementation of the FLHM has several advantages over WinBUGS. It is easy to run parallel chains, meaning that convergence can now be assessed quantitatively (e.g. using diagnostics such as the Gelman-Rubin statistic, Gelman and Rubin, 1992), in addition to visual inspection of trace plots. Although still computationally burdensome, the JAGS FLHM runs ~9 10 times faster e.g. taking around 16 days to run two chains with iterations each on a 64-bit Operating System machine with 32.0 GB of installed RAM. The runjags R package (Denwood, 2016) was used to implement parallel computing.

2 ICES WGBAST REPORT Correction of errors In the course of moving the FLHM from WinBUGS to JAGS, several errors were corrected. Most of these changes are inconsequential for model results and status evaluations, with exceptions described here. As noted last year (ICES, 2017), rates of natural mortality for Emån and Mörrumsån were inconsistent with those for other stocks, since Emån and Mörrumsån lacked natural mortality during the coastal fishery. This has now been corrected, so that natural mortality rates for these stocks are consistent with those for other stocks and with the forward-projection (scenarios) code. The number of months of natural mortality applied to wild and reared post-smolts was corrected to 12 instead of eleven and nine, respectively. While this is not expected to have made a large difference to model results (except to estimates of annual instantaneous post-smolt mortality) it resulted in inconsistency between the assessment model and projections (scenarios) code, with higher levels of post-smolt mortality applied in the latter. Hatchery-reared salmon juveniles stocked as parr in Torne River and Simojoki and returning to spawn were not available to counting in those rivers; this has now been corrected. Since releases of reared salmon have now been phased out in these rivers, this is only expected to affect estimates of historical stock development. New stock recruitment parameterization The JAGS FLHM uses an updated stock recruitment parameterization (Model 3 from WKBALTSalmon, ICES, 2017d). This comprises an eggs-per-recruit (EPR) calculation as a function of vital rates (survival, maturity, fecundity etc.), priors on maximum egg survival (1/α) instead of steepness, and transferal of stock-specific priors for unfished equilibrium smolt production (R0 or PSPC) to maximum recruitment (i.e. the stock recruitment carrying capacity).under this parameterisation, R0 varies by year (see ICES, 2017d for details). Since the benchmark, it was discovered that the stock recruitment alpha prior used there did not closely resemble that from Pulkkinen and Mäntyniemi (2012). The alpha prior has therefore been updated since the benchmark (Figure ). The alpha prior used in 2018 implies a steeper (median) slope at the origin (higher maximum egg survival) than in both the 2017 assessment and benchmark (Figure ). The new (2018) and old (2017) priors on R0 are shown in Figures a c. Slight modifications to the eggs per recruit calculation used in the benchmark were needed to account for model developments that affect population dynamics, namely river- and year- specific sex ratios, and the proportion of fish passing the ladder in Ume/Vindelälven (see below). A final consideration when using the new stock recruit parameterisation is the fact that status is assessed using the ratio of smolt production to unfished equilibrium smolt production (PSPC or R0). Now that R0 varies by year, an annual R0 or average thereof must be used as reference points for status evaluations. Year 2017 status is evaluated against R0 in that year, while we have chosen to use the average of the R0 estimates during the final five years of the assessment period ( ) in calculations of reference points and stock status in future years. Inclusion of recreational trolling catches and effort Updated recreational trolling catch estimates for the period (by expert elicitation ) are now accounted in the model as a part of the offshore longline fishery, by pooling the annual estimated trolling catches together with the reported longline catches (see Section for details of the estimation procedure for trolling catches).

3 196 ICES WGBAST REPORT 2018 The longline effort was increased proportionally, by annual trolling catch estimates divided by the cpue for reported longline catches and effort (all countries). Updates to prior distributions for the proportion of salmon that finds the Ume/Vindelälven fish ladder counter The hierarchical Bayesian mark recapture model used to obtain priors for the annual proportion of salmon that finds the fish ladder in Ume/Vindelälven was updated to account for the fact that in 2017 many of the tagged fish are believed to have ceased their upstream migration after tagging (cf. Section 3.4.3). This would otherwise result in a very low estimated proportion that finds the fish ladder in The model was modified by adding prior distributions for the annual proportion of salmon that continues migrating after tagging, based on expert opinion. The sample size of the Binomial observation model for the number of salmon counted in the ladder then becomes the number of tagged fish that continue their migration, rather than the total number of tagged fish. Priors for the proportion of salmon that continues migrating after tagging had a mean and standard deviation of 0.99 (0.01) from 1996 to 2016 and a mean and standard deviation of 0.10 (0.03) in FLHM priors (posteriors from the mark recapture model) and posteriors are shown in Figure Changes to population dynamics for Ume/Vindelälven Owing to observations of a decreasing proportion of females among spawers in Ume/Vindelälven over time, not seen in e.g. Torneälven/Tornionjoki (Figure ), the FLHM was modified to accommodate river- and year-specific spawner sex ratios. Spawner sex ratios are unchanged for all other rivers, but updated for Ume/Vindelälven. There have also been increased levels of mortality among spawners in Ume/Vindelälven in recent years (Sections and 3.4.3). This was accounted for in the FLHM by multiplying the annual proportion of females among spawners by the modal proportion that survives after counting. Annual proportions of salmon dying shortly after counting were set to zero until in 2013, whereas, based on local information and expert judgement, they were increased to 0.20, 0.02, 0.20 and 0.10 for years In addition to the changes listed above, the 2018 assessment also includes revised priors for the stock recruitment carrying capacity for Lögdeälven, Öreälven and Piteälven (ICES, 2017), and updated smolt production estimates for Mörrumsån, Emån (ICES, 2017d), and Piteälven (Section 4.2.2). Effect of changes on results and status evaluations In order to analyse and understand the effect of different changes made since the 2017 assessment, several additional comparison runs of the JAGS FLHM were made. These results were discussed in detail at the working group meeting. For brevity, they are not presented here, although they inform the following discussion. The extra comparisons were as follows: 1) 2017 assessment model configuration, i.e. same priors and data as in the 2017 assessment (data up to 2015) but with correction of errors noted above. 2) same model structure as the 2017 assessment but with data and priors updated for the 2018 assessment (data up to 2017) 3) updated data and priors plus new stock-recruitment parameterisation 4) as 3) but without new recreational trolling estimates. Changes in the assessment methods outlined above are expected to result in a number of changes to assessment results. The corrected (higher) natural mortality for Emån and Mörrumsån results in higher estimates of maximum egg survival (lower alpha) and stock recruit steepness and lower estimates of historical spawner abundances for

4 ICES WGBAST REPORT those rivers. It also rectifies the failure to recover with zero fishing mortality scenario in forward projections observed in earlier assessments (e.g. ICES, 2015). The corrected post-smolt mortality rates result in higher estimates of post-smolt survival, particularly for reared salmon, with more similar rates of post-smolt survival for wild and reared fish. This follows from the fact that the post-smolt mortality rate must be lower after correction when applied for 12 months, to achieve the same overall survival from natural mortality obtained when it was applied for only nine months. It should be noted that this correction is not expected to result in appreciable changes in estimates of other vital rates (adult survival, etc.) since the total annual mortality rate applied before and after the correction is comparable. It does however result in a difference in reconstructed abundances of reared salmon in scenarios (forward projections) where 12 months of post-smolt mortality have been applied also in previous years, regardless of the number of months applied in the FLHM. The new stock recruit alpha prior implies higher maximum egg survival than the prior used in last year (ICES, 2017), and together with EPR calculated from vital rates, higher stock recruit steepness for most stocks (with the notable exception of Ume/Vindelälven, which has a lower EPR because of a lower average spawner sex ratio, and the fact that not all spawners pass the fish ladder). The lower prior on PSPC or R0 (resulting from transferring the prior from R0 to K, and lower EPR in the case of Ume/Vindelälven; Figures a c) may account at least partially for lower posterior estimates of R0 for some stocks, especially those lacking smolt and spawner counting observations (e.g. Rickleån and Byskeälven, but also Ume/Vindelälven). Acting alone, this change could be expected to lead to slightly more favourable status evaluations for most rivers. However, acting together with the many other changes to data and priors between the 2017 and 2018 assessments, it is difficult to quantify the effect of individual changes on estimated stock status. Nonetheless, readers should be aware of the potential implications of changes to the assessment model in interpreting the results that follow Updated submodels The river model (hierarchical linear regression analysis) provides input about smolt production as likelihood approximations (these are sometimes called also pseudo observations in the literature, but for simplicity they are usually called smolt priors in this report) into the life cycle model, by analysing all the juvenile survey data from the rivers in AUs 1 3. For rivers in AUs 4 6, other methods are used to estimate smolt production (see Stock Annex, Section C.1.5 and ICES, 2017d). Results of the river model indicate a substantial increase in smolt abundance in AU 1 2 rivers since the late 1990s. At the moment (2017), smolt abundance is in its highest level in most of theses rivers, but the abundance is predicted to level off or even decrease during (Table ). The long-term increase in smolt production in AU 3 (R. Ljungan) is less apparent than in the AU 1 2 rivers, nevertheless smolt abundance is currently peaking also in this AU. For the rivers Tornionjoki, Simojoki, Ume/Vindelälven, Sävarån and Lögdeälven the results of the river model are more informative than for the other rivers, because of the availability of smolt trapping data. Also, smolt estimates of years without smolt trapping have become somewhat more precise in these rivers. Smolt trapping has been conducted only in one year (2016) in Lögdeälven, which increases the precision of Lögdeälven smolt abundances mainly in that specific year.

5 198 ICES WGBAST REPORT 2018 Smolt priors in River Piteälven A large part of the production areas in Piteälven (AU 2) are hard to electrofish. Therefore, this river is not included in the river model used for the other wild AU 1 3 rivers to derive input smolt estimates. Instead, in Piteälven smolt production have been estimated from numbers of eggs deposited based on the number of adults passing the fishladder at the Sikfors power plant station (which must be passed by all successful spawners). The calculated smolt numbers have been based on an assumed egg-to-smolt survival rate of 1% and constant proportions of 3 (62%) and 4 (38%) year old smolts (Annex 2). Because the fishcounter only provides total numbers of adults passing, annual information from river Ume/Vindelälven on average body size and sex ratio among spawners has been used toghether with data on size-dependent fecundities, to calculate egg numbers. A basic shortcoming with this simple approach used over the years for deriving smolt priors is that is assumes a stock recruit relationship without density-dependence. This may result in overestimation of smolt abundance when spawner numbers increase. On the other hand, the decreasing proportion of females seen in Ume/Vindelälven (Figure ) may be a river-specific problem, which when applied to Piteälven could result in underestimation of egg deposition and consequent smolt abundance. Further, the extra migration mortality among Piteälven smolts, known to occur when they have to pass the Sikfors dam, has so far not been accounted for. To accommodate for the above shortcomings, the method for calculating smolt priors for Piteälven was modified before this year s assessment. Same assumptions as previously were used for egg-to-smolt survival rate and smolt age distribution, but the information on spawners in Ume/Vindelälven was replaced with corresponding data from Torneälven/Tornionjoki (i.e. average number of eggs per spawner of both sexes and all age classes in that river). Density-dependence was accounted for by transforming estimated egg numbers into smolts following a Beverton Holt stock recruit function based on α=100 (i.e. same egg to smolt survival as before) and the recently updated carrying capacity (K=1/β) for Piteälven with associated 90% probability intervals (ICES, 2017). Based on congruent results from two telemetry studies in 2010 and 2015, smolt mortality when passing the Sikfors dam was set to 21%. The new and old smolt priors are depicted in Figure Until late 1990s, the difference is very small in absolute terms. Later, however, the new input is mostly higher than the old one, and for some years, it predicts more than additional smolts. The main reason is likely the female ratio in Ume/Vindelälven that since the late 1990s has been decreasing (Figure ). A model for M74 mortality provides input about mortality due to M74 into the life cycle model by analysing all data on incidence of M74 in the stocks (see Stock Annex, Section C.1.6). Figure shows the estimates for M74 mortality (median and 95% probability interval); within the last ten years, the mortality has decreased until the spawning year 2015 when it increased to the level of magnitude of 5 20%. The results from the 2016 spawning (Figure ) and the predictions made for 2017 spawning (Section 3.4) indicate similar level of mortality as in In general, the percentage of females with offspring affected by M74 overestimates the M74 mortality due to the fact that part of the offspring will die due to normal yolk-sac-fry mortality, unrelated to M74. Also, not all offspring necessarily die when affected by M74. Because of the decreasing trend in mortality among offspring of females affected by M74, the data on proportion of females affected by M74 especially overestimate M74 mortality in recent

6 ICES WGBAST REPORT years. Data on the total average yolk-sac-fry mortality are much better at tracking the general trend but overestimate the actual M74 mortality, because these data do not distinguish between normal yolk-sac-fry mortality and yolk-sac-fry mortality caused by the M74 syndrome. Table shows the actual values of the M74 mortality for the different salmon stocks. Figure illustrates the probability that offspring of M74-affected females would die, which has been possible to calculate for Simojoki, Tornionjoki and an unsampled salmon stock Status of the assessment unit 1 4 stocks and development of fisheries in the Gulf of Bothnia and the Main Basin The full life-history model (FLHM) was run with two chains for iterations after an adaptive phase of iterations. The first iterations were discarded as burn-in and the chains were thinned with an interval of 150 to yield a final sample size of 2000 (1000 iterations from each of two chains). Using the JAGS FLHM, convergence can now be assessed using metrics such as the Gelman-Rubin diagnostic, in addition to visual inspection of trace plots. Gelman-Rubin diagnostics indicated convergence for ~85% of model parameters (Gelman-Rubin diagnostic <1.2). Among key life-history and stock-status parameters, some annual post-smolt natural mortality rates (both wild and reared) as well as adult natural mortality rates (both wild and reared) had not reached convergence. Gelman-Rubin diagnostics further indicated lack of convergence for stock recruitment function alpha parameters for Torne River, Simojoki and Vindelälven, and several annual stock recruitment steepness estimates for the same rivers. Poor convergence was also noted for the estimated proportion of wild salmon in offshore catches in some years (corresponding to years in which post-smolt mortality rates showed poor convergence), maturation rates in some years (both wild and reared) and offshore abundances of salmon on May 1st. Caution must therefore be taken in the interpretation of results for these parameters/quantities. In the text and figures that follow, medians and 90% probability intervals (PI s), are used where possible as statistics of posterior probability distributions. The results indicate a decreasing long-term trend in the post-smolt survival until mid- 2000, after which survival has somewhat improved (Figure ). The lowest overall survival (median estimate around 6 8% among wild and 5% among reared smolts) was estimated for salmon that smolted in years and Low survivals were estimated for either wild or reared smolts also in some of the years , however, as pointed out above some of these parameters (mainly smolt years ) are not reliably estimated owing to the limited amount of MCMC iterations. After the last decade the survival has increased to 13 23% for wild smolts and 11 19% for reared smolts (median estimates in ). Survival improved especially among salmon that smolted in Currently survival is slightly lower than in the early 2000s, and less than half of the estimated survival level prevailing two decades ago. According to this year s assessment, the relative difference in survival of wild vs. reared post-smolts is much smaller than according to the earlier years assessments (see Section 4.2.1). The adult natural annual survival of wild salmon (median 93%, PI 88 96%) is estimated to be clearly higher than that of reared salmon (median 75%, PI 71 83%). Thus, the difference in total sea survival back to the spawning/stocking site for wild and reared salmon remains large also in this assessment, despite the smaller estimated relative difference in the survival of wild vs. reared post-smolts. Maturation of 1-sea winter salmon (grilse) has in most years been around 20% and 30 50% among wild and reared individuals, respectively (Figure ). Among 2-sea winter salmon maturation is estimated to have been mostly 30 60% (wild) and 40 75%

7 200 ICES WGBAST REPORT 2018 (reared) salmon. Also for 3- and 4-sea winter salmon the maturation rates of wild salmon have on average been somewhat lower that those of reared salmon, but the difference is small. The estimated maturation rates of 4-sea winter are on average lower than those of 3-sea winter salmon. This is against intuition but might be an artefact due to the inconsistency between current model assumptions (no repeat spawners, all fish mature at latest after five sea winters) and the biology of salmon (some repeat spawners exist and some salmon have a longer lifespan than five years at sea). The maturation rates were generally on low level around , but higher than average around and again around The full life-history model allows estimation of steepness of the stock recruit relationship (Table ) and the PSPC (Table ) for different salmon stocks. Figure gives an indication of river-specific stock recruit dynamics. The blue clouds in the figure panels indicate posterior probability distributions of all the historical estimates of yearly egg deposition and corresponding smolt abundance (the density of the cloud indicates the probability). Curves added in the figure panels are draws from the posterior distribution of the Beverton Holt stock recruit function. Adding the latest information about spawner and smolt abundance together with the latest changes in the model structure and priors of PSPC s has resulted in several changes in posterior probability distributions of the PSPC's, as compared to in last year (Figure , Table ). PCPC s of several rivers were significantly updated from last year s assessment especially in the AU 2. The largest updates were in the PSPC s of Öreälven (287% increase in median), Lögdeälven (252% increase), Piteälven (117% increase) and Sävarån (101% increase). In all these rivers, except in Sävarån, the priors for the PSPC s/k s have been updated much upwards, which probably explains most of increase in the posterior PSPC s (Section 4.4.2). Other remarkable updates (>10% change in median) to the PSPC s are seen in Emån (35% decrease), Rickleån (25% decrease), Ume/Vindelälven (20% decrease) and Kågeälven (18% decrease). There are no remarkable updates in the PSPC s of the rivers in AU 1. As pointed out above, care should be taken in the interpretation of these results because of changes in the assessment methodology (new stock recruit parameterization; see ICES, 2017d and Section 4.2.1). The total combined AU-specific PSPC estimates changed from the last year s assessment only by a few percent in AUs 1 and 2 (Table ). The PSPC estimates of AU 4 decreased by 15%, apparently mostly due to the various changes made to the input data of the AU 4 rivers. Total PSPC for AUs 5 and 6 were not updated in this year s assessment. The estimated grand total PSPC of AUs 1 6 (median 4.11 million) is only 3% (6000 smolts) higher than the corresponding estimate from the last year s assessment. Since the mid-1990s, the status of many wild salmon populations in the Baltic Sea has improved, and the total wild production has increased from less than 0.5 to over three million smolts (Figure , Table ). There are significant regional differences in trends in smolt production. For the wild salmon stocks of AUs 1 2, the very fast recovery of smolt production indicates high steepness for stock recruit relationships in these rivers. The recovery is most pronounced in the largest rivers, but recently also the salmon stocks spawning in the smaller forest rivers of the region (Åbyälven, Rickleån, Sävarån, Öreälven, Lögdeälven) have speeded up their recovery. However, their stock status (current production level against the potential) is generally assessed to be lower than that of the larger salmon rivers, as discussed below.

8 ICES WGBAST REPORT The only wild stock in AU 3 currently evaluated in the assessment model (Ljungan) has also recovered, but the estimates of both the current and the potential smolt production of this river are highly uncertain. Following the revision of the time-series of smolt production estimates used as input in the FLHM (Section 4.2.2), the perception about the development of AU 4 stocks has changed: the Mörrumsån stock has stayed relatively stable with only slight improvement seen towards the most recent years, while the abundance in Emån has been gradually increasing. Most of the AU 5 stocks are showing a decreasing trend in smolt abundance, but the stocks of AU 6 show improvements similar to in AU 3 (see Section 4.2.4). Smolt production in the AU 1 4 rivers is estimated to have jumped again to a higher level since 2017, which is a reflection of the further increase in the number of spawners in and beyond (Figure ). By comparing the posterior smolt production (Table ) against the posterior PSPC it is possible to evaluate current (year 2017) status of the stocks in terms of their probability to reach 50% or 75% of PSPC (i.e. R0 in 2017, Figures and , Table ). Table contains also AU 5 6 stocks and Testeboån, which are currently not included in the FLHM. These stocks have not been analytically derived, but expert judgments are used to classify their current status; see Sections (AU 5 6) and (Testeboån). The perception about the overall status of stocks (amount of stocks in different status classes) has markedly changed compared to the last year s assessment, which is probably a combined result of numerous changes made in the assessment model and by adding two more years of data ( ). All stocks in the AU 1 are estimated to have very likely reached 50% of their PSPC s, and three out of four stocks have likely or very likely also reached 75% of their PSPC s. The stock of Tornionjoki had very likely reached even 75% of its PSPC in 2017, when the smolt production in the river is estimated to have reached its all-time high. The lowest status in the AU 1 has been assessed for Simojoki: it is uncertain if the stock has reached 75% of its PSPC (Table ). Six out of nine stocks in the AU 2 are likely or very likely to have reached 50% of their PSPC s, but only three have (likely) reached the 75% target. The stock of Lögdeälven has unlikely reached even the 50% target, and Rickleån and Öreälven are uncertain to have reached this target. In AU 3, Ljungan is likely and uncertain to have reached 50% and 75% of PSPC, respectively, whereas Testeboån is uncertain and unlikely. In AUs 4 5, only Mörrumsån has likely or very likely reached both of the targets, whereas all the remaining 13 stocks are uncertain or unlikely to have reached even the 50% target (Table ). Out of the 41 assessed wild and mixed-stocks in Table , 34% (14 stocks) are likely or very likely to have reached 50% of PSPC, and 22% (nine stocks) are likely or very likely to have reached 75% of PSPC. The corresponding proportions calculated only for the 28 wild stocks are 50% and 32%. Generally, the probability to reach targets is highest for stocks in the largest northern rivers. A total of nine wild and 12 mixed-stocks are unlikely to have reached 50% of PSPC, i.e. they are considered to be weak. All except one of the weak stocks are located in AUs 5 6. While most of the AUs 1 2 stocks show strong indications of recovery over the years, the stocks in AUs 4 5 have mostly been unable to recover. Stocks in rivers situated between these areas (i.e. AU 3 and AU 6 stocks) have mostly shown modest indications of recovery (Figures , and Section 4.2.4).

9 202 ICES WGBAST REPORT 2018 The model captures quite well the overall historic fluctuation of catches in various fisheries (Figure ). However, the offshore catches from the early and mid-2000s become underestimated, and there is some tendency for the older part of time-series of the coastal catches to become overestimated. The model also does not fully capture the high river catches of the years The model is fitted to the proportion of wild and reared salmon (separately for ages 2SW and 3SW) in the offshore catches. The posterior estimates of wild vs. reared proportions follow rather closely the observed proportions (Figure ). However, for the MCMC was not able to reach convergence, which is reflected by the much more uncertain posterior estimates of those years than in other years. An increasing trend in the number of spawners is seen in most of the rivers of the AUs 1 4 (Figure ). Spawner abundance has increased, particularly in the years In Simojoki, the very high estimates of spawners around the turn of the millennium are a result of very intensive stocking of hatchery-reared parr and smolts in the river during the late 1990s. The model captures trends seen in fishladder counts, even short-term variation in rivers where the data are not used for model fitting (e.g. Byskeälven). Annual variation in river conditions affect the success of fish to pass through ladders and therefore the ladder counts themselves are not ideal indices of spawner abundance. For Ume/Vindelälven, however, fish counts are good approximations of the total amounts of fish reaching the spawning grounds, and the model based spawner estimates follow closely these observations. The good agreement between observations and estimates in the Ume/Vindelälven is expected because of the assumption that all spawners are counted in this river (Section 4.2.1). In Piteälven, the agreement is not particularly good, however, although all spawning grounds are located upstream the counter. In this river, spawner counts are currently not used directly for the assessment. One reason for the discrepancies may therefore be that the proportion of ascending spawners being counted fluctuates much between years (currently unknown) and this may result in mis-matches between model estimated and observed spawner numbers. In Kalixälven, Åbyälven and Rickleån the development of spawner abundance estimated by the model appears more optimistic than the development observed in the fishladder counts. In Kalixälven, the counter is located about 100 km from the river mouth with large spawning areas downstream. In Åbyälven and Rickleån fishladders are built up around the turn of the millennium and salmon are gradually repopulating the upstream sections of these rivers. Therefore, counts in these rivers account for a small fraction of the total spawner population and the counts may not well represent the actual development of the salmon stocks. The general synchronous drops and increases in the observed spawner counts are well-captured by the model, also the most recent drop observed from 2016 to This is probably a consequence of fitting the model to spawner counts in combination with assuming annually varying maturation rates; maturation rates are estimated to be low preceding poor spawning runs and high preceding high spawning runs (Figure vs. Figure ). Despite some fluctuations, there was a strong long-term decreasing trend in the harvest rate of driftnets until the total ban of this gear type in 2008 (Figure a). The combined harvest rate of longlining and trolling has been fluctuating much with peaks around the years 1990, 2000 and In the last 5 6 years this harvest rate has, however, stayed on the long-term average level without any clear trend. Recreational salmon trolling has been increasing (Section 2.1.2), especially during the 2010s, and it currently accounts for roughly half of the combined harvest rate of longlining and

10 ICES WGBAST REPORT trolling fishing (cf. ICES, 2017). Since the early 2000s the coastal harvest rate has decreased almost continuously, and after 2015 the harvest rate has stayed on its all-time low without any further decrease (Figure b). Estimates of harvest rates in the rivers are inaccurate and lack trends (Figure c). River-specific data indicate that there can be substantial variation in the harvest rate between rivers (Section 3.2.1), which is currently not taken into account in the model. The overall harvest rates (all gear types regionally combined) have been decreasing in both offshore and coastal fisheries (Figure ). However, most recently these trends have levelled off and the harvest rate in the offshore fishing shows even a slight increase in the last two years Status of the assessment unit 5 6 stocks Smolt production in relation to PSPC in the AU 5 stocks shows a negative trend in almost every wild and mixed river (Figures and ). During the last decade, smolt production dropped from 50% or higher to below 50% of PSPC. Thereafter smolt production has stayed on this low level except for in , when a sudden temporal increase was observed in most rivers (Figure ). In 2017, most AU 5 rivers were estimated to produce just about 10 30% of their PSPCs and they are therefore either unlikely or uncertain to reach 50% (given the associated uncertainties in estimation; Table ). In river Pärnu the smolt production has shown small signs of improvement. The second river in AU 5 which shows limited positive development is Nemunas. This is a large watercourse with several tributaries, and many of them have been subject to long-term restoration efforts (habitat restorations, restocking, etc. see Sections and 3.2.2). Despite the positive trend, the observed smolt production in the Nemunas in relation to PSPC is still far below 50% level. Rivers Salaca (AU 5) and Mörrumsån (AU 4) are both well-known salmon rivers with the most extensive and longest time-series of monitoring data in the Main Basin area (Sections and 3.1.5). The developments of parr densities in these two rivers roughly resemble each other since the early 1990s; an increase in the densities from the early to the late 1990s and a subsequent decrease starting in the early 2000s. Smolt production in the AU 6 stocks shows positive trends in most rivers but also a large interannual variation, especially in the smallest rivers (Figures to ). Among wild (Figure ) and mixed (Figure ) Estonian stocks the clearest positive trend exists in least two of the wild ones (Keila and Kunda) which have reached 75% of their PCPCs. However, smolt production in wild Vasalemma remained below 50% of PSPC in 2017 (Figure ). In the small Estonian mixed-stocks the trend has also been is also positive in recent years (Figures and ). However the current PSPC in some of these rivers is severely limited by migration barriers and there is also a lot of annual variation in these small populations. PSPC in mixed River Valgejõgi has increased since 2016 (from 1500 smolts to ) because salmon regained access to all potential historical spawning and rearing areas. The size and quality of spawning areas will be investigated in detail during In the Finnish mixed river Kymijoki no clear positive trend can be seen, although occasional stronger year classes have occurred. The smolt production has nevertheless remained far below the 50% level. In Russian river Luga wild smolt production is stable but low, and it has remained below 10% of PSPC despite large-scale annual smolt releases using salmon of local origin (Figure ).

11 204 ICES WGBAST REPORT Harvest pattern of wild and reared salmon in AU 6 Salmon originating in the Gulf of Bothnia and Baltic Main Basin contribute to the catches in the Gulf of Finland (Bartel, 1987; ICES, 1994). Salmon from the Main Basin stocks migrate to the Gulf of Finland for feeding, and salmon from Gulf of Bothnian stocks visit the Gulf of Finland area in early summer during their spawning migration to the Gulf of Bothnia. In (excluding and 2016) samples has been collected from Finnish commercial fisheries in the Gulf of Finland. These catch samples have been aged and wild/reared origin have been determined by scale reading. Stock proportions were also estimated by DNA-analysis (MSA). The MSA results from the earlier years ( ) suggested that the largest stock contribution (50 60%) was from locally released reared Neva salmon, whereas the average proportion of wild stocks originating in the Gulf of Bothnia was 40 55% (Section 2.8). In 2015 and 2017 the overall proportion of reared Neva salmon has been higher (up to 67%) whereas the share of wild GoB salmon was lower (10 15%). It should be noted that there were pronounced differences between sampling sites and sampling times between the years (Section 2.8). The share of Gulf of Bothnian salmon was clearly higher during the early fishing season (June), whereas the share of GoF Neva salmon was high later in the season. So far, apart from Neva salmon, the proportion of other Gulf of Finland stocks (Russia and Estonia) in genetically analysed catch samples from the area have been estimated to zero or close (<0.5% Kunda in 2017, Table 2.8.3). The numbers of feeding salmon from these wild and mixed rivers are expected to be low, and the probability to observe them is probably minimal in samples collected from fisheries in the feeding area in the Gulf of Finland (and the Main Basin). According to Carlin tag recaptures from releases made in Estonian rivers in the area (smolt cohorts ), only 19% of the stocked fish are harvested outside Gulf of Finland, 68% are harvested in the Gulf of Finland s Estonian coast and 13% of the recaptures originate in the Finnish side of the gulf (ICES, 2014). A substantial share of these returns, however, came from recreational fishery off the coastal area (trolling, etc.). The reduction of harvest rate in the Main Basin in the last few years has had a positive effect on the AU 6 wild stocks. The harvest rate in the Main Basin (driftnetting and longlining combined) was estimated to be 30 60% in 1990s, while currently the harvest rate (longlining and trolling combined) is estimated to be around 15 20% (Figures a and ). Most Estonian stocked parr and all stocked smolts have been adipose finclipped since late 1990s. The share of adipose finclipped salmon in Estonian coastal catches is monitored by gathering catch samples. If comparing the relative production of wild and reared smolts with the share of finclipped fish in coastal Estonian catch samples, it shows that the share of finclipped fish is clearly smaller than expected, and that the situation has remained relatively stable since 2010 (Figure ). This indicates that reared fish have had very low survival since 2010, and that wild fish are harvested in significant numbers. However, the river origin of the wild fish is not known. To further reduce the harvest rate on the regions stocks, the closed area at Estonian river mouths was in 2011 extended to 1500 m during the main spawning migration period (from 1st September to 31st October) in all wild (Kunda, Keila, Vasalemma) and most mixed rivers (Selja, Loobu, Valgejõe, Pirita, Vääna, and Purtse). Harvesting in the Main Basin has declined, particularly since Taking into account a rather large proportion of salmon from the Gulf of Bothnia observed in Finnish catch samples from the Gulf of Finland (Section 2.8) the exchange of salmon between areas

12 ICES WGBAST REPORT is considered to be significant, although the total magnitude still remains to be quantified. Comparison of the spatial distribution of tag recaptures from Gulf of Bothnian and Gulf of Finland stocks provides a qualitative overview on the rate of exchange (ICES, 2014), although this information is dependent not only on the distribution of salmon but also on the distribution of fisheries. As shown above, status of Estonian wild and mixed salmon stocks has shown improvements since 2005, followed by recent declines for mixed stocks since 2015 (Figures and ). 4.3 Stock projection of Baltic salmon stocks in assessment units Assumptions regarding development of fisheries and key biological parameters Table provides a summary of assumptions on which the stock projections are based. The basis has been kept as similar to the last full assessment (ICES, 2017) as feasible, in order to allow for a review of how the new information is affecting projections. However, inclusion of salmon trolling in offshore fisheries and some other changes made to the FLHM required further consideration in the scenarios, as described below. Fishing scenarios The base case scenario (scenario 1) for future fishing (2019 and onwards) equals to the commercial catch adviced by ICES for 2018, i.e. the median commercial removal would equal to salmon. Scenarios 2 and 3 correspond to a 20% increase and 20% decrease from the scenario 1, respectively. Scenario 4 equals to an F=0.1 harvest rule, applied for total commercial removals. Scenario 5 illustrates how recreational fishing alone would affect stock development. Finally, scenario 6 illustrates stock development in case all fishing (both at sea and in rivers) was closed. Similar to in previous years, fisheries in the interim year (2018) follow the scenarios, except for longline fishing during the first months of the year, which is estimated based on the effort observed during the corresponding months of Scenarios were computed by searching an effort that results in a median catch that corresponds to the desired total sea catch (depending on the scenario) in the advice year (2019). For example, in scenario 1 the total sea catch ( salmon) consists of total commercial sea catch ( salmon) and total recreational sea catch ( salmon). The recreational sea catch in 2019 is the same in all scenarios (Table ) except in scenario 6, which assumes closure of all fisheries. In scenarios 1 5, recreational fishery has the same effort in future years as in 2019, meaning that the annual recreational catches are proportional to the abundance. The recreational catch of salmon in 2019 consists of an estimated three year average ( ) offshore trolling catch ( salmon) and reported recreational catches other than offshore trolling in 2017 (9400 salmon). As the current model framework does not allow inclusion of recreational fisheries as a separate fishery, it is technically included as a part of offshore longline fishery, as described in Section As the scenarios are technically defined in terms of future fishing effort, the predicted catches have probability distributions according to the estimated population abundance, age-specific catchabilities and assumed fishing effort. Scenarios 1 4 assume the same fishing pattern in commercial fisheries (division of effort between fishing

13 206 ICES WGBAST REPORT 2018 grounds) as realized in Figure a b shows the harvest rates prevailing in the scenarios. In all scenarios it is also assumed that the commercial removal reported under the TAC covers 55% of the total commercial sea fishing mortality, whereas 45% of this mortality consists of discards, misreported, and unreported commercial removals. This corresponds to the situation assessed to prevail in 2017 (Figure ). According to expert evaluation, misreporting has increased significantly from 2016 to The share of misreporting out of the total commercial catch was evaluated to be 16% in 2016, whereas the corresponding share is evaluated to be as high as 29% in 2017 (Section 2.3.3). The share of other sources of extra mortality (dead discards and unreporting) is evaluated to remain roughly the same from 2016 to Because of the change in misreporting, the share of total removal originating from the reported commercial catches results become 13 percentage units smaller in 2017 than in Survival parameters In both the M74 and the post-smolt mortality (Mps) projections, an autoregressive model with one year lag (AR(1)) is fitted at the logit-scale with the historical estimates of the survival parameters. Mean values of the mean of the post-smolt survival over years (19%), variance over the same time-series and the autocorrelation coefficient are taken from the historical analysis into the future projections. The method for M74 is similar, but the stable mean for the future is taken as the mean over the whole historical time-series (85%). In addition, the forward projection for Mps is started from 2017 to replace the highly uncertain model estimate of the last year of the historical model. The starting point of M74 projections is Time-series for Mps and M74 survival are illustrated in Figure Adult natural mortality (M) is assumed to stay constant in future, equalling the values estimated from the history. Different fisheries occur at different points in time and space, and many catch only maturing salmon, which has been subject to several months natural mortality within a year. Thus, in order to increase comparability of abundances and catches, the abundances at sea have been calculated by letting M first to decrease the PFA (stock size at the beginning of year) of multi-sea-winter salmon for six months. Moreover, the stock size of grilse has been presented as the abudance after the period of post-smolt mortality and four months of adult natural mortality. This period is considered because the post-smolt mortality period ends in April, after which eight months of that calendar year remain during which grilse are large enough to be fished. Half of that period, i.e. four months, is considered to best represent the natural mortality that takes place before the fishing. Calculations for the F=0.1 scenario (Scenario 4) are also based on stock sizes which are first affected by M, as described above. Maturation Annual sea-age group-specific maturation rates are given as the average level computed over the historical period, separately for wild and reared salmon. This projection starts from 2019, as the maturation rates of 2018 can be predicted based on sea surface temperature (SST) information from early 2018 (ICES, 2014, Annex 4). The time-series of maturation rates are presented in Figure Releases of reared salmon The number of released reared salmon per assessment unit is assumed to remain at the same level in future as in 2017 (Table 3.3.1).

14 ICES WGBAST REPORT Results According to the projections, stock size on the feeding grounds (pre-fishery abundance, PFA) will be about 1.66 ( ) million salmon (wild and reared, 1SW and MSW fish in total) in 2019 (Figure a b). Of this amount, MSW salmon (i.e. fish which stay on the feeding area at least one and half years after smolting) will account for 0.76 ( ) million salmon. These MSW fish will be fully recruited to both offshore and coastal fisheries in From the predicted amount of 1SW salmon (0.84 million, million) at sea in spring 2019, a fraction (most likely 20 40%) is expected to mature and become recruited to coastal and river fisheries, while the rest of the 1SW salmon will stay on the feeding grounds and will not become recruited to the fisheries until next winter. The abundance of wild salmon at sea has fluctuated without any apparent trend until During the current decade the abundance has on average been higher than before, at or above one million (according to median values for 1SW and MSW wild salmon combined) (Figure ). Within the range of the scenarios, the abundance of wild salmon is predicted to stay with high probability on this elevated level in future. As one of the simplifying assumptions of the life cycle is that all salmon die after spawning, a lower maturation rate will increase the survival of the cohort to the next year compared to years with the same abundance but with average maturation. Similarly, a high maturation rate will decrease the abundance of MSW salmon in following years. Because of this feature, it is important to note that the predicted abundance may easily become over- or underestimated because of the (predicted) development of maturation rates. In contrast to wild salmon, the abundance at sea of reared salmon strongly decreased from the mid-1990s to the late 2000s, mainly due to the decline in post-smolt survival. In some occasional years in the early 2010s, substantial amounts of reared salmon have been assessed to recruit to the fisheries (which may be an artefact due to the poor estimation of e.g. Mps in those years, see Section 4.2.3), but thereafter the abundance has stayed on a rather low level, and it is predicted to stay low also during the coming years. Further reduction in the amount of reared salmon may take place in future, if the long-term declining trend since the early 2000s in the amount of stocked smolts will continue (Table 3.3.1). The combined wild and reared abundance (PFA) also declined substantially from mid-1990s until late 2000s, but thereafter the total abundance has increased and is expected to stay on this elevated level in future (Figure ). Table shows the predicted total catch by scenario for 2019, divided into the following components: commercial wanted sea catch, consisting of reported, unreported and misreported; commercial unwanted sea catch, consisting of discarded undersized and seal damaged salmon; recreational sea catch; and catch in the rivers. The table also shows the predicted fishing mortality (separate F of commercial fishing and F of all sea fisheries) as well as the predicted number of spawners in 2019 for the given fishing scenarios. The amount of unreporting, misreporting and discarding in 2019 is based on the expert evaluated share of those catch components compared to the reported catches in 2017

15 208 ICES WGBAST REPORT 2018 fisheries. In 2017, the wanted catch reported (commercial) accounted for about 55% from the corresponding estimated total commercial sea catch, this percentage being remarkably smaller than the one estimated for the three previous years (64 72%). Unreporting, misreporting and discarding in 2017 are considered to take, respectively, about 6%, 29% and 10% share of the total commercial sea catch. The share of the total catch by its components for the period is illustrated in Figure It is important to keep in mind that future changes in either fishing pattern or in fisheries control may easily lead to changes in the share of catch caught under the quota regulation. With the given set of scenarios (excluding the scenarios zero fishing and recreational fishing only ), the predictions indicate that the wanted catch reported (commercial) in year 2019 would be 56 95% ( salmon) compared to the TAC of 2018 (Table ). The corresponding total sea removal (including recreational fishing) would range from salmon. The harvest rule of F0.1 for commercial catch (scenario 4) results in the highest catch among the examined scenarios, indicating a wanted catch reported of salmon and about 8% smaller spawning stock than under Scenario 1. The amount of spawners would be about 5% higher in Scenario 3 than in Scenario 1, and the zero fishing scenario indicates about 57% increase in the number of spawners compared to the scenario 1. The scenario recreational fishing only illustrates the magnitude of the current level of the recreational fishing which is predominantly angling in rivers and trolling at sea: recreational fishing alone would decrease the number of spawners by 22% compared to the zero fishing scenario. Figure illustrates the longer term development of (reported) future catches given each scenario. Figure a d presents the river-specific annual probabilities to meet 75% of the PSPC under each scenario (note that river Testeboån is left out from river-specific results because it is currently not included in the FLHM; but see Table for Testeboån s current status). Under the scenarios 1 4, different amount of fishing has some influence on the level but not on the trend of the probability of meeting 75% over time. Only the zero fishing and recreational fishing only scenarios diverge clearly; several of the weakest rivers show a stronger positive effect in trends than for the other scenarios. As expected, changes in fishing has the smallest effect to those stocks that are close to their PSPC. As the overall level of fishing effort is rather low in these scenarios compared to in history, the examined range of fishing mortality only results in modest impacts on the chances of reaching the management objective. Table compares the probabilities of reaching 75% target around the years , which are approximately one full generation ahead from now. Evidently, the probabilities are higher for effort scenarios with low exploitation, but differences between scenarios are small except for the recreational fishing only (Scenario 5) and zero fishing (Scenario 6) scenarios. Figure a b illustrates by scenario the rate and the direction of change in smolt abundance in 2023/2024 compared to the smolt abundance in Future predictions about smolt abundance are naturally more uncertain than the estimated abundance in However, in those stocks that are close to their PSPC, also the predictions are rather certain, indicating that smolt abundance will stay close to PSPC in these rivers under different fishing scenarios. Figures a d show longer term predictions in the river-specific smolt and spawner abundances for three scenarios (1=removal which corresponds to ICES advice for 2018; 4=harvest rule of F0.1 for commercial catch; and 6=zero fishing). The two most

16 ICES WGBAST REPORT extreme scenarios (4 and 6) illustrate the predicted effects of contrasting amounts of fishing. 4.4 Additional information affecting perception of stock status Independent empirical information is important for the evaluation of model predictions and their key parameters. Over the years, repeated comparisons with different kinds of such independent information have been performed, and in several cases, these comparisons have prompted modifications or extensions to the full life-history model. For example, some years ago sea temperature data were introduced as a covariate of age-specific maturations rates, based on the analyses and development work carried out in the last inter-benchmark protocol (ICES, 2012b) and thereafter. Also, comparisons between model predictions and empirical results from genetic mixedstock analyses (MSA) have been used over the years to verify model performance (e.g. ICES, 2014). This section focuses on other auxiliary information important to complete evaluation of the current stock status. In particular, we highlight information about diseases and other factors that may affect development in stock status, but which are not fully taken into consideration in the current modelling. Likewise, weaknesses in input data used in the assessment model might affect the precision of status evaluations, and in the worst case introduce biases. Such shortcomings in the current assessment model, when it comes to input data and ways of handling those, are also discussed under this section. An example is the ongoing work of updating prior information on production areas and potential smolt production levels in salmon rivers, which may affect status evaluations of individual stocks Potential effects of M74 and disease on stock development If the increase observed in M74 in (and predicted in 2018; Section 3.4.1) should last for several years, this may gradually result in decreased stock status and reduced fishing possibilities. Occurrence of M74 more than half a year ahead (thiamine level in the spawned eggs indicates quite well M74 mortality among offspring hatching from these eggs, see Section 3.4) cannot currently be predicted, but many of the M74- fluctuations seen since the early 1990s have tended to last for some years before changing in direction (Figure 3.4.3). Also, the disease outbreaks reported in several rivers in recent years (Section 3.4) is a concern for the future. In contrast to M74, the cause(s) of the disease is still unknown, and to accurately quantify the amount of affected or dead salmon in a river appears difficult, if at all possible. Although future development and effects of M74 and other health issues are hard to predict, the development of individual stocks could depend on their current stock status. In populations where smolt production is approaching PSPC, density-dependent mortality is expected to become higher. Hence, in a recovered stock (with high status) elevated fry mortality may partly be compensated by the reduced density-dependent mortality among the offspring not affected by M74. For the same reason, stronger stocks may be less sensitive to a reduced number of deposited eggs due to adult female mortality. In contrast, among weaker populations the effects of M74 and other diseases on future smolt production could be more pronounced, because of the lack of the above described buffer effect of density-dependence in the reproduction dynamics. It should be emphasized, however, that so far empirical evidence in support for compensation of M74-related mortality is lacking.

17 210 ICES WGBAST REPORT 2018 Despite the recent increase in M74 and disease outbreaks, average salmon 0+ parr densities in have remained at seemingly normal levels in most AU 1 4 rivers, with the exceptions of Vindelälven and Ljungan. In Vindelälven the average 0+ density dropped drastically, from ca. 40 parr/100 m 2 in 2015 to only ca. 1 parr/100 m 2 in 2016, the lowest density observed since the 1990s (Table ). In 2017, the average 0+ density remained at a very low level (ca. 4/100 m 2 ). The reason for the decline is unclear, but likely reflects a combination of factors. In 2015, only 790 females were counted in the Norrfors fish ladder, which represented just 11% of the spawning run (18% among MSW salmon, if assuming 6% females among grilse). There was no indication of such a skewed sex ratio in the sea or at the river mouth. Hence, the recent disease problems in Ume/Vindelälven may for some reason have prevented particularly females from reaching the fish ladder. In 2016, the number of females counted was higher (2741; Table ), but a large proportion of the salmon passing the ladder had severe skin problems (fungus infections) and many died soon after having been counted (see Section on how this additional mortality has been handled in the assessment). Moreover, low levels of thiamine among spawners in 2015 and 2016 resulted in increased M74-mortality in the following hatching years (19% and 45% females affected in Vindelälven 2016 and 2017, respectively; Table 3.4.1). Also in Ljungan the average 0+ salmon density in 2017 was very low (<one parr/100 m 2 ) compared to in preceding years (average density of 61 in ; Table ). Notably, the collapsed parr density in 2017 followed after a year with many dead salmon observed in the river, combined with a high expected level of M74-mortality. The very low parr densities in Vindelälven ( ) and Ljungan (2017) are expected to result in a drastically reduced smolt production in However, it should be noted that the estimated pre-fishery abundance of Vindelälven salmon exploited in the fishery during the advice year (2019) is not affected by the reduced parr densities in Regardless, the situation in Ume/Vindelälven is alarming, and local management actions aimed at protecting ascending spawners appear warranted Revision of basic input data Colonization of salmon to new areas further upstream and/or restoration efforts improving or increasing river habitats will increase the potential smolt production capacity (PSPC) of rivers. If such changes are not accounted for, the status assessment will likely become biased. WGBAST is continuously revising important input data, such as e.g. production areas, to avoid such biases in status assessment. Factors affecting the PSPC include river production area, smolt production potential per unit area and mortalities during downward smolt migration. In the analytical assessment of Baltic salmon, all these quantities are used to formulate river-specific prior probability density functions (hereafter called priors ) for PSPC, which are updated by the model to posterior PSPCs when stock recruit data are included. Status of individual stocks is evaluated by comparing posterior estimates of current smolt production levels with posteriors of PSPC. In last year s report, we updated figures on production areas and information on maximal smolt production per unit of area for three Swedish rivers: Piteälven, Lögdeälven and Öreälven. The updated information on production area and smolt production potential per unit area was used in combination with information on other important factors, such as mortality during smolt migration, to formulate new priors for PSPC (ICES, 2017). These updated priors have now been included in the assessment model (Section 4.2).

18 ICES WGBAST REPORT Preparation for inclusion of Testeboån in the assessment model The work to revise model input data has continued also this year. Testeboån received the status of a wild salmon river in 2013, but the stock has not yet been included in the assessment model. A PSPC prior for Testeboån was formulated using expert opinions about relevant variables. A simple model (the same model used in previous years for e.g. Öreälven and Lögdeälven, see Annex 4 in ICES, 2015 for more information) was used to derive a probability distribution for PSPC as a function of the expert-elicited variables. The derived median value for the PSPC prior was 8895 (90% PI: ). To obtain smolt priors that would allow for inclusion of Testeboån in the assessment model, the river was included in the same river model as used to produce smolt prior estimates for Emån and Mörrumsån. For a detailed description of this Southern river model, see ICES, 2017d. Data from Testeboån comprised a time-series of electrofishing data for the period and smolt counting results from the years Derived estimates of smolt priors for Testeboån are presented in Table Because of time limitations, Testeboån was not included in the full life-history model this year, which means that we cannot properly assess status and evaluate the development of this stock in the near future (following the different fishing scenarios evaluated in Section 4.3). The river will be fully integrated in the assessment work in However, a preliminary status evaluation based on the priors for PSPC and current (2017) smolt production indicates that it is uncertain whether Testeboån has reached the 50% objective, and unlikely that the 75% objective has been reached (Table ) Updated reference points for management (stock-specific MSY targets) Suitable management reference points for Baltic salmon were explored in the Workshop on Baltic Salmon Management Plan Request (WKBALSAL, ICES, 2008); where it was proposed that the limit of natural smolt production should not be lower than 75% of the estimated R 0 for each stock. This 75% R 0 proxy was revisited at the 2016 assessment meeting (using results from the 2015 assessment), following development of a simulation algorithm to approximate MSY. However, the results from that exercise were inconclusive, e.g. estimated smolt production at MSY of 40 smolts (0% of R 0 ) for Emån (cf. 24% in ICES, 2008). This appeared to be related to problems identified with the stock recruitment model outlined above (estimated EPR 0 for Emån 2 3 times greater than that for most other stocks). The exercise was repeated during the benchmark, for a subset of stocks (Torne River, Simojoki, Emån and Mörrumsån) to investigate the effect of different stock recruitment parameterizations on estimated reference points (ICES, 2017d). While those results are comparable with each other, it is necessary to perform the exercise with the results of the full assessment model comprising all observation models (since this can affect estimates of stock recruitment parameters, survival, etc.). The procedure used to obtain reference points is described in ICES (2017d). Briefly, an optimisation routine is used to find the fishing effort (and mortality) that maximises the long-term stable catch in forward projection of stock dynamics. The distribution of effort among fisheries is assumed to reflect the status quo. Note that alternative distributions of fishing effort across fisheries for immature vs. mature fish may result in different estimates of MSY and associated measures of abundance. Recreational trolling effort was added to offshore longline fishing effort as in the scenarios for future stock development. Specifications for future vital rates are the same as in future fishing scenarios (Table ).

19 212 ICES WGBAST REPORT 2018 Results of simulations are shown in Table Note that R0 (PSPC) estimates presented below are the long-term average smolt production from projections with F = 0, and are slightly lower than assessment model estimates because of slightly lower survival rates etc. used in future projections. Estimates of the smolt production corresponding to MSY as a proportion of RR 0 ( RRMMMMMM ) were fairly variable among stocks ranging RR 0 from 49% to 80% (Table ). Thirteen stocks had an MSY proxy lower than 0.75RR 0, while three stocks had a higher MSY proxy. These results indicate that the use of a single stock-wide reference point is consistent with a precautionary approach for many but not all Baltic salmon stocks. Estimates of RRMMMMMM were generally lower than those reported in ICES, 2008 (Table , RR 0 ICES, 2008 and Table ), with the exception of Emån and Mörrumsån. This is likely the result of higher estimates of stock recruit steepness in 2018 compared with 2008 (because of higher assumed M74 and post-smolt survival in projections, higher prior steepness, higher estimated adult survival etc. in 2018). Note that in ICES (2008), the 75th percentile of the distribution for RRMMMMMM appears to have been reported together RR 0 with the 25th percentile for the harvest rate at MSY corresponding to a 25% risk level (e.g. a 25% probability that the true RRMMMMMM is greater than the estimated value). Here, RR 0 medians are reported (Table ). The results shown above are not directly comparable to those obtained in the benchmark (ICES, 2017d) owing to the fact that only a subset of stocks and incomplete assessment model were used in that earlier exercise. The variability in estimates of RRMMMMMM among stocks appears to be related largely to variability of Beverton Holt alpha parameters (since estimated EPRs are on the whole sim- RR 0 ilar among stocks, except for Ume/Vindelälven), where all else being equal, stocks with a lower posterior estimate of alpha (higher maximum egg survival) tend to have more curved stock recruitment relationship corresponding to higher RR MMMMMM and higher RRMMMMMM. RR Conclusions For most rivers included in the FLHM (i.e. rivers in AU 1 4), the smolt production is expected to stay at relatively high levels in the coming years. Also, the prefishery abundance is expected to increase slightly in the near future, indicating possibilities for maintained exploitation levels under Results from the stock projections indicate that the current exploitation rate will result in more or less positive developmental trends of all AU 1 4 stocks (Section 4.3.2). In addition, projections indicate that changes in sea removal of +/-20% have rather small effects on the development of these stocks, further indicating that fishing mortality is currently at fairly low levels compared to other (natural) sources of mortality. Obviously, probabilities to reach the objectives are higher for scenarios with lower exploitation, but differences between scenarios are small except for the ones with zero fishing and recreational fishery only. Wild stocks in AU 6 have also shown a positive development in recent years, indicating that current exploitation levels are compatible with a successive recovery of these stocks. There are, however, concerns for the development of some wild salmon stocks. In particular, a majority of the AU 5 stocks have not responded positively to previous reductions in fisheries exploitation, and many stocks in this area are still far below a good state, indicating that current exploitation and natural mortality rates do not allow for a recovery.

20 ICES WGBAST REPORT Within the current management of Baltic salmon, there are no rules or guidelines for how fast (within which time frames) weak salmon stocks should recover, or when a certain proportion of all stocks should have obtained their management goal. Therefore, under current conditions with only one TAC for SD and many stocks with variable status, any catch advice for the mixed-stock fishery on Baltic salmon will be associated with trade-offs (and some degree of subjectivity). For some weak stocks, additional measures (on top of restrictions through the TAC system) need to be directed to increase number of spawners, for example by reducing fisheries on mixedstocks in the Main basin (to reduce the exploitation of weak AU 5 stocks) and on the migration routes (e.g. close to the river mouth) where their share in catches becomes higher. Measures focused on the freshwater environment, such as work to improve river habitats and migration possibilities and actions to reduce poaching, may also be necessary to increase status. Thus, special actions (not only fishery-related ones) directed to the weakest stocks are likely required at the adviced TAC levels, especially in AU 5 but also for a few weak rivers in other AUs, to enable these stocks to recover. M74-mortality has increased in recent years, as well as reported deaths of spawners due to disease problems (Section 4.4.1). If the higher levels of M74 should prevail or increase further, this may gradually result in decreased stock status and reduced fishing possibilities, and may easily counteract any positive effects of higher-than-expected post-smolt survival, especially in weaker stocks. The two Swedish rivers Vindelälven and Ljungan have been particularly affected by disease problems, and the recruitment of parr has dropped markedly (see Section 4.4.1). National and local management organisations of these two rivers should consider introducing measures to increase number of spawners, for example by reducing exploitation rates on migrating spawners in the rivers and in coastal areas outside the river mouths. Also Mörrumsån has reported substantial disease problems (affected and dead adults) in recent years, but so far the parr densities have not decreased as dramatically as in the other rivers mentioned. Several of the northern stocks are close to or above the MSY-level (2017 smolt production; Table ), and the surplus produced by these stronger stocks could in theory be directed towards stock-specific fisheries. However, the current management system, with a single TAC for SD that is set at a relatively low level to safeguard weaker salmon stocks, prevents this surplus to be fully utilised by the commercial sea fishery. In a similar way, the surplus of reared salmon cannot be fully utilised today because reared salmon is also included in the TAC. Stock-specific management could be developed further, by implementation of more flexible systems for regulation of commercial fisheries with the aim of steering exploitation towards harvesting of reared salmon and stronger wild stocks, through e.g. areaspecific quotas and/or exclusion of certain single-stock fisheries from the quota system (such as fisheries in estuaries of rivers with reared stocks). Also, non-commercial coastal fishing, not regulated by international quotas, could be steered towards stockspecific harvesting. In contrast, the increasing recreational trolling in Main Basin is a true mixed-stock fishery where stock-specific harvesting is not possible, although regulations that only allow landing of finclipped (reared) salmon, such as has been implemented in Sweden since 2013, can somewhat reduce fishing mortality for wild stocks (given that post-release mortality is relatively low). A higher degree of stock-specific exploitation will also be necessary in future, if different management objectives should be decided upon for individual stocks (e.g. if to allow for a larger number of spawners than needed to fulfil the MSY-level in certain wild rivers).

21 214 ICES WGBAST REPORT Ongoing and future development of the stock assessment Benchmark The benchmark of Baltic salmon (WKBaltSalmon) convened into two workshops at ICES (Copenhagen, Denmark), a data evaluation workshop in autumn 2016 and a method evaluation workshop early in The work carried out is presented in a separate benchmark report (ICES, 2017d). The data evaluation workshop resulted in e.g. an updated description of available river monitoring data and a time-series of preliminary trolling catch estimates trough an expert elicitation. In addition, shortcomings in fisheries data were identified, and some outlining for an improvement work plan was done. A planned transfer of commercial fisheries data into ICES database InterCatch largely failed, however, and the aim is to solve the catch data issues in connection to next data call in late The method evaluation workshop focused on three major aspects: stock recruitment model selection, development of a smolt production model for southern Baltic Sea rivers, and an evaluation of the 75% objective as a proxy for MSY. The new parameterization of stock recruitement dynamics and a new smolt production model for southern Baltic Sea rivers have been implemented in this year assessment (see Section 4.2). An evaluation of the 75% objective as a proxy for MSY is presented in Section Further developments of the methodology, which have not yet been implemented in the assessment, are presented in the section below Road map for development of the assessment The tasks listed below refer to ongoing, planned and potential updates of the assessment methodology. Issues that were included in the benchmark assessment (WKBaltSalmon; ICES, 2017d) are indicated. Ongoing and short term Add Testeboån to the FLHM. In this year s report, the status assessment of river Testeboån is only preliminary and e.g. based on expert opinions on PSPC. The plan is to include the river in the FLHM Necessary preparations for this inclusion have already been carried out (see Section 4.4.2). Inclusion of the recreational sea fishery (mainly trolling) as a separate fishery (part of WKBaltSalmon). At present, trolling catch estimates are added to the offshore commercial ones in the FLHM (see Section 4.2). Because of the increase in the recreational trolling fishery at sea, it would be desirable to model recreational trolling as a separate fishery. This will require good quality data and catch estimates from countries with a significant recreational trolling fishery. During the benchmark, a plan for how to collect data and include recreational catches in the assessment model as a separate fleet was discussed and decided upon. Adding repeat spawners to the FLHM. Salmon are currently assumed to die after first spawning in the FLHM. This assumption is known to be unrealistic (repeat spawners in some stocks now account for ~10% of all spawners). This is likely to cause bias in estimated parameters such as survival rates and stock recruitment parameters, as well as potentially management reference points. Use spawner counting observations for Piteälven in the FLHM. Use of spawner counting observations directly in the FLHM for Piteälven would constitute

22 ICES WGBAST REPORT a more consistent use of available data, requiring fewer assumptions about stock recruitment parameters, etc. external to the model. Smolt production estimates would then be treated as missing for Piteälven. Medium-term, important issues planned to be dealt with in the next 2-3 years Continuing the work of including data from established index rivers and expanding data collection in other rivers. Some of the datasets collected in index rivers are still not used in the assessment model, such as e.g. spawner count data from River Mörrumsån. To improve precision in assessment results, there is also a need to increase collection of abundance data in non-index rivers. Therefore, a rolling sampling programme that regularly collects abundance data from rivers where limited data are currently available will be established in Sweden, starting in Improved estimates of the exploitation of stocks in the coastal fishery. There is a need to replace the crude assumptions about how the coastal fisheries affect development of the stocks with more precise stock-specific estimates as input in the assessment model. Therefore, a spatially and temporally-structured Bayesian population dynamics model that tracks the migration of Baltic salmon stocks from their feeding grounds in the Baltic Sea to their natal rivers has been developed (Whitlock et al., 2018). The model uses information about the proportions of different stocks in catch samples from Swedish and Finnish coastal fisheries at different points in space and time, as well as finclipping information about the proportion of wild and reared fish in catches (also for catches where no genetic data are available). Further development of the model to be able to estimate stock-specific exploitation rates in the coastal fishery using genetic mixed-stock analysis is in progress (Whitlock et al., in prep). The model can also be used as a tool to evaluate the effect of alternative management actions (fishing effort configurations) on the exploitation and development of wild salmon stocks. This is key to developing stock-specific management. Improving precision in short-term projections by including covariates for sea survival. The potential for incorporating covariates such as herring recruitment strength and sea surface temperatures will be investigated as means to increase precision in short-term projections. Inclusion of AU 5 stocks in the full life-history model. At present, these stocks are treated separately from the AU 1 4 stocks. Inclusion in the full life-history model will require updated information regarding e.g. smolt age distributions, maturation rates, exploitation rates and post-smolt survival. In addition, increased amounts of basic biological data (e.g. smolt and spawner counts, additional electrofishing sites) may be needed for some rivers. The smolt production model for southern stocks that has been developed could be used also for AU 5 stocks in future, to produce smolt production priors and estimates for the life-history model. Development of an analytical assessment of AU 6 stocks. Development of an assessment model for AU 6 has been started in Model structure will be based on the same life-history model as used for AU 1 4, but it will be modified to follow the migration and fishing patterns specific to the AU 6 stocks. Also, the prior information available about the productivity (S/R dynamics) of the AU 6 spawning rivers will be incorporated. The AU 6 model will not be integrated to the AU 1 4 assessment in the first phase, but will be run as

23 216 ICES WGBAST REPORT 2018 a separate unit of stocks. However, the model will take into account migrations of salmon between the assessment units, which will to some extent link the assessments of the AU 1 4 salmon and AU 6 salmon together. The aim is to run the analytical assessment for AU 6 stocks latest in WGBAST Long-term and/or less urgent issues, good to keep in mind Refine the two river models to improve smolt priors used in the FLHM. The present river models do not account for annual fluctuations in smolt age structure, which may result in biases. Development of the river models to account for fluctuations in parr growth rates and length-specific smoltification probabilities to improve estimates of smolt age structure would solve this question. Allow for fluctuations in the stock recruitment carrying capacity (K) over time in rivers. Changes in physical river characteristics (e.g. habitat restoration and removal of obstacles to migration) have likely led to increases in K over the assessment period for some rivers. The current model version cannot handle this which may lead to biases when using old stock recruit data. Inclusion of data on composition of stocks at sea: The life-history model has already been fitted to information on proportions of wild and reared salmon in Main Basin as determined from scale readings. The next step would be to include genetic information on proportions of fish from different AUs, separating also wild and reared salmon from those areas. Subsequently, information on the representation of single stocks may be included. See more on future MSA in ICES (2015), Section 4.7. Further use of scale-reading data: In addition to wild/reared proportions, age data from catch samples could be used to get improved knowledge of yearclass strength, maturation and natural mortality rates. 4.7 Needs for improving the use and collection of data for assessment Because requirements for data will always exceed available resources, preferences must be given. The identification and prioritisation of new data collection, or modifications to ongoing monitoring work, should be based on end-user needs, particularly ICES assessment needs, and is of importance with respect to the European data collection framework (EU-MAP). Over the years, the WG has repeatedly highlighted and discussed various needs for data collection (e.g. ICES, 2014; 2015; 2016). For example, three years ago (ICES, 2015) the need for genetic analysis to study stock composition in catch samples (MSA) was reviewed, with suggestions provided regarding future studies. In the WGBAST 2016 report (ICES, 2016) comments were also given to a comprehensive list of proposals for Baltic salmon data collection produced at an earlier ICES workshop in Further, the need for at least one wild index river per assessment unit was highlighted, with suggestions given on potential candidates in AUs 5 6 (where full index rivers are still missing). Finally, as part of the recent benchmark for Baltic salmon (WKBALTSalmon, ICES, 2017d) all different types of information needed as input for the Baltic salmon stock assessment (fisheries statistics, biological data, etc.) were reviewed with respect to needs, availability and quality. Data issues and questions listed in the report are rather extensive and prioritizations will be needed. However, this can be used as a basis for decisions about data collection included in EU-MAP.

24 ICES WGBAST REPORT In brief, WKBALTSalmon highlighted the below data needs and development areas, and WGBAST encourage member states to include these elements subjects into their national data collection programmes. River data Biological monitoring Expansion of networks for electrofishing sites, to cover also newly populated river stretches; Updates of estimates for river-specific reproduction areas using standardised methodology; Inventories of habitat quality, particularly in weak salmon rivers (i.e. with low status); Compilation of stocking data on young life stages combined with information that enables estimation of survival of these releases to smolts; Counting data of ascending spawners from additional rivers. Guidelines to assure the comparability of such data should also be compiled. In rivers where counting is ongoing but data are yet not used in the assessment, additional information may be needed (e.g. from tagging studies). River fisheries The amount and quality of catch statistics vary considerably between rivers and countries. There is a general need for improvement and harmonisation of methods used for data collection, including estimates of unreporting; River-specific salmon catches should be included in InterCatch (ICES database); Available effort data from river fisheries should be evaluated. Sea fisheries data The level of misreporting of salmon (as sea trout) in the Polish offshore fishery may still be underestimated. For the Polish coastal fishery, no misreporting is accounted so far, although it potentially occurs there too. Data on proportions of sea trout and salmon (separately) in offshore and coastal catches are needed to facilitate a more precise estimation of the misreporting rates; Recreational trolling open sea catches have been estimated to be higher than previously recognised. Time-series of country specific catch estimates by three main fishing areas should be estimated; Also estimates of other recreational salmon catches at sea (i.e. coastal fishing in Sweden and Finland) should be added into InterCatch; Unreporting of catches is challenging to estimate, and it is possible that higher than currently estimated unreporting takes place in some of the countries and fisheries. An expert elicitation covering all relevant fisheries is needed in order to update unreporting estimates. Also, discards may be substantially underestimated (both undersized and sea-damaged catch) and studies on these are needed; Commercial salmon and sea trout catch and effort data by fleets and half years from all countries should be added into InterCatch;

25 218 ICES WGBAST REPORT 2018 Shortcomings in currently available fisheries data may cause bias in mortality estimates (F and M). At present, the possible magnitude of such bias, and consequently its potential impact on conclusions regarding stock status and catch advice, has not been evaluated. The present assessment model is assumed to estimate the magnitude of total mortality reasonably reliably. However, an exercise exploring extra uncertainties emerging from data deficiencies (currently not accounted for), and how these may influence the catch advices (both qualitatively and quantitatively) should be carried out.

26 ICES WGBAST REPORT Table Likelyhood approximations for the wild smolt production (*1000) in the Baltic salmon rivers included in the Full Life-History Model (FLHM). The distributions are described in terms of their median, the 90% probability interval (PI) and the method on how these probability distributions have been obtained. These estimates will be updated in Section Wild smolt production (thousand) Method of estimation Assessment unit 1 1 Tornionjoki ,2 90% PI Simojoki ,2 90% PI Kalixälven % PI Råneälven % PI Total assessment unit % PI Assessment unit 2 5 Piteälven % PI Åbyälven % PI Byskeälven % PI Kågeälven NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA % PI NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA Rickleån ,2 90% PI Sävarån ,2 90% PI Ume/Vindelälven ,2 90% PI Öreälven % PI Lögdeälven ,2 90% PI Total assessment unit % PI Assessment unit 3 13 Ljungan % PI Total assessment unit % PI Assessment unit 4 14 Emån % PI Mörrumsån % PI Total assessment unit % PI Method of estimation: 1. Bayesian linear regression model, i.e. the river model (see the Stock Annex) 2. Sampling of smolts and estimate of total smolt run size. 3. Inference of smolt production from data derived from similar rivers in the region.

27 220 ICES WGBAST REPORT 2018 Table Median values and coefficients of variation of the estimated M74 mortality for different Atlantic salmon stocks (spawning years ). The values in bold are based on observation data from hatchery or laboratory monitoring in the river and year concerned. Grey cells represent predictive estimates for years from which no monitoring data were available Simojoki cv Tornionjoki cv Kemijoki cv Iijoki cv Luleälven cv Skellelteälven cv Ume/Vindelälven cv Ångermanälven cv Indalsälven cv Ljungan cv Ljusnan cv Dalälven cv Mörrumsån cv Unsampled stock cv

28 ICES WGBAST REPORT Table Posterior probability distributions for steepness, alpha and beta parameters of the Beverton Holt stock recruit relationship and eggs per recruit (EPR, millions) for Baltic salmon stocks. Posterior distributions are summarised in terms of their mean and CV (%). Assessment unit 1 Mean cv Mean cv Mean cv Mean cv 1 Tornionjoki Simojoki Kalixälven Råneälven Assessment unit 2 5 Piteälven Åbyälven Byskeälven Kågeälven Rickleån Sävarån Ume/Vindelälven Öreälven Lögdeälven Assessment unit 3 Alpha parameter Beta parameter 14 Ljungan Assessment unit 4 Steepness 15 Emån Mörrumsån EPR

29 222 ICES WGBAST REPORT 2018 Table Posterior probability distributions for the smolt production capacity (x 1000) in the AU 1 4 rivers and the corresponding point estimates in the AU 5 6 rivers. The posterior distributions are described in terms of their mode or most likely value, the 90% probability interval (PI) and the method by which the posterior probability distribution was obtained. These estimates serve as reference points to evaluate the status of the stock. For the updated estimates of the AU 1 4 rivers except Testeboån, medians as estimated by last year s stock assessment are also shown. This enables comparison of how much the estimated medians have changed compared to last year. Smolt production capacity (thousand) Method of Last year s median % change Mode Median Mean 90% PI estimation Assessment unit 1 1 Tornionjoki % 2 Simojoki % 3 Kalixälven % 4 Råneälven % Total assessment unit 1 Assessment unit % 5 Piteälven * % 6 Åbyälven % 7 Byskeälven % 8 Kågeälven % 9 Rickleån % 10 Sävarån % 11 Ume/Vindelälven % 12 Öreälven * % 13 Lögdeälven * % Total assessment unit % Assessment unit 3 14 Ljungan % 15 Testeboån % Total assessment unit 3 Assessment unit % 16 Emån * % 17 Mörrumsån * % Total assessment unit % Total assessment units 1-4 Assessment unit % 18 Pärnu % 19 Salaca % 20 Vitrupe % 21 Peterupe % 22 Gauja % 23 Daugava % 24 Irbe % 25 Venta % 26 Saka % 27 Uzava % 28 Barta % 29 Nemunas river basin % Total assessment unit % Assessment unit 6 30 Kymijoki % 31 Luga % 32 Purtse % 33 Kunda % 34 Selja % 35 Loobu % 36 Pirita % 37 Vasalemma % 38 Keila % 39 Valgejögi % 40 Jägala % 41 Vääna % Total assessment unit % Total assessment units % * River w ith recently updated prior for potential or current smolt production.

30 ICES WGBAST REPORT Table Wild smolt production in Baltic rivers with natural reproduction of salmon grouped by assessment units: posterior probability estimates derived from the Full Life- History Model (FLHM) for the AU 1 4 rivers (except Testeboån which is currently not included in the FLHM), and estimates derived by other means (inferred from parr densities, smolt trapping, etc.) for the rest of the rivers. Median estimates (x 1000) of smolts with the associated uncertainty (90% Probability interval) are shown. Also, the river-specific reproductive areas and the potential smolt production capacities (PSPCs) are shown as medians and 90% PIs. Reprod. area (ha, median) Potential (*1000) Method of Pred Pred Pred estimation Pot. Pres. prod. prod. Assessment unit, subdivision, country Category Gulf of Bothnia, Sub-div : Finland Simojoki wild % PI '3-8 '6-13 '8-17 '22-42 '37-62 '37-63 '34-59 '27-47 '22-37 '25-42 '26-40 '27-47 '20-37 '26-44 '33-47 '24-40 '32-43 '35-53 '23-44 '27-33 '31-56 '47-83 '19-65 '22-75 Finland/Sweden Tornionjoki;Torneälven wild % PI ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' Sweden Kalixälven wild % PI ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' Råneälven wild % PI '2-14 '2-13 '5-25 '6-31 '11-43 '12-49 '11-43 '12-45 '12-48 '15-57 '20-65 '18-60 '25-82 '21-73 '22-73 '24-76 '23-79 '24-78 '24-82 '27-90 ' ' ' ' ' Assessment unit 1, total % PI ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' Piteälven wild % PI '4-5 '4-5 '5-5 '3-3 '1-2 '3-3 '6-6 '17-26 '16-25 '11-15 '14-21 '16-29 '23-46 '25-54 '27-58 '26-53 '22-39 '19-31 '22-36 '27-49 '24-43 '22-35 '30-61 '21-76 '23-78 Åbyälven wild % PI '1-8 '2-10 '3-12 '3-13 '5-20 '7-24 '5-20 '6-19 '5-18 '5-17 '6-20 '6-22 '8-28 '7-24 '6-22 '7-23 '7-22 '7-24 '8-25 '8-26 '9-32 '9-33 '10-35 '7-32 '8-34 Byskeälven wild % PI '15-71 '11-48 ' '26-91 ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' Rickleån wild % PI '0-1 '0-1 '0-1 '0-1 '0-1 '1-3 '1-2 '1-2 '0-2 '0-2 '0-2 '1-2 '1-4 '1-3 '1-3 '1-3 '1-3 '1-4 '2-3 '1-5 '3-5 '4-7 '3-9 '2-8 '3-10 Sävarån wild % PI '0-1 '0-1 '1-3 '1-3 '1-3 '1-4 '1-3 '1-3 '1-4 '3-5 '3-4 '2-4 '3-6 '2-5 '2-4 '2-4 '2-7 '3-6 '3-8 '3-9 '4-11 '5-14 '6-17 '3-14 '4-17 Ume/Vindelälven wild % PI '8-33 '27-90 ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' Öreälven wild % PI '0-1 '0-1 '0-2 '0-2 '1-4 '1-5 '1-4 '1-4 '1-4 '1-5 '2-8 '2-7 '3-12 '2-10 '2-10 '3-11 '3-14 '3-15 '4-17 '4-21 '7-32 '9-39 '11-49 '8-43 '10-56 Lögdeälven wild % PI '0-1 '0-1 '0-2 '1-3 '1-4 '1-5 '1-3 '1-3 '1-4 '1-5 '2-6 '2-6 '2-8 '2-6 '2-6 '2-7 '3-9 '3-10 '3-10 '4-12 '5-10 '7-20 '8-25 '5-25 '7-34 Kågeälven wild % PI '3-57 '3-54 '2-39 '1-30 '1-27 '2-43 '3-67 '3-61 '5-85 '13-58 '12-59 '8-57 '12-72 Assessment unit 2, total % PI ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' Ljungan wild % PI '0-1 '0-2 '0-2 '1-2 '1-3 '1-3 '1-2 '1-3 '1-3 '1-2 '1-3 '1-2 '1-3 '1-3 '1-3 '1-3 '1-2 '1-3 '1-3 '1-3 '1-4 '1-4 '1-4 '1-4 '1-4 Testeboån wild % PI na Assessment unit 3, total % PI '0-14 '0-15 '0-13 '1-13 '1-17 '1-23 '1-25 '1-23 '1-31 '1-32 '1-32 '1-24 '1-23 '1-12 '1-15 '1-16 '1-17 '1-20 '2-9 '1-6 '2-6 '2-8 '1-24 '1-4 '1-4 Total Gulf of B., Sub-divs % PI ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' '

31 224 ICES WGBAST REPORT 2018 Table Continued. 0 Method of Pred Pred Pred estimation Assessment unit, subdivision, Reprod. area Potential Pot. Pres. country Category (ha, median) (*1000) prod. prod. Sweden Emån wild % PI '0-3 '0-4 '0-6 '1-6 '1-6 '1-4 '1-4 '1-5 '1-6 '1-7 '1-5 '1-7 '2-4 '1-6 '2-7 '1-6 '1-7 '1-6 '1-6 '2-11 '2-12 '3-15 '2-11 '2-13 '2-12 Mörrumsån wild % PI '25-96 ' ' '27-95 ' '27-91 '29-96 ' '28-95 '26-95 '26-92 '29-92 '27-95 '25-90 '25-86 '24-90 '23-82 '24-80 '26-95 '28-88 '29-97 '28-93 '27-91 ' '26-96 Assessment unit 4, total % PI '53-96 '25-97 ' ' '29-98 ' '28-93 '31-98 ' '30-98 '30-98 '27-94 '32-96 '30-97 '28-94 '28-89 '27-93 '27-85 '27-83 '29-98 '33-94 ' ' '32-96 ' ' Estonia Pärnu mixed , 4 Latvia Salaca wild Vitrupe wild Peterupe wild , 5 Gauja mixed , 5 Daugava*** mixed , 6 Irbe wild Venta mixed , 5 Saka wild Uzava wild Barta wild Lithuania Nemunas river basin wild , 4 Assessment unit 5, total Total Main B., Sub-divs (AU's 4-5) Method of Pred Pred Pred estimation Assessment unit, subdivision, Reprod. area Potential Pot. Pres. country Category (ha, median) (*1000) prod. prod. Finland: Kymijoki mixed 15 1) +60 2) 20 1) +80 2) Russia: Neva mixed Luga mixed SE Estonia: Purtse mixed Kunda wild 1.9 2,1(3,7) Selja mixed Loobu mixed Pirita mixed , 3 90% PI Vasalemma wild Keila wild 3.5 5,4 (12) Valgejõgi mixed Jägala mixed Vääna mixed Assessment unit 6, total Gulf of B.+Main B.+ Gulf of F., Sub-divs % PI ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' = Low and uncertain production (not added into sub-totals or totals) ++ = Same method over time series; only the extension backwards preliminary Methods of estimating production Present production 7. Estimate inferred from stocking of reared fish in the river. Potential production 1. Bayesian full life history model (section 6.3.9) 8. Salmon catch, exploitation and survival estimate. 1. Bayesian stock-recruit analysis 2. Sampling of smolts and estimate of total smolt run size. 2. Accessible linear stream length and production capacity per area. 3. Estimate of smolt run from parr production by relation developed in the same river. Reared smolts *** = Tributaries 3. Expert opinion with associated uncertainty 4. Estimate of smolt run from parr production by relation developed in another river. *=Release river not specified **** = Only Latvian part, Lithuanian part of the river needs to 4) Below the lovest dams 5. Inference of smolt production from data derived from similar rivers in the region. be added 5) Above the lowest dams 6. Count of spawners. n/a No data available.

32 ICES WGBAST REPORT Table Overview of the status of the Gulf of Bothnia and Main Basin wild and mixed-stocks (grey rows) in terms of their probability to reach 50 and 75% of the smolt production capacity in 2017 (compared to PSPC in that year). Stocks are considered very likely to have reached this objective in case the probability is higher than 90%. They are likely to have reached the objective if the probability is between 70 and 90%, uncertain when the probability is between 30 and 70 % and unlikely if the probability is less than 30%. For the AU1 4 stocks except Testeboån, the results are based on the assessment model, whereas the categorization of Testeboån and AU5 6 stocks is based on expert judgments; for those rivers there are no precise probabilities (column 'Prob'). Prob to reach 50% Prob to reach 75% Stock Category Prob V.likely Likely Uncert. Unlikely Prob V.likely Likely Uncert. Unlikely Unit 1 Tornionjoki wild 1.00 X 0.95 X Simojoki wild 0.96 X 0.68 X Kalixälven wild 0.98 X 0.81 X Råneälven wild 0.95 X 0.71 X Unit 2 Piteälven wild 0.81 X 0.15 X Åbyälven wild 0.97 X 0.78 X Byskeälven wild 0.98 X 0.80 X Kågeälven wild 0.73 X 0.38 X Rickleån wild 0.45 X 0.12 X Sävarån wild 0.82 X 0.55 X Ume/Vindelälven wild 1.00 X 0.88 X Öreälven wild 0.37 X 0.15 X Lögdeälven wild 0.22 X 0.08 X Unit 3 Ljungan wild 0.87 X 0.66 X Testeboån *) wild n.a. X n.a. X Unit 4 Emån wild 0.45 X 0.17 X Mörrumsån wild 0.98 X 0.76 X Unit 5 Pärnu mixed n.a. X n.a. X Salaca wild n.a. X n.a. X Vitrupe wild n.a. X n.a. X Peterupe wild n.a. X n.a. X Gauja mixed n.a. X n.a. X Daugava mixed n.a. X n.a. X Irbe wild n.a. X n.a. X Venta mixed n.a. X n.a. X Saka wild n.a. X n.a. X Uzava wild n.a. X n.a. X Barta wild n.a. X n.a. X Nemunas wild n.a. X n.a. X Unit 6 Kymijoki mixed n.a. X n.a. X Luga mixed n.a. X n.a. X Purtse mixed n.a. X n.a. X Kunda wild n.a. X n.a. X Selja mixed n.a. X n.a. X Loobu mixed n.a. X n.a. X Pirita mixed n.a. X n.a. X Vasalemma wild n.a. X n.a. X Keila wild n.a. X n.a. X Valgejögi mixed n.a. X n.a. X Jägala mixed n.a. X n.a. X Vääna mixed n.a. X n.a. X *) Preliminary evaluation, see section

33 226 ICES WGBAST REPORT 2018 Table Key assumptions underlying the stock projections. The same post-smolt survival scenario and M74 scenario are assumed for all effort scenarios. Survival values represent the medians to which Mps and M74 are expected to return. Scenario Total commercial removal (dead catch) for year Removal that corresponds to ICES advice for fishing year % increase to scenario % decrease to scenario 1 4 F0.1 approach (commercial removal) 5 recreational fishing only 6 zero fishing In all scenarios we assume that the commercial removal (wanted catch reported) covers 55% of the total commercial sea fishing mortality, whereas 46% of this mortality consists of discards, misreported and unreported. Recreational fisheries in 2019 are assumed to have a catch that corresponds to the average catch in these fisheries in period, whereas in future years the effort component is the same for these fisheries but the catch varies according to abundance. (See text for details) Post-smolt survival of wild salmon Average survival between (19%) Post-smolt survival of reared salmon Same relative difference to wild salmon as on average in history M74 survival Historical median (85%) Releases Same number of annual releases in the future as in 2017 Maturation Age group specific maturation rates in 2018 are predicted using january-march 2018 SST data. For other years, average maturation rates over the time series are used, separately for wild and reared salmon. Ume/Vindelälven Average proportions (no. spawners passing ladder, MSW sex ratio passing ladder, extra mortality after ladder)

34 ICES WGBAST REPORT Table Estimates (in thousands of fish) of total removal in the commercial fishery at sea by scenario, and the corresponding reported commercial catch in total and divided between these fisheries in Calculations about how the total catch is divided between reported commercial catch and discards/unreporting/misreporting are based on the situation prevailing in 2017 (see text). The table shows also the predicted total number of spawners in 2019 (in thousands). All values refer to medians unless stated otherwise. Commercial catches (thousands of fish) at sea in SD in 2019 Wanted Catch Total inst. F of Reported Unwanted Catch (Dead+Alive) commercial comm. Scenario catch at sea Catch (% of 2018 EU TAC) Undersized Seal damaged Wanted Catch Unreported Wanted Catch Misreported % % % % % % Scenario Total sea catch (comm. + recr.) 2019 inst. F of total catch at sea Recreational catch at sea 2019 River catch 2019 Spawners

35 228 ICES WGBAST REPORT 2018 Table River-specific probabilities in different scenarios to meet 75% of PSPC in 2023/2024 (depending on the assessment unit) Probabilities higher than 70% are presented in green. Probability to meet 75% of PSPC River Year of Scenario comparison Tornionjoki Simojoki Kalixälven Råneälven Piteälven Åbyälven Byskeälven Rickleån Sävarån Ume/Vindelälven Öreälven Lögdeälven Ljungan Mörrumsån Emån Kågeälven

36 ICES WGBAST REPORT Table Estimated management reference points for Baltic salmon stocks. MSY smolt production, spawner escapement and R0 are given in thousands and are median values, likewise; MSY proxy values are ratios of medians. Harvest rates in the final two columns apply to salmon aged 2 and older (i.e. not post-smolts). MSY smolt MSY Stock production escapement R 0 MSY proxy MSY proxy ICES 2008 Immature harvest rate Mature harvest rate Tornionjoki % 83% Simojoki % 62% Kalixälven % 89% Råneälven % 83% Piteälven % 85% Åbyälven % 81% Byskeälven % 82% Rickleån % 78% Sävarån % 76% Vindelälven % 91% Öreälven % 79% Lögdeälven % 79% Ljungan % 74% Mörrumsån % 75% Emån % 24% Kågeälven % NA

37 230 ICES WGBAST REPORT 2018 Figure Prior probability distributions for the Beverton Holt alpha parameter (the inverse of the slope of the stock recruitment curve near the origin).

38 ICES WGBAST REPORT Figure a. Prior distributions for R0 (PSPC) from WGBAST ICES, 2017 (dashed line) and the updated stock recruitment parameterisation used in 2018 (final year R0, solid line). Dashed vertical lines indicate the medians in 2017 (grey) and 2018 (black). Note that updated (higher) priors for carrying capacity have been used 2018 for Piteälven, Öreälven and Lögdeälven (Section 4.2.2), and that updates have been done for Ume/Vindelälven (new priors on sex ratio and proportion of tagged ascending spawners finding the fish ladder; see Section 4.2.1).

39 232 ICES WGBAST REPORT 2018 Figure b. Prior distributions for R0 (PSPC) from WGBAST ICES, 2017 (dashed line) and the updated stock recruitment parameterisation used in 2018 (final year R0, solid line). Dashed vertical lines indicate the medians in 2017 (grey) and 2018 (black). Note that updated (higher) priors for carrying capacity have been used 2018 for Piteälven, Öreälven and Lögdeälven (Section 4.2.2), and that updates have been done for Ume/Vindelälven (new priors on sex ratio and proportion of tagged ascending spawners finding the fish ladder; see Section 4.2.1).

40 ICES WGBAST REPORT Figure c. Prior distributions for R0 (PSPC) from WGBAST ICES, 2017 (dashed line) and the updated stock recruitment parameterisation used in 2018 (final year R0, solid line). Dashed vertical lines indicate the medians in 2017 (grey) and 2018 (black). Note that updated (higher) priors for carrying capacity have been used 2018 for Piteälven, Öreälven and Lögdeälven (Section 4.2.2), and that updates have been done for Ume/Vindelälven (new priors on sex ratio and proportion of tagged ascending spawners finding the fish ladder; see Section 4.2.1).

41 234 ICES WGBAST REPORT 2018 Figure Probability that returning salmon find the fishladder in river Ume/Vindel. For years in which mark recapture study has not taken place, prior distribution is the predictive distribution that is based on other years mark recapture studies. See the text concerning the exceptional year Boxplots show medians with 5%, 25%, 75% and 95% quantiles.

42 ICES WGBAST REPORT Figure Old and new (and updated) smolt production input to be used in the Full Life-history Model (FLHM) for River Piteälven (medians with 90% PI).

43 236 ICES WGBAST REPORT 2018 Figure M74 mortality among Atlantic salmon stocks within the Baltic Sea by spawning year class in Solid circles and whiskers represent the medians and 95% probability intervals of the estimated M74 mortality, respectively. Open circles represent the proportion of females with offspring affected by M74 and triangles the total average yolk-sac-fry mortalities among offspring.

44 ICES WGBAST REPORT Figure Estimated proportion of M74-affected offspring that die (i.e. mortality among those offspring that are from M74 affected females) by spawning year class in Boxplots show medians with 5%, 25%, 75% and 95% quantiles.

45 238 ICES WGBAST REPORT 2018 Post-smolt survival 0.6 Survival Year Figure Post-smolt survival for wild (black) and hatchery-reared salmon (grey). Boxplots show medians with 5%, 25%, 75% and 95% quantiles.

46 ICES WGBAST REPORT Maturation rates Grilse 2SW Proportion SW 4SW Year Figure Proportion maturing per age group and per year for wild (black) and reared salmon (grey). Boxplots show medians with 5%, 25%, 75% and 95% quantiles.

47 240 ICES WGBAST REPORT 2018 Figure a. Distributions for egg abundance (million), plotted against the smolt abundance (thousand) for stocks of assessment units 1 4. Blue dots present the posterior distributions of annual smolt and egg abundances, red curves indicate the distributions of stock recruit relationship.

48 ICES WGBAST REPORT Figure b. Distributions for egg abundance (million), plotted against the smolt abundance (thousand) for stocks of assessment units 1 4. Blue dots present the posterior distributions of annual smolt and egg abundances, red curves indicate the distributions of stock recruit relationship.

49 242 ICES WGBAST REPORT 2018 Figure c. Distributions for egg abundance (million), plotted against the smolt abundance (thousand) for stocks of assessment units 1 4. Blue dots present the posterior distributions of annual smolt and egg abundances, red curves indicate the distributions of stock recruit relationship.

50 ICES WGBAST REPORT Figure d. Distributions for egg abundance (million), plotted against the smolt abundance (thousand) for stocks of assessment units 1 4. Blue dots present the posterior distributions of annual smolt and egg abundances, red curves indicate the distributions of stock recruit relationship.

51 244 ICES WGBAST REPORT 2018 Figure a. Prior and posterior probability distributions for the potential smolt production capacity (PSPC) obtained in the assessment in 2017 (thin line) and 2018 (final year, bold line). The 2017 assessment prior distributions for PSPC are shown by the grey dashed lines, while the 2018 assessment prior distributions for PSPC (final year) are shown by the black dotted lines.

52 ICES WGBAST REPORT Figure b. Prior and posterior probability distributions for the potential smolt production capacity (PSPC) obtained in the assessment in 2017 (thin line) and 2018 (final year, bold line). The 2017 assessment prior distributions for PSPC are shown by the grey dashed lines, while the 2018 assessment prior distributions for PSPC (final year) are shown by the black dotted lines.

53 246 ICES WGBAST REPORT 2018 Figure Posterior probability distributions for the total smolt production in assessment units (AU) 1 to 4 and in total. Vertical lines within each box show the median (solid line); whiskers denote the 90% PI for potential smolt production capacity (PSPC).

54 ICES WGBAST REPORT Figure Probability of reaching 50% of the smolt production capacity for different stocks of assessment units 1 4.

55 248 ICES WGBAST REPORT 2018 Figure Probability of reaching 75% of the smolt production capacity for different stocks of assessment units 1 4.

56 ICES WGBAST REPORT Catch (in thousands) River Catch (in thousands) Coast Year Year Catch (in thousands) Offshore Catch (in thousands) Total Year Year Figure Estimated posterior distributions of catches compared with corresponding observed catches (boxplots with medians, 5%, 25%, 75% and 95% quantiles). Offshore catches cover both commercial fisheries and recreational trolling. Observed catches have been recalculated to account for unreporting.

57 250 ICES WGBAST REPORT 2018 Wild proportion Proportion SW 3SW Year Figure Estimated proportions of wild in offshore catches compared with wild proportions observed in the catch samples among 2SW and 3SW salmon. Boxplots show medians with 5%, 25%, 75% and 95% quantiles.

58 ICES WGBAST REPORT Number of spawners Torne Simo Kalix Number of spawners (1000s) Råne Pite Åby Byske Rickleån Sävärån Year Figure a. Estimated posterior distributions of the number of spawners (in thousands) in each river vs. number of observed in fish counters. Observations indicated with dots are used as an input in the full life-history model whereas the ones indicated with triangles are so far not used as an input. Boxplots show medians with 5%, 25%, 75% and 95% quantiles.

59 252 ICES WGBAST REPORT 2018 Ume Öre Lögde Number of spawners (1000s) Ljungan Mörrum Emån Kåge Year Figure b. Estimated posterior distributions of the number of spawners (in thousands) in each river vs. number of observed in fish counters. Observations indicated with dots are used as an input in the full life-history model whereas the ones indicated with triangles are so far not used as an input. Boxplots show medians with 5%, 25%, 75% and 95% quantiles.

60 ICES WGBAST REPORT Offshore driftnet HR 1SW MSW Harvest rate Year 0.0 Offshore longline HR 1SW 0.30 MSW Harvest rate Year Figure a. Estimated posterior distributions of the harvest rates (harvested proportion of the available population) in offshore driftnet and offshore longline fisheries separately for one-sea-winter and multi-sea-winter salmon. The offshore longline fishery contains now also recreational trolling (see Section for details). Note that the driftnet harvest rate in 2008 is not zero, since due to computational reasons it contains fishing effort from the second half of year Boxplots show medians with 5%, 25%, 75% and 95% quantiles.

61 254 ICES WGBAST REPORT 2018 Coastal HR AU1 Grilse MSW 0.8 Harvest rate Year Coastal driftnet HR Grilse 0.12 MSW Harvest rate Year 0.00 Figure b. Estimated posterior distributions of the harvest rates (harvested proportion of the available population) in other coastal fisheries than driftnetting in AU1 and in coastal driftnetting (all AU s together) separately for one-sea-winter and multi-sea-winter salmon. Boxplots show medians with 5%, 25%, 75% and 95% quantiles.

62 ICES WGBAST REPORT River HR Grilse MSW Harvest rate Year Figure c. Estimated posterior distributions of the harvest rates (harvested proportion of the available population) in the river fishery separately for one-sea-winter and multi-sea-winter salmon. Boxplots show medians with 5%, 25%, 75% and 95% quantiles. Combined offshore HR, Combined coastal HR, M Harvest rate Harvest rate Figure Combined harvest rates (harvested proportion of the available population, medians with 90% probability intervals) for offshore and coastal fisheries for MSW wild salmon in

63 256 ICES WGBAST REPORT 2018 Smolt production relative to PSPC (%) % level 50 % level Barta Irbe Peterupe Saka Salaca Uzava Vitrupe Nemunas Pärnu Figure Wild smolt production level in relation to the potential in AU 5 wild salmon populations. Figure Wild smolt production level in relation to the potential in AU 5 mixed salmon populations.

64 ICES WGBAST REPORT % 90% Smolt production relative to PSPC 80% 70% 60% 50% 40% 30% 20% 10% 0% Keila Vasalemma Kunda Figure Smolt production level in relation to the potential in AU 6 wild salmon populations. Note that the potential is calculated only up to the lowermost migration obstacle and that rivers have substantial rearing habitat areas above migration obstacles. Smolt production in relation to potential (%) % level 75% level Purtse Selja Loobu Pirita Valgeõgi Jägala Vääna Figure Smolt production level in relation to the potential in Estonian AU 6 mixed salmon populations. Note that the potential is calculated only up to the lowermost impassable migration obstacle and that many rivers have considerably higher total potential.

65 258 ICES WGBAST REPORT 2018 Smolt production in relation to potential (%) % level 50 % level Kymijoki Luga Figure Wild smolt production level compared to potential in river Kymijoki (Finland) and in river Luga (Russia). 100% 90% Smolt production relative to PSPC 80% 70% 60% 50% 40% 30% 20% 10% 0% Figure Average smolt production level in relation to the potential in AU 6 mixed salmon populations (with 90% probability interval). Note that the potential is calculated only up to the lowermost impassable migration obstacle and many rivers have considerably higher total potential.

66 ICES WGBAST REPORT The sahare of adipose fin-clipped salmon Figure Share of adipose finclipped salmon caught on the south coast of the Gulf of Finland.

67 260 ICES WGBAST REPORT 2018 Scenario 1 Scenario 2 LL HR for MSW wild LL HR for MSW wild Scenario 3 Scenario 4 LL HR for MSW wild LL HR for MSW wild Figure a. Harvest rates (median values and 90% probability intervals) for wild multi-sea winter salmon in offshore longline fishery (including also recreational trolling, see Section for details) within scenarios 1 4.

68 ICES WGBAST REPORT TN HR for MSW wild in AU Scenario 1 TN HR for MSW wild in AU Scenario TN HR for MSW wild in AU Scenario 3 TN HR for MSW wild in AU Scenario Figure b. Harvest rates (median values and 90% probability intervals) for wild multi-sea winter salmon in coastal trapnet fishery within scenarios 1 4.

69 262 ICES WGBAST REPORT 2018 Rate Post-smolt survival wild reared Year M74 survival Rate Year Figure Median values and 90% probability intervals for post-smolt survival of wild and reared salmon and M74 survival assumed in all scenarios.

70 ICES WGBAST REPORT Grilse 2SW Proportion mature wild reared Proportion mature Year 3SW Year 4SW Proportion mature Proportion mature Year Year Figure Median values and 90% probability intervals for annual proportions maturing per age group for wild and reared salmon in all scenarios.

71 264 ICES WGBAST REPORT 2018 Abundance (in 1000's) SW wild, scen 1 Abundance (in 1000's) SW wild & reared Abundance (in 1000's) MSW wild, scen 1 Abundance (in 1000's) MSW wild & reared Figure a. Pre-fishery abundances of MSW and 1SW wild salmon and wild and reared salmon together based on scenario 1 (medians with 90% probability intervals). PFAs reflect the abundance that is available to the fisheries. In case of MSW salmon natural mortality is taken into account until end of June of the fishing year and in case of post-smolts, until end of August (four months after post-smolt mortality phase). See text for details.

72 ICES WGBAST REPORT Abundance (in 1000's) SW wild, scen 6 Abundance (in 1000's) SW wild & reared Abundance (in 1000's) MSW wild, scen 6 Abundance (in 1000's) MSW wild & reared Figure b. Pre-fishery abundances of MSW and 1SW wild salmon and wild and reared salmon together based on scenario 6 (zero fishing) (medians with 90% probability intervals). PFAs reflect the abundance that is available to the fisheries. In case of MSW salmon natural mortality is taken into account until end of June of the fishing year and in case of post-smolts, until end of August (four months after post-smolt mortality phase). See text for details.

73 266 ICES WGBAST REPORT 2018 Sea catches Catch (1000's) Year Figure Estimates of reported commercial sea catches (all gears, black boxplots) and recreational sea catches (all gears, grey boxplots) based on scenarios 1 6. Boxplots show medians with 5%, 25%, 75% and 95% quantiles.

74 ICES WGBAST REPORT Probability of meeting 75% Tornionjoki Probability of meeting 75% Simojoki Probability of meeting 75% Kalixälven Probability of meeting 75% Råneälven Figure a. Probabilities for different stocks to meet an objective of 75% of potential smolt production capacity under scenarios 1 6. Fishing in 2019 affects mostly years

75 268 ICES WGBAST REPORT 2018 Probability of meeting 75% Piteälven Probability of meeting 75% Åbyälven Probability of meeting 75% Byskeälven Probability of meeting 75% Rickleån Figure b. Probabilities for different stocks to meet an objective of 75% of potential smolt production capacity under scenarios 1 6. Fishing in 2019 affects mostly years

76 ICES WGBAST REPORT Probability of meeting 75% Sävåran Probability of meeting 75% Vindelälven Probability of meeting 75% Öreälven Probability of meeting 75% Lögdeälven Figure c. Probabilities for different stocks to meet an objective of 75% of potential smolt production capacity under scenarios 1 6. Fishing in 2019 affects mostly years

77 270 ICES WGBAST REPORT 2018 Probability of meeting 75% Ljungan Probability of meeting 75% Mörrumsån Probability of meeting 75% Emån Probability of meeting 75% Kågeälven Figure d. Probabilities for different stocks to meet an objective of 75% of potential smolt production capacity under scenarios 1 6. Fishing in 2019 affects mostly years

78 ICES WGBAST REPORT Tornio Simo Scen1 Scen2 Scen3 Scen4 Scen5 Scen Scen1 Scen2 Scen3 Scen4 Scen5 Scen Smolt production Smolt production Kalix Råne Scen1 Scen2 Scen3 Scen4 Scen5 Scen Scen1 Scen2 Scen3 Scen4 Scen5 Scen Smolt production Smolt production Pite Åby Scen1 Scen2 Scen3 Scen4 Scen5 Scen Scen1 Scen2 Scen3 Scen4 Scen5 Scen Smolt production Smolt production Byske 2017 Scen1 Scen2 Scen3 Scen4 Scen5 Scen Rickleån 2017 Scen1 Scen2 Scen3 Scen4 Scen5 Scen Smolt production Smolt production Figure a. Predicted smolt production in 2024 (or 2023 for Emån and Mörrumsån) under fishing scenarios 1 6 (thin lines) compared to estimated production in 2017 (bold line). Vertical lines illustrate medians of the distributions.

79 272 ICES WGBAST REPORT Sävarån 2017 Scen1 Scen2 Scen3 Scen4 Scen5 Scen Ume/Vindel 2017 Scen1 Scen2 Scen3 Scen4 Scen5 Scen Smolt production Smolt production Öre Lögde Scen1 Scen2 Scen3 Scen4 Scen5 Scen Scen1 Scen2 Scen3 Scen4 Scen5 Scen Smolt production Smolt production Ljungan Mörrumsån Scen1 Scen2 Scen3 Scen4 Scen5 Scen Scen1 Scen2 Scen3 Scen4 Scen5 Scen Smolt production Smolt production Emån 2017 Scen1 Scen2 Scen3 Scen4 Scen5 Scen Kåge 2017 Scen1 Scen2 Scen3 Scen4 Scen5 Scen Smolt production Smolt production Figure b. Predicted smolt production in 2024 (or 2023 for Emån and Mörrumsån) under fishing scenarios 1 6 (thin lines) compared to estimated production in 2017 (bold line). Vertical lines illustrate medians of the distributions.

80 ICES WGBAST REPORT Figure a. Median values and 90% probability intervals for smolt and spawner abundances for rivers Tornionjoki, Simojoki, Kalixälven and Råneälven in scenarios 1 (black), 4 (red) and 6 (blue).

81 274 ICES WGBAST REPORT 2018 Figure b. Median values and 90% probability intervals for smolt and spawner abundances for rivers Piteälven, Åbyälven, Byskeälven and Rickleån in scenarios 1 (black), 4 (red) and 6 (blue).

82 ICES WGBAST REPORT Figure c. Median values and 90% probability intervals for smolt and spawner abundances for rivers Sävarån, Ume/Vindelälven, Öreälven and Lögdeälven in scenarios 1 (black), 4 (red) and 6 (blue).

83 276 ICES WGBAST REPORT 2018 Figure d. Median values and 90% probability intervals for smolt and spawner abundances for rivers Ljungan, Mörrumsån, Emån and Kågeälven in scenarios 1 (black), 4 (red) and 6 (blue).

84 ICES WGBAST REPORT Figure Share of commercial and recreational catches at sea, river catches (including unreporting and also some commercial fishing), and discard/unreporting/misreporting of total sea catches in subdivisions in years Commercial sea catch also includes recreational sea catch in Recreational sea catch is presented separately from 2001 onwards. Note that updated expert estimates of recreational trolling catches (Section 2.1.2) are included.