By-catch of grey seals (Halichoerus grypus) in Baltic fisheries : A Bayesian analysis of interview survey

RESEARCH ARTICLE By-Catch of Grey Seals (Halichoerus grypus) in Baltic Fisheries—A Bayesian Analysis of Interview Survey Jarno Vanhatalo 1 *, Markus Vetemaa 2 , Annika Herrero 3 , Teija Aho 4 , Raisa Tiilikainen 5 1. Department of Environmental Sciences, University of Helsinki, Helsinki, Finland, 2. Estonian Marine Institute, University of Tartu, Tartu, Estonia, 3. Finnish Game and Fisheries Research Institute, Helsinki, Finland, 4. Department of Aquatic Resources, Swedish University of Agricultural Sciences, O ¨regrund, Sweden, 5. Metsa¨hallitus, Savonlinna, Finland *[email protected] Abstract Baltic seals are recovering after a population decline. The increasing seal stocks cause notable damage to fisheries in the Baltic Sea, with an unknown number of seals drowning in fishing gear every year. Thus, sustainable seal management requires updated knowledge of the by-catch of seals—the number of specimens that die in fishing gear. We analyse the by-catch of grey seals (Halichoerus grypus) in Finland, Sweden, and Estonia in 2012. We collect data with interviews (35 in Finland, 54 in Sweden, and 72 in Estonia) and analyse them with a hierarchical Bayesian model. The model accounts for variability in seal abundance, seal mortality and fishing effort in different sub-areas of the Baltic Sea and allows us to predict the by-catch in areas where interview data was not available. We provide a detailed description of the survey design and interview methods, and discuss different factors affecting fishermen’s motivation to report by-catch and how this may affect the results. Our analysis shows that the total yearly by-catch by trap and gill nets in Finland, Sweden and Estonia is, with 90% probability, more than 1240 but less than 2860; and the posterior median and mean of the total by-catch are 1550 and 1880 seals, respectively. Trap nets make about 88% of the total by-catch. However, results also indicate that in one of the sub-areas of this study, fishermen may have underreported their by-catch. Taking the possible underreporting into account the posterior mean of the total by-catch is between 2180 and 2380. The bycatch in our study area is likely to represent at least 90% of the total yearly grey seal by-catch in the Baltic Sea. OPEN ACCESS Citation: Vanhatalo J, Vetemaa M, Herrero A, Aho T, Tiilikainen R (2014) By-Catch of Grey Seals (Halichoerus grypus) in Baltic Fisheries—A Bayesian Analysis of Interview Survey. PLoS ONE 9(11): e113836. doi:10.1371/journal.pone. 0113836 Editor: Andreas Fahlman, Texas A&M UniversityCorpus Christi, United States of America Received: June 29, 2014 Accepted: October 31, 2014 Published: November 25, 2014 Copyright: ß2014 Vanhatalo et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability: The authors confirm that all data underlying the findings are fully available without restriction. All relevant data are within the paper. Funding: This work was supported by the Academy of Finland (grant 266349, http://www.aka. fi, to JV), Estonian state financed project SF0180005s10 to MV, and INTERREG project ECOSEAL, (http://www.ecosealproject.eu/,toJV, MV, AH, TA, and RT). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist. PLOS ONE | DOI:10.1371/journal.pone.0113836 November 25, 2014 1/16 Introduction Baltic seals are recovering after a population decline in the late 20 th century. However, they face a changed ecosystem both in terms of human-induced mortality (hunting and by-catch in fishing gear) and in availability of food resources. On the other hand, the increasing seal stocks cause notable financial loss to coastal fisheries [1,2], especially in the northern Baltic Sea. Lack of updated information on different aspects of the interactions between seal and fish stocks and coastal fishery aggravates seal-fishery co-existence and the accomplishment of a favourable conservation status. As the conflict with coastal fishery has arisen [3], licensed hunting has been reintroduced in Finland and Sweden, and a stronger regulation of the seal population has been called for by fishermen. However, the by-catch of seals – the number of specimens that die in fishing gear is unknown, complicating the assessment of a sustainable hunting quota and the conservation actions needed. The only earlier data sources about seal by-catch in the Baltic Sea are interviews of fishermen by Lunneryd et al. [4]. The authors extrapolated that in 2001 the number (and 95% confidence interval) of by-caught seals in Swedish waters was 462 (360–575) grey seals, 52 (34–70) ringed seals and 461 (333–506) harbour seals. However, there is no estimate for the total by-catch of seals in the Baltic Sea. Moreover, in parallel to the growing number of Baltic grey seals, the number of animals drowning in fishing gears might also have increased. Hence, sustainable management both on national and international levels requires broader and updated knowledge on the by-catch of seals in the Baltic. For large-scale fisheries with high economic values by-catch data is often achieved by independent observers recording by-catch on-board, which is later extrapolated to the whole fishery. Obtaining information from small-scale fisheries like the Baltic coastal one is more problematic. The small economic value of fisheries combined with the low frequency of by-catch makes the usage of observers practically impossible. Even if few by-catch events could be detected, extrapolation of such episodic data would have low statistical reliability. Due to these constraints, by-catch data from small-scale fisheries is usually based on interviews [5,6,7], which is the chosen data collection method in this study as well. This leads, however, to greater uncertainty and lower credibility in data compared to data collected by observers and thus requires a rigorous analysis. In this work, we analysed the by-catch of grey seals (Halichoerus grypus)in coastal trap nets and gill nets used by professional coastal fisheries in Finland, Sweden, and Estonia in 2012. According to our study, these gears are responsible for most of the by-catch in the Baltic Sea. Still, trawls cause some sporadic mortality too. We collected data with interviews (35 in Finland, 54 in Sweden and 72 in Estonia) and analysed them with a hierarchical Bayesian model. The model accounts for variability in seal abundance, seal mortality and fishing effort in different sub-areas of the Baltic Sea and allows us to predict the by-catch in areas where interview data was not available. We provide a detailed description of the By-Catch of Grey Seals in Baltic Fisheries—A Bayesian Analysis PLOS ONE | DOI:10.1371/journal.pone.0113836 November 25, 2014 2/16 survey design and interview methods, and discuss different factors affecting fishermen’s motivation to report by-catch and how this may affect the results. Methods Study area and gears considered The choice of the specific types of fishing gear to be included in this study was based on the interviews; earlier experience of seal and fisheries researchers in Finland, Estonia and Sweden, and annual notifications of claims for seal-induced harm by fishermen. We excluded gear types for which no by-catch or only sporadic by-catch have been reported. Our study area covers the central and northern Baltic (approximately the ICES statistical squares 27–30 and 32), which is the main distribution area of Baltic grey seals. (Fig. 1). Since both the fishing methods, as well as the abundance of seals, vary across the Baltic Sea, we aggregated the data into 9 coastal sub-areas (Fig. 1,Table 1). The division was done so that within each sub-area the coastal fishing methods and environmental conditions are homogenous enough to assume constant (average) by-catch mortality and seal abundance. This was done based on the fisheries statistics and ecosystem-based division of the Baltic Sea in the literature (see e.g., [8]). Data The data was obtained through interviews of fishermen in 2012–2013 and from databases of national authorities. We also interviewed 5 fisheries and seal experts (2 seal monitoring researchers, 1 fisheries spokesman, and 2 conservationists) in order to elicit prior distributions for model parameters. In order to diminish scepticism and distrust in fishermen, which could lead to underreporting, we directly contacted fishermen who had the highest catches, who were classified by national standards as professionals (e.g. over 30% of annual income in Finland) and who had frequent contacts with researchers. In Finland, we also advertised our interviews among all fishermen so that anyone willing to be interviewed would be included into the panel. This resulted in 35 fishermen interviewed in Finland, 54 in Sweden and 72 in Estonia. In order to assist honest reporting, most of the interviews were conducted face-to-face, and anonymity was granted to interviewees. In Sweden, approximately 80% of the interviews were made face-toface and 20% by sending the questionnaire by mail. In Estonia, 50% of the interviews were made face-to-face and 50% by telephone. In Finland, 33 interviews were made face-to-face and one by telephone. Interviews were recorded by filling in the questionnaire form, and they were conducted by 4 people; one in Finland, one in Sweden, and two in Estonia. Three of the interviewers were female and one male. The set of questions and other details concerning interviews were mutually agreed before the interviews so that the content of all interviews was the same. By-Catch of Grey Seals in Baltic Fisheries—A Bayesian Analysis PLOS ONE | DOI:10.1371/journal.pone.0113836 November 25, 2014 3/16 The personal information gathered from the participants included their name, age, address and fishing region. This information was used to link the fishermen with national fisheries information systems from where the fishing effort in the interviewed sample was calculated. However, for the purposes of this study the individual answers and the fishing efforts were aggregated so that individual participants could not be identified. Data on fishing effort was obtained from the respective national authorities: the Finnish Game and Fisheries Research Institute, the Swedish Agency for Marine and Water Management, and the Estonian Ministry of Agriculture. The fishing effort was calculated for trap nets in gear-days and gill nets in km-days. The general questions posed to fishermen were: how many years they had fished professionally, what fishing methods (e.g. gear) they used and which species they Figure 1. The study area and its division into 9 subareas. doi:10.1371/journal.pone.0113836.g001 By-Catch of Grey Seals in Baltic Fisheries—A Bayesian Analysis PLOS ONE | DOI:10.1371/journal.pone.0113836 November 25, 2014 4/16 fished, their description of harm caused by seals to fisheries and fishing gear, what time of the year seals caused problems, had seal induced harm increased and had fishermen changed their fishing method because of seals. The specific questions on by-catch were: did fishermen catch seals as by-catch, with which gears and how often did they catch seals, how many seals did they catch in 2012, what time of the year did most by-catch occur, of what species (grey or ringed seal) the by-caught seals were and had by-catch increased during recent years. After this, fishermen were asked about seal harm mitigation: had harm caused by seals in their fishing region been mitigated by changing fishing gear or by hunting, did fishermen use bars in front of their traps, repellents or other means to protect their gear and had these methods been successful? The ages of interviewed fishermen ranged from 23 to 77 years. All provided their verbal informed consent to participate in this study and knew that their responses would be used as a part of potentially published research. The informed consent was implicitly recorded by the fact that fishermen participated in the interviews since those who did not provide consent were not interviewed. Written consent was not obtained since it was not required for this type of study by the national rules concerning ethics in research. Based on the rules of the Finnish Advisory Board on Research Integrity, this consent procedure and the study design are exempt from prior ethical approval by an ethics committee. The aggregated data is summarised in Table 1. We did not have permission to publish the raw data but Table 1 lists all data necessary to redo the analysis. We excluded sub-areas F3 and S3 from the survey for practical reasons. The funding and time budget of this research did not allow the inclusion of these subareas into the survey. For the same reason we excluded Russia, Latvia and southern parts of the Baltic Sea. However, on the basis of our results, we discuss their share of by-catch as well. Table 1. The survey and effort data and the estimated proportion of seals from the total population in each sub-area. Trap nets Gill nets proportion of seals subarea by-catch in sample effort in sample (gear-days) total effort (gear-days) by-catch in sample effort in sample (kmdays) total effort (km-days) number of interviews expected (%) spring (%) fall (%) E1 104 2458 11772 34 6.25 7.5 5 E2 59 1542 8722 25 6.25 7.5 5 E3 10 1096 3135 13 1.75 0.5 3 F1 45 23533 36636 25 1.75 0.5 3 F2 5 3823 58863 10 30 35 25 F3 0 0 5057 0 2 0 4 S1 0 0 8515 20 1402 7904 13 22.5 25 20 S2 61 2450 2563 35 1540 1611 41 16.5 15 18 S3 0 0 4806 0 2 0 4 total 284 34902 140069 55 2942 9515 161 89 91 87 Empty gill net cells represent areas where we assume gill nets do not contribute to by-catch. doi:10.1371/journal.pone.0113836.t001 By-Catch of Grey Seals in Baltic Fisheries—A Bayesian Analysis PLOS ONE | DOI:10.1371/journal.pone.0113836 November 25, 2014 5/16 Analysis We applied Bayesian methods to analyse the data. The benefits from using Bayesian analysis are that we can explicitly and transparently state our assumptions and uncertainties about the phenomenon and data in probabilistic form. The results of an analysis are posterior distributions which provide estimates for the by-catch and model parameters and describe the uncertainty in them. We have divided the total study area into sub-areas (Figure 1) so that they correspond to the different kind of gears and use of those gears in the Baltic Sea. Hence, it is reasonable to model the by-catch within each sub-area separately. We modelled the average mortality rates of a seal per unit effort (catchability) with a trap or a gill net as exchangeable between sub-areas. We also assumed that bycatch mortality is additive to other mortality sources, and that, given the catchability, the probability of a seal surviving a unit effort of fishing is independent of the fishing effort it has already survived. Then, the fishing effort a seal survives (‘‘lifetime’’ of a seal) will be exponentially distributed, and the probability for a seal in sub-area ato die via by-catch is 1{e {P 2 g~1 Eg,aCg,a , where E g,a is the total effort and C g,a is the catchability of a gear gin that sub-area. The catchability accounts for the process of a seal encountering a gear and becoming entangled in it. Since the number of seals in a sub-area may vary within a year we parameterize the model with an average (effective) number of seals in sub-areas. Then, the number of by-caught and survived seals in a sub-area will be y1,a,y2,a,sa fg ~MN Na,p1,a,p2,a,1{p1,a{p2,a fg ðÞ where N a represents the average number of seals in sub-area athat survive other reasons for mortality, y g,a is the number of seals that died in gear g,s a is the number of surviving seals, and pg,a~Eg,aCg,a P2 g~1Eg,aCg,a 1{e{P2 g~1Eg,aCg,a  is the probability of dying in gear gin one year. We gave a hierarchical prior for the catch-abilities, p(C g,a ), and number of seals, p(N a ), and calculated their posterior distribution using the Bayes rule p(Ca,Najys a,Es a)!p(ys ajCa,Na,Es a)p(Ca)p(Na) where ys a~ys 1,a,ys 2,a no and Es a~Es 1,a,Es 2,a no are the by-catch and effort in the survey respectively, and Ca~C1,a,C2,a fg collects the catch-abilities of both gears. After this we calculated the posterior predictive distribution of the by-catch by fishermen that were not interviewed p(~ yg,ajys a,Es a,~ Eg,a)~ð? 0 p(~ yg,ajCg,a,Na,~ Eg,a)p(Cg,a,Najys a,Es a)dCg,adNa where ~ Eg,ais the gear and sub-area specific fishing effort of fishermen that were not interviewed. The total by-catch is then the sum of the by-catch in the survey and the predicted by-catch by the remaining fishermen, yg,a~~ yg,azys g,a. By-Catch of Grey Seals in Baltic Fisheries—A Bayesian Analysis PLOS ONE | DOI:10.1371/journal.pone.0113836 November 25, 2014 6/16 Baltic grey seals are counted annually during their peak moulting time in early summer. The population is increasing and in 2012 there were 28255 counted individuals [9]. Hiby et al. [10] estimated that the proportion of counted individuals is 70%–85% of the total population. Hence, it is reasonable to assume that the number of counted seals is a conservative minimum and 28255/ 0.7<40000, an optimistic maximum estimate of Ntot~P 9 a~1 Na– the total number of seals that survived other mortalities than by-catch in 2012. We encoded these assumptions by a log-Gaussian prior NtoteLogN(mN~10:42,s2 N~0:1) where mN~E log(Ntot)½and s2 N~Var log(Ntot)½. This gives 95% probability for values less than 40000 and 95% probability for values more than 28000, with mean 34000<28255/0.85. The division of the total seal population between sub-areas was based on annual counts and expert assessment as follows. The number of seals in each sub-area during the survey was calculated from the survey counts, with two exceptions. The survey counts reported the total number in E1 and E2, which was evenly divided between them. In survey counts the area south from Stockholm was combined with S1 and, thus, we estimated that 30% of the counted seals in that area belonged factually to sub-area S2 of our study. We interviewed a researcher responsible for the survey counts in Finland (Markus Ahola, Finnish Game and Fisheries Research Institute) in order to estimate the change from the number of individuals during surveys to that in fall. During surveys seals are aggregated in archipelago areas, which are good for moulting, whereas in fall seals distribute throughout the northern Baltic Sea to forage. The redistribution was done roughly in proportion of the area not suitable for moulting in each sub-area. These estimates were transferred to proportions of seals from the total population by dividing them by the total number of counted animals. The larger (smaller) from survey and fall estimates was used as maximum (minimum) estimate of the number of seals in each sub-area (Table 1). The uncertainty about the sub-area proportions were encoded into the model using Dirichlet distribution as follows. The expected proportion of seals in each sub-area was the mean of the minimum and maximum estimates (Table 1). The scale parameter of the Dirichlet distribution was chosen so that approximately 90% of the prior probability mass was between the minimum and maximum estimates in the case of sub-area with the highest proportion (sub-area F2). Since, due to the properties of Dirichlet distribution, the variance relative to mean is greater for smaller proportions only about 60% of prior probability mass is within min/max values in the case of subareas with an expected proportion less than 10%. Given the total number of seals and the proportion of seals in sub-areas, the average number of seals in them was assumed to follow multinomial distribution. This led to Dirichlet Multinomial prior for the average number of seals in sub-areas By-Catch of Grey Seals in Baltic Fisheries—A Bayesian Analysis PLOS ONE | DOI:10.1371/journal.pone.0113836 November 25, 2014 7/16 N1,:::,N9 fg ~ DMN(Ntot,Qg,g), where Qdenotes the vector of prior expectation of proportions (Table 1) and the scale g~200 governs the uncertainty about the expected value. The catchability of a gear depends on many things, such as the specific type of gear and the fish species it targets. However, despite the evident variability in subarea specific catchabilities they are still related. This was modelled by giving a hierarchical prior [11] for the scaled catchabilities. For computational reasons we implemented the model by down-scaling the effort by 10 24 but all the results are reported in the original scale. The prior for the catchabilities was 104|Cg,aeLogN log(mg){ 1 2log(s2 g=m2 gz1),log(s2 g=m2 gz1)  mgeLogN(log(0:015),1) s2 geInv{x2(4,0:002) where Inv{x2(4,1)is the inverse Chi squared distribution with 4 degrees of freedom and scale 1 [11]. Here the area specific catchabilities depend on population mean mg~EC g,a  and variance s2 g~Var Cg,a  which define the prior expected catchability of all gears gused in the Baltic Sea and across sub-areas variation around it. The hyper-priors for the population parameters were set as follows. Based on expert (2 seal monitoring researchers, 1 fisheries spokesman, and 2 conservationists) assessment, the total by-catch in the Baltic Sea was of order few hundreds at minimum to few thousands at maximum. Their estimates ranged from 500 to 2000 with mean 1200 by-caught grey seals in total. A catchability of 1.5610 26 or 8610 26 (in the original scale) for both trap nets and nets in each sub-area, would lead to approximately 1200 and 5500 by-caught seals, respectively. The former represents the mean estimate of experts and the latter can be used as a pessimistic upper limit of the mean catchability since such a high number of by-caught seals is unlikely when compared to the total population size and the fact that the population is increasing. Thus, the prior was set so that with 95% probability mgis (in the original scale) below 10 25 and its prior median is 1.5610 26 . We assumed that the coefficient of variation in the catchabilities is likely over one but unlikely to be much more than ten. Hence the prior for s2 gwas set so that with 95% prior probability the coefficient of variation was more than 0.6 and less than 19 with prior median 3.4. The resulting marginal prior for Cg,a has 95% of its probability mass over 0.3610 26 and under 8610 26 . The posterior inference was then conducted for the parameters and population parameters. The hierarchical prior allowed us to predict, based on the posterior of the population parameters, the catchability in sub-areas F3 and S3, where interviews were not made. Since we had effort data for these sub-areas we could calculate the posterior predictive distribution of the by-catch there as well. We approximated the posterior distributions of model parameters with Markov chain Monte Carlo using the Metropolis-Hastings algorithm [12]. By-Catch of Grey Seals in Baltic Fisheries—A Bayesian Analysis PLOS ONE | DOI:10.1371/journal.pone.0113836 November 25, 2014 8/16 Results Gears causing by-catch In Estonia, only two types of trap net cause considerable by-catch: open-sea fykes with mouth size over 3 m, and coastal fykes with mouth size 1–3 m. These were included in the analysis in sub-areas E1-E3. Some fishermen interviewed in Estonia also described the rare drowning of pups in fyke nets with a smaller mouth size, and in salmon gill nets, but since there were no such reports for year 2012, these were excluded. In Finland practically all by-catch is caused by push-up trap-nets with mouth openings larger than 3 m. Hence, in sub-areas F1-F3 we considered only this type of gear. In Sweden, the by-catch is mainly caused by similar push up trap nets to those in Finland, but in sub-areas S1 and S2 also by several types (different mesh sizes) of gill nets. In sub-area S1 (Fig. 1), the gill nets are mainly used for fishing cod and flatfish whereas in sub-area S2 they target mainly whitefish and herring. Trap nets were included in the analysis in sub-areas S1-S3 and gill nets in sub-areas S1 and S2. Few fishermen in Sweden also reported sporadic by-catch in trawls, but due to the very low share of this type of by-catch, trawls were excluded from the analysis. Recreational fishing causes negligible bycatch at most since recreational fishermen do not use trap nets. Moreover, in subareas S1 and S2 their nets are weaker than those used by professional fishermen and, hence, seals do not drown in them to the same extent. Number of by-caught seals The total by-catch by trap and gill net fisheries in our study area was, with 90% probability, more than 1240 but less than 2860, and the posterior median and mean were 1550 and 1880 seals respectively. The posterior distribution was highly right skewed (Fig. 2) and the 80% quantile of the total by-catch was 2130. With 90% probability, the total by-catch with trap nets was more than 1100 but less than 2600, and that of gill nets more than 70 but less than 410. The posterior mean of the proportion of by-catch with trap nets from the total by-catch was 88%. With 90% probability the by-catch in Estonia was more than 780 but less than 930, and in Finland more than 130 but less than 270. In Sweden, the posterior distribution of the by-catch was highly right skewed and there the by-catch was with 90% probability over 210 and under 1790. However, the 80% quantile of the posterior of the by-catch in Sweden was 1060. The main reason for the heavily right skewed posterior of the by-catch in Sweden was the uncertainty in the posterior distributions of catchabilities of gill nets and trap nets in S1 (Fig 3.), where the average seal abundance was second highest (Fig. 4). The posterior of the gill net catchability in S1 was concentrated in smaller values than in S2. However, since the gill net effort in the sample relative to the total effort in S1 was small compared to that in S2, the posterior of gill net cacthability in S1 was more uncertain leading to heavy right tail. This caused the posterior of the gill net by-catch in S1 to have heavy right tail as well. Similarly, By-Catch of Grey Seals in Baltic Fisheries—A Bayesian Analysis PLOS ONE | DOI:10.1371/journal.pone.0113836 November 25, 2014 9/16 10. Hiby L, Lundberg T, Karlsson O, Watkins J, Ju¨ssi M, et al. (2007) Estimates of the size of the Baltic grey seal population based on photo-identification data. NAMMCO scientific publications, 6: 163–175. 11. Gelman A, Carlin JB, Stern HS, Dunson DB, Vehtari A, et al. (2013) Bayesian Data Analysis Third Edition. Chapman&Hall/CRC. 12. Gilks ER, Richardson S, Spiegelhalter DJ (1996) Markov chain Monte Carlo in practice. Chapman & Hall/CRC. 13. Dmitrieva L, Kondakov AA, Oleynikov E, Kydyrmanov A, Karamendin K, et al. (2013) Assessment of Caspian Seal By-Catch in an Illegal Fishery Using an Interview-Based Approach. PLOS ONE 8:e67074. doi: 10.1371/journal.pone.0067074. 14. Bjørge A, Ølen N, Hartvedt S, Bøthun G, Bekkby T (2002) Dispersal and bycatch mortality in gray, Halichoerus grypus, and harbor, Phoca vitulina, seals tagged at the Norwegian coast. Marine Mammal Science 18: 963–976. doi: 10.1111/j.1748-7692.2002.tb01085.x 15. Harding K, Ha¨rko¨ nen T (1999) Development in the Baltic grey seal (Halichoerus grypus) and ringed seal (Phoca hispica) populations during the 20th century. Ambio 28: 619–627. 16. Suuronen P, Siira A, Kauppinen T, Riikonen R, Lehtonen E, et al. (2006) Reduction of seal-induced catch and gear damage by modification of trap-net design: design principles for a seal-safe trap-net. Fisheries Research, 79: 129–138. doi: 10.1016/j.fishres.2006.02.014 17. Hemmingsson M, Fja¨lling A, Lunneryd SG (2008) The pontoon trap: description and function of a sealsafe trap-net. Fisheries Research, 93: 357–359. doi: 10.1016/j.fishres.2008.06.013 18. HELCOM (2006) HELCOM recommendation 27-28/2. Helsinki Commission, 8 July, 2006. By-Catch of Grey Seals in Baltic Fisheries—A Bayesian Analysis PLOS ONE | DOI:10.1371/journal.pone.0113836 November 25, 2014 16 / 16