Recreational sea fishing in Europe in a global context — Participation rates , fishing effort , expenditure , and implications for monitoring and assessment

Kieran Hyder1 | Marc Simon Weltersbach2 | Mike Armstrong1 | Keno Ferter3 | Bryony Townhill1 | Anssi Ahvonen4 | Robert Arlinghaus5,6 | Andrei Baikov7 | Manuel Bellanger8 | Janis Birzaks9 | Trude Borch10 | Giulia Cambie1,11 | Martin de Graaf12 | Hugo M C Diogo13 | Łukasz Dziemian14 | Ana Gordoa15 | Ryszard Grzebielec14 | Bruce Hartill16 | Anders Kagervall17 | Kostas Kapiris18 | Martin Karlsson19 | Alf Ring Kleiven20 | Adam M Lejk14 | Harold Levrel21 | Sabrina Lovell22 | Jeremy Lyle23 | Pentti Moilanen4 | Graham Monkman11 | Beatriz Morales-Nin24 | Estanis Mugerza25 | Roi Martinez1 | Paul O’Reilly26 | Hans Jakob Olesen27 | Anastasios Papadopoulos28 | Pablo Pita29 | Zachary Radford1 | Krzysztof Radtke14 | William Roche26 | Delphine Rocklin30 | Jon Ruiz25 | Callum Scougal1 | Roberto Silvestri31 | Christian Skov32 | Scott Steinback33 | Andreas Sundelöf34 | Arvydas Svagzdys35 | David Turnbull36 | Tessa van der Hammen12 | David van Voorhees22 | Frankwin van Winsen37 | Thomas Verleye38 | Pedro Veiga39 | Jon-Helge Vølstad3 | Lucia Zarauz25 | Tomas Zolubas35 | Harry V Strehlow2

Recognizing the need for data to support implementation of the Common Fisheries Policy (EU, 2013), the European Commission introduced a Data Collection Framework (DCF) in 2001 placing a legal requirement for Member States to collect specified types of data, including estimates of recreational catches and releases for selected species (EU, 2001). The requirements were altered slightly in the subsequent EU DCF regulations (EU, 2008a(EU, , 2010(EU, , 2016a

MRF in Europe involves many different methods including both
active (e.g. rod and line, spear and hand-gathering) and passive (e.g. nets, traps, pots, and set-lines) approaches (Table 1). A broad range of species are targeted, including finfish (e.g. gadoids, European sea bass, mackerels (Scombridae), flatfish, seabreams (Sparidae)), shellfish (e.g. scallops (Pectinidae), mussels) and crustaceans (e.g. crabs, European lobster (Homarus gammarus, Nephropidae)), with the mix of species varying between countries (for full details, see Supporting Information). For those species defined under the DCF (EU, 2008a(EU, , 2010 or EU-MAP (EU, 2016a) (hereafter termed DCF), Atlantic cod, European eel, Atlantic salmon and sea trout are the main targets in the Baltic Sea; Atlantic cod, European eel, European sea bass, Atlantic salmon, pollack, and elasmobranchs in the North Sea, Eastern Arctic, and North Atlantic; and European sea bass, European eel, elasmobranchs and Atlantic bluefin tuna in the Mediterranean and Black Seas (Table 1). Many more species are targeted by MRF than are included on the list of species reported under the DCF, so there are numerous other marine species where recreational catches may be a significant or even dominant component of total fishing mortality (e.g. European lobster- Kleiven, Olsen, & Vølstad (2012)).
Even though recreational fishing is thought to have significant impacts on many fish stocks (Arlinghaus & Cooke, 2005), impacts on the marine environment are difficult to assess as global assessments of recreational fishing generally do not distinguish between freshwater and marine fisheries (e.g. Arlinghaus, Tillner, & Bork, 2015;Cooke & Cowx, 2004). There are a number of regional estimates of MRF globally; however, for Europe, available data are either limited or outdated. There have been several national estimates of MRF participation, effort and expenditure for European countries

K E Y W O R D S
European marine recreational fisheries, fisheries assessment and management, fishing effort and expenditure, participation, surveys and monitoring of marine recreational fisheries (e.g. UK- Armstrong et al., 2013;the Netherlands-van der Hammen et al., 2016;Germany-Strehlow et al., 2012;France-Rocklin, Levrel, Drogou, Herfaut, & Veron, 2014) covering various methods (e.g. angling- Veiga et al. (2010), nets and pots -Sparrevohn & Storr-Paulsen (2012), and spearfishing- Zarauz et al. (2015)). However, no attempt has been made to synthesize these data to generate a robust assessment of MRF in Europe or address the challenges associated with the underlying data and associated biases.
The objectives of this study were to assess the importance of MRF in Europe, highlight key knowledge gaps, make proposals of how to fill these gaps and evaluate the implications of these gaps for fisheries monitoring and assessment. To achieve these objectives, estimates are derived of the total number of fishers, participation rates, days fished and expenditure of MRF in Europe. The contribution of MRF to total fishing mortality is exemplified using western Baltic cod ( Figure 1) and European sea bass (ICES areas IVb,c and VIIa,d-h; Figure 1) as casestudies. Results are discussed in the context of global MRF, the implications for fisheries management, and proposals are made of how to address the challenges of monitoring and assessment of MRF in Europe and other regions.

| Collection and selection of MRF data
The collection and selection of MRF data are summarized in this section, but due to the complexity, the full description and justification are provided for each country in the Supporting Information. Data are available from MRF surveys for some countries in Europe, many of which are published in the grey literature and in local languages.
Each year the ICES Working Group on Recreational Fisheries Surveys (ICES WGRFS) brings together experts from across Europe and compiles the latest MRF estimates for species where statutory data collection is required (ICES, 2012b(ICES, , 2013(ICES, , 2014(ICES, , 2015b(ICES, , 2017. Here, available literature was compiled for 27 countries within Europe that had a coastline on the Atlantic Ocean, North Sea, Baltic Sea, Mediterranean Sea, and Black Sea. The only exception was Bosnia and Herzegovina, which has a very limited coastline (see Table 1 for full list of countries). For each country, the population size for 2014 was downloaded (Eurostat, 2016a)  characterized for each country, fishing modes and gears, and target species were identified where there are requirements for catches to be reported under the DCF (EU, 2008a(EU, , 2010(EU, , 2016a.
A list of studies was compiled for each country that included estimates of number of fishers, participation rates, effort (total days, days per fisher) and expenditure (total expenditure, expenditure per fisher) (see Table 2 and Supporting Information for a full description of derivation of data from studies). For some countries (e.g. Spain and Portugal) or groups of countries (e.g. UK), data were pooled from constituent states or regions to provide national estimates (see Table 2 and Supporting Information). Calculations from national data were made where estimates were not provided in survey reports (e.g. only participation rate, but no estimate of numbers). Estimates of participation for France were partitioned between Atlantic and Mediterranean regions using the relative split of sea fishing effort between regions (i.e. 60:40 split) (see Table 2 and Supporting Information).

| Estimating numbers of fishers, participation rates, fishing effort and expenditure
Estimates of numbers of recreational sea fishers, participation rates, fishing effort (total days, days per fisher) and expenditure (total, per fisher) were generated for countries bordering the Atlantic (including the Baltic Sea) and the Mediterranean, and all of Europe combined (see Table 2 for a detailed list of countries).
The numbers and participation rates of MRF were used for countries with existing data. The relationship between participation rate and GDP was examined using correlation analysis comparing arcsine transformed participation rate and per capita GDP (following Arlinghaus et al., 2015). For countries with no data, participation rates were extrapolated to the "recipient country" from the most relevant country identified by national experts, hereafter termed "donor country" (see Table 2 and Supporting Information), and used with population size to estimate numbers of fishers. This assumed that the same proportion of the population was engaged in recreational sea fishing in both recipient and donor countries. The numbers of fishers were then summed for the Atlantic, Mediterranean and the whole of Europe, and used to derive participation rates based on the total population size. To ensure reproducibility of the calculations, the equations are provided below. If F r is the number of fishers in region r, f i is the number of fishers in country i, m countries are in region r, p j is the participation rate in donor country j used for extrapolation, P r is the participation rate in region r and x i is the population of country i, then the following equations were used to estimate numbers of fishers: A similar procedure was used to derive estimates of effort, with extrapolations based on the average annual fishing days per recreational sea fisher. This assumed that, on average, recreational sea fishers in the recipient country fished the same number of days per year as in the donor country. The average days fished per year was multiplied by the numbers of recreational sea fishers in the country to derive the total effort. The total effort was then summed for Atlantic, Mediterranean and the whole of Europe and the days per fisher derived from the total population size.
For expenditure, a similar method was used with the exception that expenditures were first converted to 2015 prices using Harmonised Consumer Price Index (Eurostat, 2016b), and a correction was made for the difference in per capita GDP (IMF, 2016). This assumed that the same proportion of overall wealth is spent on MRF in the donor.
If E r is the total expenditure of marine recreational fishers in region r, e i is the per fisher amount spent in country i, e j is the per fisher amount spent in donor country j used for extrapolation, and g i and g j are the per capita GDP in country i and country j respectively, then fisher expenditure was estimated as follows: An assessment of data quality based on expert judgement was used to select specific studies for the analysis, and a semi-quantitative measure was developed to show the potential bias associated with each survey estimate. This semi-quantitative assessment of bias is similar to approaches used to provide indications of uncertainty in other fields (e.g. food safety-EFSA Scientific Committee (2015)).
Each individual country value was assessed for the magnitude and the direction of bias (b i ) which was rated on a 7-point scale, ranging between highly overestimated (+3), negligible bias (0), and highly underestimated (−3), taking into account known sources of survey bias that might affect the accuracy of the estimates including coverage, non-response, recall and avidity biases (see Pollock, Jones, & Brown (1994); ICES (2010) for general reviews). It was necessary to weight the contribution of the bias in each country (w i ), so that, for example, a large error in a small estimate did not have as much influence on the overall bias as a small bias in a large value for a country. Hence to calculate the relative bias in region r (B r ), the following equation was used: where w i was the individual country value for number, effort and expenditure, and b i was assumed to be the same for the donor and recipient countries (see Supporting Information in Table S1). The relative bias in the overall estimates was a ratio, so it was categorized by sign to indicate direction of bias (positive-overestimates, negative-underestimate) and on a categorical logarithmic scale (negligible < 0.2; 0.2 ≤ minimal < 0.4; (1) 0.4 ≤ small < 0.8; 0.8 ≤ moderate < 1.6; and 1.6 ≤ large). The relative bias ranged between −3 and +3, representing the situation where the estimates for all individual countries were highly underestimated (−3) or highly overestimated (+3) in relation to the likely actual value. Estimates of direct expenditure were corrected for inflation to 2015 prices using the World Bank annual regional inflation in T A B L E 2 Data compiled for each country on recreational sea fishing numbers, participation, activity (average effort per fisher (days) and total days fished per year), average spend per fisher and total expenditure (euro; not corrected for inflation), and country information including basin (AT -Atlantic, MED -Mediterranean and Black Sea) the Gross Domestic Product (GDP -thousands of USD per capita) and population size compiled from Eurostat (Eurostat 2016a, b)  Comparisons were then made between the estimates developed in this synthesis and other regions globally for participation rate, effort and expenditure, and used to assess the importance of the sector in Europe. Cisneros-Montemayor and Sumaila (2010) split Europe into four areas, but did not provide details of the countries in each area, so comparisons were made with north and south regions of Europe.

| Removals by MRF and comparisons with commercial fisheries
Despite the European requirement for recreational catches and releases to be reported for several species ( Comparisons of the recreational and commercial removals were made between reconstructed recreational removals and commercial catches.
The derivation of recreational removals for western Baltic cod (Strehlow et al., 2012) and subsequent inclusion in the stock assessment are well described (Eero et al., 2014). At present, only recreational removals from Germany have been included in the stock assessment (Strehlow et al., 2012), but data were available for Denmark (Sparrevohn & Storr-Paulsen, 2012) and Sweden (ICES, 2015b). Here, total recreational removals (catches and dead releases) were estimated using catches and releases multiplied by post-release mortality for all countries with MRF for western Baltic cod. Available data were grouped into sea-based (boat angling, charter boat angling, trolling) and land-based (shore angling, wading) fishing modes. Post-release mortality of cod was assumed to be 100% for shore-based releases (precautionary approach), and a mortality rate of 11.2% was applied to boat-based releases (ICES, 2016b;

| Numbers of fishers, participation rates, fishing effort and expenditure
The total number of European recreational sea fishers was estimated to be approximately 8.7 million, with 5.9 million and 2.8 million in Atlantic and Mediterranean regions, respectively ( Figure 2, Table 3).
Around 13% of the total estimate was based on extrapolation between countries, with a greater proportion of the Mediterranean region estimates subjected to extrapolation ( Table 4). The highest numbers of recreational sea fishers were from Norway and the UK in the Atlantic region, whereas the greatest numbers of fishers in the Mediterranean were from Italy ( Figure 2, Table 3). The overall participation rate for the whole of Europe was 1.6%. In the Atlantic region, participation rates ranged between 33% (Norway) and 0.22% (Germany). In the Mediterranean, Greece (2.7%) had the highest and Spain (0.61%) had the lowest participation rate, but there was considerable uncertainty about participation rates in many of the other countries ( Figure 2, Table 3). The semi-quantitative assessment of bias indicated that the estimate of the numbers of people fishing recreationally in the Atlantic region was reasonable with only a small underestimate likely, but there was potential for a large underestimate for the Mediterranean and a moderate underestimate of numbers for the whole of Europe (Table 4). Individual country surveys were generally categorized as negatively biased due to coverage issues. There was a significant positive correlation between per capita GDP and participation rate (r = 0.56, n = 19, p < 0.05), but this was driven by a single value for Norway. Removal of Norway from the analysis led to a non-significant positive relationship (r = 0.24, n = 18, p > 0.05), so more data would be needed to characterize this relationship.
The total MRF effort in Europe was estimated at 77.6 million sea fishing days, with the majority (73%) carried out in the Atlantic (56.8 million fishing days) compared to the Mediterranean (20.9 million fishing days) (Figure 3, Tables 3 and 4). Over 5 million days in total were fished each year in Norway, UK, Portugal, France and Spain (Figure 3, Table 3). This equated to on average 9.0 sea fishing days per recreational fisher in Europe, with more than 10 days fished per year in Estonia, Finland, Iceland, Latvia, Lithuania, Norway, Portugal, and Spain ( Figure 3, Table 3). About 16% of the fishing effort estimate was based on extrapolation between countries, with a much higher proportion extrapolated in the Mediterranean (Table 4). Overall, the semiquantitative assessment of bias indicated that effort estimates were reasonable for the Atlantic region and the whole of Europe (minimal and small underestimation, respectively), but there was a moderate underestimate of participation for the Mediterranean generally due to under-coverage of fishing methods (Table 4).
The expenditure by recreational sea fishers was estimated to be €5.89 billion in Europe, with around €4.97 and €0.92 billion spent in the Atlantic and Mediterranean regions, respectively (Tables 3 and 4). The UK and Norway accounted for 53% of this expenditure, with the UK having the highest annual average expenditure per recreational sea fisher at €1,732 ( Figure 4, Table 3).
On average, European recreational sea fishers spend €680 annually. In total, 42% of the expenditure estimate for Europe was based on extrapolations between countries (Table 4). Overall, the semi-quantitative assessment of bias indicated that expenditure estimates were reasonable for the Atlantic region and the whole of Europe, but were moderately underestimated for the Mediterranean (Table 4). There was no significant correlation between per capita GDP and expenditure (r = 0.25, n = 11, p > 0.05), but this was based on a small sample size, and the overall trend was positive.

| Comparison with other regions globally
European regional comparisons were possible with angling (European Anglers Alliance as cited in Pawson et al. (2008)) and sea fishing (Pawson et al., 2007) with similar estimates of numbers, but higher expenditure than estimated in this study (Table 5). Direct comparison at a European level with a global analysis of participation and expenditure in MRF (Cisneros-Montemayor & Sumaila, 2010) suggested a higher participation rate of 3.7% which was driven by much higher participation rates in northern Europe (6.21%), but a lower per fisher spend of €465 than this study. There was also a different pattern in expenditure, with a higher spend in southern than northern Europe (Table 5). Comparison with other regions of the world showed that participation rates were highest in Australia (19.5%), followed by the United States (3.26%), Africa (0.28%) and Asia (0.18%) ( Table 5). Days fished per recreational sea fisher were generally similar across all regions, but estimates were only present for the United States (6.5), Canada (9.1), Australia (6.1) and New Zealand (5.1) ( Table 5). Previous estimates of average annual expenditure by recreational sea fishers for Oceania and Europe were similar to this study, and were generally much higher than Africa and Asia, with the United States and Central America having a much greater expenditure than other regions (Table 5).

| Removals by MRF and comparisons with commercial fisheries
MRF removals accounted for a significant component of the total removals (recreational and commercial) for both western Baltic cod (ICES subdivisions 22-24; Figure 1) and northern European sea bass stock (ICES areas IVb-c and VIIa, d-h; Figure 1). The total commercial and recreational landings of western Baltic cod by Germany, Denmark, and Sweden were 17,306 t, of which, 27% was estimated to be from MRF (Table 6).
For western Baltic cod, Germany had the highest proportion of total cod removals by recreational fishers (52%), and Sweden the lowest (9%).
Release proportions (based on released fish in numbers) ranged between 32% and 48% of the total recreational catch depending on country (overall release proportion 35%). For European sea bass, recreational fishing was estimated to be responsible for 27% of the total removals of 5,401 t in 2012 for countries where survey estimates were available (Table 7).

F I G U R E 2
Estimated number of recreational sea fishers and the proportion of population that had been sea fishing in the last 12 months. Cross-hatching indicates the country used to extrapolate where no data existed. France and Spain were divided between the Atlantic and the Mediterranean region which is indicated by the dividing line (see Methods and Supporting Information for details) The proportion varied between countries, with the highest proportion of recreational removals in Belgium (53%), reflecting the low commercial catch for that country (Table 7). High release proportions (based on released fish in numbers) were observed ranging from 44% to 66% (overall release proportion 55%). Overall, the assessments of bias suggested that the recreational cod and sea bass harvest, release and removal estimates were small underestimates.

| Robustness of estimates of numbers, participation rate, effort and expenditure
This study provides for the first time a robust synthesis of survey data across Europe selected by in-country experts, to characterize the T A B L E 3 Numbers of recreational sea fishers, days fished, and expenditure (presented in constant 2015 prices) by countries in Europe. Where data were not available, data have been imputed from the % fishers (numbers) or total days fished (effort) and corrected for population size numbers, participation rates, fishing effort and expenditure by MRF.
The data used to represent the most comprehensive compilation of surveys undertaken, to date, were supported by in-depth knowledge of the strengths and shortcomings of individual studies, and included extrapolation to countries without data (see Table 2  incomplete coverage, recall bias). As a result, the most appropriate approach with the data available was to separate precision and bias, and focus on bias using a semi-quantitative method as it has the largest impact on the uncertainty. Hence, potential biases and the impact of bias on the robustness of the estimates are discussed in more detail in the rest of this section.  (ICES, 2013(ICES, , 2014(ICES, , 2015b(ICES, , 2017. As a result, we are confident that this approach has provided realistic assessments of magnitude and direction of bias in individual country surveys. The assessment of bias in combination with the proportion of the overall estimate based on extrapolation gave a good representation of the robustness of the estimates at a regional level. However, as effort and expenditure tended to be derived from extrapolations of recreational sea fisher numbers, there was additional uncertainty in these estimates. Imputations accounted for just 13% of the total estimate of recreational sea fisher numbers in Europe, suggesting that this represented a relatively minor bias overall, but has the potential  (Tables 3 and 4). The use of per capita GDP to correct for differences in expenditure was a potential additional uncertainty, as the relationship between expenditure and GDP is un-

clear. Previous studies indicated no clear trends with per capita GDP
(Cisneros-Montemayor & Sumaila, 2010) or a decline with increasing population density and GDP (Arlinghaus et al., 2015). Here, positive but non-significant correlations between per capita GDP and expenditure were found. However, per capita GDP was only used to correct between pairs of similar countries, so was likely to be a reasonable approach for this purpose. *Indicates conversion from US dollars to euro using a conversion rate of 0.89 euro; †is CPI adjusted to 2015, and converted from AU dollars to euros using a conversion rate of 0.69 euro; •80% of total effort occurred in marine waters; and +estimated from Fisheries & Oceans Canada (2012). All estimates were converted to 2015 equivalents using the compounded regional and country annual inflation rates from the World Bank.
T A B L E 6 Most recent estimates of biomass (tonnes) and/or numbers of western Baltic cod (ICES subdivision 22-24) harvested and released by recreational fishers, total removals (tonnes) by recreational and landings by commercial fisheries. The percentage of recreational removals is based on total removals

| Participation, effort and expenditure in a global context
The current estimate of participation in MRF  Edwards, 1989;Heberlein, Ericsson, & Wollscheid, 2002;Arlinghaus et al., 2015), but it is likely that complex interactions between factors drive differences in participation rate and vary between countries, making differences difficult to interpret. However, lower participation rates in Europe than Oceania and the United States could be related to past urbanization trends, as access to coast was not a significant predictor of total recreational marine and freshwater fishing rates in previous studies (Arlinghaus et al., 2015), or could also be due to increasing costs or decreasing catch rates. Despite variability in global participation rates, there was general similarity in the average days fished, ranging between 5 and 10 days per fisher each year. This suggests that the time that fishers dedicate to MRF may be driven by common elements despite the variation in environment and target species between countries and within regions.
There were differences between the expenditure for the Zealand (Henry & Lyle, 2003;Holdsworth, Rea, & Southwick, 2016) was comparable to Europe, although more recent Australian surveys suggest that average expenditure may be higher than previously estimated (Lyle, Stark, & Tracey, 2014;West, Lyle, Matthews, Stark, & Steffe, 2012). Due to sampling challenges, recreational fisheries are undervalued and so recognizing the value is a large step in considering the benefits of MRF in comparison with commercial fisheries (Lynch et al., 2016).

| Monitoring
Regular data collection is required to improve both the understanding and the management of MRF (ICES, 2013). In some countries, analysis of annual recreational fisheries monitoring has thrown light on the factors influencing the social, economic and biological dynamics (e.g. Arlinghaus et al., 2015;Brownscombe et al., 2014), and have been used to support MRF development. However, these are not available for MRF in many countries affecting the ability to develop and increase the impact on the economy (e.g. Europe-ICES, 2013). A lack of the expertise required to carry out these complex surveys and the generally held belief that MRF has minimal impact on fish stocks has slowed the start of data collection in many regions including Europe (Pawson et al., 2007). In Europe, differences in national fisheries research priorities impeded MRF data collection in some countries, resulting in large differences in MRF data quality, and highlight the importance of common data collection regulations such as the DCF.
In fact, the requirement to carry out MRF pilot studies within 2 years of the implementation of the new EU-MAP (EU, 2016a), the need for evidence to underpin derogation from delivery of national MRF data, and the provision for fisheries managers to define additional species, where required, may lead to broader monitoring in future to fill data gaps in Europe.
The frequencies of monitoring surveys vary globally, with surveys carried out every 5 years in Canada (Brownscombe et al., 2014;Fisheries & Oceans Canada, 2012), 2 years in the Netherlands (van der Hammen et al., 2016) and annually in the United States (NMFS, 2015).
However, time series of MRF catches show large variation in catchper-unit-effort and catches between years (Strehlow et al., 2012).
These variations underline the importance to collect annual estimates of catches for inclusion in stock assessments, otherwise assumptions are required to generate times series from data from either a single year (e.g. sea bass-ICES, 2012a, 2015a) or to deal with intermittent data (ICES, 2013). MRF effort is not directly related to stock size (e.g. Strehlow et al., 2012), anglers behave in different ways (e.g. Post et al., 2002), and improvements in gear or technology can improve catch rates (e.g. Brownscombe et al., 2014)  The introduction of national recreational fishers registries or licences would facilitate MRF data collection by providing representative sampling frames at low costs (Ashford, Jones, & Fegley, 2009;ICES, 2013), but may face opposition from recreational fishers and require enforcement to be a useful tool. Recreational fishers logbook smartphone applications ("apps") could provide an alternative means of collecting data to support existing monitoring programmes and deliver broader spatial data sets in real time, but only once a good understanding of the biases in app data is available, and appropriate standards are developed (Venturelli, Hyder, & Skov, 2017

| Assessment
MRF catches are routinely included in stock assessments and management in some countries (e.g. USA- Lee et al. (2017), Australia-Ryan et al. (2016)). However, this is not the case in many countries, and global fish catches have been estimated to be 14% higher if recreational fishing was included alongside commercial catches (Cooke & Cowx, 2004). Hence, inclusion of MRF in total fishing mortality is important due to the widespread and popular nature meaning that catches can be large for certain species or stocks (Coleman et al., 2004;Cooke & Cowx, 2004Ihde et al., 2011;Lewin et al., 2006;McPhee et al., 2002;Post et al., 2002).
Even where MRF data exist, there are significant challenges in including MRF catches in stocks assessments due to the irregular collection and changing survey methodologies. Europe provides a good example of this, with MRF only included in assessments of Baltic salmon (ICES, 2015c), western Baltic cod (Eero et al., 2014;ICES, 2016b) and European sea bass (ICES, 2015a). Significant assumptions have been made to include the estimated MRF removals of European sea bass of 1,500 t for 2012 from the northern stock, with MRF mortality assumed to be constant to generate a time series for the assessment (ICES, 2015a). This excluded the potential post-release mortality (Ferter et al., 2013) and countries that lacked data (e.g. Wales, Scotland, Northern Ireland, Channel Islands) (ICES, 2015a). Even in the case of western Baltic cod, assumptions were necessary despite the length of the time series (Strehlow et al., 2012) and good understanding of post-release mortality (Capizzano et al., 2016;Ferter et al., 2015a,b;. In other regions, these challenges are even greater with lack of data leading to reconstructions made based on data from other fisheries (e.g. Pauly & Zeller, 2016) or unconventional sources (e.g. Belhabib et al., 2016), and little information on release rates or post-release mortality (e.g. Ferter et al., 2013). However, comparison of novel data sources with existing surveys in Europe to understand the implications of their use (Venturelli et al., 2017), alongside the further development of the value of individual data points in data-rich situations (ICES, 2016c), will help inform development of MRF monitoring across the world.
MRF management measures have been implemented in Europe that will affect catches in the future (e.g. bag limits and seasonal closures for European sea bass (EU, 2015) and bag limits for western Baltic cod (EU, 2016b)). However, assumptions made (e.g. post-release mortality, times series) or exclusion of MRF catches from stock assessment may lead to bias in stock estimates, and a failure of stocks to respond as expected to management measures (Eero et al., 2014;Hyder et al., 2014;Ryan et al., 2016). Hence, robust methods that account for MRF removals in stock assessments and allocation decisions need to be developed even for data-poor assessments, alongside a better understanding of release rates and post-release mortality (Hyder et al., 2014).

| CONCLUSION
This synthesis showed that MRF is an important activity in Europe with significant participation rate, substantial effort, large economic impact, and important impacts on certain fish stocks. There are still significant data gaps that affect understanding, assessment, management and development of MRF within Europe (e.g. ICES, 2016b), but a large-scale single survey is not appropriate due to the diverse nature of the sector and cultural difference. The EU-MAP (EU, 2016a) provides a mechanism that will address some of the data gaps, but robust regular multispecies surveys of MRF in all countries across Europe are needed. The European situation is mirrored in many other regions with even fewer data available, so pragmatic solutions that may include novel methods for data collection need to be considered.