Journal list menu
Are fish sensitive to trawling recovering in the Northeast Atlantic?
Abstract
- The protection of sensitive species from overfishing is a key aspect of the ecosystem approach to fisheries management.
- We use life-history parameters and knowledge of fish shape and habitat to estimate the sensitivity of 270 species in the Northeast Atlantic to demersal trawling and compare sensitivity to the most recent IUCN categorization. Species classified as threatened were on average significantly more sensitive to trawling than other species. Using trawl surveys in European Atlantic waters from 36°N to 62°N, we estimated indicators of abundance of 31 highly sensitive species and compared changes in abundance to sensitivity, management measures, and value of landings.
- The abundance of 23 of the 31 sensitive species increased after year 2000 with 14 of the species showing increases significant at the 5% level. The increases were not due to specific management measures, as less than half of the species were covered by catch limits. Furthermore, sensitivity or value of landings was not related to trends in abundance.
- Three species (Atlantic wolf-fish, tusk and starry ray) declined significantly. These species are all at their southern distributional limit in the North Sea.
- Synthesis and applications. We recommend monitoring the development of sensitive species to identify species under pressure and allow rapid management actions before species enter the IUCN threatened category. Furthermore, we recommend taking precautions where species are under combined pressure from climate change and fishing.
1 INTRODUCTION
A key aspect of the ecosystem approach to fisheries management is the protection of species from overfishing (CBD, 2014). However, managing every species in large marine ecosystems is impractical and, therefore, approaches such as risk-based assessments (Hobday et al., 2011) have been proposed to ensure that the most sensitive components are the focus of management. A scoping process identifies priority objectives (Hobday et al., 2011), and a subsequent population susceptibility analysis (PSA) or similar is used to identify the most susceptible species. In Europe, priority objectives stated in the Marine Strategy Framework Directive (MSFD) aim to support ‘halting biodiversity loss, ensuring the conservation and sustainable use of marine biodiversity’ (European Commission, 2008). These objectives have been confirmed in various large-scale scoping exercises (Rindorf et al., 2017).
The Minimum Viable Population concept is often used in viability analysis of terrestrial populations (Beissinger & McCullough, 2002), and for marine fish populations, the minimum spawning stock per recruit concept plays a similar role. Basically, fishing should not reduce the spawning stock per recruit of a fish population below the minimum level necessary for replacement. For fish, this level is typically in the order of 20%–40% of the unexploited population (Brooks, Powers, & Cortés, 2010). Large, slow-growing, late maturing fish species with a low level of natural mortality are likely to reach the minimum level at a lower level of fishing mortality than small, fast-growing, early maturing species with higher levels of natural mortality (Fernandes et al., 2017; Jennings, Greenstreet, & Reynolds, 1999). Life-history characteristics such as asymptotic length L∞, growth rate K, the length at which 50% of the individuals have reached maturity Lmat, and natural mortality M are therefore often used to indicate the sensitivity of different species to fishing (Greenstreet et al., 2012; Hobday et al., 2011). When species-specific information is missing, an estimate of the maximum length Lmax of a species can be used to infer asymptotic length, growth rate, natural mortality, and proportion mature at length (Walker, García-Carreras, Le Quesne, Maxwell, & Jennings, 2019).
Following this approach, Le Quesne and Jennings (2012) used a life-history model and asymptotic length to predict the sensitivity of different fish species in the Celtic Sea to fishing by estimating the fishing mortality required to reduce the stock to a specific level of spawning stock biomass per recruit relative to the unfished status, using a general relationship between size of the individual and the size selectivity in the fishery. However, in a subsequent study, gear efficiency expressed as the probability of being retained by a particular gear was shown to have a larger effect on the estimated reference points than the uncertainty associated with general life-history relationships (García-Carreras, Jennings, & Le Quesne, 2015). Gear efficiency depends on the habitat, individual size, body shape and behaviour of the fish (Table S1), as well as the properties of the gear (Reid et al., 2007). In 2010–2012, bottom trawl and beam trawl fisheries accounted for 98% of the discards in EU member countries (Catchpole, Ribeiro-Santos, Mangi, Hedley, & Gray, 2017). Hence, while by-catches taken in other fisheries may be important for particular species, the probability of being retained in, or killed by, the bottom trawl and beam trawl fisheries is likely to be most important for demersal fish.
The objective of this study is to identify species sensitive to fishing in the Northeast Atlantic and provide an overview of changes in their abundance in the light of recent effort changes and current management. To achieve this aim, we ranked fish species according to their sensitivity to bottom and beam trawling in the Northeast Atlantic using gear efficiencies estimated by Walker, Maxwell, Le Quesne, and Jennings (2017). We then investigated whether species sensitivity was related to IUCN species status (Nieto et al, 2015), and whether management measures are in place to limit the impact of fishing on sensitive species. Finally, we examined the temporal development in the abundance of sensitive species in survey catches throughout the Northeast Atlantic and compared changes in abundance to sensitivity, current management measures, the decreases in fishing effort from 2000 onwards and market value of the species.
2 MATERIALS AND METHODS
2.1 Species sensitivity
Species sensitivity to trawl-generated mortality was estimated by applying a length-based Spawning stock per Recruit model (Beverton & Holt, 1957; Gislason, Pope, Rice, & Daan, 2008) to species recorded in the DATRAS survey database covering the Northeast Atlantic from the coast of Portugal to the North Sea and from the Baltic Sea to west of Scotland (http://www.ices.dk/data/data-portals/Pages/DATRAS.aspx). The maximum length of the species, Lmax, the asymptotic length, L∞, the von Bertalanffy growth rate parameter, K, length at first maturity, Lmat, and the length at which the individuals entered the sensitivity model (length at birth for chondrichthyan and length at metamorphosis for bony fish), Lmin, were derived from the literature whenever possible (Tables S2 and S3), thereby minimizing the variation induced by using one parameter to predict the remaining parameters whenever possible (Thorson, Cope, & Patrick, 2014).
General relationships between life-history parameters and species lengths were used to provide missing values whenever information could not be found in the literature (Table 1). When no growth parameters were available, L∞ was estimated from Lmax using the equation provided by Froese and Binohlan (2000), K was estimated from a regression of ln(K) versus ln(L∞) based on the species in the dataset for which K and L∞ estimates were available (p < 0.0001, adj.R2 = 0.43, df = 186). When Lmat was missing, a regression of ln(Lmat) versus ln(L∞) (p < 0.0001, adj.R2 = 0.91, df = 155) based on species for which estimates were available was used to derive the missing value. For chondrichthyans, regressions of ln(Lmin) versus ln(L∞) differed significantly between egg-laying and live-bearing species (ANOVA, p(>F) = 0.0004, n = 21). Separate regressions of ln(Lmin) versus ln(L∞) were therefore used for egg-laying (R2 = 0.68, df = 8) and live-bearing species (R2 = 0.81, df = 10). For bony fish, insufficient information about length at metamorphosis was available and an overall Lmin of 2 cm was adopted. The life-history relationships were entered into the model of Spawning Stock Biomass per recruit. The model was insensitive to changes in the value of Lmin as long as Lmin was below the length at which the species is either exploited or mature.
| Life-history component | Equation | Reference/assumption/data |
|---|---|---|
| Deterministic relationships | ||
| Asymptotic length, L∞ (cm) |
|
Froese and Binohlan (2000) |
| Von Bertalanffy parameter, K (year−1) |
|
Data from Table S2 |
| Length at first maturity, Lmat (cm) |
|
Data from Table S2 |
| Minimum length, Lmin (cm) | ||
| Egg-laying elasmob. |
|
Data from Table S2 |
| Live-bearing elasmob. |
|
Data from Table S2 |
| Actinopterygians |
|
Assumed |
| Body weight at length L, WL (g) |
|
Gislason et al. (2008) |
| Model equations | ||
| Proportion mature at length, PL |
|
Assuming L75% = 1.2 × Lmat |
| Natural mortality at length, ML (year−1) |
|
Charnov, Gislason, and Pope (2013) |
| Fishing mortality at length, FL (year−1) |
|
Walker et al. (2017) |
| Total mortality, ZL (year−1) |
|
|
Numbers at length  |
|
Gislason et al. (2008) |
| Average numbers in length interval |
|
Gislason et al. (2008) |
| Spawning stock per recruit |
|
Gislason et al. (2008) |
Bottom and beam trawls retain a variable proportion of the fish found in the path of the gear. Some species, such as shallow water or reef-associated species, are difficult to catch because their habitat is not consistently sampled during trawl surveys, while pelagic species, such as Atlantic herring Clupea harengus, are found in the water column and may pass above the headline of the gear without being caught. Furthermore, among the individuals that enter the mouth of the gear, the smaller individuals and species may escape through the meshes. Towed gears will therefore generate species- and size-specific fishing mortalities in the area where they operate. Walker et al. (2017) estimated the relative efficiency by which different species groups and sizes of fish were retained by the major towed commercial fishing gears operating in the North Sea. Dividing the species according to their body shape and typical vertical position in the water column, catch efficiency at length (exploitation pattern) was calculated for seven groups (Table S1). We allocated the species to these groups and used the average efficiency of small and large meshed commercial beam trawls and otter trawls estimated by Walker et al. (2017) to provide the relative catch efficiency for each centimetre of length for each of the seven groups. The resulting exploitation pattern was used to estimate the relative efficiency of towed gears for each of the species and size groups and provided the relative fishing mortality at size by which they would be affected for a given level of overall effort.
As a measure of the sensitivity of each species to fisheries generated mortality, we used the relative level of effort needed to reduce the biomass of mature individuals (the spawning stock biomass) to 25% of its unfished level, F25%SSB, (Table S2). Sensitive species can only tolerate relatively low levels of fishing mortality before their spawning biomass is reduced to 25% of the unfished level. The lower the F25%SSB, the higher the sensitivity. The exact percentage (25%) is of little consequence to the subsequent ranking, as the different F indicators are highly correlated (Le Quesne & Jennings, 2012). To distinguish between sensitive and non-sensitive species, we used the F25%SSB of the most sensitive of the commercial target species which has shown sustained high catches. Less sensitive species may also be severely impacted by a targeted fishing effort, but here we concentrated on species which are mostly bycaught in mixed fisheries rather than targeted directly. Ana- and catadromous species were excluded from the study.
2.2 Comparison of sensitivities and IUCN ratings, L∞ and taxonomy
The estimated values of F25%SSB were analysed to determine if F25%SSB differed significantly between IUCN ratings (ANOVA) and if there was a linear effect of asymptotic length (L∞) and a categorical effect of taxonomy (Chondrichthyes, Agnatha and Actinopterygia) in a generalized linear model with a linear effect of L∞ and a categorical effect of taxonomy. The natural log of F25%SSB was taken prior to analyses as variance increased with the mean of F25%SSB.
2.3 Management of sensitive species in the EU and value of landing
Current management measures for EU fleets were derived from the latest available decisions on fishing opportunities (European Commission, 2018a, 2018b). Price per kg was derived from the European Market Observatory for Fisheries and Aquaculture Products (www.eumofa.eu) and supplemented where prices were lacking with prices from Hanstholm Auction (www.hanstholmfiskeauktion.dk/prices). Species were ranked as high value if their value exceeded that of Atlantic cod Gadus morhua, medium value if their value was between 50% and 100% of that of Atlantic cod, and low value if their value was not listed or was <50% of that of Atlantic cod.
2.4 Development in abundance of sensitive species
Species abundance indices were derived from average survey catch rates in the ICES coordinated international bottom and beam trawl surveys in the area in which the species has historically been reported. Data were downloaded from the ICES DATRAS database (http://www.ices.dk/marine-data/data-portals/Pages/DATRAS.aspx) on 9 July 2019. The time range for each survey is given in Table S4.
For the Baltic Sea, North Sea and west of Scotland, only hauls in which all fish species caught were recorded were analysed. Catch rates of beam trawls in the North Sea, where several beam widths are used, were standardized to a beam width of 4 m. In the Baltic Sea, catches with the large and small versions of the trawl were considered as two different surveys.
Species which were not accurately identified to species were joined in species groups. Among the sensitive species, this affected the common skate complex (comprising records for Dipturus spp., D. batis, the invalid synonym D. flossada and D. intermedius) and smooth-hounds (Mustelus asterias and M. mustelus). Annual estimates of mean catch rates were estimated for species observed in at least 50% of the years. The estimated mean included only hauls taken in realized habitat defined as all ICES statistical rectangles in which the species had been recorded during a survey at any time in the data. Consequently, the total number of hauls varied between species.
To provide integrated species indicators, two types of difference between surveys were considered: differences in mean catch rate and differences in interannual variation in catch rate. Differences in mean catch were accounted for by scaling all survey catches to a mean of 1 in the years from 2009 onwards. Species-level indicators were derived by estimating the weighted average of all indicators for the given species. As the study was focused on population abundance indicators for long-lived species, we expect a smooth development of the underlying population size over the years. Survey catches may appear more or less smooth, depending on the consistency of the catchability and relative local abundance of the species in the particular survey. To ensure that more weight was placed on surveys showing consistent population development, the combined indicators were weighted with the reciprocal variance around a Loess smoother in time fitted to the logged non-zero catch rates of each species in each survey. Furthermore, to ensure a minimum precision level for the derived indicators, combinations of surveys and species for which the CV of the residuals from the Loess smoother exceeded 0.75 were excluded. Years in which the catch rates exceeded five times the long-term mean were excluded in the analysis of integrated indicators. This removed occasional very large hauls, which occurred for some of the species (e.g. Raja clavata and Squalus acanthias). Indicators were integrated by region and across all regions.
After a period of increasing or stable high fishing pressure, demersal fishing effort in the Northeast Atlantic area has decreased since the 2000s (ICES, 2018a, 2018b, 2018c). We therefore estimated the trends in species abundance from 1980 to 2000 and from 2000 onwards as the linear trends in log-transformed species-level indicators. In addition, the number of surveys showing positive trends in the two periods was investigated (regardless of significance level). The categorical effect on trend in abundance of value (low, medium or high) and management (species TAC, group TAC or no TAC) was investigated in an ANOVA.
3 RESULTS
3.1 Species sensitivity
The estimated values of F25%SSB for all 270 taxa are given in Table S2. The sensitivities of species from Greenstreet et al. (2012) and Le Quesne and Jennings (2012) were well correlated with those derived by the present method (correlations = –0.78 and 0.74, respectively, on a log-log scale). Some of the difference between the ratings can be attributed to differences in the assumed values of the life-history parameters. For example, Greenstreet et al. (2012) assumed L∞ of spurdog Squalus acanthias to be 90 cm, whereas Le Quesne and Jennings (2012) and this study assumed 120 and 121 cm respectively. The most sensitive of the commercial target species which has shown sustained high catches was saithe Pollachius virens, with a value of F25%SSB of 0.43. Using this species to define the level below which species can be considered particularly sensitive, 59 species with F25%SSB ≤ 0.43 were defined as sensitive species/taxa and used in subsequent analyses. These species were dominated by 33 chondrichthyan taxa, including skates (Rajidae; Amblyraja hyperborea, A. radiata, common skate-complex, Dipturus nidarosiensis, D. oxyrinchus, Leucoraja circularis, L. fullonica, L. naevus, Raja brachyura, R. clavata, R. microocellata, R. montagui, R. undulata, Rajella bathyphila, Rajella fyllae, Rajella lintea and Rostroraja alba), squaliform sharks (Somniosus microcephalus, Deania calcea, Squalus acanthias, Dalatias licha and Etmopterus princeps), catsharks (Scyliorhinidae; Galeus melastomus, Scyliorhinus canicula and S. stellaris), hound sharks (Triakidae; Galeorhinus galeus and Mustelus spp.) and various other species (Hexanchus griseus, Lamna nasus, Torpedo marmorata, T. nobiliana, Dasyatis pastinaca, D. tortonesei and Chimaera monstrosa). The remaining species comprised a variety of gadiforms (Brosme brosme, Coryphaenoides rupestris, Macrourus berglax, Molva molva, M. dypterygia, M. macropthalma, Mora moro and Phycis blennoides) and other species (Conger conger, Lophius budegassa, L. piscatorius, Brama brama, Argyrosomus regius, Anarhichas lupus, A. minor, Dicentrarchus punctatus, Ephippion guttifer, Epigonus telescopus, Sebastes spp., Lepidorhombus whiffiagonis, Hippoglossus hippoglossus, Polyprion americanus, Scophthalmus rhombus, Scorpaena scrofa and Synaphobranchus kaupi).
3.2 Comparison of sensitivities and IUCN assessments, L∞ and taxonomy
Comparing F25%SSB to the IUCN assessments, species sensitive to fishing occurred in all IUCN categories. However, average F25%SSB was significantly higher for species in the Least Concern category (p < 0.0001) (Figure 1) indicating that less sensitive species were, on average, more likely to be categorized as Least Concern. There was no significant difference in the average F25%SSB between the Critically Endangered, Endangered, Vulnerable and Near Threatened categories (p = 0.2594). None of the species in the Critically Endangered and Endangered categories showed an F25%SSB > 0.75 (Figure 2; Table S2). The two species classified as Endangered but not sensitive to demersal fishing pressure were basking shark Cetorhinus maximus and thresher shark Alopias vulpinus. Six data deficient species (Anarhichas lupus, Argyrosomus regius, Dasyatis tortonesei, Ephippion guttifer, Epigonus telescopus and Phycis blennoides) were identified as sensitive species. For the remaining 51 sensitive species, 24 are currently listed by the IUCN as Least Concern.
3.3 Management of sensitive species and value of landings
Among the 59 sensitive taxa, landings of four species (6.8%) were subject to species-specific quota management for defined management units, and a further five taxa (8.5%) by genus or family-based quota management (Table S5). Additionally, 13 species of skate (22.0%) would be included within quotas largely set for the complex (Rajiformes). While 13 (22.0%) of the taxa were subject to conservative management (e.g. a combination of prohibited listings and near-zero by-catch TACs), the remaining 24 species (40.7%) are not currently subject to any management measures.
Five high value species had landing values per kg greater than cod (Dicentrarchus labrax, Hippoglossus hippoglossus, Lepidorhombus whiffiagonis, Scophthalmus rhombus, Lophius budegassa and L. piscatorius). Another 10 medium value species attained values between that of cod and half this value (Anarhichas lupus, Etmopterus princeps, Macrourus berglax, Molva dypterygia, Molva macropthalma, Molva molva, Scyliorhinus canicula, Scyliorhinus stellaris, Sebastes spp. and Squalus acanthias). The remaining species were categorized as low value species.
3.4 Development in abundance of sensitive species
Of the taxa classified as sensitive, 31 were recorded with sufficient frequency and accuracy in the surveys to allow estimation of trends (Table 2). For some species, indices were highly correlated, whereas for others, the surveys differed in development (Figure 3). The abundance of 24 of the 31 species increased from 1980 to 2018 (Table 2; Figures 3 and 4). Among the species showing significant trends, 15 increased and three (starry ray Amblyraja radiata, wolf-fish Anarhichas lupus and tusk Brosme brosme) decreased. These three are at their southern boundary in the surveyed area (Figure 2) whereas this was not the case for the remaining species (except halibut Hippoglossus hippoglossus; Figure S1). Starry ray and tusk decreased at a significantly greater rate in the later period than in the period as a whole (p < 0.0181 in both cases). Among the 10 species showing a significant change in the trend between 1980–1999 and 2000–2018, six exhibited an increasing trend while four deteriorated. Fourteen species showed no significant change between time periods while the remaining six species had insufficient historical data to test for changes in trends. Categorizing the surveys individually according to the trend estimated, trends in 42 of 78 combinations of species and surveys from 1980 to 1999 were positive, whereas the corresponding number after 2000 was 161 of 209, a change from 54% to 77%.
| Species | Annual change from 1980 to 2018 | Annual change from 2000 to 2018 | Probability of same trend from 1980 to 1999 and 2000–2018 | Number of surveys with positive trend 1980–1999/all surveys | Number of surveys with positive trend 2000–2018/all surveys |
|---|---|---|---|---|---|
| Amblyraja radiata | −0.032 (0.008, p = 0.0002) | −0.069 (0.009, p < 0.0001) | 0.0181 | 1/2 | 2/5 |
| Anarhichas lupus | −0.070 (0.007, p < 0.0001) | −0.081 (0.022, p = 0.0019) | 0.7383 | 0/2 | 1/4 |
| Brosme brosme | −0.025 (0.007, p = 0.0010) | −0.078 (0.015, p < 0.0001) | 0.0042 | 2/3 | 2/5 |
| Chimaera monstrosa | 0.045 (0.017, p = 0.0152) | 0.067 (0.015, p = 0.0003) | 0.1080 | 2/2 | 4/5 |
| Conger conger | 0.052 (0.008, p < 0.0001) | 0.000 (0.008, p = 0.9936) | 0.0786 | 2/3 | 5/10 |
| Dasyatis pastinaca | −0.000 (0.027, p = 0.9972) | 0.006 (0.036, p = 0.8714) | 2/2 | ||
| Dipturus batis complex | 0.075 (0.012, p < 0.0001) | 0.092 (0.009, p < 0.0001) | 0.5196 | 4/4 | 5/6 |
| Galeorhinus galeus | −0.011 (0.019, p = 0.5632) | 0.039 (0.034, p = 0.2759) | 0.6018 | 2/3 | 4/4 |
| Galeus melastomus | 0.065 (0.012, p < 0.0001) | 0.064 (0.011, p < 0.0001) | 0.5192 | 2/4 | 8/8 |
| Hexanchus griseus | 0.028 (0.015, p = 0.0890) | 0.028 (0.015, p = 0.0890) | 1/1 | ||
| Hippoglossus hippoglossus | 0.012 (0.009, p = 0.2041) | −0.029 (0.027, p = 0.2940) | 0.4556 | 1/2 | 0/2 |
| Lepidorhombus whiffiagonis | 0.023 (0.004, p < 0.0001) | 0.009 (0.008, p = 0.2674) | 0.4411 | 3/5 | 14/14 |
| Leucoraja circularis | 0.035 (0.020, p = 0.0959) | 0.074 (0.017, p = 0.0006) | 3/4 | ||
| Leucoraja fullonica | −0.008 (0.013, p = 0.5550) | 0.023 (0.018, p = 0.2077) | 0.9217 | 1/1 | 2/3 |
| Leucoraja naevus | 0.030 (0.006, p < 0.0001) | 0.001 (0.007, p = 0.9004) | 0.0551 | 4/5 | 8/10 |
| Lophius budegassa | 0.028 (0.009, p < 0.0053) | 0.048 (0.008, p < 0.0001) | 0.0410 | 1/3 | 7/9 |
| Lophius piscatorius | 0.026 (0.008, p = 0.0046) | 0.002 (0.010, p = 0.8562) | 0.0014 | 3/5 | 6/12 |
| Molva macropthalma | 0.122 (0.039, p = 0.0054) | 0.054 (0.037, p = 0.1597) | 3/3 | ||
| Molva molva | 0.016 (0.007, p = 0.0213) | 0.049 (0.012, p = 0.0007) | 0.0146 | 1/4 | 5/8 |
| Mustelus spp. | 0.095 (0.010, p < 0.0001) | 0.122 (0.010, p < 0.0001) | 0.2205 | 3/4 | 8/8 |
| Phycis blennoides | 0.059 (0.015, p = 0.0006) | −0.017 (0.018, p = 0.3785) | 0.0010 | 2/3 | 4/7 |
| Raja brachyura | 0.011 (0.014, p = 0.4289) | 0.059 (0.012, p < 0.0001) | 0.0005 | 0/4 | 6/7 |
| Raja clavata | 0.016 (0.008, p = 0.0656) | 0.086 (0.007, p < 0.0001) | <0.0001 | 2/6 | 11/13 |
| Raja microocellata | 0.009 (0.009, p = 0.3126) | 0.000 (0.012, p = 0.9995) | 1/3 | ||
| Raja montagui | 0.054 (0.009, p < 0.0001) | 0.050 (0.013, p = 0.0015) | 0.8078 | 2/6 | 10/10 |
| Raja undulata | 0.069 (0.019, p = 0.0009) | 0.099 (0.043, p = 0.0339) | 0.8856 | 1/1 | 3/3 |
| Scophthalmus rhombus | 0.046 (0.008, p < 0.0001) | 0.052 (0.026, p = 0.0593) | 0.0955 | 3/5 | 12/15 |
| Scyliorhinus canicula | 0.041 (0.005, p < 0.0001) | 0.069 (0.005, p < 0.0001) | <0.0001 | 4/6 | 11/12 |
| Scyliorhinus stellaris | 0.039 (0.012, p = 0.0027) | 0.079 (0.020, p = 0.0012) | 0.0014 | 1/1 | 4/4 |
| Squalus acanthias | 0.039 (0.008, p < 0.0001) | 0.067 (0.021, p = 0.0060) | 0.6783 | 0/4 | 8/11 |
| Torpedo marmorata | −0.044 (0.016, p = 0.0140) | −0.018 (0.018, p = 0.3358) | 1/1 |
- Bold values indicate significant at the 5% level.
Among the species listed by IUCN as Critically Endangered, Endangered or Vulnerable, four taxa increased significantly after 2000 (Dipturus spp., Leucoraja circularis, Mustelus spp. and Squalus acanthias) whereas four showed no significant change (Dasyatis pastinaca, Galeorhinus galeus, Leucoraja fullonica and Hippoglossus hippoglossus). Among the species assessed as Least Concern, starry ray, tusk and marbled electric ray Torpedo marmorata all decreased significantly from 1980 to 2018.
There was no relationship between sensitivity, the presence of a catch limit, the value at landing and the population index trend of the species (p > 0.29). For example, the two least sensitive of the species (starry ray and wolf-fish) declined whereas the most sensitive (the common skate complex) increased. The three species showing the greatest increase in the past 20 years had very different management measures: the common skate complex (to be returned to the sea unharmed), smooth-hounds (no management) and undulate ray Raja undulata (individual species TAC). Starry ray and wolf-fish showed comparable declines, despite the former being required to be returned to the sea unharmed (since 2014) whereas the latter can be landed. High and medium value species included both declining/stable species (wolf-fish and halibut) and increasing species (Lophius budegassa and Squalus acanthias), and the same was true of low value species (starry ray and tusk declining and smooth-hounds and the common skate complex increasing).
4 DISCUSSION
Using life-history parameters and knowledge of fish shape and habitat to identify the likely sensitivity for each species, we demonstrated that in general, sensitive demersal marine fish increased in abundance in the Northeast Atlantic since 2000 and in many cases had been doing so since 1980. The analysis identified potentially sensitive species along the Northeast Atlantic shelf, and the trawl survey data integration method allowed the temporal changes in species abundance to be monitored for more than half of these sensitive species. The increase in abundance of most of the sensitive species analysed did not appear to be due to species-specific management measures, as less than half of the species were covered by catch limits and there was no strong evidence of an increase in relative abundance as fishing effort decreased after year 2000. Furthermore, neither sensitivity nor market value appeared to be related to trends in abundance. However, the three species which declined significantly were all at their southern limit in the North Sea, indicating that conditions in the North Sea may be affecting these species more than others, possibly due to the greater changes in water temperature (Hobday & Pecl, 2014).
The accuracy of the identification of sensitive species depends on the accuracy of the parameters included and the method used to derive sensitivity. Thorson et al. (2014) demonstrated the importance of a proper literature review to derive the best available life-history parameter estimates. While we performed a detailed literature review, our focus was on the development of the species across all of the sampled parts of the NE Atlantic, and therefore, we did not use local values of life-history parameters. If the analysis is repeated for a specific area, local values could be used instead, but as information is often sparse for species identified as sensitive, using local values increases the risk of basing the analyses on very limited data, and hence, it may be preferable to use the values given here for analyses even at a more limited spatial scale. While our results on species sensitivity ranking are in broad agreement with findings from other studies, in that larger, slower growing, late maturing species are the most vulnerable to fisheries generated mortality (e.g. Jennings et al., 1999; Ravard, Brind'Amour, & Trenkel, 2014), sensitivities based on the life-history model were not well explained by a linear combination of life-history parameters causing our results to differ from those of Greenstreet et al. (2012). Our results also differed from those of Le Quesne and Jennings (2012) perhaps as a result of including group-specific catchabilities. The methods used to derive catch efficiencies assumed all individuals within a certain size and habitat group are equally likely to be caught in the trawl (Walker et al., 2017). This assumption may not be appropriate for species with variable behavioural responses to the trawl and species that occur on non-trawlable or deeper grounds, such as rocky bottoms, reefs and wrecks. For example, conger eel Conger conger inhabits rocky areas of the continental shelf and shelf slope (Xavier, Cherel, Assis, Sendão, & Borges, 2010). Once settled, it has a relatively sedentary lifestyle. Walker et al. (2017) assumed conger eel to be caught with the same catch efficiency at size as other eel-like species encountered near the seabed, which may be an overestimate. Similar considerations may apply to other species favouring untrawlable habitats. Furthermore, our study did not consider sensitivity of the species to pelagic fisheries or earlier targeted fisheries. Species like basking shark and thresher shark would owe their current IUCN Endangered status to these factors, rather than demersal fisheries.
Information about the survival of individuals that are discarded from different gear types was not available. However, many by-catch species are discarded and, if they survive, their populations may be more resilient to fishing than their life-history parameters would indicate. For most fish species, discard mortality will increase with the duration and depth of the tow, the time spent on deck and handling/discarding practices, but survival can also vary in response to size, sex, temperature, catch composition and handling (Ellis, McCully Phillips, & Poisson, 2017). Recent work on capture–release mortality could be used to improve the knowledge of the degree to which released fish survive to re-enter the population (Dapp, Walker, Huveneers, & Reina, 2016). Needless to say, discard survival will not impact species which are landed.
Spatially explicit models and management measures are potentially useful alternatives for managing sensitive species (Duggan, Eichelberger, Ma, Lawler, & Ziv, 2015; Walker et al., 2019). Spatially explicit models may provide indication of whether some species are likely to be at greater risk due to fishing than expected from our analyses and whether catches of sensitive species observed to be in decline can be limited by spatial measures (Shephard, Gerritsen, Kaiser, & Reid, 2012). However, such analyses must be based on a reliable distribution of the sensitive species, and this was only available for about half the sensitive species identified here.
Le Quesne and Jennings (2012) suggested that significant reductions of fishing effort in the Celtic Sea were required to allow sensitive species such as the common skate complex, spurdog and spotted ray Raja montagui to increase. The same conclusion was reached for the North Sea by Walker et al. (2019) for wolf-fish, anglerfish, brill, starry ray, spurdog, smooth-hounds, spotted ray and thornback ray (excluding the species not listed here as sensitive). Among these eight species, five were observed as increasing in this study. This may indicate either that these species are less frequently caught outside the heavily fished North and Celtic Sea, providing the possibility for them to increase in less fished areas, or that they are experiencing an increase in productivity due to, for example, warming waters or other ecosystem changes.
The IUCN ratings of the species were in broad agreement with the sensitivity of the species, and species ranked in the Critically Endangered, Endangered or Vulnerable categories were generally classified as sensitive in our analysis. However, the IUCN ranking of different species was not a reliable indicator of the potential for the species to sustain fishing pressure – the Least Concern and Data Deficient species included starry ray, tusk, marbled electric ray and wolf-fish, which were all ranked as sensitive and all decreased significantly from 1980 to 2018, indicating the potential error of assuming that species listed in these categories are not under pressure. Ideally, the status of these species should be updated during the next IUCN assessment cycle using survey data under the IUCN criteria (IUCN Standards and Petition Subcommittee, 2017). On the positive side, just above half of the species listed by IUCN as Critically Endangered, Endangered or Vulnerable increased significantly since 2000, hence providing some confidence that these species may in time move out of these categories.
The lack of a direct management of catches of a large proportion of the sensitive species means that some of the species in decline can continue to be caught. We stress here that while the assumption that the fishing mortality affecting non-target species is unlikely to exceed the fishing mortality affecting target species (Pope, Macdonald, Daan, Reynolds, & Jennings, 2000) is likely to hold for many species; this is not necessarily the case for species with restricted distributions. Furthermore, the fishing mortality predicted to generate MSY for the more productive species may not be sustainable for sensitive species. Species experiencing negative effects of other pressures or drivers (e.g. climate change) may also be more sensitive to fishing mortality than their general life-history parameters would suggest. Due to these caveats, species like wolf-fish, tusk and starry ray, which are at the southern limit of their distributions in the North Sea, are likely to be more sensitive to fishing than indicated in our sensitivity ranking. We recommend further monitoring of sensitive species and more regular assessments of their status to spur management actions before species enter the Critically Endangered, Endangered or Vulnerable IUCN categories. Furthermore, we recommend taking additional precaution where species are under combined pressure from climate change and fishing.
ACKNOWLEDGEMENTS
A.R. and D.R. received funding from the PROBYFISH project (EASME/EMFF/2017/022). D.R. also received funding from the Department of Agriculture, Food and the Marine's Competitive Research Funding Programmes. The funding bodies accept no responsibility for the paper.
AUTHORS' CONTRIBUTIONS
H.G. and A.R. conceived the ideas and designed the analyses; H.G. conducted the literature review and estimated species sensitivity, A.R. analysed species abundance and led writing of the manuscript. All authors contributed critically to the development of the analyses and manuscript and approved the final draft.
Open Research
DATA AVAILABILITY STATEMENT
The data used in this work are life-history parameters as given in Tables S2 and S3 in the Supporting Information and are available in Dryad Digital Repository https://doi.org/10.5061/dryad.8sf7m0cjd (Rindorf, 2020).
