Shedding light on the river and sea lamprey in western European marine waters

: Lampreys are ancestral jawless vertebrates with particularly complex life histories. Population declines resulting from increased anthropogenic pressure have been observed. For semelparous diadromous lampreys, the marine phase remains largely a black box, making targeted management and conservation measures difficult to implement. Here, we collated a database of 168 904 hauls from both fisheries-dependent and fisheries-independent surveys between 1965 and 2019. Lampreys were observed in only 254 hauls (<1% lamprey presence); 421 sea lamprey Petromyzon marinus and 300 European river lamprey Lampetra fluviatilis were identified. Sizes ranged from 13 to 92 cm and from 14 to 42 cm, respectively. The majority of lampreys (61%) were caught by mobile demersal gear types. The highest presence of both species was recorded within the Greater North Sea, followed by the Bay of Biscay. L. fluviatilis was observed closer to the coast than P. marinus . For both lampreys, there was an increase in size with distance from the coast. P. marinus were predominantly <60 cm and observed from August to February, indicating that these were sexually immature juveniles migrating out to sea. For L. fluviatilis , the majority were thought to be adults (>20 cm) and occurred in autumn, indicating inshore migration. Our observations provide insight into the ecology of lampreys at sea and highlight study locations and gear types, which may be more pertinent for future research. Greater awareness is needed during surveys to collate catch information on lampreys and improve understanding of their ecology and phenology at sea.


INTRODUCTION
Lampreys are ancestral jawless vertebrates with complex life histories (Kelly & King 2001, Potter et al. 2015. They occur within temperate waters of both the Northern and Southern hemispheres (Renaud 1997, Kelly & King 2001. Three semelparous anadromous lamprey species have been identified in north-western European waters: the sea lamprey Petromyzon marinus (Linnaeus, 1758); the Arctic lamprey Lethenteron camtschaticum (Tilesius, 1811), which has only been observed in Sweden and therefore not considered hereinafter) (Maitland 1980, Potter et al. 2015; and the European river lamprey Lampetra fluviatilis (Linnaeus, 1758) (Potter et al. 2015). Very little is known about the distribution at sea of anadromous lamprey species.
The population status of P. marinus and L. fluviatilis is of major concern. During the late 20 th century, the, the combined impacts of increased targeted fishing in estuaries, river pollution, freshwater habitat destruction and engineering works such as the construction of dams led to their decline (Kelly & King 2001, Beaulaton et al. 2008, Mateus et al. 2012, Lasne et al. 2015, Maitland et al. 2015. Climate change may also aggravate lamprey populations during their freshwater habitat occupancy (Lassalle & Rochard 2009, Maitland et al. 2015.
Both P. marinus and L. fluviatilis are listed under the Habitat Directive (92/43/EEC), the Bern Convention (Appendix III) and the Barcelona Convention. P. marinus is also listed under the OSPAR convention (Maitland et al. 2015). While considered as Least Concern in the IUCN Red List in Europe (Freyhof & Brooks 2011), both species are listed as vulnerable or endangered in many European countries (Mateus et al. 2012, Maitland et al. 2015, https://www.national redlist.org). For example, in France and Spain, where significant populations have been exploited (Beaulaton et al. 2008, Mateus et al. 2012, P. marinus is listed as endangered and vulnerable, respectively, and L. fluviatilis is listed as vulnerable and regionally extinct, respectively (Doadrio 2001, UICN Comité français et al. 2019.
The freshwater stages of the P. marinus and L. fluviatilis life cycle (spawning in rivers, larval stages, metamorphosis and downstream migration) are well characterised (e.g. Kelly & King 2001, Maitland, 2003, Docker & Potter 2019). The precise timings of migration to and from the sea vary with factors such as latitudinal clines and environmental conditions (e.g. temperature, rainfall), in addition to stream and river characteristics (Moser et al. 2015, Pavlov et al. 2017, Docker & Potter 2019). However, ecological information on the marine phase of P. marinus and L. fluviatilis (e.g. host species and size preference, movement and distribution at sea, mortality, return migration cues) largely remains a black box. Such knowledge gaps make targeted management and conservation measures difficult to implement (ICES 2015, Hansen et al. 2016.
Adult P. marinus upstream spawning migration takes place from February to June in north-western European waters (Maitland 2003, Moser et al. 2015, Hansen et al. 2016. In south-western European waters, migration has been observed from December to June, with peaks from February to April (depending on populations) (Moser et al. 2015, Hansen et al. 2016. L. fluviatilis upstream migration can extend from July to June the following year, though they appear to have distinct autumn and spring runs (Maitland 1980, 2003, Moser et al. 2015. P. marinus and L. fluviatilis larval (ammocoete) stage duration in European waters is approximately 3 to 5 yr (Maitland 2003, Dawson et al. 2015, Hansen et al. 2016. During this phase, they feed on detritus and microorganisms within the soft sediment of rivers and streams (Taverny et al. 2012, Potter et al. 2015. Post-metamorphic P. marinus (approximately 10 to 22 cm) downstream migration in north-western Europe takes place between late autumn and early winter (Maitland 1980, Bird et al. 1994, Quintella et al. 2003, Silva et al. 2013b, Hansen et al. 2016. Within south-western European waters, downstream migration has been ob served between October and May, with peaks in February and March (Silva et al. 2013b, Hansen et al. 2016. Post-metamorphic L. fluviatilis (9−17 cm) downstream migration has been recorded from midwinter through to April (Maitland 2003, Dawson et al. 2015, Pavlov et al. 2017.
Given the recent declines in lamprey abundance, a better understanding of their ecology during their marine phase is needed. This would help determine threats posed to these species while at sea and identify conservation measures required to improve populations (Maitland et al. 2015). Here, we collated a substantial database of 168 904 hauls which occurred in north-western European waters. The data were obtained from fisheries-independent (scientific surveys) and French fisheries-dependent data (from fishing vessels; Tables S1, S2 and Fig. S1 in the Supplement at www. int-res. com/ articles/ suppl/n044 p409 _ supp .pdf). Given the existing literature on the life history of lampreys, we expected to see spatial and size differences, with P. marinus having a more dispersed distribution and larger size ranges than L. fluviatilis (Maitland 2003, Potter et al. 2015. We also expected to observe periods during the year when lampreys migrate to and from marine waters and are more likely to be caught.

Survey data
Fisheries-dependent and fisheries-independent data were collected within European waters (Greater North Sea, Celtic Sea, Bay of Biscay and Iberian coast, and Metropolitan French waters within the Mediterranean; Fig. 1 and Tables S1 & S2). Fisheries-independent data from ICES were extracted from the Database of Trawl Surveys (DATRAS) portal (https:// www. ices.dk/data/data-portals/ Pages/ DATRAS .aspx). Metropolitan French scientific surveys (excluding data available through DATRAS) were collated from the Institut Français de Recherche pour l'Exploitation de la Mer (IFREMER). Information from Regimbart et al. (2018), which highlights survey information on diadromous fish, was used to gather French scientific surveys (https:// campagnes. flotte oceano graphique .fr/ campaign; Table S1).
Fisheries-dependent data came from the observation of on-board fishing vessels, referred to as ObsMer (Cornou et al. 2015). ObsMer data are a collection of on-board catch data held by IFREMER and available on request from the French Ministry of Fisheries and Aquaculture (Direction des pêches maritimes et de l'aquaculture). ObsMer data provide targeted and bycatch, landed and discarded data from fishing vessels throughout the year within the Greater North Sea, the Celtic Sea, the Bay of Biscay and the Mediterranean. According to the sampling plan for ObsMer data, observers randomly sample professional fishing vessels and fishing operations when on board (Fauconnet et al. 2015).
Only fully processed and fully marine hauls, downstream of transitional waters, were taken into account. Due to missing data and insufficient information on the length of hauls, size of vessels, mesh size, etc., it was not possible to calculate catch per unit effort. Equally, the capture of lampreys at sea is directly related to the capture of their host and is therefore susceptible to vary depending on when in the haul the host was captured. Details on gear types, number of hauls and years the surveys were undertaken are outlined in Table S1. To help evaluate lamprey capture from the different surveys, gear cate-gories were identified from the type of gear they were caught with, whether the gear was static or mobile, and the water zone (pelagic, demersal or benthic) in which the gear was employed (Table S2).
For all surveys, the number of lampreys captured and some biological information (species, number of fish per haul, length and in some cases weight) were provided. However, because of the lack of information to quantify survey effort and variability in catch between gears and seasonal and spatial effort, only presence−absence and length data per haul were considered in the analysis.

Frequency of occurrence
As a result of the limited number of presence observations and associated difficulties in taking into account spatio-temporal heterogeneity, only simple robust statistical analyses were performed to avoid over-interpretation of the data. The initial data set contains less than 1% presence, making the analysis of presence−absence data very difficult. To understand the effect of distance from the coast on lamprey presence, all gear types, surveys and ICES divisions which did not contain lamprey presence were removed from the dataset, thus improving the balance ratio between presences and absences (Fielding & Bell 1997). To further reduce zero inflation, the nonrandom spatial distribution of fisheries observer data, and the lack of knowledge of the precise location of capture (beginning or the end of the haul), the study area was downscaled and divided into a regular grid (20 km 2 ). Grid cells of 10, 20 and 30 km 2 were tested, but 20 km 2 was found to be best to reduce zero inflation and yet not lose too much detail. For each grid cell, the central point was assigned a value of 1 if it contained a presence and a value of zero if no presence was recorded in the cell, as recommended by Aarts et al. (2012), Keil et al. (2013) and Pointin et al. (2018). This process brought the percentage of zeros to 98% for Petromyzon marinus and 96% for Lampetra fluviatilis.
Distance from the coast (km) and season (spring [March− May], summer [June−August], autumn [September− November] and winter [December− February]) were used to examine spatial variation of lamprey presence− absence, with gear and year included as random effects when significant (Eqs. 1 & 2). Depth was not taken into account since it was not possible to obtain the precise depth at which lamprey presence was recorded. Binomial generalised linear models (GLMs) were used to quantify the effect of distance from the coast for both lamprey species. A complementary log-log link function was used since it is better adapted to data with more zeros (Zuur et al. 2009). Eqs. (1) & (2) outline the model of best fit for both species. (1) ( 2) where PA i,j is the complementary log-log probability of presence (link function), β 0,j and β 1,j are the parameters that depend on species j, i refers to the number of samples and RE refers to the random effect. A post hoc Tukey's HSD test was used to identify differences between seasons. To identify whether one species presence was closer to the coast than the other, an ANOVA test was performed followed by a post hoc Tukey's HSD test on the species-specific parameters β 0,j and β 1,j .

Size
For both species, length data were analysed using GLMs to explore the effect of distance from the coast and the existence of potential interactions between seasons and distance. To explore potential seasonal length differences in lampreys, all individuals were first modelled against season with distance from the coast. Since there were few large P. marinus, smaller individuals were then modelled against season. In line with literature on P. marinus post-metamorphic sexually immature sizes (e.g. Maitland 1980, Hansen et al. 2016, smaller individuals were characterised as less than 50 ± 10 cm. Since these smaller individuals were not observed in June and July, seasons were then categorised into 3 groups: summer to autumn (August−November), winter (December−February) and spring (March−April). To assess the robustness of our results for the definition of seasons, different combinations of months within the seasonal categories were also tested. Latitudinal variations in lamprey length were also explored using the above method to understand potential seasonal migration timing differences. Gear and year were included as random effects for significance. Eqs. (3) & (4) outline the model of best fit for P. marinus and L. fluviatilis, respectively. (3) For all GLMs, the model of best fit was identified by the lowest Bayesian's information criterion (BIC). BIC was used instead of Akaike's information criterion since it performs better when there is heterogeneity in the dataset and the sample size is smaller (Brewer et al. 2016). Model significance was tested against the null hypothesis using a log-likelihood ratio test. All mapping and statistical analysis was undertaken in R CRAN free software (version 3.3, http://cran.r-project.org).

RESULTS
Between 1965 and 2019, 721 lampreys (421 Petromyzon marinus and 300 Lampetra fluviatilis) were identified from 68 287 hauls from scientific trawl surveys and 100 617 hauls from French fishing vessels (  Tables S3 & S5). Although far fewer lampreys were caught by the fisheries-dependent data, a wider range of gear types targeting a variety of fish species caught lampreys (Tables S3 & S6).

Frequency of occurrence
Presence of L. fluviatilis was highest in the Greater North Sea along the Dutch, Swedish and western German coasts (ICES divisions 3.a.21, 4.b and 4.c) ( Fig. 2; Table S7). P. marinus presence was highest in the Greater North Sea, where samples were widely dispersed among open-sea and coastal locations (Fig. 2), followed by the Bay of Biscay (ICES division 3.a, 4.b−c and 8.a−b; Fig. 2; Table S7). L. fluviatilis occurrence was higher closer to the coast (24.29 ± 35.50 km, mean ± SD) than for P. marinus (36.21 ± 49.91 km; Tukey's HSD test p < 0.05), whose distribution was more dispersed (p < 0.001, df = 2, deviance explained 7%; Figs. 2 & 3; Table S3). L. fluviatilis was also ob served closer to the coast in summer and autumn than in winter or spring ( Fig. 3a.ii, Tukey's HSD test p < 0.01 and p < 0.05, respectively).

Size
P. marinus length ranged between 13 and 92 cm, whereas L. fluviatilis length ranged between 14 and 42 cm (Table S3). An increase in P. marinus and L. fluviatilis length with distance from the coast was detected (p < 0.001, df = 5, deviance explained 4% and p < 0.001, df = 5, deviance ex plained = 3%, respectively; Fig. 3b). On average, larger P. marinus were caught by mobile demersal gear than by other gear types (Fig. 3b). No clear statistical seasonal or latitudinal length differences were observed for either species (Fig. 3b).
Across the range of latitudes in which P. marinus were present, the smallest individuals were observed in winter. Too few large individuals were present to observe seasonal latitudinal differences. Since the majority of L. fluviatilis were observed in the North Sea, no seasonal latitudinal variation was detected.
Two length modes were detected for P. marinus, with the presence of larger individuals (greater than approx. 50 cm) scattered across the course of the year and the presence of smaller individuals (less than approx. 40 cm) with a higher frequency from August through to February (Fig. 4a.i, b.i). A higher percentage of P. marinus was observed at the beginning of the year (Fig. 4b.i), whereas a higher percentage of L. fluviatilis can be seen in autumn, with no clear seasonal size trend (Fig. 4a.ii, b.ii).  Table S2 for more detail

DISCUSSION
The presence of lampreys was reported in less than 1% of hauls. Given the wide range of depths, gears types and regions where surveys were undertaken, the very low percentage of presence indicates a combination of rarity, low detectability and poor reporting rate. These results therefore highlight the difficulty in understanding the marine life history phase of lampreys.

Capture
Lampreys at sea live as an external parasite of host species, and their capture is almost entirely dependent on catching the host species. Lampreys have been observed to parasitise species which the fisheries-dependent data were targeting (e.g. gadoids, mullet, hake), indicating a link between the abundance of the targeted fish and the abundance of lampreys (Silva et al. 2014, Renaud & Cochran 2019. However, clear information on lamprey hosts is sparse, with no definitive trends other than attacking species which may be more abundant and have thinner skin (Renaud & Cochran 2019). The lack of confirmed lamprey host trends hinders targeted lamprey surveys. In our database, presence of both lamprey species was higher where mobile demersal gear was used, indicating that the majority of lamprey hosts may be demersal species. Lança et al. (2014) also found a relationship between benthic hosts and Petro myzon marinus, possibly as a result of declines in pelagic-preferred hosts such as shads and salmonids (Merg et al. 2020). Insufficient survey information was available to calculate catch per unit effort for each gear type, making it difficult to accurately quantify rates of gear or métier capture (which considers the gear and target species).
A greater proportion of lampreys were observed from the fisheries-independent data than from the fisheries-dependent data. Since lampreys may detach themselves from their hosts at the time of capture (Halliday 1991), the higher presence of lampreys from scientific surveys may be as a result of the shorter hauls in scientific surveys and, hence, reduced likelihood of disturbance and the greater likelihood of the parasitic lamprey remaining attached to the host fish. Alternatively, parasitic species may not have necessarily been recorded within fisheriesdependent surveys. Thiel et al. (2009) analysed lamprey catch from commercial published records and fisheries research data within the Baltic Sea. Al though very few observations were made, more lampreys were recorded from fisheries-dependent data (5 records, 9.3%) than from scientific surveys (1 re cord, 1.8%). From our database, more lampreys were caught from mobile demersal gear across northwestern European waters. The greater presence of lampreys by scientific data will therefore probably have been because of the larger number of scientific hauls within the southern North Sea. Information on lamprey stocks across Europe is sparse (Beaulaton et al. 2008, Thiel et al. 2009, Mateus et al. 2012, ICES 2015, Silva et al. 2017).

Frequency of occurrence
Although our database contains trawls from across the European distribution of both species (Maitland 2003), neither of the 2 species was found in all ICES ecoregions. Lampetra fluviatilis presence was closer to the coast than P. marinus, which was more dispersed; this finding is in line with existing literature (e.g. Maitland 2003, Thiel et al. 2009, Potter et al. 2015. L. fluviatilis was recorded in only 2 ICES ecoregions (the Greater North Sea and the Bay of Biscay and Iberian coast), whereas P. marinus was also found to occur in the Celtic Sea and the Mediterranean. The higher presence of both species in the Greater North Sea and the Bay of Biscay may have been a result of the greater sampling effort within these regions.
The presence of L. fluviatilis was particularly high along the coast of Germany and Holland within the south-eastern North Sea. Despite this being a heavily fished sea (Berg et al. 1996) few published papers have referred to L. fluviatilis presence within this area (e.g. Admiraal et al. 1993, Thiel & Salewski 2003, Pavlov et al. 2017. Conversely, few L. fluviatilis were observed off the coast of the Humber estuary, even though significant captures and a commercial fishery exist further upstream within the River Derwent and the River Ouse (Jang & Lucas 2005). The majority of surveys within the North Sea came from DATRAS bottom trawl surveys and NS-IBTSs, which are undertaken between August and February. According to Jang & Lucas (2005), the main pre-spawning up stream migration period for L. fluviatilis in the River Derwent is between No vember and February. This slight mismatch in dates between surveys and lamprey migration may therefore ex plain the few presences observed in this area of the UK.
No lampreys were observed along the Iberian coast where populations are known to exist (Quintella et al. 2003, Mateus et al. 2012, Silva et al. 2017). This may be because the majority of hauls were under-taken within the Greater North Sea. Lança et al. (2014) found that P. marinus migrate to deeper oceanic regions off the Iberian west coast. The majority of hauls analysed within this study were closer to the coast, with few samples off the Iberian coast, which may explain the lack of either lamprey species observed within this region. Furthermore, the surveys undertaken within this region were from Portuguese IBTSs, which take place between September and November, before seaward and up stream P. marinus migration takes place (Silva et al. 2013a, Moser et al. 2015, Hansen et al. 2016).

Size
The majority of P. marinus caught were less than 40 cm, with a higher presence of small P. marinus between autumn to spring. These observations indicate that the P. marinus were most likely post-metamorphic sexually im mature individuals migrating into marine waters (Quintella et al. 2003, Silva et al. 2013b, Hansen et al. 2016). Most P. marinus were small individuals, most likely because the majority of surveys took place in waters closer to the coast, where small individuals would be present in greater numbers, prior to any dispersal of the host fish.
Although no statistical seasonal latitude variations in small P. marinus were observed, the wide range in size of P. marinus less than 40 cm may partially be due to latitudinal variations of post-metamorphic migration to marine waters. More northerly populations are known to migrate in autumn to winter and more southerly populations migrate in winter to spring (Silva et al. 2013a, Hansen et al. 2016. The wide range in size of small P. marinus may also indicate that juveniles stay close to the coast for these initial seasons before migrating further offshore. Alternatively, some individuals may choose to migrate into marine waters later in the year and at a larger size (King & O'Gorman 2018).
A bimodal length tendency was observed for P. marinus, with the majority of small individuals (< 40 cm) present from autumn to spring, when metamorphosed seaward migration is considered to take place (Silva et al. 2013a, Hansen et al. 2016). The few large (> 60 cm) individuals were found further from the coast and caught over the course of the year. These results corroborate with existing literature (e.g. Halliday 1991, Silva et al. 2013a) that P. marinus occupy marine waters for over a year, given that large and small individuals were observed in autumn and winter.
Although fewer hauls were undertaken during summer months, proportionally far fewer lampreys were observed, albeit the ratio of gear types remained the same. Mortality of adult lampreys at sea (e.g. from potential lack of hosts) is poorly understood (Maitland et al. 2015, Hansen et al. 2016), but bycatch fishing mortality at sea appears to be low (Stratoudakis et al. 2016) as opposed to targeted fishing mortality within estuaries (Beaulaton et al. 2008). Fewer individuals were identified in summer, potentially as a result of offshore migration to deeper waters further from the coast (Lança et al. 2014). Larger P. marinus may parasitise larger species (e.g. cetaceans and larger elasmobranchs) (Halliday 1991), which are rarely caught in fisheries-dependent and fisheriesindependent surveys. Cetaceans and elasmobranchs also often migrate long distances. Lamprey wounds on marine mammals are nonetheless rare (Renaud & Cochran 2019, W. Dabin unpubl. data), and lamprey host size selectivity tendencies are not clear (e.g. Swink 1991).
The majority of L. fluviatilis were observed in autumn, with a mean length of 30 cm. This corresponds with adults returning to freshwater to spawn (Maitland 2003, Dawson et al. 2015, given that adult size ranges from 20 to 50 cm (Maitland 1980). A bimodal length frequency for L. fluviatilis was not evident. Since the majority of L. fluviatilis were caught in autumn within 50 km from the coast and appeared to be adults, our results support the hypo thesis that L. fluviatilis occupy marine waters for considerably less time (less than a year) than P. marinus (Docker & Potter 2019, Renaud & Cochran 2019. Furthermore, few individuals were observed over the course of the summer and L. fluviatilis is not known to migrate far from the coast (Maitland 2003, Thiel et al. 2009, Potter et al. 2015. Increased coastal sampling from February to May, when P. marinus and L. fluviatilis return to rivers, may help improve the understanding of growth at sea. Tagging studies may be better adapted to understand lamprey migration timing and mortality at sea (Silva et al. 2013a). However, the probability of recapture has been observed to be low (e.g. Silva et al. 2013a), potentially because both lamprey species discussed here adopt the suitable river strategy rather than homing to natal streams (Moser et al. 2015). Furthermore, microacoustic tags have only recently been developed (Mueller et al. 2019), and minimum lamprey size for acoustic implants is 14 cm (Mueller et al. 2019).
Conservation measures have been set up across the globe to try and restore lamprey populations (e.g. rebuilding programs, improved river connectivity, removal of barriers and weirs, spatial areas of conservation) (Renaud 1997, Thiel et al. 2009, Maitland et al. 2015. As for many diadromous fish, most of those conservation measures concern the freshwater phase of the life cycle. Setting efficient conservation measure at sea is difficult because of our lack of understanding of the ecology and phenology of lampreys at sea. From our database, there were too few presence observations to identify a critical spatio-temporal window to be protected. Fishing mortality from bycatch does, however, seem to be low. Although lamprey sighting from hauls was low, greater awareness is needed in both fisheries-dependent and fisheries-independent surveys to collate more data on non-target species such as lampreys. On-deck examinations of landed fish for signs of lamprey scarring could provide substantial additional information (King 1980, King & O'Gorman 2018. This might be feasible for scientific surveys but the logistics would be difficult with large samples on board commercial fishing vessels. Nonetheless, such information could provide additional information and help unravel important aspects of lamprey life history during their marine life history phase. This could help improve our understanding of threats to lamprey survival and implement more targeted conservation measures, such as temporal spatial closures within estuaries when adults migrate back to freshwater.