Glass eel (Anguilla anguilla ) recruitment to the river Lis: Ingress dynamics in relation to oceanographic processes in the western Iberian margin and shelf

Received: 8 September 2017 | Accepted: 20 February 2018 DOI: 10.1111/fog.12274 ORIGINAL ARTICLE Glass eel (Anguilla anguilla) recruitment to the river Lis: Ingress dynamics in relation to oceanographic processes in the western Iberian margin and shelf Yorgos Stratoudakis1 | Paulo B. Oliveira1 | Ana Teles-Machado2 |  Manuel Oliveira1 | Maria Jo~ Jose ao Correia3 | Carlos Antunes4 1 IPMA, Lisboa, Portugal 2 ^ncias, Instituto Dom Luiz, Faculdade de Cie Universidade de Lisboa, Lisboa, Portugal 3 Abstract To provide new information on the ingress dynamics of glass eels at the Portuguese MARE-Marine and Environmental Sciences ^ncias, Centre, Faculdade de Cie Universidade de Lisboa, Lisboa, Portugal coast, the study analyzes catch per unit effort (CPUE) data from river Lis, biometric 4 CIIMAR, Terminal de Cruzeiros do Porto ~es, Matosinhos, Portugal de Leixo from the Iberian margin and shelf. Biometric data of glass eels in western Portugal Correspondence Y. Stratoudakis Email: yorgos@ipma.pt and show clear seasonality, with longest and heaviest individuals arriving from Octo- data from recruits in basins of western Portugal and meteo-oceanographic data are congruent with the latitudinal size gradient of leptocephali in oceanic surveys ber to December. Generalized additive models fitted to CPUEs from two experimental fishing periods (1996–1997 and 2013–2014) show that availability of glass eels Funding information European Union, PNAB/Data Collection Framework this is an EU funding scheme; ^ncia e a Tecnologia, Fundacß~ ao para a Cie Grant/Award Number: SFRH/BPD/100720/ 2014 to the sampling gear was inversely related to lunar phase and sea wave height. CPUE was lowest from May to September, increased towards the end of the year and peaked in February–March, when recruits were almost 40% lighter than autumn arrivals. Commercial CPUE during the 1989–1990 fishing season was significantly higher than experimental fishing data in 1996–1997. Higher variance and smaller sample size in 2013–2014 prevent conclusive interdecadal comparisons. Finally, CPUE was significantly higher during the prevalence of northward current flow off western Portugal and during strong cross-shelf westerly winds. Findings support the suggestion that eel recruitment in Portugal occurs mainly by leptocephali travelling along the Azores Current, deflected northwards through the Iberian Poleward Current, with river ingress, and possibly oceanic metamorphosis, modulated by seasonal dynamics in coastal hydrology and shelf/upper margin oceanography. KEYWORDS Azores Current, biometry, catch per unit effort, cross-shelf transport, generalized additive models, Iberian Poleward Current 1 | INTRODUCTION of marine animals (Righton et al., 2016). Spawning mainly takes place in March and April (Miller et al., 2015), or slightly earlier (Righton European eels (Anguilla anguilla) spawn panmictically in the southern et al., 2016). Young leptocephali, transparent and laterally com- Sargasso Sea within the Subtropical Convergence Zone (Kleckner & pressed, develop eyes and teeth for exogenous food intake, possibly McCleave, 1988; McCleave et al., 1998; Miller et al., 2015; Munk particulate organic matter (Miller, 2009; Miller & Tsukamoto, 2017). et al., 2010; see also Figure 1a). Maturing silver eels swim 5,000– The duration of the larval oceanic phase remains uncertain and an 10,000 km away from continental waters in the northeast Atlantic issue of controversy. Ageing by some otolith microstructure studies and the Mediterranean Sea in one of the longest known migrations point to less than a year, while length cohort analysis and other types Fisheries Oceanography. 2018;1–12. wileyonlinelibrary.com/journal/fog © 2018 John Wiley & Sons Ltd | 1 2 | STRATOUDAKIS ET AL. F I G U R E 1 (a) Mean (1993–2016) North Atlantic geostrophic flow in the 50–300 m layer from satellite and in situ observations (Global ARMOR3D analysis). Boxes in broken lines indicate approximate eel spawning area in the Sargasso Sea (left) and Iberian Peninsula and margin considered in the regional circulation model (right). Ellipse (centered at 28°W, 39°N) highlights the position of the Azores archipelago. Dotted lines and arrows indicate prevailing mean flow trajectories from the eel spawning area towards Europe (see also figure 24 in Miller et al., 2015). (b) West Iberia circulation model solution for the average (50–300 m) salinity (shade; scale in panel (d) and flow (lines) in December 2006 (Teles-Machado et al., 2016)). Box (39.7–40.1°N) indicates the marine area off river Lis (shaded basin) used to extract meteooceanographic data for monthly climatologies and for explanatory variables in GAMs. Dotted line and arrow indicate prevailing flow trajectory for the Iberian Poleward Current. The main river basins and Atlantic mouth locations in the Iberian Peninsula are presented (Gdr, Guadalquivir;  n). The positions of hauls with leptocephali larvae in the 1991 oceanic Gda, Guadiana; Te, Tejo; Mo, Mondego; Do, Douro; Mi, Minho; Na, Nalo survey (Antunes & Tesch, 1997b) are marked by stars. Model salinity (shade) and temperature (contour) vertical sections for the west Iberian shelf and slope at 40°N in (c) December 1995 and (d) December 2006 of studies suggest a minimum of 1.5 years (Bonhommeau, Cas- Depending on the relative importance of the Azores Current in the cir- , & Le Pape, 2010). Lagrangian models simulattonguay, Rivot, Sabatie culation model, a secondary peak may appear off northern Africa at ing the oceanic transport of European eel leptocephali from the 35°N (Blanke et al., 2012; Bonhommeau et al., 2009; Kettle & Heines, Sargasso Sea favour the longer duration hypothesis and suggest a 2006; Pacariz et al., 2014; Zenimoto et al., 2011). An alternative main pathway through a western/northerly entrainment within the transport route is through entrainment to eastward flowing frontal jets Antilles/Florida Current and subsequent advection through the Gulf and drifting to the northeast, but to the south of the Azores, to reach Stream and the North Atlantic drift (Blanke, Bonhommeau, Grima, & the Azores Current is possible (Miller & Tsukamoto, 2017; Miller et al., Drillet, 2012; Bonhommeau et al., 2009; Kettle & Heines, 2006; Miller 2015), particularly for eggs hatching at the eastern region of the et al., 2015; Pacariz, Westerberg, & Bjork, 2014; Zenimoto, Sasai, spawning area (Munk et al., 2010; Pacariz et al., 2014). Sasaki, & Kimura, 2011). This pathway results in a peak of arrivals to Larval size distribution in oceanic waters of the eastern North the European continental margin at around 50°N (northern Bay of Bis- Atlantic contradicts expectations of dispersal dynamics exclusively cay) and minimal returns at 40°N (western Iberian Peninsula). driven by the Gulf and North Atlantic currents (Bast & Strehlow, STRATOUDAKIS | ET AL. 1990; Tesch & Niermann, 1992). Apart from the increase with distance from the Sargasso Sea, individual survey (Bast & Strehlow, 1990) and aggregate data (McCleave et al., 1998) reveal a persistent latitudinal gradient in leptocephali length: larvae at 35–38°N at any longitude from the Azores eastwards are up to 10 mm smaller 3 2 | MATERIALS AND METHODS 2.1 | Oceanography and hydrology in the study system than those further north. Plankton surveys off the European conti- The western Iberian coast forms the eastern boundary of the North nental shelf from 1975 to 1991 have confirmed that leptocephali Atlantic between 36 and 44°N, extending south-north along the arriving south of 43°N, at the western Iberian margin, are in earlier 9°W meridian. This region is characterized by absence of persistent developmental stages and significantly smaller overall and per stage upper-layer currents, in contrast with its western counterpart where than leptocephali further north at the Bay of Biscay (Antunes & the Gulf Stream (GS) detaches from the American continental shelf Tesch, 1997a; Bast & Strehlow, 1990; Tesch, 1980; Tesch & Nier- and spreads eastwards (Figure 1a and b). The vertically integrated mann, 1992). Bypassing the debate of active versus passive trans- mean currents in the upper layer (50–300 m), computed using port (Bonhommeau et al., 2009; McCleave et al., 1998), the 24 years (1993–2016) of monthly observation-based estimates of persistent latitudinal difference in size of pre-metamorphic lepto- the global 3D geostrophic circulation (CMEMS Global ARMOR3D L4 cephali suggests that larvae at lower latitudes arrive faster to the Reprocessed dataset—Mulet, Rio, Mignot, Guinehut, & Morrow, European margin, or follow a pathway that results in slower growth 2012), show a GS branch at approximately 40°N, 50°W where part (McCleave et al., 1998). of the flow recirculates southward and another fraction re-assumes The oceanic phase ends with metamorphosis into glass eels, an an eastward direction south of 34°N, feeding the Azores Current ontogenetic process that results in a sudden reduction of buoyancy (Klein & Siedler, 1989), whose axis is remarkably zonal and attains (Tsukamoto et al., 2009). Plankton surveys have not found lepto- highest mean speeds at the longitudes of the Azores archipelago cephali in the continental shelf (Antunes & Tesch, 1997a; Tesch, (dotted ellipse in Figure 1a). 1980; Tesch & Niermann, 1992), and showed that metamorphosis Off western Portugal, the shelf width is approximately 40 km in into glass eels starts in the slope region, possibly around the the area north of Lisbon and half this width to the south. Circulation 1,000 m isobath (Antunes & Tesch, 1997a,b; Tesch, 2003). The trig- is seasonally dominated by the IPC, a slope-trapped flow from the ger of metamorphosis remains unknown, but only happens once an Gulf of Cadiz northwards (dashed arrow in Figure 1b) transporting appropriate size limit has been reached (Antunes & Tesch, 1997b) warm and salty waters from the south (Frouin et al., 1990; Peliz and may involve cues of land proximity able to guide glass eels et al., 2005; Teles-Machado et al., 2016). The IPC starts developing towards appropriate continental habitats (Miller, 2009). During meta- during late summer as an undercurrent with the core at around morphosis, leptocephali lose the early teeth, enlarge the olfactory 200 m depth, when the shelf and upper slope are still occupied by organ and rearrange the position of the gut, interrupting feeding the southward upwelling jet. In December and January, under south- until settlement (Miller, 2009). This leads to substantial weight loss erly wind conditions, the IPC core migrates from the shelf break and some shrinking, with a concomitant reduction in body condition, depths to the surface, becoming a surface intensified jet. From while larvae modify the height to width body ratio and alter the February to March the current deintensifies and part of it propa- swimming mode to adapt to bottom dwelling (Tesch, 2003; Tsuka- gates offshore (Teles-Machado et al., 2016). The IPC also extends to moto et al., 2009). the northern coast of the Iberian Peninsula, usually in December or Here, we describe the ingress dynamics of glass eels to river January (Pingree & Le Cann, 1990), and can occasionally be detected Lis (small catchment in central Portugal) using catch per unit effort up to southern France. In addition to its pronounced seasonality, the (CPUE) data from experimental and commercial fishing with a current exhibits considerable inter-annual variation, both in cross- hand-net at a beach next to the river mouth. Generalized additive shelf position and salinity (Figure 1c and d). models (GAMs) are applied to biometric and CPUE data to model The river Lis (shaded basin in Figure 1b) is situated approxi- statistically and interpret ecologically ingress dynamics unobstructed mately 100 km to the north of the westernmost tip of continental by subsequent processes within the estuary and during upstream  Canyon, one Europe and is the first river to the north of the Nazare migration. Environmental catchment data, remotely-sensed and of the largest submarine canyons in Europe bringing the shelf break observation-based model outputs of oceanic circulation are used as to less than a mile from the shore. The river Lis has its origin at the explanatory variables to explore the potential influence of the central region of Portugal (part of the Mondego hydrographic region) Azores Current (Klein & Siedler, 1989) and the Iberian Poleward draining an area of approximately 850 km2 and flowing for 40 km za, Ambar, & Boyd, 1990; Peliz, Dubert, Current (IPC—Frouin, Fiu along its heavily modified watercourse into the Atlantic Ocean, Santos, Oliveira, & Le Cann, 2005; Teles-Machado, Peliz, McWil- immediately to the north of the Vieira de Leiria beach. The area liams, Couvelard, & Ambar, 2016) in the migration of leptocephali  littoral cell of linear sandy coast forms part and the Douro-Nazare and the recruitment of glass eels in Portugal. Knowledge gaps with common behaviour of alongshore sediment transport due to related to the stimulus of metamorphosis and mechanisms of similar wave exposure and shelf morphology (Silva, Taborda, Bertin, detrainment are highlighted and discussed in the context of cross- & Dodet, 2012). Annual mean rainfall is 956 mm, with rainy and cold shelf circulation dynamics. winter (from October to March, but mainly up to December) and hot 4 | STRATOUDAKIS ET AL. and dry summer. There is a prevalence of northerly, NW winds espe- signature (values >35.9). These high salinity values, not observed cially during summer and a predominant swell from the northwest from June to September, appear from October to January within the (Silva et al., 2012). outer continental shelf, while from January to March expand further Strong seasonal patterns are also observed in the cross-shelf offshore in the upper slope (Figure 2f). dynamics off Lis (Figure 2). The mixed layer depth is <30 m from April to October (Figure 2a), when surface warming leads to vertical stratification offshore and persistent northerly winds promote 2.2 | Glass eels data coastal upwelling inshore, with surface layer Ekman transport of Glass eels fishing in Portugal increased during the 1970s (Weber, colder and less salty water to the mid and outer shelf (Figure 2b and 1986) and by 1985 a negative trend in abundance was evident, lead- c). The mixed layer depth increases from November to February/ ing to tighter fishery regulation (Domingos, 1992). Monitoring at March. In February maximum values are attained (>80 m) and the river Lis started in 1989 to assist anguilliculture planning by seeking offshore extension of turbid waters of river origin (Fernandez-Novoa to identify major sources of variation in the availability of glass eels et al., 2017) is maximal, reaching the outer shelf (Figure 2d). Ekman (Bessa, 1992; Bessa & Castro, 1994, 1995). In 1996 an experimental transport is minimal from October to January, when recurrent south- fishing program was initiated with the collaboration of the local erly wind episodes may force convergence of the surface waters fisher association. Three fishers were provided with a license to fish onshore and, especially during the late months of the year, turbid beyond the November–February stipulated fishing season, on the waters from early rains are confined to the region near the coast. condition of providing data and samples. Within the fishing season Contrary to the surface layer (0–50 m depth), seasonal variation is fishers operated as frequently as possible, but for the rest of the limited in the deeper layer (50–300 m; Figure 2e and f), with the year potential activity was restricted to two consecutive nights at most pronounced seasonal pattern related to the IPC salinity each quarter of the moon cycle. The program started in February F I G U R E 2 Twenty-year (1989–2008) average seasonal variability of cross-shore profiles from the West Iberia circulation model solution in the coastal area off river Lis (see also Figure 1b for location and Figure 1c and d for examples of vertical profiles). (a) mixed layer depth (m), (b) upper (0–50 m) layer water temperature, (c) upper (0–50 m) layer water salinity, (d) remote sensing reflectance at 645 nm (for the complete satellite period, 2003–2016), (e) deeper (50–300 m) layer water temperature and (f) deeper (50–300 m) layer water salinity. In each panel, upper axis represents distance from coast (km) and solid triangles over the horizontal axis indicate the position of selected isobaths (labels at top axis of panel [e]). Superposed dotted surface in salinity panels (c, f) indicates the 4 9 10 4/sr fully normalized remote sensing reflectance at 645 nm, selected to represent the offshore extension of the river Lis turbid plume STRATOUDAKIS | ET AL. 5 1996 and ended in December 1997, reporting data for 242 fishing remote sensing reflectance at 645 nm data from MODIS (MODer- nights (80 within the fishing season and 162 outside—Table S1). ate resolution Imaging Spectrometer) were used as proxy of turbid Additionally, biometric data on fresh glass eels were collected plumes of riverine origin (Fernandez-Novoa et al., 2017); multi- monthly throughout the year 2000 in river Lis and opportunistically year (2003–2016) monthly averages were computed for the Ibe- in other years and other small river systems off western Portugal rian area from the “GlobColour 8-day MODIS product” available along 1995–2001. Each sample usually comprised 50–100 individu- at 4 km resolution (Maritorena, Fanton d’Andon, Mangin, & Siegel, als, for which total length (mm), body height and width at the first 2010). dorsal fin area (mm) and total wet weight (g) were obtained prior to fixation or release. Meteorological data for the study area and the period from 1996 to 2014 were extracted from ERA interim—the global atmospheric Fishing for glass eels was banned across Portugal in 2001, with reanalysis produced by the ECMWF (European Centre for Medium- the exception of the international river Minho that is jointly man- range Weather Forecast—Dee et al., 2011). Ocean temperature, aged with Spain and where a commercial fishery is still in operation salinity and currents down to 300 m depth, for the same period, with stow nets (Antunes, 1994; Iglesias, Lobon-Cervia, Costa-Dias, & were extracted from ARMOR3D. These observation-based data were Antunes, 2010). As part of the eel management plan for Portugal, complemented with high resolution model solutions (ROMS) from a IPMA launched a pilot project for monitoring glass eels during 2013– realistic simulation that spans the period of 1989–2008 (Teles- 2014 with the intention to replicate the sampling conditions of the Machado et al., 2016). The data from the realistic model were used 1996–1997 study in the river Lis. The same fishers that performed to produce vertically-averaged horizontal salinity and circulation the experimental fishing during 1996–1997 were contacted and two maps for selected months and water column layers (Figure 1b–d) were able to use the original gear and the traditional fishing method. and linearly interpolated for the glass-eel sampling dates to serve as A special license was emitted by the Portuguese fisheries administra- environmental variables in the preliminary analysis (model compar- tion, permitting the operation of hand-nets targeting glass eels under ison for 1996–1997 data set, for which environmental variables the strict condition of simultaneous presence of IPMA observers. could be obtained from two distinct sources). The meteo-oceano- Experimental fishing was not performed during summer months and graphic data from remote-sensing, ERA interim and ARMOR3D, concentrated around the new moon. A sample of 200 glass eels was were spatially averaged for the study area delimited in Figure 1b transported live to the laboratory of IPMA for biometric analysis in (vertically averaged in the depth layers 0–50 and 50–300 m in the March 2013. In all other occasions, glass eels were weighed and case of the ARMOR3D variables) to assess seasonal variability (Fig- released back at sea by IPMA staff at some distance from the sam- ure 2), and linearly interpolated for all sampling dates to serve as pling location. All biometric and experimental fishing data described environmental variables in the final statistical analysis. above were used in this study, together with the commercial yield data from the 1989–1990 fishing season in Lis reported in Bessa (1992). 2.4 | Statistical analysis Generalized additive models (GAMs) were used to summarize the 2.3 | Environmental data seasonal pattern of biometric variables, to describe variation in glass eel yield at different temporal scales and to identify environmental Auxiliary environmental data from the river catchment area and variables that could account for part of the seasonality in the relative the sea conditions off Lis were made available for each day of abundance of glass eels recruiting to river Lis. GAMs can be seen as experimental fishing. Conditions at sea during fishing were evalu- generalized linear models in which part of the linear predictor is ated in situ, providing data on variables empirically known to specified as a sum of smooth functions of predictor variables and affect glass eel catch (mean wave height: numeric values with where the challenge is to find suitable parametric representations for 0.5 m reporting precision; coastal current direction: nominal values the smooth functions and to control the degree of smoothness of prevailing northerly or southerly current direction; coastal cur- appropriately (Wood & Augustin, 2002). The significance of each rent strength: ordinal values of weak, moderate or strong). Lunar partial effect in the model (i.e., the influence of an explanatory vari- phase was subsequently estimated from daily tide charts (days able to the response after removing the fitted effects of all other since new moon). Air temperature (daily minimum and maximum), explanatory variables) was evaluated in a backwards elimination pro- precipitation (daily, and mean daily integrated along the last week, cedure, considering the estimated degrees of freedom for the term fortnight and month) and atmospheric pressure were obtained (does it approximate 1?), the confidence region for the smooth (does from the meteorological station closest to the study area (Monte it include zero everywhere?) and the AIC score for the model (does Real military base). it go down when the term is removed?). Satellite, meteo-oceanographic data from observation analyses Mean length and condition of glass eels were modeled with a and numerical model solutions were used to describe the main cir- Gaussian distribution and an identity link function, using month as culation features in the Iberian margin and shelf, compute monthly the explanatory variable to depict seasonal patterns. Within-sample climatologies off river Lis and obtain environmental observations variation was ignored and interannual and spatial variation were con- for dates with glass eel experimental fishing. Fully normalised sidered as random. Biomass caught per sampling night was modeled 6 | with an over-dispersed Poisson distribution and a log-link function, STRATOUDAKIS ET AL. 3 | RESULTS using the logarithm of fishing effort (hours fishing per night) as an offset. A correction factor was added to the offset to account for During 1995–2001, 5,637 glass eels were measured live from sam- seasonal variation in body weight (i.e., dividing by a distinct mean ples collected off western Portugal (83% from the same hydro- weight estimate for autumn, winter and remaining months). Data graphic region in central Portugal and the remaining from from 1996 to 1997 were not available by fisher, so the response southwestern Portugal). Body height (mean = 2.79 mm; range 1.45– variable refers to the number of glass eels caught per hour fishing 4.1 mm) and width (mean = 2.19 mm; range 2–3 mm) had a strong per night, without considering between fisher variation. Fishing correlation with total length (mean = 67.91 mm; range 50–80 mm; hours per night were not available in the commercial data from Pearson Rho = 0.71 and 0.74, respectively; p < .001) and fresh 1989 to 1990 (data reported in Bessa, 1992), so the response vari- bodyweight (mean = 0.276 g; range 0.07–0.59 g; Rho = 0.90 and able for comparisons in CPUE across the three decades was catch 0.87, respectively; p < .001). However, the body height to width per fisher per night. ratio (index of body ellipticity; mean = 1.27; range 1.02–1.98) was A slightly different selection process was followed for the choice uncorrelated with length (Rho = 0.039; p = .079) but was highly cor- of environmental variables used to account for seasonal variation in related with weight and condition (Rho = 0.14 and 0.23, respec- CPUE. The correlation between all available environmental variables tively; p < .001). Despite interannual and within-region variation, was calculated and, for each group of highly correlated variables, length, weight and condition of glass eels recruiting in western Por- only the one with lowest correlation coefficients was considered in tugal showed a strong seasonal pattern, with longest, heaviest and in the model. Term selection was performed as previously described highest condition glass eels recruiting from October to December and model performance after the inclusion of each environmental (Figure 3, and Tables S2 and S3). variable was compared with the AIC for the model where seasonality During experimental fishing in the river Lis, 184.1 kg of glass eels was included as a smooth function of sampling week. Final models were caught over 289 nights and 808 operations (2,628 hr fishing were checked through standard inspection plots and summary by one to three fishers per night). In comparison, commercial fishing tables. in the 1989–1990 season yielded 595.1 kg of glass eels over 120 All data analysis and plotting was performed in R 3.3.1 (R Core nights and 1,518 operations (median 10.5 fishers per night, range 0– Team, 2016) and GAM fitting was performed using the library mgcv 39). Median duration of experimental fishing per night was 3 hr (version 1.8–12). Non-linear biometric relationships were explored (usually during rising tide, median starting hour at 22:00 hr), with no using the library nlme and glass eel condition was estimated as in significant difference between 1996–1997 and 2013–2014 (Wil- Iglesias et al. (2010). Non-parametric Wilcoxon tests were used to coxon test W = 6,175; p = .346). Sampling effort per night (total compare variables related to sampling effort between the two exper- hours fished) was significantly higher in the early period (W = 9,643; imental fishing periods and Pearson tests were used to explore cor- p < .001) due to the regular operation of three fishers. Total sam- relations among biometric variables. pling effort was considerably larger (80% of total) and more evenly F I G U R E 3 Seasonality in mean body length (left) and body condition (right) of glass eels recruiting to western Portugal (1994–2001, open circles), in comparison to published values from France (large closed circles: Atlantic—Guerault et al., 1992; Desaunay & Guerault, 1997; small closed circles: Mediterranean—Maes et al., 2009). Lines and grey shades result from GAM. Body condition for France was estimated on mean values (see also Tables S2 and S3), using the allometric relation estimated from Portuguese samples (3.21) STRATOUDAKIS | ET AL. 7 F I G U R E 4 Ingress dynamics of glass eel recruitment to river Lis from GAM fitted to experimental fishing CPUE from 1996 to 1997 and 2013 to 2014: partial effects depicting (a) seasonal (sampling week), (b) fortnightly (lunar day) and (c, d) diel (start sampling hour and wave height) levels of variation in CPUE. Two-standard error bands (grey shade) and partial residuals (points) are superposed in all panels. Sampling hour >24 refers to following day distributed within the year (4%–13% of effort across all months) in Finally, the GAM fitted to the CPUE from experimental fishing the 1996–1997 period, when commercial fishing was still authorized was reanalyzed substituting sampling week by environmental vari- from November to February. ables available for the dates of sampling within the study domain or Figure 4 depicts the partial effects of the GAM fitted to CPUE within the Lis catchment area (Table 1). Correlation was very high from experimental fishing considering explanatory variables that cor- among precipitation variables and air temperature (all reflecting the respond to different temporal scales (54% of deviance explained). Mediterranean seasonality in the Lis basin), while salinity at 50– There was a highly significant seasonal pattern, with highest yield 300 m was correlated with meteo-oceanogaphic variables of rele- around February-March and lowest in the summer (Figure 4a). vance to the IPC (water temperature, along-shore current and wind Hand-net catchability was influenced by lunar phase, with maxima a stress). The final model (44% of deviance explained) included three few days before and after new moon and minimum during full moon highly significant environmental variables (Figure 6). CPUE off river (Figure 4b). The negative phototaxis is also shown by the significant Lis was significantly higher when associated with relatively high effect of the starting hour of sampling, with lower yields in fishing water temperature in the deeper layer (Figure 6a), during high east- events that started too early in the evening (Figure 4c). Daylight ward (shoreward) wind stress (Figure 6b) and when alongshore cur- could not explain the higher yield in (the few) fishing events that rent in the Iberian margin was northward (Figure 6c). The high started after midnight. There was also a significant effect of sea correlation of the latter variable with water salinity is a good indica- state condition, with lower yield under higher wave action (Fig- tion of a significantly higher relative abundance of ingressing glass ure 4d). Finally, there was also evidence for significant interannual eels during stronger IPC. variation, but with no clear pattern between the two decades (see Figure 5d). Figure 5 extends the comparison of experimental fishing yields 4 | DISCUSSION with that of the commercial fishing period of 1989–1990 from November to February (Figure 5a). The resulting GAM (43% of Glass eels are known to enter river basins of the European Atlantic deviance explained) depicts some intraseasonal variation in daily coast throughout the year, but, even before the onset of manage- yield and the anticipated effect of lunar phase, but mainly shows a ment restrictions, fisheries only took place from October to March/ significant interannual effect (Figure 5c). After accounting for the April in central and northern Portugal (Domingos, 1992; Weber, effect of other temporal variables, modeled CPUE in 1989–1990 is 1986), northern Spain (Iglesias et al., 2010), and southern France (De significantly higher than in 1996–1997, which again is higher than in Casamajor, Prouzet, & Lazure, 2000). Modeling ingress dynamics of 2013–2014. However, the latter effect is not statistically significant glass eels in the river Lis also revealed strong seasonal patterns in due to larger variation in the recent data set. This variation is more abundance and biometry, in addition to effects related to other evident in the entire experimental fishing data set (Figure 5b), where scales of temporal variation (daily, fortnightly and interannual). Pat- few very high yields in early 2013 and 2014 turn the partial effect terns were synchronous for body size, weight and condition and out of sampling year not significant for 2013 and 2014 (Figure 5d). of phase with CPUE seasonality. Although glass eels were found to 8 | STRATOUDAKIS ET AL. F I G U R E 5 Decadal comparisons in glass eel relative abundance (CPUE with hand-net) recruiting to river Lis. (a) boxplot for catch per fisher per day in the November–February fishing season period of 1989–1990, 1996–1997 and 2013– 2014), (b) boxplot for catch per fisher per hour per day in experimental fishing sampling years 1996, 1997, 2013 and 2014 and (c, d) corresponding partial effect of year in GAMs fitted to glass eel CPUE. Width of boxes is proportional to the number of observations T A B L E 1 Pearson correlation coefficients between all pairs of environmental variables available at the Lis basin and surrounding coastal area during experimental fishing periods (absolute Rho values >0.4 are bold for better visualization) Environmental variable AIC (2) (1) Daily precipitation (mm) 1.449 0.41 (2) Weekly precipitation (mm) (3) Fortnigthly precipitation (mm) (3) (4) 1.414 (*) (11) (12) 0.35 0.42 0.22 0.18 0.17 0.16 0.00 0.01 0.36 0.49 0.44 0.26 0.28 0.03 0.16 0.23 0.53 0.36 0.48 0.29 0.33 0.30 0.18 0.27 0.68 0.08 0.13 0.22 0.40 0.29 0.35 0.26 (10) W-E current (m/s) (11) Water temperature (daily, °C) (10) 0.42 (7) S-N wind stress (N/m2) 1.348 (*) (9) 0.03 0.12 (9) S-N current (m/s) (8) 0.17 1.416 (8) W-E wind stress (N/m ) (7) 0.41 (5) Max air temperature (daily, °C) 2 (6) 0.76 (4) Min air temperature (daily, °C) (6) Atmospheric pressure (daily, hP) (5) 0.24 0.12 0.45 0.31 0.21 0.27 0.16 0.13 0.13 0.06 0.13 0.04 0.00 0.46 0.39 0.24 0.43 0.20 0.34 0.24 0.44 0.12 0.46 0.13 0.22 1.318 (*) 0.20 0.40 (12) Water salinity (daily) AIC is shown only for the variables chosen among those highly correlated to be considered in the model. AIC value is tabulated as a fraction of the AIC for the model that includes the environmental variable over the AIC of the model where seasonality is depicted by a smooth function of sampling week (asterisks indicate terms that fulfill the three conditions described in the methods for an explanatory variable to be considered significant). ingress into the river Lis throughout the year, they were longest and peaks. Peaks of abundance in the estuaries of Mondego and Minho heaviest from October to December and progressively reduced in rivers were observed from November to February, although in some size, weight and condition until March–April. Relative abundance years large densities could also be observed in March (Domingos, peaked in February–March, when recruits were almost 40% lighter 1992; Weber, 1986). The period of high ingress is possibly wider in than early arrivals, while both biometric and abundance indices were southern Iberia, where a secondary seasonal peak has been observed minimal during the summer months. in April and May (Arribas et al., 2012). Monthly catch data (totals Fishery-independent samples have confirmed glass eel presence and standardized by effort) from the river Minho across the fishing across the year in the estuaries of Guadalquivir (Arribas, Fernandez- season over the period 1990–2012 (C. Antunes, unpublished data), Delgado, Oliva-Paterna, & Drake, 2012), Mondego (Domingos, demonstrate that the peak month of catch and CPUE may not coin- 1992), and Minho (Weber, 1986) rivers, with minimal densities from cide within a fishing season and can both vary considerably (from June to September and some interannual variation in the seasonal November to March) among years. This suggests that although there STRATOUDAKIS | ET AL. 9 F I G U R E 6 Environmental variables contributing to the seasonal variation in the relative abundance of glass eels recruiting to river Lis (GAM fitted to all experimental CPUE using oceanographic and meteorological variables indicated in Table 1 and corresponding to the Lis box area at Figure 1b). (a) partial effect of seawater temperature in the IPC layer (50–300 m), (b) cross-shelf wind stress (positive values for shoreward direction) and (c) along-shelf current (positive values correspond to currents from southern origin—IPC). Two-standard error bands (gray shade) and partial residuals (points) are superposed. Tick marks at the bottom horizontal axis indicate the distribution of observations along the environmental variable range is a clear seasonality, with higher ingress in the winter period, the pronounced indicating a higher condition for the more southerly exact month of the peak in abundance may vary among years. samples (Iglesias et al., 2010). The glass eels caught off western Portugal showed peak biomet- These differences are in line with the latitudinal gradient ric values towards the end of the year and progressive reduction reported in the oceanic distribution of leptocephali in the eastern until the following summer, which are consistent with published data part of the north Atlantic (Bast & Strehlow, 1990; McCleave et al., from rivers Minho (Antunes, 1994; Iglesias et al., 2010; Santos & 1998), and during arrival to the European margin (Antunes & Tesch,  n (Iglesias et al., 2010), Adour (De Casamajor Weber, 1992), Nalo 1997a). Hence, recruits to rivers in northern Africa, southern Spain, et al., 2000; Guerault et al., 1992), and Villaine (Desaunay & Guer- southern and western Portugal and those recruiting into the ault, 1997; Guerault et al., 1992) in the European Atlantic coast, but Mediterranean Sea may all come from a pool of smaller leptocephali also in the Mediterranean coast of France (Maes, van Vo, Crivelli, & at lower latitudes with distinct dispersal dynamics from the larger Volckaert, 2009). Despite interannual and ingress distance variation counterparts recruiting further north following the North Atlantic in sampling (which in the Atlantic coast of France is accentuated by currents (Tesch, 2003; Tesch & Niermann, 1992). Given the circula- a steep increase in continental shelf width with increasing latitude), tion patterns in the north Atlantic and the Iberian margin, it is plausi- these data also suggest some latitudinal cline in mean size at a given ble that eel larvae arrive to western Portugal from the south, month (see also Figure 3, and Tables S2 and S3), with smaller values following an oceanic trajectory along the Azores Current (Klein & for glass eels recruiting further south and in the Mediterranean Sea. Siedler, 1989; Miller et al., 2015) and deflecting northwards by This is more evident when cohorts can be compared during simulta- entrainment to the IPC in the Gulf of Cadiz (Peliz et al., 2005; Teles- neous arrival from the sea, as between concurrent samples in the Machado et al., 2016). Some of these smaller larvae may even reach  n rivers in 2003–2005 (Iglesias et al., estuaries of Minho and Nalo the Bay of Biscay, especially during December–January when the 2010), or between the rivers Lis (this study) and Adour (De Casama- IPC is frequently detected in the Cantabrian Sea (“Navidad” current), jor et al., 2000) in November–December 1997: in both cases, mean and further north (Pingree & Le Cann, 1990). sizes in the same month were 2–5 mm longer in the Bay of Biscay The present study cannot distinguish between alternative mecha- than off western Portugal, while differences in weight were less nisms leading to the smaller size of the southern leptocephali, but 10 | STRATOUDAKIS ET AL. allows an ecological interpretation of the seasonal dynamics of 1989–1990 (5 years after the introduction of restrictive manage- ingress in western Portugal. The distribution of leptocephali in the ment measures in the fishery due to signs of population decline in Iberian margin (Antunes & Tesch, 1997a; Tesch, 1980) coincides with Portugal—Domingos, 1992; Weber, 1986), than during the 1996– the bathymetric and geographical distribution of the IPC off western 1997 experimental fishing period. Nevertheless, the decline from and southern Iberia (see Figure 1b and Teles-Machado et al., 2016). 1989 onwards seems lower to that reported across the stock area Minima of ingress during the summer can result either by lower den- (ICES, 2016), or in comparison to basins further north where dra- sities of leptocephali arriving from the oceanic migration or by lower matic declines were observed (Golloc, Curnick, & Debney, 2011; probability of northward deflection into the IPC and subsequent Henderson, Plenty, Newton, & Bird, 2012). The lack of significant detrainment and cross-shelf transport. The former would require a differences between the two experimental fishing periods (1996– strong seasonal signal of oceanic arrivals into the Gulf of Cadiz and 1997 versus 2013–2014) may be due to lower statistical power the Strait of Gibraltar resulting from the seasonality of spawning in associated with a smaller sample size in recent years (20% of effort the Sargasso Sea (Miller et al., 2015), attenuated, but not eliminated, in latter period), or to an inherent increase of ingress variability with by oceanic drift; the latter would require a seasonally-varying trigger lowering abundance (largest CPUE value in 2014, year with the low- of environmental origin in the Iberian margin (Arribas et al., 2012; est median value—Figure 5b), or both. Alternatively, this might be Miller, 2009). Given that the IPC is weakest during summer months, due to the “early arrivals effect” that has been reported off western when it is also bounded below the thermocline and river plumes off Scotland (Adams, Godfrey, Dodd, & Maitland, 2013), suggesting that western and southern Iberia are minimal (Fernandez-Novoa et al., oceanic arrivals are still sufficient to guarantee recruitment to the 2017), and that an oceanic drift of more than a year can eliminate continental habitats that are first to be encountered in Europe. This any seasonal pattern in arrivals to Europe (Blanke et al., 2012), the could be further explored by decadal comparisons in other geograph- latter alternative seems more plausible. ical areas of likely early arrival (e.g., Iceland, Azores, Madeira, etc.), With the break of stratification in October and the penetration together with the verification of latitudinal cline in biometry.  (Telesof the IPC to the outer shelf north to the Canyon of Nazare Machado et al., 2016), the cross-shelf distance to river Lis becomes smaller, permitting the ingress of younger metamorphic stages with ACKNOWLEDGEMENTS higher body condition. In the early months of the year, the IPC Experimental fishing in the river Lis during 2013–2014 was part of a moves offshore, increasing the cross-shelf distance to travel. Never- pilot project of the National Program for Biological Sampling (PNAB/ theless, the surface signature of IPC, the mixing of the water column Data Collection Framework) in Portugal. ATM was funded by the and the extension of terrestrial signatures further offshore with the ^ncia e a Tecnologia (FCT) through grant SFRH/ Fundacß~ao para a Cie accumulation of a turbid plume through river run-off in the early BPD/100720/2014. We thank the fishers that participated in both months of the year, may maximize the terrestrial signal and ingress study periods, Manuela Azevedo (IPMA, PNAB coordinator) for sup- of glass eels in more advanced metamorphic stages, and with lower port in the planning and execution of the 2013–2014 pilot project, condition. Interannual variation in the hydrological cycle, the inten- Alvaro Gueif~ao (IPMA) for assistance in the field work and the mete- sity and cross-shelf position of the IPC may modify the month of orology department of IPMA for making available the data of the peak ingress without substantially changing the biometric seasonal- Monte Real base. This study has been conducted using E.U. Coperni- ity. cus Marine Service Information. The GlobColour data (http://globc Plankton observations within the Azores Current or near the olour.info) used here have been developed, validated, and distributed Strait of Gibraltar along the year could clarify the above alternatives. by ACRI-ST, France. Finally, we thank the two referees for criticisms Additionally, backward simulations (Pacariz et al., 2014), possibly and suggestions that helped to improve the original manuscript. coupled to individual-based models reversing the process of metamorphosis, and eventually extending back to the spawning area could test our hypothesis of southerly origin for eels recruiting in Portugal. For that, behavioural (e.g., negative phototaxis and rheo- ORCID Yorgos Stratoudakis http://orcid.org/0000-0002-7801-4336 taxis) and oceanographic processes at the inner shelf and near the bottom (e.g., the influence of tides: Marta-Almeida & Dubert, 2006; or the littoral drift: Silva et al., 2012) create challenges that are not faced in Lagrangian models that remain within the oceanic domain. Such modeling would also benefit from additional research attention into the trigger of metamorphosis and the possible mechanisms of recognition of freshwater turbid plumes for the cross-shelf migration of glass eels (Arribas et al., 2012). Finally, the analysis suggests some reduction in mean eel recruitment at the river Lis over the past three decades and, possibly, some increase in its variance. Mean daily yield was significantly higher in REFERENCES Adams, C. E., Godfrey, J. D., Dodd, J. A., & Maitland, P. S. (2013). Is proximity to the North Atlantic drift and the continental shelf current sustaining freshwater European eel populations in western Scotland? Freshwater Biology, 58, 1–9. https://doi.org/10.1111/fwb.12021 ~o e metamorfose de Anguilla anguila Antunes, C. (1994). Estudo da migracßa L. por analise dos incrementos de sagittae, em leptocefalos e enguias de vidro. PhD Thesis, ICBAS, University of Porto. 294 pp. Antunes, C., & Tesch, F.-W. (1997a). Eel larvae (Anguilla anguilla L.) caught by R.V. “Heincke” at the European continent slope in autumn STRATOUDAKIS ET AL. 1991. Ecology of Freshwater Fish, 6, 50–52. https://doi.org/10.1111/j. 1600-0633.1997.tb00142.x Antunes, C., & Tesch, F.-W. (1997b). A critical consideration of the metamorphosis zone when identifying daily rings in otoliths of European eel, Anguilla anguilla (L.). Ecology of Freshwater Fish, 6, 102–107. https://doi.org/10.1111/j.1600-0633.1997.tb00150.x andez-Delgado, C., Oliva-Paterna, F. J., & Drake, P. Arribas, C., Fern (2012). Oceanic and local environmental conditions as forcing mechanisms of the glass eel recruitment to the southernmost European estuary. Estuarine, Coastal and Shelf Science, 107, 46–57. https://doi. org/10.1016/j.ecss.2012.04.024 Bast, H.-D., & Strehlow, B. (1990). Length composition and abundance of eel larvae, Anguilla anguilla (Anguilliformes: Anguillidae), in the Iberian Basin (northeastern Atlantic) during July-September 1984. Hel€nder Meeresuntersuchungen, 44, 353–361. https://doi.org/10. gola 1007/BF02365472 Bessa, R. (1992). Apanha de meix~ao com “sarrico” na safra de 1989/ rios Cientıficos e Tecnicos INIP 57. 1990 mo rio Liz. Relato Bessa, R., & Castro, M. (1994). Evolucß~ao das capturas de meix~ao ao longo ~ es ambientais. do ano no rio Liz e a sua relacß~ao com as condicßo rios Cientıficos e Tecnicos INIP 2. Relato triBessa, R., & Castro, M. (1995). Variacß~ao sazonal dos par^ametros biome cos de meix~ao capturado no rio Liz. Relatorios Cientıficos e Tecnicos INIP 4. Blanke, B., Bonhommeau, S., Grima, N., & Drillet, Y. (2012). Sensitivity of advective transfer times across the North Atlantic Ocean to the temporal and spatial resolution of model velocity data: Implications for European eel larval transport. Dynamics of Atmospheres and Oceans, 55–56, 22–44. https://doi.org/10.1016/j.dynatmoce.2012.04.003 Bonhommeau, S., Blanke, B., Treguier, A. M., Grima, N., Rivot, E., Vermard, Y., . . . Le Pape, O. (2009). How fast can the European eel (Anguilla anguilla) larvae cross the Atlantic Ocean? Fisheries Ocenanography, 18, 371–385. https://doi.org/10.1111/j.1365-2419.2007. 00453.x , R., & Le Pape, O. Bonhommeau, S., Castonguay, M., Rivot, E., Sabatie (2010). The duration of migration of Atlantic Anguilla larvae. Fish and Fisheries, 11, 289–306. https://doi.org/10.1111/j.1467.2979.2010. 00362.x De Casamajor, M. J., Prouzet, P., & Lazure, P. (2000). Identification des flux de civelles (Anguilla anguilla) a partir des relations d’allometrie em foncion des conditions hydrodynamiques de l’estuaire de l’Adour. Aquatic Living Resources, 13, 411–420. Dee, D. P., Uppala, S. M., Simmons, A. J., Berrisford, P., Poli, P., Kobayashi, S., . . . Vitart, F. (2011). The ERA-Interim reanalysis: Configuration and performance of the data assimilation system. Quarterly Journal of the Royal Meteorological Society, 137, 553–597. https://doi.org/10. 1002/qj.828 Desaunay, Y., & Guerault, D. (1997). Seasonal and long-term changes in biometrics of eel larvae: A possible relationship between recruitment variation and North Atlantic ecosystem productivity. Journal of Fish Biology, 51(A), 317–339. https://doi.org/10.1111/j.1095-8649.1997. tb06106.x Domingos, I. (1992). Fluctuations of glass eel migration on the Mondego estuary (Portugal) in 1988 and 1989. Irish Fisheries Investigations Series A (Freshwater), 36, 1–4. Fern andez-Novoa, D., deCastro, M., Des, M., Costoya, X., Mendes, R., & G’mez-Gesteira, M. (2017). Characterization of Iberian turbid plumes by means of synoptic patterns obtained through MODIS imagery. Journal of Sea Research, 126, 12–25. https://doi.org/10.1016/j.seares. 2017.06.013 za, A., Ambar, I., & Boyd, T. (1990). Observations of a poleFrouin, R., Fiu ward surface current off the coasts of Portugal and Spain during winter. Journal of Geophysical Research, 95, 679–691. https://doi.org/10. 1029/JC095iC01p00679 | 11 Golloc, M., Curnick, D., & Debney, A. (2011). Recent recruitment trends of juvenile eels in tributaries of the River Thames. Hydrobiologia, 672, 33–37. https://doi.org/10.1007/s10750-011-0754-9 Guerault, D., Lecomte-Finiger, R., Desaunay, Y., Biagianti-Risbourg, S., Bellois, P., & Grellier, P. (1992). Glass eel arrivals in the Villaine estuary (northern Bay of Biscay) in 1990: Demographic features and early life history. Irish Fisheries Investigations Series A (Freshwater), 36, 5–14. Henderson, P. A., Plenty, S. J., Newton, L. C., & Bird, D. J. (2012). Evidence for a population collapse of European eel (Anguilla anguilla) in the Bristol Channel. Journal of the Marine Biological Association of the United Kingdom, 92, 843–851. https://doi.org/10.1017/ S002531541100124X ICES (2016). Report of the Working Group on Eels (WGEEL), 15–22 September 2016, Cordoba, Spain. ICES CM 2016/ACOM:19. 107 pp. Iglesias, T., Lobon-Cervia, J., Costa-Dias, S., & Antunes, C. (2010). Variation of life traits of glass eels of Anguilla anguilla (L.) during the colo n and Minho estuaries. Hydrobiologia, 651, 213– nization of Rios Nalo 223. https://doi.org/10.1007/s10750-010-0300-1 Kettle, A. J., & Heines, K. (2006). How does the European eel (Anguilla anguilla) retain its population structure during its larval migration across the North Atlantic Ocean? Canadian Journal of Fisheries and Aquatic Science, 63, 90–106. https://doi.org/10.1139/FO5-198 Kleckner, R. C., & McCleave, J. D. (1988). The northern limit of spawning by Atlantic eels (Anguilla spp.) in the Sargasso Sea in relation to thermal fronts and surface water masses. Journal of Marine Research, 46, 647–667. https://doi.org/10.1357/002224088785113469 Klein, B., & Siedler, G. (1989). On the origin of the Azores Current. Journal of Geophysical Research, 94(C5), 6159–6168. https://doi.org/10. 1029/JC094iC05p06159 Maes, G. E., van Vo, B., Crivelli, A. J., & Volckaert, F. A. M. (2009). Morphological and genetic seasonal dynamics of European eel Anguilla anguilla recruitment in south France. Journal of Fish Biology, 74, 2047–2068. https://doi.org/10.1111/j.1095-8649.2009.02294.x Maritorena, S., Fanton d’Andon, H. O., Mangin, A., & Siegel, D. A. (2010). Merged satellite ocean color data products using a bio-optical model: Characteristics, benefits and issues. Remote Sensing of the Environment, 114, 1791–1804. https://doi.org/10.1016/j.rse.2010.004.02 Marta-Almeida, M., & Dubert, J. (2006). The structure of tides in the western Iberia region. Continental Shelf Research, 26, 385–400. https://doi.org/10.1016/j.csr.2005.11.011 McCleave, J. D., Brickley, P. J., O’Brien, K. M., Kistner, D. A., Wong, M. W., Gallacher, M., & Watson, S. M. (1998). Do leptocephali of the European eel swim to reach continental waters? Status of the question. Journal of the Marine Biological Association of the United Kingdom, 78, 285–306. https://doi.org/10.1017/S0025315400040091 Miller, M. J. (2009). Ecology of anguilliform leptocephali: Remarkable transparent fish larvae of the ocean surface layer. Aqua-BioScience Monographs, 2, 1–94. Miller, M. J., Bonhommeau, S., Munk, P., Castonguay, M., Hanel, R., & McCleave, J. D. (2015). A century of research on the larval distributions of the Atlantic eels: A re-examination of the data. Biological Reviews, 90, 1035–1064. https://doi.org/10.1111/brv.12144 Miller, M. J., & Tsukamoto, K. (2017). The ecology of oceanic dispersal and survival of anguillid leptocephali. Canadian Journal of Fisheries and Aquatic Science, 74, 958–971. https://doi.org/10.1139/cjfas2016-0281 Mulet, S., Rio, M.-H., Mignot, A., Guinehut, S., & Morrow, R. (2012). A new estimate of the global 3D geostrophic ocean circulation based on satellite data and in-situ measurements. Deep Sea Research Part II: Topical Studies in Oceanography, 77–80, 70–81. https://doi.org/10. 1016/j.dsr2.2012.04.012 Munk, P., Hansen, M. M., Maes, G. E., Nielsen, T. G., Gastonguay, M., Riemann, L., . . . Bachler, M. (2010). Oceanic fronts in the Sargasso Sea control the early life and drift of Atlantic eels. Proceedings of the 12 | Royal Society of London B: Biological Sciences, 277, 3593–3599. https://doi.org/10.1098/rspb.2010.0900 Pacariz, S., Westerberg, H., & Bjork, G. (2014). Climate change and passive transport of the European eel larvae. Ecology of Freshwater Fish, 23, 86–94. https://doi.org/10.1111/eff.12048 Peliz, A., Dubert, J., Santos, A., Oliveira, P., & Le Cann, B. (2005). Winter upper ocean circulation in the Western Iberian Basin – Fronts, eddies and poleward flows: An overview. Deep Sea Research Part I: Oceanographic Research Papers, 52, 621–646. https://doi.org/10.1016/j.dsr. 2004.11.005 Pingree, R., & Le Cann, B. (1990). Structure, strength and seasonality of the slope currents in the Bay of Biscay region. Journal of the Marine Biological Association of the UK, 70, 857–885. https://doi.org/10. 1017/S0025315400059117 R Core Team (2016). R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. Retrieved from https://www.R-project.org/ Righton, D., Westerberg, H., Feunteun, E., Okland, F., Gargan, P., Amilhat, E., . . . Aarestrup, K. (2016). Empirical observations of the spawning migration of European eels: The long and dangerous road to the Sargasso Sea. Science Advances, 2, e1501694. https://doi.org/10.1126/ sciadv.1501694 Santos, A. M., & Weber, M. (1992). Growth studies on monthly captured glass eels from the river Minho in two recirculation systems. Irish Fisheries Investigations Series A (Freshwater), 36, 35–39. Silva, A. N., Taborda, R., Bertin, X., & Dodet, G. (2012). Seasonal to decadal variability of longshore sand transport at the northwest coast of Portugal. Journal of Waterway, Port, Coastal, and Ocean Engineering, 138, 464–472. https://doi.org/10.1061/(ASCE)WW.1943-5460. 0000152 Teles-Machado, A., Peliz, A., McWilliams, J. C., Couvelard, X., & Ambar, I. (2016). Circulation on the Northwestern Iberian Margin: Vertical structure and seasonality of the alongshore flows. Progress in Oceanography, 140, 134–153. https://doi.org/10.1016/j.pocean.2015. 05.021 Tesch, F.-W. (1980). Occurrence of eel Anguilla anguilla larvae west of the European continental shelf. Environmental Biology of Fishes, 5, 185–190. https://doi.org/10.1007/BF00005354 Tesch, F.-W. (2003). The eel: Biology and management of anguillid eels (5th ed.). Oxford, UK: Blackwell Science. STRATOUDAKIS ET AL. Tesch, F.-W., & Niermann, U. (1992). Stock density of eel larvae Anguilla anguilla (L.) on the European continental slope, based on collections made between 1985 and 1989. Irish Fisheries Investigations Series A (Freshwater), 36, 110–113. Tsukamoto, K., Yamada, Y., Okamura, A., Kaneko, T., Tanaka, H., Miller, M. J., . . . Tanaka, S. (2009). Positive buoyancy in eel leptocephali: An adaptation for life in the surface ocean layer. Marine Biology, 156, 835–846. https://doi.org/10.1007/s00227-008-1123-8 Weber, M. (1986). Fishing method and seasonal occurrence of glasseels (Anguilla anguilla, L.) in the Rio Minho, west coast of the Iberian Peninsula. Vie et Milieu, 36, 243–250. Wood, S. N., & Augustin, N. (2002). GAMs with integrated model selection using penalized regression splines and applications to environmental modelling. Ecological Modelling, 157, 157–177. https://doi.org/ 10.1016/S0304-3800(02)00193-X Zenimoto, K., Sasai, Y., Sasaki, H., & Kimura, S. (2011). Estimation of larval duration in Anguilla spp., based on cohort analysis, otolith microstructure and Lagrangian simulations. Marine Ecology Progress Series, 438, 219–228. https://doi.org/10.33454/meps09255 SUPPORTING INFORMATION Additional supporting information may be found online in the Supporting Information section at the end of the article. How to cite this article: Stratoudakis Y, Oliveira PB, TelesMachado A, Oliveira JM, Correia MJ, Antunes C. Glass eel (Anguilla anguilla) recruitment to the river Lis: Ingress dynamics in relation to oceanographic processes in the western Iberian margin and shelf. Fish Oceanogr. 2018;00: 1–12. https://doi.org/10.1111/fog.12274