Route Choices, Migration Speeds and Daily Migration Activity of European Silver Eels Anguilla Anguilla in the River Rhine, North-West Europe

Journal of Fish Biology (2009) 74, 2139–2157 doi:10.1111/j.1095-8649.2009.02293.x, available online at www.interscience.wiley.com

Route choices, migration speeds and daily migration activity of European silver eels Anguilla anguilla in the River Rhine, north-west Europe

A. W. Breukelaar*†, D. Ingendahl‡, F. T. Vriese§, G. de Laak¶, S. Staas** and J. G. P. Klein Breteler†† *Rijkswaterstaat Waterdienst, P. O. Box 17, 8200 AA Lelystad, The Netherlands, ‡Bezirksregierung Arnsberg, Heinsbergerstrasse 53, 57399 Kirchhundem-Albaum, Germany, §Visadvies bv, Twentehaven 5, 3433 PT Nieuwegein, The Netherlands, ¶Sportvisserij Nederland, Postbus 162, 3720 AD Bilthoven, The Netherlands, **Rheinfischereigenossenschaft im Lande Nordrhein-Westfalen, Römerhofweg 12, 50374 Erftstadt, Germany and ††Vivion bv, Händelstraat 18, 3533 GK Utrecht, The Netherlands

Downstream migration of Anguilla anguilla silver eels was studied in the Lower Rhine, Germany, and the Rhine Delta, The Netherlands, in 2004–2006. Fish (n = 457) released near Cologne with implanted transponders were tracked by remote telemetry at 12 fixed detection locations distributed along the different possible migration routes to the North Sea. Relatively more A. anguilla migrated via the Waal than the Nederrijn, as would be expected from the ratio of river discharges at the bifurcation point at Pannerden. Downstream migration from the release site to Rhine-Xanten, close to the German–Dutch border, generally occurred in the autumn of the year of release but migration speeds tended to be low and variable and unaffected by maturation status or river discharge rates. Detection frequencies were not significantly related to discharge peaks or lunar cycles, but there was a minor detection peak 1–6 h after sunset. Between 2004 and 2009, 43% of the 457 A. anguilla released were never detected and of the 260 detected entering the Netherlands, 83 (32%) were detected escaping to the sea, 78 (94%) via the Nieuwe Waterweg and three (4%) and two (2%) via the sluices in the Haringvlietdam and Afsluitdijk, respectively. Possible causes of non-detections are discussed and it is suggested that many A. anguilla temporarily ceased migration, but that fishing mortality could have been important during passage through the Dutch parts of the Rhine. Practical implications of the results for predicting emigration routes, timings and magnitudes and use in management initiatives to promote escapement of A. anguilla silver eels to the sea are critically discussed. © 2009 The Authors Journal compilation © 2009 The Fisheries Society of the British Isles Key words: escapement; lunar cycle; maturation status; river discharge; telemetry.

INTRODUCTION The European eel Anguilla anguilla (L.) stock has shown a strong decline and recruitment is as low as 1% of historic levels (Dekker, 2004; ICES, 2004). Concerns

†Author to whom correspondence should be addressed. Tel.: +31 653 776 397; fax: +31 320 249 218; email: [email protected] 2139 © 2009 The Authors Journal compilation © 2009 The Fisheries Society of the British Isles 2140 A. W. BREUKELAAR ET AL. about declines in recruitment and the status of stocks of A. anguilla have led to the European Union (EU) adopting a regulation establishing measures for the recovery of the stock (EU, 2007). The member states have to make eel management plans (EMP) in 2008 for each river basin district defined according to the Water Framework Directive (WFD). They have to specify the measures they will take to comply with a 40% escapement target of adult seaward migrating (‘silver’) A. anguilla biomass under undisturbed conditions and to make a time schedule for the attainment of that target level. Detailed data on a river basin level are usually missing in most European countries, specifically on target and current levels of escapement of silver A. anguilla to the sea and on factors responsible for the losses of emigrants. Known anthropogenic factors directly influencing A. anguilla survival during seaward migration are fisheries, and hydropower and pumping stations (Dekker, 2004; ICES, 2004). Acou et al. (2008) also suggest indirect mortalities can occur in A. anguilla with delayed migration because of barriers such as weirs and sluices. The efficiency of measures against these anthropogenic factors partly depends on the ability to predict downstream migration of the fish, both with regard to timing and migration routes chosen by the migrating A. anguilla. It also depends on migration speeds of the fish, because longer residence times underway increases risks of natural mortality or being captured by fisheries. Timing of downstream migration of A. anguilla depends on both internal and environmental variables. Reported environmental variables in timing of migration include seasonal factors (month, water temperature in current or preceding months and photoperiod), diurnal factors, weather influences (precipitation, rainfall, wind, microseismic changes, atmospheric depressions and sunshine hours), hydrographic and hydrological conditions (water level, flow rate, river discharge and turbidity) and lunar phase (Vøllestad et al., 1986; Cairns & Hooley, 2003; Haro, 2003; Tesch, 2003; ICES, 2005; Durif & Elie, 2008). Internal metamorphic changes towards ‘silvering’ of A. anguilla include ones in endocrine profiles (Van Ginneken et al., 2007a) and morphological and metabolic variables (Van Ginneken et al., 2007b)andare associated with changes in external morphological characteristics such as the silvery colour on the belly, pectoral fin length (LPF), total length (LT), mass (M)andeye diameter (DE)(Durifet al., 2005). Haro (2003) proposes a damped time distribution of downstream migration in the mainstem of large rivers due to superposition of different migrational peak events in the tributaries at different distances from the sea. In rivers with only one mainstem, there is only one route to the sea, but in braided rivers with networks of river branches and multiple discharge points to the sea, A. anguilla may choose between different escapement routes. This is the situation in the lower River Rhine. Recent studies in such river systems (Winter et al., 2006; Klein Breteler et al., 2007) suggest that the fish just ‘go with the flow’ because silver A. anguilla, except from tidal areas where they actively use the ebb current, seem to migrate downstream in the middle depths of rivers and in the main current (Tesch, 2003). Downstream migration of A. anguilla may be inhibited at bifurcation points in rivers due to locally operating sensory cues (Haro, 2003) such as ambient light, visual obstacles, noise or tactile stimuli from racks at hydropower stations (Hadderingh et al., 1992; Adam et al., 1997). The present study formed part of a project with the ultimate goal of determining key factors that could be used to optimize escapement of silver A. anguilla from the

River Rhine to the sea to comply with the 40% escapement target (from the pristine level) set by the EU (2007). Telemetry results of the 2004 and 2005 cohorts of female silver A. anguilla released in Cologne (Germany) showed that most of the ﬁsh chose the route via the Waal and Nieuwe Waterweg and that 23 and 15% escaped to the sea, respectively (Klein Breteler et al., 2007). The present study focused on telemetry of these cohorts, and additionally of the 2006 cohort, to determine migration timings, choice of routes and migration speeds of individual ﬁsh in relation to river discharges and lunar cycles, factors thought to be relevant for A. anguilla management, including appropriate water and hydropower management.

MATERIALS AND METHODS

STUDY AREA The study was carried out in the lower River Rhine (Fig. 1). A detailed description of the lower study area is given in Klein Breteler et al. (2007). Names and numbers of the detection stations are given in Fig. 1. In normal years, the yearly mean discharge of the River Rhine at Lobith varied between 1500 and 3000 m3 s−1 during the last century (www.waterstat.nl). Peak discharges mainly occurred in winter and spring during the study period, but some high river discharges also occurred in summer and autumn (Fig. 2). During these periods of high river discharge, the sluices are opened in the Nederrijn and Haringvlietdam (site 7, 7 , Fig. 1). This affects the distribution of the discharged water of the River Rhine over its three main discharge routes the Waal, the Nederrijn-Lek and the IJssel, and is reﬂected in the ratio of river discharges between the Waal and the Pannerdens Kanaal at Pannerden and between the IJssel and Nederrijn at IJsselkop (Fig. 2).

TELEMETRY SYSTEM The telemetry system (NEDAP Trail System; http://www.nedaptrail.com/) used detection stations with antenna cables laid across the river summerbed at 5–22 m depth on average and is described by Bij de Vaate et al. (2003). Transponders implanted in fish ‘fired’ an identification number when activated by the antenna of a detection station and had a battery life of 1·5–2·0 years (Breukelaar et al., 1998). Each individual fish was traced on its migration route to the sea or to the detection station where it was detected for the last time. The data from the detection stations were screened for passing fish from this study up to 1 January 2009, for longer than the possible lifetime of the transponder batteries. Data records obtained from the system and used in this study contained detection station, transponder number and date and time of detection. When the River Rhine discharged >4735 m3 s−1, as measured at Lobith, the ends of the cables at the Rhine-Xanten 1 station were flooded and undetected passages were probably possible. The same applied to the Waal-Vuren station 2 at river discharges >3850–5000 m3 s−1 (depending on tidal movement). Such situations rarely occurred but potentially affected the results.

SURGICAL PROCEDURES AND IMPLANTATIONS Anguilla anguilla were caught in the River Moselle and in the River Sieg (Fig. 1), henceforth denoted as ‘Moselle’ and ‘Sieg’ A. anguilla, respectively. Records of fishing methods are incomplete, but most fish were fyke-netted in the Moselle or electrofished in the Sieg. The fish were delivered in batches well spread through August to December in the years 2004 to 2006. Individuals were selected as being apparently silver from external characteristics. Anguilla anguilla of 640–1100 mm LT were chosen, as surgical incisions

12 Afsluitdijk

IJsselmeer 6 Detection stations

Weir 11 Release locations

North Sea IJssel

10 Nieuwe Waterweg Nederrijn 9 IJsselkop 4 8 Lek Pennerdens Kanaal 2 Pannerden 7 3 6 Waal Lobith Rhine 5 1 Haringvliet Germany Meuse

Auter Belgium Rhine Roer Sieg

Cologne

Moselle

0 10 20 30 nm

FIG. 1. Study area with release locations (1, Rhine-Xanten; 2, Waal-Vuren; 3, Beneden Merwede; 4, Oude Maas; 5, Dordtsche Kil; 6, Spui; 7, Haringvlietdam; 8, Noord; 9, Nederrijn; 10, Lek; 11, IJssel-Kampen; 12, Afsluitdijk) of Anguilla anguilla ( ) ‘Cologne’ and ‘Sieg’ detection stations ( 1 , etc.) and sluices ( ). could not be closed completely in smaller fish. Transponders were surgically implanted in each fish (457 in total, 150–157 per year), according to Klein Breteler et al. (2007). After surgery, each fish was weighed (M,g),LT measured (mm) and LPF and DE in horizontal and vertical directions were measured by callipers (0·1 mm). After recovery of normal swimming behaviour, fish were released, this occurred in seven batches in the Rhine at Cologne in 2004, four batches at Cologne and three in the Sieg near its confluence with the Rhine in 2005 and four at Cologne and two in the Sieg in 2006. The distance from Cologne to the sea depends on the migration routes chosen in the Rhine Delta

30 (a) 7000

25 6000 ) –1 s 3 5000 20 C) °

4000 15

3000 Temperature ( 10 2000 River discharge at Lobith (m 5 1000

0 0

12 (b)

2 River discharge ratio at bifurcation points

0 01 Aug 04 01 Feb 05 01 Aug 05 01 Feb 06 01 Aug 06

FIG. 2. (a) Daily mean river discharge ( ) and temperature ( ) of the River Rhine at Lobith (Dutch–German border) from August 2004 to January 2007 and (b) river discharge ratios (the ratios of river discharges in the two river sections downstream of a bifurcation point) at the bifurcation points Pannerden ( )and IJsselkop ( ) during the study.

and ranges from 360 km to the Haringvlietdam 7 to 400 km to the Afsluitdijk 12 (Fig. 1). The distance from the Sieg to the sea is c. 30 km longer than from Cologne to the sea.

DATA ANALYSIS

Values of M,LT,DE and LPF were used for calculation of the maturation stages SF-II, SF-III, SF-IV or SF-V of the female ﬁsh according to the criteria of Durif et al. (2005). These maturation classes also represent the physiological status of the ﬁsh and SF-IV and

30 Frequency (%)

0 SF-II SF-III SF-IV SF-V Maturation index

FIG. 3. Relative frequency of maturation stage of Anguilla anguilla used for implantation of transponders in the years 2004 ( ), 2005 ( ) and 2006 ( )(n = 427). Classes are according to Durif et al. (2005).

SF-V are considered to be real migrants (Durif et al., 2005). Most fish belonged to the SF-III and SF-IV class and in 2004 relatively more fish fell into the SF-III class (Fig. 3), few fish were classified as stage SF-V. The number of escapees to the sea at the Afsluitdijk (12) and the Haringvlietdam 7 was determined at the sluices where they entered the sea (Fig. 1). It is physically impossible to locate a detection station near the mouth of the Nieuwe Waterweg, so the combined detections of the three most downstream stations were used instead [Noord 8 , Oude Maas 4 and Lek (10)] to estimate escapement to the sea via the Nieuwe Waterweg (Fig. 1). No corrections were made to estimates of escapement for fishery mortalities downstream from these detection stations because catches and effort of these fisheries are unknown. Detection rates during migration were calculated from detections and from obvious missing detections, e.g. each A. anguilla that passed Waal-Vuren 2 must have passed Rhine-Xanten 1 , even if not detected at Rhine-Xanten 1 . The percentage missing detections at Rhine- Xanten 1 (Fig. 1) was determined from the number of fish detected in more downstream detection stations compared with those missed at Rhine-Xanten 1 . The percentage missing detections of the whole telemetry infrastructure in the Netherlands (system performance) was determined from numbers of fish detected at the combined stations furthest downstream before entering the sea [Haringvlietdam 7 , Oude Maas 4 , Noord 8 ,Lek 10 , and Afsluitdijk 12 ] compared with the number not detected upstream in the Netherlands. Migration routes of individual fish were determined from the sequence of detections and, in cases of no further progress to the sea but repeated detections at some stations without a clear route choice (nine fish), by the last detection station encountered. River discharge and temperature (daily mean) data were obtained from the Dutch Ministry of Transport, Public Works and Water Management (www.waterstat.nl). Daily mean discharges of the River Rhine at Lobith and the Waal at Pannerden were used to calculate the daily discharge ratio ‘Pannerden’, i.e. the ratio between the discharges of the Waal and the Pannerdens Kanaal. Daily mean discharges of the IJssel at IJsselkop and of the Nederrijn at Driel were used for the calculation of the ratios between these discharges (discharge ratio ‘IJsselkop’). In a similar way, A. anguilla ratios were calculated at Pannerden and IJsselkop, representing the ratios of numbers of fish migrating via these bifurcation points.

For each individual ﬁsh that passed between two detection stations, migration speeds were calculated by subtraction of year, day and time at both stations and division by the distance between the stations. For the same periods, the average river discharges of the Rhine at Lobith were calculated from the daily mean river discharges. Migration activity after sunset (Tesch, 2003) was analysed by subtracting the individual detection time of each ﬁsh from the time of last sunset (on the same day or the day before). For the analysis of the effect of lunar cycle on migration speeds, the days were sine transformed with full moon =−1 and new moon =+1. Means of these transformed data were calculated for the periods for which the migration speeds were calculated. Data were statistically analysed with non-parametric tests using SPSS (www.spss.com) with P ≤ 0·05.

RESULTS

DETECTION RATES From all A. anguilla released with a transponder in 2004–2006, 43% were never detected up to 1 January 2009 and 38 ± 4%, 43 ± 9% and 48 ± 19% (mean ± s.d. of different batches of fish released) in 2004, 2005 and 2006, respectively, were not detected (Table I). The batches released [missing data for batch September 2004 (Sep-04) in Table I] differed significantly in average maturation stage (Kruskal–Wallis test, χ 2,d.f.= 18, P<0·001) and so did the years of release (Kruskal–Wallis test, χ 2,d.f.= 2, P<0·01). The detected and non-detected fish did not, however, differ in their maturation stage when released (Mann–Whitney U-test, n = 427, P>0·05). Escapement from Germany, as measured from detections at Rhine-Xanten 1 , ranged from 46 to 62% (Table II). The escapement to the sea via the Netherlands varied from 13 to 23% in 2004–2006 (Table II) and detection rates ranged from 28 to 38% of the fish passing Rhine-Xanten 1 . The non-detection rates per river stretch were relatively stable through the years but increased slightly in 2006 (Table II).

MIGRATION ROUTES During the study, 83 ± 2% (yearly mean ± s.d.) of the detected A. anguilla (49–57 peryear)migratedvia the Waal. The route via the IJssel was chosen by only four to ten fish per year (11 ± 5%) and zero to seven fish per year (6 ± 4%) migrated via the Nederrijn–Lek (Fig. 4 and Table I). The majority of fish migrating via the Waal chose the Beneden Merwede 3 and proceeded via the Noord 8 to the Nieuwe Waterweg. All fish, except one in 2005, passing Dordtsche Kil 5 or Spui 6 , initially migrated to the Haringvliet but did not proceed to the sluices in the Haringvlietdam 7 and chose the route to the Nieuwe Waterweg instead. Nine fish temporarily settled in the Rhine Delta for a time in the year after release and were repeatedly detected at some stations there, or before the first weir in the Nederrijn 9 . There were no differences with regard to the route chosen by the fish at the Pannerden bifurcation (either the Waal or the Pannerdens Kanaal) between the years of release (χ 2,d.f.= 2, P>0·50). The bifurcation ratios of the fish (A. anguilla ratios) exceeded the river discharge ratios at Pannerden each year (Table III), and the fish seemed to show a greater preference to migrate via the Waal (instead of the Pannerdens Kanaal) than could simply be explained by the river discharge ratios at Pannerden each year. At the bifurcation IJsselkop, more fish chose the IJssel than the Nederrijn in 2005 and 2006, but the A. anguilla ratio was lower than the river

© 2009 The Authors Journal compilation © 2009 The Fisheries Society of the British Isles, Journal of Fish Biology 2009, 74, 2139–2157 2146 A. W. BREUKELAAR ET AL. up to , Waal-Vuren 1 13 25 Anguilla anguilla Rhine-Xanten via 17 17 50 m numbers released minus not detected) the Waal, but not known in detail in more downstream river sections via Waal) 1-2 1-2-3 1-2-3-8 1-2-3-4 1-2-3-5 1-2-7 1-2-5 1-2-5-4 1-2-6-4 1-9 1-9-10 1–11 1-11-12 Waal’ is known to be oute 1 - Route Route Route Route Route Route Route Route Route Route Route Route Route via via . The route ‘1 5 Sieg 58 45 44 0 16 3 16 0 0 3 3 9 3 0 0 3 0 Releases Chosen migration routes (% fro Release Totals Not detected Route R Cologne 399 43 25 3 25 7 16 2 0 1 2 4 1 0 4 7 1 and Dordtsche Kil I. Relative frequencies (% of numbers released minus numbers not detected) of migration routes followed by silver Totals 457 43 28 3 23 7 16 2 0 1 2 4 1 0 4 7 1 2 Subtotal 2004Subtotal 2005Subtotal 2006 150 157 150 38 43 48 27 33 22 4 1 3 17 17 39 12 7 16 1 21 10 1 4 0 0 1 0 0 0 4 3 2 1 6 0 6 1 1 1 0 1 0 8 0 4 3 10 8 1 1 0 ABLE Releaselocation batch and date (numbers) released released) (% from (%) 1 (%) ( (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) (%) c-4959211 13 6 1866 T 714 CC Oct-042635126181218 COct-0423394377147 C Sep-04 Oct-042945196251919 SOct-04104067 25 32 24 6 24 6 24 12 6 Sp03541707 5 7 23 2936 5 7CCCSSep-0530504713720 Nov-04S Nov-04CSep-05474531128198 Aug-05 27S 10COct-05303038245195 Sep-05 30COct-05135433173317 721 S 37 Oct-05C 40 4CSep-063053 40S Aug-06 3 24CSep-0627701350 Aug-06 33 50C 22 6 Sep-06 26 33 Nov-06 6 5 17 12 50 50 56 33 11 12 20 60 17 23 18 38 17 50 50 50 28 6 31 20 6 15 50 6 6 8 54 20 15 6 12 20 8 6 6 12 20 8 3 11 8 6 8 8 last detection in thein River the Rhine table that in correspond 2004–2006. with Releases detection occurred station numbers at in either Fig. Cologne 1; (C) ‘route or 1-2-5’ the for River example Sieg means the (S). route Routes are indicated by numbers

TABLE II. Percentage detection rates of Anguilla anguilla released in the River Rhine near Cologne in the years 2004–2006 and non-detection rates per river stretch (%) 2004 2005 2006 Totals Detection rates of released A. Anguilla Germany 62 57 46 55 The Netherlands 23 17 13 18 Non-detection rates per river stretch From release to Rhine-Xanten 1 38 43 54 45 From Rhine-Xanten 1 to Waal-Vuren 2 , 27 33 23 28 Nederrijn 9 and IJssel-Kampen 11 From Waal-Vuren 2 and Nederrijn 9 to sea 47 50 60 52 ∗ From IJssel-Kampen 11 to sea 75 90 100 90

∗Numbers ≤ 10. discharge ratio in 2004 and 2006. Considering the river discharges, the data with regard to route choice of ﬁsh are not consistent and conclusive for the bifurcation at IJsselkop.

MIGRATION SEASONS, SPEEDS AND TIMES Most of the fish detected (97, 93 and 90% in 2004, 2005 and 2006, respectively) migrated downstream in the year of release and from the moment of release until the winter. Detections rates were relatively low between January and August (Fig. 5). The mean ± s.d. of the migration speeds of all fish over all stretches was 0·34 ± 0·44 m s−1 (n = 490). The average migration speed to Rhine-Xanten − 1 was 0·30 ± 0·46 m s 1, and from Rhine-Xanten 1 to Waal-Vuren 2 it was − 0·50 ± 0·57 m s 1. The lower mean migration speed up to Rhine-Xanten 1 ,as compared with the more downstream river stretch, was mostly due to 14% more fish in the migration speed class 0·0–0·3ms−1. The maximum speed in the 154 km − stretch Cologne–Rhine-Xanten 1 was 2·20 m s 1, and the maximum speed in the − 128 km stretch Rhine-Xanten 1 and Waal-Vuren 2 was 1·45 m s 1. Migration speeds tended to be highly skewed (Fig. 6). For those individual fish for which data were available from both release locations to Rhine-Xanten 1 and for the stretch Rhine-Xanten 1 to Waal-Vuren 2 , migration speeds were not correlated (r = 0·013). There were only a few fish moving fast on both stretches. On the stretch from release locations to Rhine-Xanten 1 , fish from different maturation classes showed no significant differences in migration speeds (Kruskal–Wallis test, χ 2,d.f.= 2, P>0·05): fish from the maturation class SF-III migrated on average at 0·37 ± 0·56 m s−1, SF-IV fish migrated at 0·25 ± 0·38 m s−1 and SF-V fish migrated at 0·07 ± 0·07 m s−1.

ENVIRONMENTAL FACTORS The migration speeds of the ﬁsh were not correlated with average river discharges when the ﬁsh migrated from the location of release to one of the detection stations Rhine-Xanten 1 , Waal-Vuren 2 , Nederrijn 9 and IJssel-Kampen 11 (Fig. 7, r = 0·08). The individual migration speeds of silver A. anguilla in these river stretches showed little relation to the mean of the lunar cycle (r = 0·15).

Rhine−Sieg release 457

Rhine-Xanten 260

(Pannerden) (Usselkop)

Waal-Vuren Nederrijn Ussel-Kampen 157 11 20

Beneden Merwede Lek 66-73 10

Dortsche Kil Noord 1 17−24 42

Spui 0 3−10

Oude Maas 26

Haringvliet (Nieuwe dam Waterweg) Afsluitdijk 3 78 2

Escapement 83

FIG. 4. Migration routes, survival and escapement to sea deduced from telemetry of silver Anguilla anguilla released in the Rivers Rhine and Sieg near Cologne in August 2004 to January 2007. No detection stations were located at Pannerden, IJsselkop and Nieuwe Waterweg. Escapement numbers are minimum estimates and calculated from the last stations before entering the sea. Ranges of numbers at Beneden Merwede 3 , Dordtsche Kil 5 and Spui 6 indicate uncertainties about which of these stations had been passed. Left-hand panel numbers at Dordtsche Kil 5 and Spui 6 indicate migration from Beneden Merwede 3 to Haringvlietdam 7 ; right-hand panel numbers migrating from Waal-Vuren 2 via Haringvliet to Oude Maas 4 .

The migrating fish were detected throughout the day (Fig. 8) and only showed a slightly increased detection peak, indicating higher activity, in the first 6 h after sunset. The total daily number of fish detections at all detection stations in the months August to December, as an indication of activity, did not seem to show a clear relationship with river discharges at Lobith during the seasons. Only in 2004 were

TABLE III. Ratios of river discharges and of migrating silver Anguilla anguilla at the bifurcation points Pannerden (Waal–Pannerdens Kanaal) and IJsselkop (IJssel–Nederrijn) in the years 2004–2006. River discharge ratios are based on daily mean discharges of the watercourses involved from 1 August to 31 December. Anguilla anguilla ratios are based on total numbers of detected ﬁsh in the watercourses in the year of release River discharge ratio A. anguilla ratio Pannerden 2004 3·05·2 2005 3·34·5 2006 3·04·9 IJsselkop 2004 5·10·6 2005 6·510·0 2006 4·62·0 there indications for coinciding peaks (Fig. 9), but the datasets were signiﬁcantly correlated (r = 0·184, P<0·001). The daily number of detections were not related to the lunar cycle (Fig. 9, r = 0·07, P>0·05).

DISCUSSION

NON-DETECTION Thirty-two, 23 (19%) of the downstream migrating A. anguilla were not detected at Rhine-Xanten 1 , Waal-Vuren 2 and Beneden Merwede 3 , respectively. The failure rates at Rhine-Xanten 1 amounted to 30, 32 and 35% in the successive years 2004–2006. Detection failures of the whole detection system between Rhine-Xanten 1 and the last stations before entering the sea in the Netherlands (Haringvlietdam 7 , Oude Maas 4 , Noord 8 ,Lek 10 and Afsluitdijk 12 ) were 21, 8 and 0% in successive years. Missing detections made determination of the exact migration route in the delta part of the River Rhine impossible for six ﬁsh. All of these ﬁsh

90 80 70 60 50 40 detections 30 20 Number of releases and 10 0 Aug-04 Feb-05 Aug-05 Feb-06 Aug-06

FIG. 5. Number of releases ( )ofsilverAnguilla anguilla in the River Rhine at Cologne or Sieg in the years 2004–2006 and of detections of these ﬁsh at all detection stations in the River Rhine ( ). Repeated detections of one ﬁsh at the same station were counted as one detection.

80 70 60 50 40 30 Frequency (%) 20 10 0 0·0– 0·3– 0·6– 0·9– 1·2– 1·5– 1·8– 2·1– 0·3 0·6 0·9 1·2 1·5 1·8 2·1 2·4 Migration speed (m s–1)

FIG. 6. Relative frequency distributions of migration speeds of silver Anguilla anguilla in the River Rhine over the river reaches between release location (Cologne or Sieg) and Rhine-Xanten (1) [Release-Xanten ( ); n = 171] and over the reach between Rhine-Xanten 1 and Waal-Vuren 2 [Xanten-Vuren ( ); n = 71] in 2004–2006. n is the number of measurements of migration speeds of different individual ﬁsh.

1·80

1·60

1·40 )

–1 1·20

1·00

0·80

0·60 Migration speed (m s

0·40

0·20

0·00 0 1000 2000 3000 4000 River discharge (m3 s–1)

FIG. 7. Migration speeds of individual silver Anguilla anguilla (n = 180) in the transects between location of release and detection stations where they had been detected for the first time, in relation to the average river discharge of the River Rhine at Lobith (here the Rhine enters The Netherlands) when the fish passed these transects. had chosen the Waal, however, and were used for the analysis of route choice at Pannerden. Why some fish were not detected after release is unknown, but technical problems, behavioural factors and natural or anthropogenic mortality could be

10 Number of detections

0 0 4 8 12 16 20 Hours after sunset

FIG. 8. Number of detections of silver Anguilla anguilla per hour (all detection stations and years combined). The timescale is scaled to hours after sunset. Note that sunrise is undefined in this figure because it changes in the study area between 8 and 17 h after sunset during the seasons. involved. Transponder failures could have occurred or some fish may have lost their transponders, but this did not occur in a 19 week controlled experiment by Winter et al. (2005). There were some technical failures at the Beneden Merwede 3 detector station but flooding of antenna cables at some other stations did not appear to have had significant effects on detection rates. Some fish may have died or moved upstream or settled in the neighbourhood of the release location, naturally or because of stress due to handling, surgical procedures or the effects of implanted transponders. In a similar tracking study in the River Meuse, one fish from 150 released moved upstream (Winter et al., 2006) and preliminary data of telemetered fish in the River Rhine and in the River Berwijn, a small tributary of the River Meuse, showed that some tagged silver A. anguilla did not migrate downstream (unpub. data). It is possible that some fish may not have been at a sufficiently advanced stage of silvering to show full migratory activity. Detected and undetected fish, however, did not differ greatly in estimated maturation stage when released. Moreover, fish released in Cologne in August, September and October 2005 (Table I) showed a progression in maturation status during autumn with increasing levels of haematocrit, oestradiol, oocyte diameters, fat in oocytes and gonado-somatic index (A. Palstra, pers. comm.), suggesting that maturation continued during the season. It is possible, however, that reversal of the silvering process occurred in some fish, as it is known that silver A. anguilla are capable of resuming feeding and delaying migration for several years (Vøllestad et al., 1994; Feunteun et al., 2000; Durif et al., 2005). The maximum delay in first detection of the 2004 cohort was 10 months, two fish from the 2005 releases were found to delay their migration for up to 14 months and one fish of the 2006 cohort did so for 15 months. Longer delays might have occurred but this cannot be confirmed by the detection date because of the limited transponder battery life (1·5–2·0 years). Another reason why the released fish might differ is because of their capture locations and methods of capture. The migration routes chosen and the percentage

2004 4000 18 (a)

3000 14

10 2000

6 1000 2 0

4000 18 (b)

) 3000 14 y –1

10 uenc 2000 q

Discharge (m s 1000

2 Detection fre 0

4000 18 (c)

3000 14

10 2000

6 1000

2 0 Aug Oct Dec

FIG. 9. Detection frequency ( )ofsilverAnguilla anguilla in relation to river discharges at Lobith ( )and to the lunar cycle (sine transformed and on an arbitrary scale, maximum = new moon, minimum = full moon) ( ), during downstream migration in Autumn (a) 2004, (b) 2005 and (c) 2006. of undetected ‘Sieg’ and ‘Moselle’ A. anguilla, however, did not seem to differ in general terms and where they did on speciﬁc routes (Table I), the numbers were too low to reach ﬁrm conclusions. Anthropogenic mortality may be another cause of non-detections before silver A. anguilla escape to the sea, as discussed further below.

MIGRATION ROUTES AND MIGRATION OBSTACLES At lower river discharges, most of the water from the Rhine flows via the Waal and enters the North Sea via the Nieuwe Waterweg. Results show that 84% of the fish migrated via the River Waal towards the sea. The Rivers IJssel and Nederrijn–Lek were less preferred (11 and 6%, respectively). The choice for the Waal as the primary migration route was not fully explained by the difference in river discharges in the Waal and the Pannerdens Kanaal. Assuming that the fish passively drift with the flow, more fish (26–42% per year) migrated via the Waal than could be expected from the river discharge ratio at Pannerden. At the IJsselkop bifurcation, the choice of fish was unclear because of low detection rates. The yearly differences in numbers of A. anguilla migrating via the Nederrijn (Fig. 4), however, may be explained by differences in the timing of the operation of the sluices in the peak period of migration. The sluices in the Nederrijn were simultaneously opened during relatively long time periods in 2004 (22 October to 30 November and 23 to 31 December) and 2006 (16 August to 9 September, 21 September to 18 October, 27 to 30 October, 24 November to 1 December and 8 to 21 December). These sluices were only opened from 25 August to 20 September (and on 9 October) in 2005 and only one fish was detected at Nederrijn 9 in that time period, but this fish evidently returned upstream and was detected 20 days later at IJssel-Kampen 11 . Two other fish detected at Nederrijn 9 later in that year also returned and were detected at IJssel-Kampen 11 after 8 days or at Waal-Vuren 2 after 45 days (the latter first must have taken Pannerdens Kanaal upstream). Such searching or hesitating behaviour of silver A. anguilla upstream of a weir (and power station) was also observed in the River Moselle (Behrmann-Godel & Eckmann, 2003) and under controlled laboratory conditions (Adam et al., 1997). Evidently, the first weir in the Nederrijn at Driel functions as an obstruction to the downstream migration of silver A. anguilla. The sluices in the Haringvlietdam 7 must also form a major barrier, through which only three fish escaped. There is a tidal influence and saltwater intrusion in this delta region up to Waal-Vuren 2 and Lek 10 , but in the Haringvliet the tidal range is reduced to a few decimetres and saltwater intrusion is prevented by the management of the sluices in the Haringvlietdam 7 . Most of the fish passing Waal- Vuren 2 migrated via Beneden Merwede 3 (66–73 in total, Fig. 4) and Noord 8 towards the Nieuwe Waterweg, the remainder (20–27 in total) migrated via the Haringvliet but returned and migrated via Dordtsche Kil 5 and Spui 6 towards the Nieuwe Waterweg. The data obtained from this telemetric study suggest that the current water management in the main course of the River Rhine in the Netherlands does not preclude the silver A. anguilla to find their way to the sea. There are indications, however, that the existing sluices in the Nederrijn and the sluices at the Haringvliet induce a prolonged stay of some of the fish upstream of these obstacles, enhancing the risk for being caught by fishermen in the delta region of the River Rhine or reducing the risk for being entrained by hydropower stations in the Nederrijn.

FACTORS AFFECTING MIGRATION Higher commercial catches of silver A. anguilla generally occur at night, at low lunar light conditions or in deeper water with lower light conditions (Tesch, 2003). Locally operating sensory cues are considered to be important therefore for

© 2009 The Authors Journal compilation © 2009 The Fisheries Society of the British Isles, Journal of Fish Biology 2009, 74, 2139–2157 2154 A. W. BREUKELAAR ET AL. downstream migrating anguillid eels (Haro, 2003). Silver A. anguilla tend to avoid artificial light at hydropower stations (Hadderingh et al., 1992). Local ambient light conditions might affect, therefore, both depth and lateral position of the fish in the cross section of the watercourse at a bifurcation point. Subsurface light penetration, however, will be low in turbid and deep rivers such as the River Rhine (yearly average Secchi-depth range 50–60 cm with s.d. = 12–21 cm in the time period 1987–2005). Thus, light would be expected to have little effect on A. anguilla migrations in the River Rhine, both with regard to route choice and timing, because fish can easily find lower light levels in deeper water. This is possibly why, contrary to many other studies (Cairns & Hooley, 2003; Haro, 2003; Tesch, 2003), silver A. anguilla migration in the River Rhine occurred frequently during the day and showed no clear relationships with the lunar cycle. Downstream migration speeds of anguillid eels range from 0·05 to 1·10 m s−1 in telemetric studies (Haro, 2003) and depend on flow rates as fishes make both active and passive use of the currents (Richkus & Dixon, 2003). Most individual A. anguilla migrated downstream with highly varying speeds in the River Rhine, and the frequency distribution was shown to be highly skewed towards <0·30 m s−1 (Fig. 6). These results match with those from a comparable study in the River Meuse, where several fish also showed stepwise migrations with large intervals up to 20 months (Winter et al., 2006). In the present study, few fish consistently migrated at high speeds on two successive routes towards the sea (location of release to Rhine-Xanten 1 and Rhine-Xanten 1 to Waal-Vuren 2 ), but many fish did so on one of these routes. Most of the fish migrated more slowly downstream than can be explained by passive drift in the main current (Fig. 6) and thus they must have shown intermittent migrations. Such behaviours could explain why river discharges and lunar cycles did not show a clear effect on detection frequency of the A. anguilla in this study. Reasons for the delays in the migrations between detection stations, however, are unknown.

ESCAPEMENT TO SEA Of the 2004, 2005 and 2006 cohorts of silver A. anguilla that passed Rhine-Xanten 1 and immigrated into The Netherlands, 38, 29 and 29%, respectively, passed the last detection stations before escaping to the sea. All others probably ceased migration for a prolonged period or died. Silver A. anguilla fisheries and passage through hydropower plants may have been anthropogenic causes of mortality. The only river stretch where hydropower stations might have affected survival, however, was in the Nederrijn and here, only one to seven eels per year from the 2004–2006 releases were detected (Fig. 4). The total losses of detections in this river stretch during this study were low, and the effect of hydropower appeared to be insignificant. The fish in this study seemed to show a relatively continuous migration activity from August to December, with a series of successive smaller peaks spread over the whole period. Migration also occurred during daytime nearly as intensively as during night hours, thus the use of discharge data, lunar cycles and night-times in forecasting possible peaks of silver A. anguilla migration to aid escapement will therefore be limited. For example, temporary closures of hydropower plants in the Nederrijn reaches to minimize turbine mortalities cannot be planned in advance with the current knowledge.

Fishing mortality might have been a cause of non-detections and affected escapement rates. The intensive A. anguilla fishery in Lake IJsselmeer (Dekker, 2000) may have been a principal cause of the 75–100% decrease in detection rates over 2004–2006 between IJssel-Kampen 11 and the sea, but the numbers of fish involved (four to 10 per year) were small (Table II). On all other Dutch river stretches, the Lek and Nieuwe Waterweg included, the only known anthropogenic causes of mortality are A. anguilla fisheries. These are estimated (in orders of magnitude) by the Ministry of Agriculture, Nature Management and Food Quality (MANMFQ, 2008) in The Netherlands Eel Management Plan to cause a loss of 100 t (25%) of the current annual escapement to the sea of 400 t. Winter et al. (2006) estimated fishing mortality of silver A. anguilla released with transponders in the River Meuse to be 22–26%. Silver A. anguilla migration from the German part of the River Rhine to the Dutch part is estimated in The Netherlands Eel Management Plan to be in the order of 300 t year−1, of which 100 t are landed, i.e. a fishing mortality of c. 33% per year. These estimates contrast with an overall non-detection rate between Rhine-Xanten 1 and the sea of 71% (Fig. 4), suggesting that perhaps half of the fish not detected may have been lost to the fisheries. The transponder and telemetry systems deployed in the present study proved useful in tracking large female silver A. anguilla in the River Rhine. Many fish with implanted transponders, however, were not detected after release, possibly due to technical problems but also due to fish not migrating downstream within the study period and within the 1·5–2·0 year lifespan of the transponder batteries. Natural and anthropogenic mortality factors could also be important but more data are needed on causes of non-detections to reach reliable estimates of anthropogenic mortalities due to hydropower plants and fisheries. For example, the escapement rate to the sea of silver A. anguilla detected at Rhine-Xanten 1 in this study appears to be c. 29% (Fig. 4), but it remains to be determined to what extent this apparent failure to reach the 40% EU target is due to mortalities rather than technical or behavioural factors. This study has shown the telemetry system has potential in large rivers like the River Rhine for studying routes used and the effects of environmental factors on migration speeds and timings and escapement to the sea. More detection stations and tracking in local environments of individual fish, however, are needed for a fuller understanding of the migration behaviour of silver A. anguilla. The lifetime of the transponder batteries should also be doubled at least in order to detect all delayed migrants. Despite these provisos, this study has shown that most of the silver A. anguilla detected chose the River Waal and the Nieuwe Waterweg on their downstream migration to the sea, and consequently relatively few were exposed to the hydropower stations in the Nederrijn, the fishery in Lake IJsselmeer or the Haringvliet sluices. The current management of the sluices in the Haringvlietdam 7 and Afsluitdijk 12 does not seem to prevent fish from escaping to the sea. This study also showed that silver A. anguilla in the Rhine showed more or less continuous migration activity from August to December, with a series of successive smaller peaks spread over the whole period. Fish did not seem to migrate at constant speeds or follow the main flow patterns at bifurcation points and relations of migration speed to river discharge and lunar cycles were weak and absent, respectively. There was a small migration peak during the first 6 h after sunset but fish migrated all day in the turbid River Rhine. It, therefore, appears that the use

© 2009 The Authors Journal compilation © 2009 The Fisheries Society of the British Isles, Journal of Fish Biology 2009, 74, 2139–2157 2156 A. W. BREUKELAAR ET AL. of discharge data, lunar cycles and night-times in forecasting possible peaks in silver A. anguilla migration are limited. This has practical implications, because if such peaks could be accurately predicted in advance in rivers like the Rhine, temporary protection measures could be imposed, such as temporary closures of hydropower turbine intakes and water diversions. In conclusion, although there are some limitations that need addressing, this study has shown it is feasible to use the NEDAP Trail telemetry system for tracking female silver A. anguilla in large and highly modiﬁed rivers like the Rhine to develop understanding of their emigration and assess management options to achieve escapement targets.

The project was made possible by using the eels caught in the yearly trap-and-release actions by the Struktur und Genehmigungsbehorde¨ Nord (Rheinland-Pfalz). LOBF-NRW,¨ OVB, Sportvisserij Nederland and Visadvies contributed to the field work. The Universitat¨ zu Koln¨ made a field and laboratory facility available in Cologne. RWS Waterdienst provided the telemetry infrastructure and the labour for its maintenance. OVB and LOBF¨ co-ordinated the project and are acknowledged for funding the project. Another part of the project was funded by EU through the FIFG to Rheinfischereigenossenschaft. The authors acknowledge the participation of E. Winter and A. Palstra for personal comments on maturation of the eels, B. Knights and C. Belpaire for useful comments on a draft paper and the staff of all partners to the field work, especially G. Feldhaus, A. Hehenkamp and J. Merkx.

References Acou, A., Lafaille, P., Legault, A. & Feunteun, E. (2008). Migration pattern of silver eel (Anguilla anguilla, L.) in an obstructed river system. Ecology of Freshwater Fish 17, 432–442. Adam, B., Schwevers, U. & Dumont, U. (1997). Beitrage¨ zum Schutz abwandernder Fische–Verhaltensbeobachtungen in einem Modellgerinne. Bibliothek Natur & Wis- senschaft Band 16 . Solingen: Verlag Natur & Wissenschaft. Behrmann-Godel, J. & Eckmann, R. (2003). A preliminary telemetry study of the migration of silver European eel (Anguilla anguilla L.) in the River Mosel, Germany. Ecology of Freshwater Fish 12, 196–202. Bij de Vaate, A., Breukelaar, A. W., Vriese, T., de Laak, G. & Dijkers, C. (2003). Sea trout migration in the Rhine delta. Journal of Fish Biology 63, 892–908. Breukelaar, A. W., Bij de Vaate, A. & Fockens, K. T. W. (1998). Inland migration study of sea trout (Salmo trutta) into the Rivers Rhine and Meuse (Netherlands), based on inductive coupling radio telemetry. Hydrobiologia 371/372, 29–33. Cairns, D. K. & Hooley, P. J. D. (2003). Lunar cycles of American eels in tidal waters of the Southern Gulf of St. Lawrence, Canada. In Biology, Management, and Protection of Catadromous Eels (Dixon, D. A., ed.), pp. 265–274. American Fisheries Society Symposium 33. Dekker, W. (2000). Impact of yellow eel exploitation on spawner production in Lake IJsselmeer, The Netherlands. Dana 12, 25–40. Dekker, W. (2004). Slipping through our hands. Population dynamics of the European eel. PhD Thesis, University of Amsterdam, The Netherlands. Durif, C. M. F. & Elie, P. (2008). Predicting downstream migration of silver eels in a large river catchment based on commercial fishery data. Fisheries Management and Ecology 15, 127–137. Durif, C., Dufour, S. & Elie, P. (2005). The silvering process of Anguilla anguilla:anew classification from the yellow resident to the silver migrating stage. Journal of Fish Biology 66, 1025–1043. EU (2007). Establishing measures for the recovery of the stock of European eel. Council regulation (EC) No 1100/207 of 18 September 2007. Official Journal of the European Union L 248, 17–23.

Feunteun, E., Acou, A., Laffaille, P. & Legault, A. (2000). European eel (Anguilla anguilla): prediction of spawner escapement from continental population parameters. Canadian Journal of Fisheries and Aquatic Sciences 57, 1627–1635. Hadderingh, R. H., Van der Stoep, J. W. & Habraken, J. M. P. M. (1992). Deflecting eels from water inlets of power stations with lights. Irish Fisheries Investigations Series A (Freshwater) 36, 78–87. Haro, A. (2003). Downstream migration of silver-phase Anguillid eels. In Eel Biology (Aida, K., Tsukamoto, K. & Yamauchi, K., eds), pp. 215–222. Tokyo: Springer. Klein Breteler, J., Vriese, T., Borcherding, J., Breukelaar, A., Jorgensen,¨ L., Staas, S., de Laak, G. & Ingendahl, D. (2007). Assessment of population size and migration routes of silver eel in the River Rhine based on a two-year combined mark–recapture and telemetry study. ICES Journal of Marine Science 64, 1450–1456. MANMFQ (2008). The Netherlands Eel Management Plan. The Hague: Department of Fisheries, Ministry of Agriculture, Nature Management and Food Quality. Richkus, W. A. & Dixon, D. A. (2003). Review of research and technologies on passage and protection of downstream migrating catadromous eels at hydroelectric facilities. In Biology, Management, and Protection of Catadromous Eels (Dixon, D. A., ed.), pp. 377–388. American Fisheries Society Symposium 33. Tesch, F.-W. (2003). The Eel, 3rd edn. Oxford: Blackwell Science Ltd. Van Ginneken, V., Durif, C., Dufour, S., Sbaihi, M., Boot, R., Noorlander, K., Doornbos, J., Murk, A. J. & van den Thillart, G. (2007a). Endocrine profiles during silvering of the European eel (Anguilla anguilla L.) living in saltwater. Animal Biology 57, 453–465. Van Ginneken, V., Durif, C., Balm, S. P., Boot, R., Verstegen, M. W. A., Antonissen, E. & van den Thillart, G. (2007b). Silvering in European eel (Anguilla anguilla L.): seasonal changes of morphological and metabolic parameters. Animal Biology 52, 63–77. Vøllestad, L. A., Jonsson, B., Hvidsten, N. A., Naesje, T. F., Haraldstad, Ø. & Ruud-Hansen, J. (1986). Environmental factors regulating the seaward migration of European silver eels (Anguilla anguilla). Canadian Journal of Fisheries and Aquatic Sciences 43, 1909–1916. Vøllestad, L. A., Jonsson, B., Hvidsten, N. A. & Naesje, T. F. (1994). Experimental test of environmental factors influencing the seaward migration of European silver eels. Journal of Fish Biology 45, 641–651. Winter, H. V., Jansen, H. M., Adam, B. & Schwevers, U. (2005). Behavioural effects of surgically implanting transponders in European eel Anguilla anguilla.InAquatic Telemetry: Advances and Applications (Spedicato, M. T., Marmulla, G. & Lembo, G., eds), pp. 1–9. Rome: COISPA Technoloia & Riverca. Winter, H. V., Jansen, H. M. & Bruijs, M. C. (2006). Assessing the impact of hydropower and fisheries on downstream migrating silver eel, Anguilla anguilla, by telemetry in the River Meuse. Ecology of Freshwater Fish 15, 221–228.

Electronic References ICES (2004) Report of the ICES/EIFAC Working Group on Eels. International Council for the Exploration of the Sea, Sukarrieta, Spain, 7–11 October 2003. ICES CM 2004/ACFM:09. Available at http://www.eaa-europe.org/ﬁleadmin/templates/uploads/ Eels/ICES_WGEEL_2004.pdf ICES (2005) Report of the ICES/EIFAC Working Group on Eels. International Coun- cil for the Exploration of the Sea,Galway, Ireland, 22–26 November 2004. ICES CM 2005/ACFM:01 Available at http://www.eaa-anglingineurope.com/ﬁleadmin/ templates/uploads/Eels/ICES_WGEEL_2003.pdf