Behavioural responses of eel (Anguilla anguilla) approaching a large pumping station with trash rack using an acoustic camera (DIDSON)

European eel, Anguilla anguilla L., is in strong decline since the 1970s (ICES, 2018) as a result of various factors, such as migration barriers, fisheries, habitat loss and deterioration, pollution, parasites and changes in oceanic conditions (Buysse, Mouton, Stevens, Neucker, & Coeck, 2014; Drouineau et al., 2018; Feunteun, 2002; Moriarty & Dekker, 1997; Palstra, Heppener, Ginneken, Székely, & Thillart, 2007; Westerberg et al., 2018). European eel migrates from inland waters to the sea during their adult stage (silver eel) to head for their spawning area in the Sargasso Sea (Tesch, 2003). During the migration, they encounter man-made structures such as pumping stations, sluices, weirs and hydropower stations that obstruct their migration route. In the Netherlands, discharge sluices and large pumping stations have been built for water level control of the catchment areas, protection against the sea and to limit saltwater intrusion. Fish migration can be hampered by pumping stations by inducing additional mortality when the turbines are not designed in a fish-friendly manner (Buysse, Mouton, Baeyens, & Coeck, 2015; Fjeldstad, Pulg, & Forseth, 2018) or by delaying fish resulting in additional energy loss and potential mismatch in Received: 10 December 2019 | Revised: 25 February 2020 | Accepted: 28 February 2020 DOI: 10.1111/fme.12427

. European eel migrates from inland waters to the sea during their adult stage (silver eel) to head for their spawning area in the Sargasso Sea (Tesch, 2003).
During the migration, they encounter man-made structures such as pumping stations, sluices, weirs and hydropower stations that obstruct their migration route. In the Netherlands, discharge sluices and large pumping stations have been built for water level control of the catchment areas, protection against the sea and to limit saltwater intrusion. Fish migration can be hampered by pumping stations by inducing additional mortality when the turbines are not designed in a fish-friendly manner (Buysse, Mouton, Baeyens, & Coeck, 2015;Fjeldstad, Pulg, & Forseth, 2018) or by delaying fish resulting in additional energy loss and potential mismatch in
Trash racks are placed in front of pumping and hydropower stations to prevent large debris getting into the turbines and are the first physical obstacles fish encounter while approaching these structures. Most studies on the effects of pumping or hydropower stations with trash racks focus on the risk of entrainment, but more focus should be directed to behavioural responses of fish in the vicinity of these structures in relation to entrainment risk (Harrison et al., 2019). An eel approaching these structures perceives multiple cues that might trigger a change in its behaviour, for example, at larger distances underwater sound emitted by the stations (Behrmann-Godel & Eckmann, 2003) and near the trash rack and pumping stations water flows change in direction and increase in speed (Piper et al., 2015). Close to the structure, visual detection and eventually physical encounters could trigger eels to swim away from the structure, as has been shown in laboratory (Piper et al., 2015;Russon et al., 2010) and field experiments (Travade et al., 2010).
The behaviour of downstream migrating eels approaching a trash rack at the entrance of a large pumping station in IJmuiden, the Netherlands, was investigated using a dual-frequency identification sonar (DIDSON). The aim of this study was (a) to assess how eels were positioned in the water column while approaching the rack; (b) how the eels reacted in the vicinity of the trash rack and entrance of the pumping station; and (c) to use these observations to discuss management options.

| MATERIAL S AND ME THODS
The study site was a large pumping station at IJmuiden (52˚28′N, 4˚36′E), the Netherlands, that is part of a sluice complex that discharges fresh water from Lake Markermeer, the Amsterdam-Rhine The behaviour of migrating eels near the trash rack was studied during the first hours after sunset when migrating eels are most active (e.g. Winter et al., 2006). Migrating eels were observed for two periods: five evenings in November-December 2009 and five evenings in November-December 2010. In both periods, the turbines were in operation.
To study migrating eel behaviour in front of a trash rack at small scale, a DIDSON "dual-frequency identification sonar" was used of the eels while approaching the trash rack; and (c) behaviour of the eels during the encounter with the trash rack. After passing the trash rack, the eels enter the supply channel between the rack and the turbine. Here, they continue towards the turbine and eventually passing it or they turn around and swim against the water current, passing the trash rack again in an opposite direction.
Observations of eels derived from the DIDSON software files were carried out separately by two observers to validate the categorisation of eel movements. When lacking consensus between both observers, a third observer was consulted and the movement was discussed. When no consensus could be reached, the eel movement was scored as unknown. Distance of an eel to the canal bottom was measured using the DIDSON software measuring tool.
Differences in behaviour were tested using a two-sided t test, and changes in swimming depth before and after the approach were tested using a two-sided paired t test (SAS version 9.3: SAS Institute Inc, 2011).

| RE SULTS
A total of 376 eel movements were recorded during 35.5 hr spread over five nights in 2009. In 2010, 88 eel movements were recorded during 7.36 hr spread over five nights. Eels could have been recorded more than once, because they could swim in and out of the DIDSON field of view as it did not cover the entire trash rack. Out of the 464 eel movements recorded by the DIDSON, 46.7% (n = 217) were within the DIDSON field of view when approaching the trash rack from upstream moving with the water current towards the turbine.
In 22.8% (n = 106) of the movements, eels were observed while passing the trash rack in an upstream direction (Table 1). In 30.3% (n = 141) of the movements, eels were observed in a sideways direction in front of the trash rack, not passing it, coming into the field of the DIDSON from the side, with no clear approach of the trash rack from upstream.
The observed eels displayed different body positions while approaching the trash rack with the water current, coming from the trash rack and pumping station ( Figure 1). Of the 217 detected, 118 eels swam "with the current" and head first (negative rheotaxis), 42 swam actively "against the current" and tail first (positive rheotaxis), 27 swam "sideways," and 27 were "curled up" into a small ball and passively floating with the current. The exact body position for three eels was not clear from the DIDSON observations. Eels approaching the trash rack from the canal side showed three behaviours near the trash rack: 40.5% of the eels swam through the trash rack towards the turbine, 14.7% showed turning behaviour in front of the trash rack and 44.7% at the trash rack, thereby swimming away from the trash rack in an upstream direction. The number of eels turning in front of or at the trash rack was significantly more than eels swimming through the trash rack (t = 2.750, p < 0.01, n = 217); Of the eels that swam with the current and against the current during their approach to the trash rack, 47% and 45%, respectively, swam through the rack, while 53% and 55% turned either in front of or at the rack ( Figure 2). Both were not significantly different (t = 0.542, i < 0.59, n = 118 and t = 0.851, p < 0.56, n = 42). With eels swimming sideways, the percentage of eels swimming through the trash rack declined to 37% (1.308, p < 0.20, n = 27), while for curled up eels, this percentage was only 11% (t = 3.192, p <0.01, n = 27), with most eels (81%) that approached the trash rack curled up turning at the trash rack. For eels approaching sideways and going through the rack (n = 10), it could not be assessed if the eels eventually went head first, tail first or body first through the rack from the DIDSON images as the reflection of the trash rack dominated in the images.
Differences in swimming depth of the eels when entering the DIDSON field of view, at or near the trash rack and leaving the field of view (Figure 3), showed no significant change in swimming depth for eels going through the trash rack towards the turbine or eels approaching the trash rack before turning near or at the trash rack (Table 2). However, eels turning both at (p < 0.001) and in front (p = 0.02) of the trash rack showed a significant change in swimming depth after turning with eels swimming deeper towards the bottom.
Also eels passing the trash rack in an upstream direction showed a significant (p < 0.001) change in swimming depth, swimming deeper towards the bottom after passage through the rack.

| D ISCUSS I ON
There were clear behavioural responses and patterns in eel approaching and swimming near the trash rack and pumping station.
Near the trash rack, migrating eels showed different behaviour with significantly more eels turning near the trash rack or after direct contact with the trash rack than going through the rack, indicating trash rack avoidance behaviour. During the approach of the trash rack, eels turned at and in front of the trash rack and eels were also seen swimming alongside the trash rack and out of the trash rack after having turned closer to the entrance of the turbine.
The results suggest a stepwise response to different cues when approaching the trash rack and pumping station. Firstly changing their body position to curled up, sideways drifting or swimming tail first and secondly showing avoidance or countercurrent swimming away from the structures. It might well be that a perceived cue triggers the eel to a more wary state, and a second or stronger cue triggers an avoidance or flee response. Eels that are curled up, sideways drifting or swimming tail first can change quicker to upstream sprinting than downstream swimming eel. A similar stepwise TA B L E 1 Number of eels per approaching direction (upstream coming from canal side and downstream coming from the turbine side), movement near the trash rack and body position of the eels change in behavioural response, first a change body position to tail first drifting eventually followed by fleeing by swimming or sprinting against the water current, was found for salmon smolts approaching unnatural structures (Kemp & Williams, 2009).
In a laboratory experiment of eels approaching a trash rack, the majority of eels released in front of a trash rack tended to maintain regular contact with the channel floor (91.7%) and walls (95%)  When there are no alternative routes other than migrating through the pumping station, there is a higher chance of these eels eventually going through the turbines. Mortality rates then depend on how "fish friendly" the type of turbine is, the size of the fish and also on the position of the eel. An elongated species like eel has a higher chance of being struck by one of the blades than an eel that is curled up into a small ball or drifting sideward, and an eel that is swimming slowly against the current has a higher expected mortality rate than eel swimming with the current due to a shorter presence in the striking zone. Mortality rates can be Mortality rates can also be reduced when eels are hindered from entering the turbines, for example covering the entrance with racks with narrow bar spacings and at the same time supplying safe bypass routes (Dainys, Stakenas, Gorfine, & Lozys, 2018;Økland et al., 2019;Travade et al., 2010). A large, fine-mesh trash rack, however, is technically challenging to construct and operate, and could result in fish mortality due to impingement Fjeldstad et al., 2018).

| IMPLI C ATI ON S FOR MANAG EMENT
After turning in front or at the trash rack, eels showed a downward movement towards the bottom. Eels that passed the trash rack from the turbine also showed a downwards swimming direction against the water current. This is probably because water velocities are slower at the bottom than in the water column, and it is therefore easier to use for upstream escapement. Providing alternative migration routes or fishways near the bottom near objects such as a trash rack could enhance the redirection of eels away from the pumping station, thereby contributing to eel migration survival. Bottom bypasses have been suggested over surface bypasses for eel (Dumont & Hermens, 2012;Gosset, Travade, Durif, Rives, & Elie, 2005;Klopries, Deng, Lachmann, Schüttrumpf, & Trumbo, 2018); however, surface bypasses have also been reported to benefit eel migration (Travade et al., 2010). Redirection would, however, only work if there are other possibilities to migrate and it should be taken into account that eels follow migration routes with the largest portion of water flow (Jansen et al., 2007;Økland et al., 2019;Trancart et al., 2018;Travade et al., 2010).
Only 5%-9% of the eels migrating through a river power station in Germany (Økland et al., 2019) entered a specially built side bottom bypass and 1% a side bypass. Most eels, however, passed over a spillway gate (59%-49% in two study years) or continued the migration towards the turbines (24%-27%), where they were redirected to a flushing channel. Of the eels migrating downstream through a hydropower plant in Lithuania with fish passage for upstream migrating salmon (Dainys et al., 2018), 34% of eels used the fish passage to migrate past the plant. Eels migrating through a hydropower station in Germany (Egg et al., 2017) did not use an eel bypass system, but used an opening of an undershot sluice gate to pass the complex. Higher current velocities in front of this sluice gate were identified as the most important trigger to use this gate instead of the bypass. Giving alternative fish passages next to the rack, especially in combination with effective screening and higher velocities at the bypass, may be an effective measure to reduce mortality rates substantially among migrating eels Egg et al., 2017;Travade et al., 2010).

R E FE R E N C E S
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.