1

EVALUATION OF SURFACE AND BOTTOM BYPASSES TO PROTECT EEL

MIGRATING DOWNSTREAM AT SMALL HYDROELECTRIC FACILITIES IN
FRANCE

F. TRAVADEa , M. LARINIERb , C. GOSSETc , S. SUBRAa, C. DURIFd , J. RIVESb , P. ELIEd

a : EDF, Recherche et Dveloppement, Dpartement LNHE, 6 Quai Watier, B.P. 49, 78401 Chatou Cedex, France.
b : CSP (National Council of Inland Fisheries), Cemagref (French Institute of Agricultural and Environmental Engineering).
Institut de Mcanique des Fluides - Avenue du Professeur Camille Soula 31400 Toulouse, France
c : INRA, UMR ECOBIOP, Ecologie Comportementale et Biologiedes Populations de Poissons Station dHydrobiologie
de Saint-Pe-sur-Nivelle, 64310, France.
d : Cemagref (French Institute of Agricultural and Environmental Engineering), Unit Ressources Aquatiques Continentales,
50 Avenue de Verdun, 33612 Cestas Cedex, France.

ABSTRACT

Efficiencies of bypasses for downstream migrating European eels (Anguilla anguilla) were
tested at two hydroelectric power plant in Southwest France. At Halsou 74 eels were
radiotracked and a total of 637 eels were trapped during the three-year study. At Baigts, 79
eels were radiotracked. These experimental studies have shown that a downstream migration
device composed of a bypass with a discharge of 2% to 5% of the turbine discharge located
near a trashrack with 3 cm bar spacing could be partially efficient for adult eels and that, for a
small hydroelectric plant, a bottom bypass was preferable to a surface bypass. The efficiency
of such a device is only partial (3% to 60%) and not sufficient for most power plants given the
high mortality induced by the passage into the turbines. Efficiency could be improved by
reducing the bar spacing of the trashrack (close to 2 cm) which would block the major part of
the eel population or by using behavioural devices to repel eel from the trashrack area.

KEY WORDS: downstream migration; hydroelectric power plant; eel; Anguilla anguilla;
bypass; radiotracking; behaviour.

INTRODUCTION
The problem of obstacles in rivers for the European eel (Anguilla anguilla) is the subject of
increasing requirements because the eel population numbers have dropped considerably over the last
few years in France and Europe. Although the installation of upstream fishways does not seem to
generate any major problems, the protection of adults migrating downstream from mortalities in the
turbines turbine mortality appears to be a problem since this mortality is usually high, because of the
large size of the eels (EPRI, 2001 ; Desrocher, 1984 ; Monten, 1985 ; Larinier and Dartiguelongue,
1989 ; Hadderingh and Blaker, 1998)..
Many solutions have been tested to prevent eels from passing through the turbines and to direct
them to bypasses which will enable their safe return to the river: physical barriers (screens and louvers
EPRI, 2001) and behavioural barriers using light (Haddering, 1982; Haddering et al., 1992; Haddering
et al., 1999), sound (Sand et al., 2000), electricity (Gleeson, 1997) or hydrodynamic (Fielder and
Ghl, 2005).
However, none of these methods have resulted in truly operational industrial installations.
For some 15 years now, research has been conducted in France to develop specific downstream
migration devices for Atlantic salmon smolts. They led to the definition of surface bypasses combined
with conventional water intakes trashracks for which we now have comprehensive feedback (Larinier
and Travade, 1996, 1999, 2002; Chanseau et al., 1999; Croze et al., 1999, 2001; Travade et al., 1999;
Travade and Larinier, 2006). The water intake trashracks act as a behavioural barrier for the juveniles
which have been observed to be, under certain conditions, naturally reluctant to pass through.
Efficiency depends on the bar spacing of the trashrack which should be a sufficiently deterrent to fish,
and on the hydrodynamic conditions immediately upstream of the trashrack (in particular, importance
of velocity and whether or not there is any current tangential to the trashrack liable to guide the fish
toward the bypass), as well as the position, geometry and flow of the bypass. Very good efficiency
was observed for smolts, with bar spacing ranging between 2.5 cm and 3 cm on power plants were the
turbine flow rate does not exceed 100 m3/s.
Starting in 1999, experiments began to determine whether such devices could be used for eels. For
the purpose, two hydroelectric power plants in south-western France (Halsou on the Nive and Baigts
on the Gave de Pau) at which devices for salmon had already been developed and tested, were chosen
as the experiment sites because of their complementary sizes and configurations. Because of the
benthic behaviour of the eel, research focused on comparing the efficiency of bottom and surface
bypasses. So, from 1999 to 2001 two types of bypasses were tested simultaneously at the Halsou plant:
a surface bypass used for the migration of salmonids and a bottom sluice, both located adjacent to the
trashrack. Starting in 2004, studies continued for two years at the Baigts plant where were tested
successively a surface bypass and a bottom bypass. These experiments were conducted each year
during the eel migration season: from the beginning of October to the end of January.
This paper presents the results obtained on these two experimental sites. The main results for
Halsou experiments come from Gosset et al, 2005.

MATERIAL AND METHODS

Study areas
This first study was conducted at the lectricit de France (EDF) hydroelectric power plant at
Halsou, located on the Nive River in south-western France. The Nive is 80 km long and flows into the
Adour estuary at Bayonne (Figure 1). Flow (mean annual flow: 36 m3.s-1)) varies widely depending on
meteorological conditions, from 6 m3.s-1 in low-flow periods to 300 m3.s-1 during spates.
The Halsou plant (Figure 1, 2) is on the lower Nive, some 23 km from the sea. A gravity dam 172
m long and 2.50 m high diverts the water into a headrace 925 m long, 3 m deep, and 11 m wide. The

3
dam has a Denil fish pass with bottom baffles. The plant, equipped with three double horizontal
Francis turbines (180 r.p.m.), generates a maximum of 30 m3.s-1 over a 4.25-m vertical drop. Turbines
T1 (near the left bank) and T2 (at the centre of the intakes) generate a maximum flow of 8 m3.s-1 each.
Turbine T3, on the right bank and the closest to the surface bypass, generates a maximum flow of 14
m3.s-1. The trashrack (length: 20 m; height: 3 m) in front of the intakes is angled at 25 in relation to
the vertical. Spacing between the bars is 3 cm. Maximum water velocity is 1.6 m.s-1 in the headrace
and approximately 0.5 m.s-1 in front of the trashrack, angled at 15 in relation to the headrace axis
(Figure 3). The flood lights which usually illuminate the forebay area at night were turned off during
the study period.

Atlantic
Ocean

Figure 1 : Geographical location of Halsou and Baigts

power plants

Figure 2 : General view of the Halsou power plant

The second study was conducted at the lectricit de France (EDF) hydroelectric power plant at
Baigts (Figure 1), located on the Gave de Pau River (about 120 km northeast of the Halsou plant). The
Gave de Pau river is 175 km long and flows into the Adour River. The catching area is 2575 km2 and
the flow (mean annual flow 77.3 m3/s) varies from 30 m3/s in August-September to more than 100
m3/s in May-June and December-January.
The Baigts plant is located 72 km from the sea. The plant comprises a gravity dam long on the left
bank and a powerhouse on the right bank(Figure 3). The dam, 57 m long is equipped with 2 flood
gates 20 m wide (bottom opening) and a vertical gate, located on the right bank, with a bottom
opening and a surface flap gate on the top. The powerhouse is equipped with 3 Kaplan turbines
generating 30 m3/s each (90 m3/s maximum turbine flow). The trashrack (length: 41 m; height: 5 m;
max velocity: 0.45 m/s) in front of the intakes is angled at 25 in relation to the vertical and at 18 to
the axis of the river. Spacing between the bars is 30 mm. The top of the trashrack is 2m under the
water level and the bottom at 3m over the bottom of the impoundment. The intake for the upstream
migration device (fish lift) is located on the left bank. The discharge is 2.5 m3/s.

Downstream migration bypasses and trapping device

At Halsou, a surface bypass (flap gate with length: 1.38 m; width: 0.90 m), used for several years
for migration of salmonid smolts, was located on the right bank of the forebay at the end of the
trashrack protecting the turbines (Photo 1). Each time the bypass was opened, flow was set at 0.6 m3.s1 1
, but, given variations in water level due to spates and suspension of turbine operations, it varied
between 0.4 m3.s-1 and 1.0 m3.s-1. Opening of the bypass caused a surface velocity parallel to the
trashrack, particularly high at the turbine entrance closest to the bypass (T3).

4
A motorized bottom gate (width: 1.30 m; height: 1.20 m) on the right bank of the forebay, 2.5 m
from the trashrack (Photo 1), and used to drain the headrace, was used as a bottom bypass. To prevent
high velocity and small opening of the gate, a discharge tower was built at the outlet of the bottom
sluice, of the same height as the headrace wall. It was equipped with a vertical flap gate at the top
which enabled setting the flow through the bottom gate at the same rate as through the surface bypass
(0.6 m3.s-1), which for a 0.50-m vertical opening of the gate, gave a section for passage of 0.65 m2 and
water velocity of around 1 m.s-1 through this bottom bypass.
Fish passing through the bottom gate and the discharge tower fell into a reception pool 1.20 m deep
where an inclined dewatering screen were installed to guide the fish into a trap.
Surface and bottom bypasses were opened alternately every other day for one 24-hour test. The trap
was sampled twice a day for the first two years, at around 8:30 am and 6:00 pm, and at least once a
day (8:30 am) in the third year. Fish captured were identified and their main morphological
characteristics were recorded (weight and head thickness measured horizontally, but also total length
and vertical and horizontal eye diameters). These data were used in a complementary study (Gosset et
al., 2002; Durif, 2003). Captures were compared on a daily basis.

A
B
C
Photo 1 : Bypasses tested. A: Surface and bottom bypasses at Halsou plant. B: Surface bypass at
Baigts plant. C : Bottom bypass at Baigts plant
The Baigts intake is equipped with a surface downstream bypass which is efficient (92.5%) for salmon
(Gosset and Travade 1999). It is located on the left side of the intake at the end of the trashrack
(Figure 3, Photo 1) and consists in a broad weir with the following characteristics: width 2m, depth 1.4
m, discharge 2.2 m3/s, or 2.4% of the turbine discharge. As the level upstream the dam is maintained
constant; the discharge in the bypass was very stable. No trapping device has been installed on the
Baigts site. The downstream fish transfer is by a conduit 45 m long and 0.8 m wide, directing the fish
20 or so metres downstream of the dam. This bypass was used for the first study (2004-2005). For the
second study (2005-2006), the surface bypass was modified into a deep bypass by installing a semicylindrical steel mask in front of the surface bypass (Photo $1). The lower part of the steel mask,
located at the level of the bottom of the trashrack (- 7m) was open so that the flow to the bypass was
drawn off at some depth. Side openings 50 cm wide and 80 cm high (Photo $$) were made in the face
of the mask next to the grill so that eels swimming along the trashrack might enter them. The flow rate
was approximately the same as that of the surface bypass (2.2 m3.s-1) and the flow velocity at the inlet
to the bypass was around 1 m.s-1. Both bypasses were continuously in operation throughout each of the
two experiments.

Radiotracking and PIT tag detection

At Halsou, most of the radio-tagged eels had been trapped at the power station after their passage
into one of the bypasses. Some individual eels were only caught upstream of the dam by
electrofishing, enabling tagging of silver eels prior to the first catches. Potential migrants were
identified on the basis of their external measurements (Durif et al. (2000).
Transmitters were surgically implanted in the abdominal cavity of eels anaesthetised with clove oil
(Baras and Jeandrin, 1998). The trailing antennae transmitters used were an uncoded "ATS 10/28"

5
model (frequency: 48 to 49 MHz; length: 45 mm; diameter: 11 mm; weight: 8 g) with mortality
switches. The ratio of transmitter weight to the weight of the eels was equal or inferior to 2% (except
in 3 cases). An exit hole was made for the antenna with a hollow needle through the body wall 2 cm
behind the incision stitched with nylon thread.
Bottom bypass
Surface bypass

Figure 3 : General view of the Halsou (A) and Baigts (B) power plants and radiotracking zones
Displacement was monitored by manual (loop antenna) and automated radiotracking throughout
the study site. From 9 (in 1999) to 16 (in 2001) fixed ATS (2000B) and LOTEK (SRX_400)
dataloggers operating in the 48-50 MHz band were used for automated monitoring. Twelve zones
were defined (Figure 3) to be monitored with dataloggers linked to different types of antennas
(immersed short wire antennas, immersed long twin-lead antennas and aerial loop antennas). Each
monitoring zone was precisely calibrated (signal-gathering distance determined by adjusting the gain)
by manually moving a transmitter in the water.
All the radiotracked eels were released within hours of being tagged: 15 in 1999 (from 20 October
to 22 November); 25 in 2000 (from 10 October to 26 November); 30 in 2001 (from 9 November 2000
to 23 January 2002). To ensure the reliability of the automatic receiver recordings, the number of radio
frequencies memorised in the recorders had to be limited. Accordingly, the releases (1 to 8
individuals) were staggered according to the passages downstream the plant to maintain a limited
number of eels (maximum of 8 individuals) upstream of the plant. The characteristics of the radiotagged eels are given in table 1. The first 9 eels (1999) were released in the headrace 350 m upstream
of the plant; all others (65 individuals including 4 eels trapped after tagging and released a second time
in 2001) were released in the forebay, on the right bank near the spillway. After the first tests, it was
observed that most fish returned up the headrace immediately after being released and could therefore
not be observed in the forebay.
Table 1: Characteristics of radiotracked eels in Halsou and Baigts experiments: L =total length, HW = head
width
Site
Year Lmin Lmax Lmean HWmin HWmax HWmean
Nb indiv.
Nb indiv.
mm
mm
mm
mm
mm
mm
HW<30mm
HW>30mm
Halsou 1999
570
930
725
22
42
30
10
5
2000
550
950
699
22
44
27
18
7
2001
560
740
669
20
34
27
26
4
Baigts 2004
540
750
610
19
32
24
38
2
2005
500
990
646
16
46
27
31
8

At Baigts, because it was impossible to trap the eels at the plant, they were purchased from a
professional fisherman operating a downstream migration fishery on the outlet of a lake 60 km from
the site. The advantage of this solution over electric fishing is that the individuals are in active

6
migration while electric fishing leaves certain doubts about the maturity of the eels and their ability to
migrate. On the other hand, it is possible that the use of eels not coming from the water course where
the experiment is being conducted could induce some bias in the experiment.
We used the same transmitters as for the Halsou study and the same type of surgical implanting
method. However, anaesthesia was by electricity (galvanonarcosis) according to the method proposed
by Gosset (Gosset and Garaicochea, 1974; Gosset and Rives, 2005) with current ranging from 20 to 60
volts. This method makes it much easier to install the transmitter because the eel stays perfectly
immobile during the surgical operation while maintaining normal breathing allowing it to be kept in
this condition for a longer time if necessary. It immediately recovers its swimming ability as soon as
the electric current is removed.
To be sure of detecting the passage of fish into the bypass, we equipped the radio-tracked eels with
a PIT tag (TIRIS RI-TRP-RR2B, 32 mm long, 3 mm diameter, 0.8 g), inserted into the general cavity
at the same time as the radio transmitter.
Radio tracking was monitored in the same way as at Halsou with manual tracking of the fish
in the Gave, both upstream and downstream of the plant, and with automatic recording (ATS and
LOTEK equipment similar to that used at Halsou) on the Baigts site as well as in a plant located 5 km
downstream (Puyoo plant) to detect the continuation of migration after going through the Baigts plant.
Seventeen fixed recording stations were set up and connected to a set of aerial loop antennas
or immersed wire antennas to establish 12 reception zones (8 upstream and 4 downstream of the
device) as shown in figure $$.
Two PIT tag detection antennas (rectangular shape) each connected to a recorder were
installed in series (2.6 m intervals between two antennas) in the channel connecting the bypass to the
downstream side of the dam. Evaluation tests showed that the detection by each of the antennas was
95% and that detection by the set of two antennas came to 100%.
After marking, the eels were observed for at least 24 hours in fish baskets immersed in the
headpond then transported and released upstream of the device. In 2004-2005, 40 eels were released
2.8 km upstream of the plant, between 9 October and 11 January. In 2005-2006, 39 eels were released
partly (25 individuals) 7 km upstream of the plant and partly (14 individuals) 2.8 km upstream of the
plant during the period between 10 November and 2 February. In the same way as at Halsou, the
releases (2 to 4 individuals) were staggered so that there were never more than 8 eels upstream of the
plant at a given time. The biometric characteristics are given in table 1.
The main environmental parameters were recorded on site or obtained from environmental Offices:
temperature, conductivity, turbidity, atmospheric pressure, rain, daylight, river flow, water levels.

RESULTS

Migration rates
The migration rhythms were monitored at Halsou by trapping and on the two sites by radio
tracking. At Halsou, a total of 637 silver eels were captured in the trap after passing through the two
bypasses in 3 years (66 in 1999, 75 in 2000 and 496 in 2001). The vast majority of captures (90% to
98%) were at night and during twilight, i.e. between 6:00 pm and 8:30 am. Downstream migration
takes place in successive runs (for periods of 4 days maximum) between which there is very little
passage: 74% of captures were during 2 runs in 1999, 87% in 4 runs in 2000 and 95% in 3 runs in
2001 (Figure 3). In 2001, with hydrological conditions particularly favourable to downstream
migration through the headrace (an increase in flow in the Nive with no spillage over the dam), 436
individuals were caught in a single night, accounting for 88% of the captures that season. These
downstream migration peak figures tie in closely with the flow rate variations of the Nive
accompanied by turbidity increases and decreases.

7
These rates were confirmed by radio-tracking both in the Nive and in the Gave de Pau. It was
observed that most of the eels remain immobile during periods when there are few variations in flow,
turbidity and conductivity parameters and that the active upstream migration phases correspond to
periods when these parameters fluctuate (Durif et al., 2003; Subra et al., 2005). In the Gave de Pau,
the distances covered by the eels are significantly greater when the flow increases (ANOVA, p < 0.01)
and when the turbidity is higher than 10 NTU (ANOVA, p < 0,01).

General migrating behaviour and effects of the hydroelectric facilities

Halsou experiments
More than half (51%) of the 74 tagged eels released in the forebay and in the headrace during the 3
years, swam up the headrace 1hour to 16 days after release and remained in the Nive upstream of the
dam (Table 2). A portion of the eels (20%) swam up the headrace without exploring the water intake
trashrack. Another part (31%) explored the trashrack for a variable period of time (5 minutes to 16
days) before swimming upstream. They took up waiting positions (displacement of very small
amplitude) in the Nive, 2 to 3 km upstream of the dam for periods of between 2 days and 65 days.
They then migrated downstream, mainly over the dam (36/38), during environmental windows
corresponding to increased flow and turbidity, and reduced conductivity (Durif et al., 2003). The time
between release and migration activity depended only on these specific environmental conditions.
Only three eels (7%) which initially swam back up the headrace migrated back downstream through
the same channel. Two of them passed through the power plant and one swam back up the headrace a
second time before passing over the dam.
Less than half the released eels (49% or 36 individuals) passed through the plant where they took
one of the 5 passage ways: turbines, surface bypass, bottom bypass, upstream fishpass and plant
spillway.
Table 2 : Passage of radio tagged eels at the Halsou power plant
Released
Passed at the powerhouse *

Total

Swam through the powerhouse Bottom bypass

directly or after a stay in the Surface bypass
forebay **
Turbines
Forebay weir
Fishway
Total efficiency of the 2 bypasses
Swam upstream of the raceway

Dam
Abandoned or lost
Dead

1999
15
5 (33%)

2000
25
11 (44%)

2001
34
20 (59%)

Total
74
36 (49%)

4
0
1
0
0

5 to 9
1 to 3
1 to 3
0 to 1
0

6
2
8 to 9
3
0 to 1

15 to 19 (42% to 53%)
3 to 5 (8% to 14%)
10 to 13 (28% to 36%)
3 to 4 (8% to 11%)
0 to 1 (0% to 3%)

80%

72% to 80%

40%

56% to 64%

10 (66%)
0
0

13 (52%)
1 (4%)
0

5 (15%)
8 (40%)
1

28 (38%)
9 (12%)
1 (1.4%)

* : Percentage calculated according to the number of tagged fish

** : Percentage calculated according to the number fish that passed through the powerhouse
Furthermore, it was observed, by comparing tracked eels and trapping results that downstream
migration of ratio-tagged eels occurred at the same time as for the untagged eels.
Baigts experiment
After being released into the Gave, the behaviour of the fish is characterized by periods of
immobility that are sometimes long, especially after return to the water and periods of movement. In
2004, 50% of the eels (20 out of 40) began to move within 24 hours after being put back in the water.
80% swam upstream within the 5 days that followed and 90% within 10 days. Three individuals
remained immobile for more than 20 days, and one for almost 50 days. Some individuals (11) began
by swimming up the Gave de Pau over distances of between several tens (9 individuals) to several
hundreds of meters (2 individuals) before migrating downstream again. Upon reaching the plant at

8
Baigts, most of the eels (73%) pass through during their first incursion and while remaining right next
to the plant. The others come up from 2 to 4 times, swimming up a few hundred meters upstream of
the plant.

Behaviour in the forebay area and passages at the plant

Halsou experiment
Out of the 74 individuals radio tracked over the three years, 63 (85%) stayed in front of the water
inlet (the section at least once near the water inlet grill) while the others swam directly up into the
headrace then passed over the dam. Among the individuals that stayed in front of the inlet, 27
individuals (43%) swam up the headrace and most of them (25/27 93%) passed over the dam on their
downstream migration. 36 individuals (57%) stayed in front of the inlet and took one of the plant
passage ways.
The time spent in front of the water inlet for the individuals that then swam up into the headrace
varied enormously (between one minute and 14 days): 7 individuals (26%) stayed for less than 10
minutes, 14 (52%) less than 24 hours and 6 (22%) more than 24 hours. The stopping time of the
individuals that then passed through the plant is also very variable (30 seconds to 22 days). 7
individuals (19%) stayed for less than 10 minutes, 16 (42%) less than 24 hours and 14 (39%) for more
than 24 hours.
The plant passage ways are reported in table 2. In 2000, the particularly unfavourable hydrological
conditions (numerous spates) limited monitoring precision and uncertainties remain with respect to the
path taken by some individual. Accordingly the evaluation of the distribution of the passage channels
is affected by some inaccuracy. It can be seen that over the three years of the study (36 radiotracked
eels passing through the plant), the passages are in decreasing order through the bottom bypass (42%
to 53%), by the turbines (28% to 36%), by the surface bypass (8% to 14%) then over forebay spillway
(8% to 11%). There is some uncertainty about the possibility of the passage of one individual (0% to
3%) by the fishway. The combined efficiency of the two bypasses (proportion of eels passing through
either bypass compared to the total that passed through the plant) is included between 56% and 64%
for the three years considered together. It varies from one year to another (80% in 1999, 72% to 80%
in 2000 and 40% in 2001).

Figure 4 : Repartition of mean residence time and number of approaches for each area (see fig. 3)
of Halsou experiment
The behaviour in the forebay was generally characterized by alternating phases of exploration of
the forebay area and rest in areas where velocity gradients were very low, mainly the counter-current
in RB, or T1 when the turbine T1 was not operating. The number of approaches and time spent in the
different monitoring zones were highly variable for each eel and reveal no preferential movements.
The attendance in the various zones in terms of the number of approaches and the duration of presence
for all the fish and for each period of the survey is given in figure 4. Note that zones T1 (Turbine 1)
and RB are more frequented whereas the zone near the surface bypass (SS) is the least frequented.
During the three years, the zone in front of the surface bypass (SS) was used infrequently,
regardless of the hydrological conditions. On the other hand, while time spent near the bottom bypass
entrance (BS) was low, approaches were 4 to 6 times more frequent than those observed near the
surface bypass. It would therefore appear that the eels seek passage at the bottom rather than on the

9
surface.
No significant delay was observed during the transit through the discharge tower were eels are
entrained or swim up a 3-m water column.
Most of the passages (83%) take place at night whichever passage is used. It was also observed that
the passages were synchronized with the flow variations of the Nive.
Baigts experiment
The stopping time upstream of the Baigts plant (time elapsing between initial detection at one of the
upstream recording stations and effective crossing) vary between 1 minute and 7 days. In general the
times are short because 25% of the eels take less than 10 minutes to pass over the dam and 50% take
less than one hour.
The passage ways for the two years of experiment are given in table 3.
Table 3 : Passage of radio tagged eels at the Baigts power plant
Zones
2004
Surface bypass
N
%
Bypass
7
18%
Turbines
24
60%
Flood gates
5
13%
Vertical gate
1
3%
Upstream fishpass
3
8%
TOTAL
40

2005
Deep bypass
N
1
21
13
3
1
39

%
3%
54%
33%
8%
3%

During the two experiments, most of the eels passed through the turbines (respectively 60% and 54%
in 2004 and 2005). A large proportion passes through the flood gates (13% in 2004 and 33% in 2005).
This variability is due to the higher river flow in 2005 causing the gates to open more often.
The efficiency of the surface bypass in 2004 is 18% whereas it was only 3% for the bottom bypass in
2005.
Most of the passages at the plant take place at night: in 2004-2005, 88% (n = 35) of them are between
5 p.m. and 9 a.m.
The distribution of the standing times (total times obtained by adding the time spent to by each of the
eels in the various zones) in the various water inlet points (trashrack, bypass and flap gate) shows that
the bypass area is where they spend the least time.
Radio-tracking, carried out more accurately than at Halsou reveals the behaviour patterns in front of
the trashrack.
This behaviour is characterized either by direct passage through the trashrack (35% of individuals) or
by multiple incursions (2 to 28 per individual) before moving away from the trashrack or passing
through it. The time spent in front of the trashracks for each of the 218 incursions made by the
individuals varies enormously: 1 second to 2 hours and 37 minutes, but most of the incursions of very
short and 59.1% are no longer than 30 seconds and 91.7% last less than 3 minutes.
Most of the eels do not really search much in front of the grill area. In 68.8% of cases (n = 150), the
fish only frequents one of the five predefined areas, either by moving along the trashrack for not more
than 5 m to 13 m. 20.6% of the incursions resulted in movements in two to three different zones, that
is prospecting of the trashrack area not exceeding a distance of 20 m or so. For only 10.6% of the eels,
prospecting extends over more than four zones (prospecting between 25 m and 40 m along the
trashrack).
Longer incursions are related significantly (ANOVA, F = 23.7; p < 0.0001) to extensive prospecting of
the trashrack area, suggesting an active prospecting of the trshrack area rather than "being blocked" at
a set point of the trashrack.
For the eels prospecting several zones along the trashrack, trashrack prospecting is either in the
direction of the flow (upstream to downstream) all in the opposite direction. We see that 35.4 % (n =
25) of the movements are from upstream to downstream (directions A, B, C, D, E see Fig 3), 38.5%
from downstream to upstream (direction E, D, C, B, A) and that 26.2% of the eels move backwards
and forwards in front of the trashrack area.

10
Passage through the bars of the trashracks is significantly linked to the time spent prospecting in front
of the trashrack area. Actual passing only occurs after a relatively long period of prospecting in front
of the obstacle (ANOVA, F = 12.56; p = 0.0005). We see that only 4.1% of incursions last less than 10
seconds (3 out of 74) are confirmed by actual passing. This figure rises to 28.3% (17 out of 60) for
incursions lasting more than a minute.
60%

Passage
Not Passage

70%

40%

Frequency (%)

Frequency (%)

80%

Passage
Not passage

50%

30%
20%
10%

60%
50%
40%
30%
20%
10%

0%

90

80

10
0
>

Turbidity (NTU)

70

60

50

40

30

20

10

<

Velocity in front of the trashrack (m/s)

4
0,

0,
35

0,
3

0,
25

0,
2

0,
15

0,
1

0,
05

>

<

0,
05

0%

Figure 5 : Passage of radio-tracked eels through the Baigts trashrack according to velocity in front of
the trashrack and water turbidity
The passing through the trashrack appears to be influenced by the current velocity (figure 5). Indeed,
we see that velocity higher than values facilities passage through the trashrack (ANOVA, F = 35.99; p
< 0.0001). The permeability factor of the trashrack therefore changes from 5.5% (9 out of 163) for
current velocities of less than 0.2 m/s to 32.7% (18 out of 55) for higher velocities.
The crossing of the trashracks also appears to depend on how turbid the water is (figure 5). The
turbidity of the grill area increases with the turbidity (ANOVA, F = 32.03; p < 0.0001). For turbidity
values that are less than 10 NTU (covering 68.8% of incursions in front of the trashrack), the trashrack
passage factor is only 8%. This rate increases to 22.1% for turbidity values in excess of 10 NTU.
The width of the eel head (range 22 mm to 29.9 mm) has no real influence on passage through the
trashracks (ANOVA, F = 1.06; p = 0.3).
DISCUSSION

Evaluation and comparison of efficiency of bypasses

The original protocol of the Halsou study envisaged a comparison of the efficiency of the two
bypasses, the surface type and the bottom type, opened alternately every other day, using the
downstream migration capture technique for unmarked individuals and the radiotracking methods.
Both methods in fact came up against problems inherent in the nature of the eels downstream
migration and in the environmental conditions in which this migration occurred. Migration on the
Nive, as has been observed on other rivers (Voellestad et al. 1994, Haddering et al. 1999, Watene et
al. 2003, Durif et al. 2003, Haro et al.2002)) is characterized by short and infrequent peaks linked to
variations in the river flow. The peak distribution presents a problem for evaluation by either method,
seeing that the results depend on whether one of the bypasses is open or closed at the time of a
migratory peak. For the capture evaluation method this difficulty is confounded by a more severe
problem of trapping device clogging with the surface bypass than with the bottom bypass owing top
the fact that it resulted from drifting of floating leaves, carried down mainly by floodwaters periods
when downstream migration peaks occur. Because of this, the results coming from the capture method
could not be used for comparing the two bypass types. In the radiotracking method, the bias stemming
from the peaks persists and therefore the raw results for passage of radiotracked fish in the two devices
cannot be used. Consequently we have taken the combined efficiency of the two bypasses, which is
around 60 % over the three years studied (56 to 65%). Moreover, by considering the converging
elements of the most intense passages in the bottom bypass evaluated either by capture or by

11
radiotracking, as well as the most intense frequentation by the radiotracked eels of the zone near the
bottom bypass, the Authors concluded that the efficiency of the bottom bypass was greater. The
bottom bypass would have an efficiency 3 to 4 times as high as the surface type.
This efficiency can be variable with the year (40 to 80% during the three years in question), although
it is not possible to demonstrate reliably the influencing factors. However, 2001, when the efficiency
of the bypasses was lowest, stands out from the previous two years by extremely low flow on the
River Nive. Finally, the combined efficiency of the two bypass devices is very similar to that obtained
with just the surface bypass for salmonid juveniles: an average of 56% over the period 1995-1999
(Gosset et al., 2005).
At Baigts the problem of experimental bias does not exist because the Authors used the same bypass
throughout each of the experiments. The efficiency of the surface bypass tested in 2004 (18 to 23%
depending on whether the passages through the bypass are considered in relation to the total number of
passages in the system or in relation only to the passages in the turbines) appears markedly lower than
what is observed at Halsou with the two bypasses combined. It is moreover strikingly lower than that
obtained previously for salmon (92.5%). That can be explained by the larger size of the water intake
trashrack (40 m long instead of 18 m at Halsou) which would raise the probability of passage through
the trashrack and by the greater depth of the upper pond (12 m instead of 3.5 m) which would make
perception of the surface bypass more difficult. A non-negligible proportion of eels find their way to
this surface bypass, which confirms the fact that, as has been brought to evidence by several authors
(Mc Grath et al., 2003; Watene et al., 2003; Haro et al., 2000), the eels going downstream explore the
whole of the water mass when they come close to an obstacle.
It may seem surprising that the bottom bypass tested at Baigts is markedly less efficient (3%) than the
surface bypass, which contradicts the results obtained at Halsou. Two causes can be invoked. The first
is that the entrance into the bypass is not located at the bottom of the upper pond as at Halsou but at
the bottom of the trashrack situated 3 m above the floor of the upper pond. The second is that the eels
behaviour during experimentation in 2005 was different from that in 2004. According to the first
results of the study (undergoing analysis), the eels appear to have done less prospecting or sweeping
of the trashrack in 2005 than they had in 2004 and had passed through the trashrack more quickly.
That could result from the greater proportion of individuals that arrived at the Baigts site during the
spates of 2005 than in 2004. Evidence was found in 2004 that the individuals that most explored the
trashrack without passing through had arrived on the site during a period without variation of the river
flow. These individuals showed an exploratory behaviour with a repulsion against going through the
trashracks that the active ones arriving during spates did not have.
If the two sites are compared, the Halsou site appears highly favourable to ensure good efficiency of
the bottom bypass owing to its small extent (and notably to the shallowness of the water in front of the
water inlet 3.5 m), its configuration (trashrack angled to the axis of the headrace), the location of the
bottom bypass in a cul de sac at the downstream end of the inclined trashrack. Furthermore, the
position of this bypass, on a level with the bottom immediately next to a zone of counter-current with
considerable sedimentation and serving as a resting area for the eels, also appears highly favourable
for this species.
In any case, the position of the bypass and the hydrodynamic characteristics in the area around its
entrance appear to be important factors, as has been brought into evidence for salmon. These will have
to be optimised for the eels.
Exploratory behaviour in the forebay and effect of the trashrack
At Halsou, eels demonstrated foraging behaviour in the forebay. Total time spent in this area varies
in a wide range (30 sec to 14 days), and the number of approaches and time spent in the different
zones monitored were extremely variable. It is very difficult to estimate the exact time spent searching
for a bypass, as some areas (in front of the trashrack and near the river bottom) can occasionally be
favourable for resting, when hydrological conditions are favourable (low or no turbined discharge).
Resting periods near the spillway can be explained by the virtual absence of current in this area, as
well as by the presence of a large sand and mud tumulus created by counter-currents. Acoustic

12
tracking of 10 eels (Begout-Anras, 2001) confirmed these results and showed that most displacement
was near the bottom. It is probable that, particularly at night, there is movement also between the
bottom and the surface as observed on other sites. This parking behaviour is partly due to the
experimental protocol of releasing the radio tracked eels in front of the water inlet. The behaviour is
different at Baigts where the eels arrive during active migration. It results in faster exploration and
crossing of the water intake (50% in less than an hour). Staying for long periods (up to seven days) is
the result of a minority of individuals all arriving when there are no significant flow rate variations in
the river. Like at Halsou, these individuals appear to carry out "exploration" and are capable of
swimming up the water course (up to 3 km at Halsou and 1 km at Baigts) to take up a new position in
the favourable zones.
The deterrent effect of the water inake trashrack with 3 cm bar spacing (through which almost all
the tested individuals can physically pass) is undeniable because on both sites, a very small minority
of individuals passed directly through the trashrack without any evident pausing. This deterrent effect
is evidenced by incursions along the trashrack followed by returns upstream or by "sweeping" across
the trashrack. In most cases, there is rather evidence of multiple incursions of short duration along the
trashrack.
This repulsion with respect to the trashrack appears to be small compared to what is observed for
the salmon smolts in that a large fraction of the eels (69% at Baigts) only prospect a small portion of
the trashracks (distance 5 to 13 m) before swimming through them. It is probable that as observed in
the laboratory (Amaral et al., 2003; Adam and Schwevers, 1997), it is more the contact with the
trashracks than a hydrodynamic effect (as is the case for salmon smolt) that generates this repulsive
effect.
The depth of trashrack exploration could not be determined because this information is not
accessible using the transmitters that we used. The fact that there are only very brief incursions in the
area near the surface bypass suggests that exploration is rather on depth, but that considering the small
size of these detection areas it is nevertherless difficult to conclude.
Most passages through the Baigts trashrack are in the downstream portion, near the upstream
migration bypass. This could be caused by the guiding effect of the trashrack (combined with the
repelling effect of the trashrack and the flow tangent to the trashrack) but also to the fact that a fair
proportion (30%) of the individuals appears to pass at a depth under the water intake trashrack and
arrive directly in the area downstream of the trashrack. It is also noteworthy that prospecting along the
trashrack for a high proportion of individuals (38.5%) is from downstream to upstream i.e. against the
flow of the current. This makes it difficult to reach any conclusion about the guiding of the eels into
the bypass by tangential flow along the trashrack.
The importance of the water velocity across the trashrack plane is not clear and appears to be
contradictory according to the site: at Halsou, the most intense passages across the trashrack took
place when the turbine discharges were low (and therefore at low water velocities) whereas at Baigts
the opposite was observed because the passages through the trashrack went up as soon as water
velocities reached 0.2 m/s. The fact that the repulsion regarding the trashrack depends also on the
water turbidity (repulsive effect higher when turbidity was lower at Baigts), a factor linked to the river
discharge rate and therefore to water velocity across the trashrack, prompts the question of a possible
direct influence of one of these two parameters and of the way it acts. It could be that these two
parameters are only a reflection of a different migratory behaviour associated in turn with the
environmental conditions in the rivers.
Finally, contrary to what has been observed in smolts, the repulsion effect of the grill does not seem to
be influenced by the size of the individuals (head width) in that there is no significant difference in
size between the individuals that have passed through the grill and those which have not. That agrees
with the laboratory observations (Adam and Schwevers, 1997) which showed that the individuals try
to pass through the trashrack by force and confirms the idea that it is the trashrack itself that induces

13
a small repulsive effect on eel behaviour rather than a hydrodynamic or visual factor, as is the case for
smolts.
To prevent the passage through the trashrack by physical screening, a bar spacing smaller than the
eels head width is necessary. This spacing depends on the eel size distribution which can be different
from a river to another. The eels trapped during the Halsou experiments have the following
characteristics : 20% of them have a head width larger than 30 mm, 47% larger than 25 mm and 12%
larger than 20 mm. So, a trashrack with a bar spacing of 20 mm is able to prevent the passage through
the trashrack of 88% of the eel population.
CONCLUSION
These experiments recently undertaken to test surface and bottom bypasses have shown that their
efficiency is poorer than that obtained with salmon, essentially because eels are less repelled by the
water intake trashracks. Improving such devices will necessarily involve using finer screens which
cannot be crossed by the majority of the eel population. Installing such screens is, however, impossible
on most existing plants without enlarging the screen surface or lowering the turbine discharge: the
flow velocities, which are generally too high, pose the risk of eel mortality due to impingement and
create operating difficulties for the plant due to the loss of head and the challenge of adequately
cleaning the screens.
The experiments should be pursued in various directions: optimum spacing of trashrack bars,
characteristics of the bypass, testing of behavioral devices, predicting the periods of runs so as to
perform the necessary maneuvers at the plant to allow passage of migrants.

14
REFERENCES
ASCE. 1995. Fish passage and protection. In: Guidelines for design of intakes for hydroelectric plants, American Society of
Civil Ingineers, New York, N.Y., 469-499.
Adam B., Schwevers D.U., 1997. Behavioral surveys of eels (Anguilla anguilla) migrating downstream under laboratory
conditions. Rapp. Institute of Applied Ecology.
Amaral S.V., Winchell F.C., Mc Mahon B.J., Dixon D.A., 2003. Evaluation of angled bar Racks and louvers for guiding sicer
phase American eels. In: Biology, Management, and Protection of Catadromous Eels, Dixon DA (ed.). American
Fisheries Society, Symposium 33: Bethesda, Maryland, 367-376.
Baras E, Jeandrin D. 1998. Evaluation of surgery procedures for tagging eel Anguilla anguilla (L.) with telemetry
transmitters. Hydrobiologia, 371/372: 107-111.
Bgout Anras ML. 2001. Etude comportementale de l'anguille de dvalasion devant les exutoires par pistage sonique. Rapp.
22C07986, CREMA L'Houmeau, 9 p.

Chanseau M., Larinier M., Travade F., 1999. Efficacit dun exutoire de dvalaison pour smolts de saumon atlantique et
comportement des poissons au niveau de lamnagement hydrolectrique de Bedous sur le Gave dAspe tudis par
la technique de marquage-recapture et radiotlmtrie. Bull. Fr. Pche Piscic. 353/354: 99-120.
Croze O., Chanseau M., Larinier M., 1999. Efficacit dun exutoire de dvalaison pour smolts de saumon atlantique et
comportement des poissons au niveau de lamnagement hydrolectrique de Camon sur la Garonne. Bull. Fr. Pche
Piscic. 353/354: 121-140.
Croze O., Breinig T., Pallo S., Larinier M., 2001. Etude de lefficacit de trois dispositifs de dvalaison pour smolts de
saumon atlantique. Usines de Guilhot, Las Rives et Campagna (Arige). Rapport GHAAPPE RA 01-07. 73 p.
Desrochers D., 1994. Suivi de la migration de languille dAmrique au complexe Beauharnois. Monitoring of American eel
at Beauharnois generating station. Report, Vice prsidence Environnement, Hydro Qubec, 107 p.
Durif C, Elie P, Dufour S, Marchelidon J, Vidal B. 2000. Analysis of morphological and physiological parameters during the
silvering process of the European eel (Anguilla anguilla) in the lake of Grand-Lieu (France). Cybium 24: 63-74.
Durif C. 2003. The downstream migration of the European eel Anguilla anguilla: Characterization of migrating silver eels,
migration phenomenon, and obstacle avoidance. PhD Thesis, University Paul Sabatier, 347 p.
Durif C, Gosset C, Rives J, Travade F, Elie P. 2003. Behavioral study of downstream migrating eels by radio-telemetry at a
small hydroelectric power plant. In: Biology, Management, and Protection of Catadromous Eels, Dixon DA (ed.).
American Fisheries Society, Symposium 33: Bethesda, Maryland; 343-356.
Euston ET, Royer DD, Simmons CL. 1997. Relationship of emigration of silver american eels (Anguilla rostrata) to
environmental variables at a low head hydro station. Proceedings of the International Conference on Hydropower,
Atlanta, Georgia, 549-558.
EPRI. 2001. Review and documentation of research and technologies on passage and protection of downstream migrating
catadromous eels at hydroelectric facilities. Report, Electric Power Research Institute (EPRI), Palo Alto, CA.
1000730, 73 p.
Field K., Ghl , 2006.
Gleeson L. 1997. Use of electric eel passage. Presented at the Fish Passage Workshop (sponsored by Alden Research
Laboratory, Inc., Conte Anadramous Fish Researh Center, Electric Power Research Institute, and Wisconsin
Electric), Milwaukee, WI (May 6-8, 1997).
Gosset C., Garaicochea C., 1974. Cuve galvanonarcotique pour la mensuration des poissons. Bull. fr. Pche. Piscic., 225 :
72-76.
Gosset C, Travade F. 1999. Etude de dispositifs d'aide la migration de dvalaison des Salmonidae: barrires
comportementales. Cybium 23: 45-66.

15
Gosset C., Rives J., 2005. Anesthsie et procdures chirurgicales pour limplantation de radio metteurs dans la cavit
ventrale de truites communes adultes (Salmo trutta). Bull. fr. Pche. Piscic, (2005) 374 : 21-34.
Gosset C, Travade F, Durif C, Rives J, Elie. 2005.Tests of two types of bypass for downstream migration of eels at a small
hydroeectric power plant. River Res. Applic.21 : 1095-1105 (2005).
Hadderingh RH. 1982. Experimental reduction of fish impingement by artificial illumination at Bergum Power Station. Int.
revue Ges. Hydrobiol. 67(6) : 887-900.
Haddering RH, Van Der Stoep JW, Habraken JMPM. 1992. Deflecting eels from water inlets of power stations with lights.
Irish Fisheries Investigations Serie A (Freshwater). 36: 78-87.
Hadderingh RH, Baker HD. 1998. Fish mortality due to passage through hydroelectric power stations on the Meuse and
Vecht rivers. In: Fish Migration and Fish Bypasses, Jungwirth M, Schmutz S, Weis S (eds.). Fishing News Books:
Oxford; 315-328.
Hadderingh RH, Van Aerssen GHFM, De Beijer RFLJ, Van der velde G. 1999. Reaction of silver eels to artificial light
sources and water currents: an experimental deflection study. Regulated Rivers - Research & Management 15: 365371.
Haro A, Castro-Santos T, Boube J. 2000. Behavior and passage of silver-phase American eels, Anguilla rostrata (LeSueur)
at a small hydroelectric facility. Dana 12: 33-42.
Haro A, Castro-Santos T, Mclaughlin L, Whalen K, Wippelhauser G. 2002. Simulation of migration and passage of american
eels at riverine barriers. In: Biology, Management, and Protection of Catadromous Eels, Dixon DA (ed.). American
Fisheries Society, Symposium 33: Bethesda, Maryland, 357-365
Larinier M, Dartiguelongue J. 1989. La circulation des poissons migrateurs: le transit travers les turbines des installations
hydrolectriques. Bulletin Franais de la Pche et de la Pisciculture 312: 1-94.
Larinier M, Travade F. 1996. Smolt behavior and downstream fish bypass efficiency at small hydroelectric plants in France.
2me Symposium international sur l'hydraulique et les habitats, Ecohydraulique 2000, Association Internationale de
Recherches Hydrauliques. Qubec, vol B, 891-902.
Larinier M, Travade F. 1999. The development and evaluation of downstrean bypasses for juveniles salmonids at small
hydroelectric plants in France. In : Innovations in Fish Passage Technology, Dixon DA (ed.). American Fisheries
Society, Bethesda, Maryland, 25-42.
Larinier M, Travade F. 2002. Downstream migration: problems and facilities. Bulletin Franais de la Pche et de la
Pisciculture 364: 181-207.
Monten E. 1985. Fish and turbines. Fish injury during passage through power station turbines. Vattenfall, Stocklom, 111 p.
Nilo P, Fortin R. 2001. Synthse des connaissances et tablissement d'une programmation de recherche sur l'anguille
d'Amrique (Anguilla rostrata). Universit du Qubec Montral, Dpartement des Sciences biologiques pour la
Socit de la faune et des parcs du Qubec, Direction de la recherche sur la faune, Qubec, 298 p.
Sand O, Enger PS, Karlsen HE, Knudsen F, Kvernstuen T. 2000. Avoidance responses to infrasound. In : Downstream
migrating european silver eels, Anguilla anguilla. Environmental Biology of fishes 57 : 327-336.
Subra S., Gomes P., Vighetti S., Thellier P., Larinier M., Travade F., 2005. Etude de dispositifs de dvalaison pour languille
argente. Comportement de languille et test dun dispositif de dvalaison lusine hydrolectrique de Baigts de
Barn (Gave de Pau 64). Rapport EDF R&D HP-76/05/025/A. Rapport GHAAPPE RA-05-03. 89 p. + annexes.
Travade F, Larinier M. 1992. La migration de dvalaison: problmes et dispositifs. Bulletin Franais de la Pche et de la
Pisciculture 326/327: 165-176.
Travade F., Gouyou C., De Faveri N., 1999. Efficacit dun exutoire de dvalaison et dune barrire acoustique pour les
smolts de saumon atlantique (Salmo salar L.) lamnagement hydrolectrique de St Cricq sur le Gave dOssau
Bull. Fr. Pche Piscic. 353/354: 157-180.
Travade F, Larinier M. 2006. French experience with downstream migration devices. International DWA symposium on
water resources management, Berlin 3-7 April 2006.

16
Voellestad LA, Jonsson B, Hvidsten NA, Naesje TF. 1994. Experimental test of environmental factors influencing the
seaward migration of European silver eels. Journal of Fish Biology 45: 641-651.
Watene EM, Boube JAT, Haro A. 2003. Downstream movement of mature eels in a hydroelectric reservoir in New Zealand.
In: Biology, Management, and Protection of Catadromous Eels, Dixon DA (ed.). American Fisheries Society Symposium
Series 3: Bethesda, Maryland, USA; 295-305.