ArticlePDF Available

Overwintering behaviour of yellow-stage European eel (Anguilla anguilla) in a natural marine fjord system


Abstract and Figures

Like many animals, northern temperate eel can enter a hibernation-like state and become dormant during the winter. Knowledge of overwintering behaviour in eel is sparse and mainly based on anecdotal observations and a few experimental studies on thermal tolerance. We studied European eel (Anguilla anguilla) overwintering behaviour in a Skagerrak fjord in Southern Norway, during three consecutive years, using an array of acoustic receivers and acoustic tags with depth and temperature sensors. We obtained results from 55 yellow eel, of which 19 were studied for one winter, 35 for two winters and one for three winters. Dormancy was inferred to begin in September for the earliest individuals and lasted until May for the last, with the majority of eel dormant from at least late October–November until mid-April. The timing of dormancy was mainly related to photoperiod and less to temperature. More than 50% of eel became dormant when day length was <9 hours and became active when day length was >14 hours. Approximately 10% of eel remained active during the winter and 31% of eel changed their pattern between consecutive years. Some dormant individuals exhibited activity periods that interrupted their dormancy. Eel in the outer fjord nearer the open sea became dormant before eel in the inner more freshwater part of the fjord, and were dormant longer.
Content may be subject to copyright.
Estuarine, Coastal and Shelf Science 276 (2022) 108016
Available online 11 August 2022
0272-7714/© 2022 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (
Overwintering behaviour of yellow-stage European eel (Anguilla anguilla) in
a natural marine fjord system
Mehis Rohtla
, Even Moland
, Anne Berit Skiftesvik
, Eva B. Thorstad
, Sebastian Bosgraaf
Esben M. Olsen
, Howard I. Browman
, Caroline M.F. Durif
Institute of Marine Research, Ecosystem Acoustics Group, Austevoll Research Station, Sauganeset 16, Storebø, Norway
Estonian Marine Institute, University of Tartu, Tartu, Estonia
Institute of Marine Research, Flødevigen Research Station, His, Norway
Norwegian Institute for Nature Research, Trondheim, Norway
Acoustic telemetry
North sea
Winter ecology
Like many animals, northern temperate eel can enter a hibernation-like state and become dormant during the
winter. Knowledge of overwintering behaviour in eel is sparse and mainly based on anecdotal observations and a
few experimental studies on thermal tolerance. We studied European eel (Anguilla anguilla) overwintering
behaviour in a Skagerrak fjord in Southern Norway, during three consecutive years, using an array of acoustic
receivers and acoustic tags with depth and temperature sensors. We obtained results from 55 yellow eel, of which
19 were studied for one winter, 35 for two winters and one for three winters. Dormancy was inferred to begin in
September for the earliest individuals and lasted until May for the last, with the majority of eel dormant from at
least late OctoberNovember until mid-April. The timing of dormancy was mainly related to photoperiod and less
to temperature. More than 50% of eel became dormant when day length was <9 h and became active when day
length was >14 h. Approximately 10% of eel remained active during the winter and 31% of eel changed their
pattern between consecutive years. Some dormant individuals exhibited activity periods that interrupted their
dormancy. Eel in the outer fjord nearer the open sea became dormant before eel in the inner more freshwater
part of the fjord, and were dormant longer.
1. Introduction
The winter season poses several challenges for aquatic animals, such
as low light, short days (at high latitudes), low prey abundance, low
temperature and, in some systems, low oxygen (Shuter et al., 2012).
Some species are adapted to these conditions and continue acquiring
energy during the winter. Others subsist mainly on energy stores and
reduce their activity (Shuter et al., 2012). Foraging during the winter is
more common in juvenile than adult sh because they have lower en-
ergy stores and a higher metabolic rate (Shuter and Post, 1990; Ultsch,
1989). Winter foraging is also more common at high latitudes where
winters are long, and the risk of starvation is higher. In temperate en-
vironments, it is common for sh to enter a period of dormancy: a
hibernation-like state, in which organisms are inactive, cease feeding,
and have a low metabolic rate (Speers-Roesch et al., 2018; Ultsch,
1989). Fish that exhibit dormancy take shelter, minimize movements,
and rely on lipid reserves accumulated during the previous feeding
season (Shuter et al., 2012; Reeve et al., 2022). Overwintering may
involve different levels of activity, from sleep through quiet wakefulness
to alert, and active (Crawshaw, 1984). Some species exhibit facultative
dormancy while, for others, dormancy is obligatory.
The European eel (Anguilla anguilla) is widely distributed from north
Africa and the Mediterranean to northern Norway. The species is
tolerant of large differences in temperature. Eel are catadromous, but
some individuals skip the freshwater phase and reside in marine coastal
habitats or brackish water lagoons and estuaries (Tesch, 2003; Daverat
et al., 2006). Eel reduce their activity as temperature decreases during
the winter (Bertin, 1951; Dannevig, 1945; Tomie et al., 2017; West-
erberg and Sj¨
oberg, 2015). Apart from a few experimental studies on eel
thermal tolerance (Dannevig, 1945; Sadler, 1979; Walsh et al., 1983),
observations of winter dormancy in the wild are anecdotal. Eel are rarely
observed during the winter. It is assumed that they bury themselves in
the sediment or hide in crevices (Bertin, 1951; Tesch, 2003). Most of the
commercial catches during the fall are silver eel, which argues that they
* Corresponding author. Institute of Marine Research, Austevoll Research Station, 5392, Storebø, Norway.
E-mail address: (C.M.F. Durif).
Contents lists available at ScienceDirect
Estuarine, Coastal and Shelf Science
journal homepage:
Received 11 February 2022; Received in revised form 29 July 2022; Accepted 5 August 2022
Estuarine, Coastal and Shelf Science 276 (2022) 108016
are active during this period of the year whereas other life stages are less
so (Durif and Skiftesvik, 2019; Durif and Elie, 2008). A study in the
freshwater Lake M¨
alaren in Sweden conrmed that most eel reduced
their activity during the winter (Westerberg and Sj¨
oberg, 2015), but it is
unknown whether eel residing in the marine environment also display
this behaviour or migrate to freshwater for overwintering as do some
American eel (Anguilla rostrata) (Thibault et al., 2007). The mechanisms
that trigger the initiation and termination of dormancy are poorly
known, but are hypothesised to be related to temperature, photoperiod,
food availability and body condition (Andrews et al., 2019; van Deurs
et al., 2010; Winslade, 1974).
The objective of this study was to describe the general overwintering
patterns of eel residing in a marine environment and to evaluate the
inuence of temperature and day length on the timing of dormancy. To
this end, we used data from yellow-stage European eel tagged with
acoustic transmitters whose movements were recorded by an array of
stationary receivers extending from freshwater to saline environments in
a fjord system in southern Norway.
2. Materials and methods
2.1. Study area
Sandnesfjord together with Songevannet and Nævestadfjord form a
~15 km long coastal fjord system located in southern Norway (Fig. 1). A
1.5 km long and 3 m deep channel, Lagestraumen, physically separates
these two areas, which display contrasting salinities (Tjomsland and
Kroglund, 2010). Driven by tidal inuences and signicant freshwater
input from the Storelva River and other sources, this fjord has a highly
dynamic and vertically layered salinity prole, with average salinities
Fig. 1. Study area in Sandnesfjord, Southern Norway. Deployed receivers are numbered from 1 to 26. Different fjord zones (IV) are identied by blue gates. (For
interpretation of the references to colour in this gure legend, the reader is referred to the Web version of this article.)
M. Rohtla et al.
Estuarine, Coastal and Shelf Science 276 (2022) 108016
increasing from Songevannet to the outer reaches of the fjord system
(Tjomsland and Kroglund, 2010; Kroglund et al., 2011).
Freshwater is only found in the surface waters of Songevannet and
Nævestadfjord, and the extent of it varies seasonally. The study area was
divided into ve zones based on salinity and fjord topography (Fig. 1):
Zone I (Songevannet) layered salinity marine water zone with fresh
water occurring in the surface layers; Zone II (Nævestadfjord) layered
salinity marine water zone with freshwater occurring near the surface
only seasonally; Zone III layered salinity marine water zone with direct
brackish water inow from Zone II through the Lagestraumen tidal
stream and a tributary fjord in the north; Zone IV - layered salinity
marine water zone with no direct brackish water inow; Zone V -
layered salinity marine water zone, with opening to the ocean. Flowing
between Zones II and III is the Lagestraumentidal stream where the
currents turn daily with the tides during periods with limited freshwater
discharge from Storelva River. Depth varies, often on small spatial
scales, throughout the fjord system with densely vegetated shallow
habitats situated close to trenches down to 70 m deep. Eel are abundant
throughout the fjord system, usually associated with the bottom of the
fjord as they are caught with fyke nets (Durif and Skiftesvik, 2019).
2.2. Receiver array
An array of 14 acoustic telemetry receivers (VR2W, Innovasea Sys-
tems Inc., US), which were previously used for studying Atlantic cod
(Gadus morhua) movements (see Kristensen et al., 2021), was used for
this study from August 2018 to August 2021 (Fig. 1). Twelve additional
receivers were deployed in September 2019 to increase the coverage
(Fig. 1). The receivers were deployed along the salinity gradient from
Storelva River to the open sea and attached to moorings at about 2.5 m
from the surface with the hydrophone facing down. Data were down-
loaded and receiver batteries replaced in early August each year.
2.3. Fish sampling and tagging
Eel (N =100) were tagged in 2018 and 2019 on three occasions
throughout the fjord system (Table 1). The sh were captured by an
experienced commercial eel sher using fyke nets. Eel were anaes-
thetized immediately after capture using a 1:1 mixture of clove oil and
ethanol (40 mL/L). Eel were measured (total length, pectoral n length,
and eye diameters) to determine sex and silver stage (Durif et al., 2005,
2009). All tagged eel were female (body length >500 mm) and all were
classied to be yellow eel, except one individual that was at the silver
Acoustic transmitters equipped with temperature and pressure sen-
sors (V9TP or V13TP, Innovasea Systems Inc., US) were implanted
through a 10 mm surgical incision into the intraperitoneal cavity of each
eel (Baras and Jeandrain, 1998; Durif et al., 2003; Thorstad et al., 2013).
The incision was closed with 23 sutures (Novosyn braided, coated,
absorbable thread). The surgical procedure took less than 5 min per sh.
All tagged eel were released at the site of capture within 10 min after
they recovered from anaesthesia. Transmitters emitted a signal at
110250 s intervals at a frequency of 69 kHz. The tags were cylindrical,
the V9TP model with diameter of 9 mm, length of 44 mm, mass of 63 g
(3.5 g in water) and estimated lifetime of 675 days, and the V13TP
model with diameter of 13 mm, length of 48 mm, mass of 13 g (6.5 g in
water) and estimated lifetime of 1484 days. Temperature measurement
accuracy and resolution given by the manufacturer for both transmitter
models was ±0.5 C and 0.1 C, respectively. Depth measurement ac-
curacy and resolution given by the manufacturer for V9TP and V13TP
transmitters at room temperature were ±1 m and 0.3 m, and ±3.4 m
and 0.3 m, respectively. Maximum potential depth that could be
recorded by each of the transmitter models was 68 m. Negative depth
values were occasionally recorded in shallow areas and these were
replaced with 0.01 m.
2.4. Determination of dormancy
We opted for a broad denition of dormancy, inactive for a period of
at least seven days. Our purpose was to describe a general pattern for
overwintering behaviour of eel in the marine environment which
included dormant periods, disappearances (from the receiver array) and
active periods. The actual physiological state of eel was not known. The
period from September 1st to May 1st was analysed. Eel activity was
assessed from logged detections of transmitter depths (i.e., their vertical
movements and the receiver where they were detected, following the
approach taken by Villegas-Rios et al. (2020)). Dormant, disappeared
and active eel were determined using the following criteria:
Eel were considered dormant when detected at the same depth (±1
m for V9TP transmitters and ±3.4 m for V13TP transmitters), by the
same receiver for at least seven days (Fig. 2a, b, e), (following from
the criteria for dormancy applied by Westerberg and Sj¨
Some eel remained undetected for several weeks. If such disap-
pearances occurred within the inactive period of nearby eel, and
reappearance occurred at the same receiver, we interpreted these
disappearances to be episodes of dormancy, assuming that they had
disappeared because they had found a shelter which blocked the
transmission of the acoustic signals from the tag.
Eel were dened as active when detected at different depths (Fig. 2)
(e.g. in neither of the above categories).
2.5. Data analysis
We analysed dormant and disappeared eel separately and together.
Eel with ambiguous behaviour - brief periods of activity in between
episodes of dormancy or disappearances - were removed from the ana-
lyses (n =7). Dormancy and disappearance episodes are identied in all
the Figures. To identify differences in overwintering patterns within the
fjord, we applied a Principal Component Analysis (PCA) on the
following variables: total eel length, Julian day (when dormancy started
and ended), day length (in hours, when dormancy started and ended), T
(temperature when dormancy started and ended). Plots were labelled
according to the fjord zone where eel overwintered and according to
disappearance/dormancy assignment.
Logistic regression was used to explore whether the start and end of
dormancy was associated with specic environmental factors. The
probability that an eel was dormant was modelled according to water
temperature (measured by the transmitters) and day length (calculated
based on latitude and date). Data from all three years were combined.
Dormancy periods were considered as independent observations since
the same eel could display different behaviours between years. The
probability of an eel waking from dormancy was also modelled using the
same explanatory variables. Duration of dormancy was compared be-
tween the inner part (Zones I-II) and the outer part of the fjord (Zone III-
V) using a linear model. Final models were validated using standard
procedures (Zuur et al., 2010). Data exploration, visualization and an-
alyses were done in R version 4.0.3 (Team, 2021). In the gures, the
timing of dormancy is depicted using the number of days after August
Table 1
Overview of eel (Anguilla anguilla) tagged in Sandnesfjord, Norway. n =sample
size, TL =total length, SD =standard deviation, FII =yellow eel growth stage,
FIII =premigration stage, and FV =migration stage (according to Durif et al.,
2005). Proportion of tagged eel detected by the receiver array is also given.
Release date n TL (mm) ±SD Stage Proportion of detected eel
9-Aug-18 29 650 ±70 FII, FIII 72%
11-Oct-18 22 650 ±50 FII, FIII 100%
17-Sep-19 49 640 ±70 FII, FIII, FV 94%
Total 100 650 ±60 FII, FIII, FV 89%
M. Rohtla et al.
Estuarine, Coastal and Shelf Science 276 (2022) 108016
1st rather than Julian days to avoid discontinuous axes between
December and January. Maps were made using Manifold System 8.0
using Shape les downloaded from the NVE database (Norwegian Water
Resources and Energy Directorate).
3. Results
3.1. Detection summary
Some eel turned silver during the study period. This was inferred
from their abrupt change in behaviour: diel vertical migrations and
subsequent escapement from the fjord. These data were excluded. A
total of 1 028 318 (temperature) and 1 029 780 (depth) eel detections,
were recorded by 25 receivers during the study period. No eel trans-
mitters were detected at the innermost freshwater site or in the outer-
most part of the fjord (receiver 1 and zone V, Fig. 1). Nine of the 100
tagged eel were never detected by any of the receivers. Two eel were
removed from the dataset because they had either expelled the trans-
mitter or died 12 months after tagging (the transmitter was detected at
constant depth). Eel with <10 detections during the study period were
excluded. This resulted in a nal sample of 55 eel representing 92 in-
dividual overwintering/study periods across the three years.
3.2. Behaviour of eel
There was no detectable effect of the tagging procedure on eel
behavior; all showed motor activity. Some eel resumed swimming right
after tagging and release, whereas some were only briey detected and
then disappeared (Fig. 2). Out of 92 overwintering periods, 11% were
active (six eel) that is, these eel maintained activity throughout the
winter (Figs. 2 and 3). The rest of the periods (89%) corresponded to
dormancy (61%), or disappearances (39%). Some eel displayed spurts of
activity between episodes of dormancy. These temporary active periods
lasted between one day to a few weeks and included vertical movements
of up to 56 m. Seven eel displaying these ambiguous behaviours were
not included in the statistical analysis.
Most eel (64%) were observed for two winters (one eel, #15175, for
three winters, Fig. 3). However, they did not necessarily display the
same behaviour during those two years (Figs. 2 and 3). For instance,
some eel were dormant during the rst winter, but were active during
the second winter, and vice-versa (Fig. 2d). Of the 36 eel with data on
two or three consecutive winters, 69% displayed the same behaviour
during consecutive years.
During dormancy, eel were usually stationary at depths between two
and 10 m (96.5% of detections at <10 m), which corresponded
approximately to the same range as during the rest of the year (99.8% of
detections at <10 m). However, four dormant individuals were regis-
tered at 15, 19, 51 and 53 m.
Fig. 2. Examples of depths (black circles) experienced by individual yellow eel (Anguilla anguilla) tagged with acoustic transmitters. a) deep-water dormancy episode,
which was preceded by periods of disappearance and activity; b) dormancy during two consecutive years with decrease in variation of depth use; c) disappearance
from the receivers during the two consecutive winter periods; d) constant activity during the rst winter period, but not during the second; e) dormancy and a
preceding high-amplitude depth variation in autumn; and f) constant activity during the two winter periods.
M. Rohtla et al.
Estuarine, Coastal and Shelf Science 276 (2022) 108016
The average duration of dormancy (including dormancy and disap-
pearances) was 138 days (min: 35 and max: 224 days, Fig. 3). The
duration of dormancy was more strongly correlated with the start date
than the end date (start date: R
=0.60; end date: R
=0.27); the later
the dormancy started, the shorter it lasted. The start date was more
variable (range =158 days, Fig. 4) than the end date (range =92 days,
Fig. 4b). By OctoberNovember, most individuals (>50%) had entered
dormancy (Fig. 4). In mid-April, most eel (>50%) had become active
again, but during the last two years some eel were still dormant until
May (Fig. 4).
Fig. 3. Activity timelines of European eel (Anguilla anguilla) (N =55) tagged with acoustic transmitters. Red line activity; green line no depth variation
(interpreted as a dormancy event); black line disappearance (interpreted as a dormancy event). Individual eel are sorted along the y-axis by total length from 530
mm (ID 5807) to 820 mm (ID 5751 and 15169). (For interpretation of the references to colour in this gure legend, the reader is referred to the Web version of
this article.)
Fig. 4. Dormancy behaviour of eel tagged with acoustic transmitters monitored between August 2018 and August 2021. The dashed lines represent the median
number of days for the end of dormancy. Day one was assigned as August 1st. (For interpretation of the references to colour in this gure legend, the reader is referred
to the Web version of this article.)
M. Rohtla et al.
Estuarine, Coastal and Shelf Science 276 (2022) 108016
Only one eel overwintered in the innermost part of the fjord (Zone I).
The percentage of eel that started their dormancy was related to a
decrease in day length and temperature and an increase in Julian days
(Fig. 5). The percentage of eel that ended their dormancy increased with
an increase in all three variables: Temperature, day length and Julian
days (Fig. 5). Eel located in the mid and outer parts of the fjord (Zones III
and IV) became dormant before eel in the inner (more freshwater) part
of the fjord (Zone II). The reverse occurred for the end of dormancy: eel
located toward the inner freshwater zone (Zone II) were the rst to
become active again. This pattern was consistent between years (Fig. 5).
Length of eel increased toward the outer part of the fjord (Fig. 5).
Duration of dormancy was signicantly longer in the outer part of the
fjord: by 54 days on average (t =5.16, P <0.0001, linear model
Only day length was signicant as an explanatory variable (Table 2).
When including eel that has disappeared, almost all models were sig-
nicant, except for the temperature variable in the combined tempera-
ture and day length model (Table 2). Dormancy occurred at a wide range
of water temperatures (6.515 C, Fig. 6). Overall, day length showed a
better t than temperature for explaining overwintering patterns
(Fig. 6). The estimated number of hours of daylight corresponding to
half of the eel entering dormancy (start) or awakening (end) are 9.4 and
14.1 h respectively (Table 3).
4. Discussion
Low temperatures induce inactivity, fasting and a low metabolic rate
in sh (Shuter et al., 2012; Speers-Roesch et al., 2018). For eel, this
occurs at about 5 C. Below that temperature, oxygen consumption
decreases and eel enter metabolic torpor (Walsh et al., 1983; Reeve
et al., 2022). Eel in our study rarely experienced such low temperatures.
Sandnesfjord is a temperate marine environment in which dynamic
water mixing occurs throughout the year due to ebb and ood tides
resulting in higher average and more variable water temperatures
compared to inland waters at the same latitude (Tjomsland and Kro-
glund, 2010). Despite these relatively mild temperature conditions, eel
were inactive at approximately the same time in both study years.
Overall, overwintering pattern in Sandnesfjorden varied little from year
to year: from OctoberNovember to mid-April, and were best predicted
by day length (Table 3). Temperature had much less power than day
length in explaining the timing of the inactive period, which we refer to
as dormancy. We conclude that eel may enter dormancy even under mild
Fig. 5. a) Plot of individual scores of the Principal
component analysis (PCA) carried out on data from
overwintering eel in Sandnesfjord (Norway). Zones
are numbered from 1 (innermost part of the fjord) to
4 (outermost part of the fjord). Centroids were
calculated to group individuals from these four
different fjord zones. Individual eel are also labelled
according to their overwintering behaviours (disap-
pearance or dormancy). b) Plot of the variables used
in the PCA, JD: Julian days (s: start and e: end), %s:
Proportion of individuals that started overwintering,
%e: Proportion of individuals that ended over-
wintering, DL: day length (s: start and e: end), T:
temperature (s: start and e: end), Length is the body
length of eel. (For interpretation of the references to
colour in this gure legend, the reader is referred to
the Web version of this article.)
M. Rohtla et al.
Estuarine, Coastal and Shelf Science 276 (2022) 108016
temperature conditions (in our study this could occur at temperatures
around 14 C), and that timing of entering dormancy is correlated with
photoperiod or ecological processes associated with photoperiod.
Similar overwintering periods were observed in eel tagged with data
storage tags in Lake M¨
alaren in Sweden (Westerberg and Sj¨
oberg, 2015):
between early November until the end of May. Since both study loca-
tions are at approximately the same latitude (58.7and 59.4) and have
similar photoperiods, this is consistent with our observations that eel
may enter a state of dormancy even at mild temperatures and that this is
correlated with day length. Westerberg and Sj¨
oberg (2015) also reported
that the start of dormancy was associated with a wide range of tem-
peratures (4.512.4 C, our study: 6.515 C) and attributed this to a
residual effect of tagging. In other words, tagging surgery and handling
caused the eel to be inactive for a period. However, in our study, tagging
had little effect as eel rapidly resumed normal activity after the pro-
cedure. Thus, we argue that overwintering levels of activity occur even
Table 2
Overview of the logistic regressions carried out on data from overwintering eel
in Sandnesfjord, Norway. The proportion of eel that became dormant (start) or
awoke from dormancy (end) according to either day length or temperature or
the combination of both variables. Two groups of eel were tested according to
how dormancy was dened by either including disappearances or not
(dormancy). AIC: Akaike Information Criteria.
Model Z value P AIC
Dormancy (n =27)
~ day length 2.153 0.03 24
~ temperature 0.1537 0.12 29
~ day length +temperature 1.642 (day length) 0.10 25
0.330 (temperature) 0.74
~ day length 2.194 0.028 22
~ temperature 1.863 0.062 29
~ day length +temperature 1.602 (day length) 0.11 23
0.632 (temperature) 0.53
Dormancy and disappearances (n =69)
~ day length 3.758 0.00017 54
~ temperature 2.999 0.0027 74
~ day length +temperature 2.657 (day length) 0.0079 56
0.106 (temperature) 0.915
~ day length 3.365 0.00077 58
~ temperature 3.061 0.0022 72
~ day length +temperature 2.342 (day length) 0.019 60
0.483 (temperature) 0.63
Fig. 6. Timing of eel (Anguilla anguilla) dormancy according to photoperiod and temperature. The graphs show the proportion of eel that fell into dormancy (start)
and awoke from dormancy (end). Logistic regressions were tted to the data. Grey shading represents condence interval of the GLM model using a binomial
distribution. (For interpretation of the references to colour in this gure legend, the reader is referred to the Web version of this article.)
Table 3
Day length (number of hours of daylight ±standard error) estimated for 25%
(L25), 50% (L50), and 75% (L75) of eel (Anguilla anguilla) to enter dormancy
(start) or to awake (end). Estimates are based on logistic regressions of the
percentage of dormant (or awakened) eel according to day length.
DL25 DL50 DL75
Dormancy start 11.3 ±0.7 9.4 ±0.5 7.5 ±0.7
Dormancy end 13.2 ±0.4 14.1 ±0.2 15.0 ±0.3
M. Rohtla et al.
Estuarine, Coastal and Shelf Science 276 (2022) 108016
at relatively mild temperatures. Whether light/day length is the trigger
of overwintering behavior is unknown. However, it seems unlikely
because eel often take refuge in the mud during the dormant period
(Tesch, 2003). The trigger could also be one or more ecological pro-
cesses that are correlated with photoperiod; for example, the decrease in
productivity and subsequent disappearance of prey items during the
winter (Larsen et al., 2004).
There was a clear difference in the timing of dormancy between
zones II and III-IV (Fig. 5). Dormancy started later and ended earlier in
the inner part of the fjord (zone II) and was also shorter by an average of
approximately 54 days compared to the outer part of the fjord. This
indicates that photoperiod is not the only factor modulating the timing
and duration of dormancy. What the other factors are is unclear but may
be related to seasonal changes in salinity. In mid-April when eel awake
from dormancy river ow generally increases as snow melts. Thus, the
increased freshwater input from Storelva river causes the salinity to
decrease, initially, in zones I and II, then later in the outer fjord system
(Tjomsland and Kroglund, 2010).
Dormancy of eel observed in this study seems to be facultative rather
than obligatory; several eel (~11% of the overwintering periods)
remained active, or were intermittently active, during the winter. The
reasons why some eel remained active while other became dormant
require further investigation, for example, at the physiological level.
Some caution is warranted regarding our interpretation that all
disappearance episodes represent periods of dormancy. As the receiver
area did not provide total coverage of the entire study area, we cannot
exclude the possibility that some eel moved to areas between receivers
in which they would not have been detected. However, nearly all the eel
that disappeared reappeared at the same or a nearby receiver. This in-
dicates that they stayed in the area but were not detected by the re-
ceivers. Temperate eel bury themselves into sediments or hide in
crevices during winter (Bertin, 1951; Tesch, 2003; Tomie et al., 2017).
As the transmissions of acoustic waves are blocked by rocks, and can be
severely weakened by sediments and vegetation, it is possible that the
signals emitted by transmitters of buried and sheltered eel had a reduced
detection range and, therefore, were less likely to be recorded by the
stationary receivers.
Our observations of the dormancy period are useful for deciding the
timing of surveys for eel, at least in marine coastal areas. Logistic re-
gressions may also be used to adjust abundance estimates a posteriori
according to the time of the year. For example, a survey conducted when
daylength is 10 h, should consider that approximately 50% of eel may be
dormant. This would be especially important during annual surveys
carried over several years at the same location as part of monitoring of
this endangered species.
CRediT authorship contribution statement
Mehis Rohtla: Writing original draft, Visualization, Methodology,
Investigation, Formal analysis, Conceptualization. Even Moland:
Writing review & editing, Supervision, Methodology, Investigation,
Conceptualization. Anne-Berit Skiftesvik: Writing review & editing,
Supervision, Investigation, Funding acquisition, Conceptualization. Eva
B. Thorstad: Writing review & editing, Supervision, Methodology,
Investigation, Conceptualization. Sebastian Bosgraaf: Methodology,
Investigation. Esben M. Olsen: Writing review & editing, Methodol-
ogy, Investigation, Conceptualization. Howard I. Browman: Writing
review & editing, Supervision, Investigation, Funding acquisition,
Conceptualization. Caroline M.F. Durif: Writing review & editing,
Visualization, Project administration, Methodology, Investigation,
Funding acquisition, Formal analysis, Conceptualization.
Declaration of competing interest
The authors declare that they have no known competing nancial
interests or personal relationships that could have appeared to inuence
the work reported in this paper.
Data availability
Data will be made available on request.
The authors thank Helge Larsen for catching eel and Susanna
Thorbjørnsen for help with eel tagging. We also thank Ørjan Karlsen for
lending us receivers. M.R. is grateful to Pieterjan Verhelst for providing
guidelines on acoustic telemetry data exploration. This study was fun-
ded by the Norwegian Research Council (MAREEL; Grant/Award
Number: 280658) and by the Institute of Marine Researchs Coastal
Ecosystems Program (15637).
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://doi.
Andrews, S.N., Buhariwalla, C.F., Fleet-Pardy, B., Dadswell, M.J., Linnansaari, T.,
Curry, R.A., 2019. Left out in the cold: the understudied overwintering ecology of
striped bass in Canada. Environ. Biol. Fish. 102 (3), 499518.
Baras, E., Jeandrain, D., 1998. Evaluation of surgery procedures for tagging eels Anguilla
anguilla (L.) with biotelemetry transmitters. Hydrobiologia 371/372, 107111.
Bertin, L., 1951. Les anguilles. Variation, croissance, euryhalinit´
e, toxicit´
hermaphrodisme juv´
enile et sexualit´
e, migrations, m´
etamorphoses. Payot.
Crawshaw, L.I., 1984. Low-temperature dormancy in sh. Am. J. Physiol. 246 (4 Pt 2),
Dannevig, A., 1945. Ålens reaksjon på miljøforandringer. Naturen 12, 366372. https://
Daverat, Francoise, Limburg, Karin E., Thibault, Isabel, Shiao, jen-Chieh, Dodson, Julian
J., Caron, Francois, Tzeng, Wan-Nian, Iizuka, Yoshiyuki, Wickstr¨
om, Håkan, 2006.
Phenotypic plasticity of habitat use by three temperate eel species, Anguilla anguilla,
A. japonica and A. rostrata. Mar. Ecol. Progr. Ser. 308, 231241.
Durif, C., Dufour, S., Elie, P., 2005. The silvering process of Anguilla anguilla: a new
classication from the yellow resident to the silver migrating stage. J. Fish. Biol. 66,
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: Dixon, D.A. (Ed.), Biology, Management, and Protection of Catadromous Eels,
vol. 33. American Fisheries Society Symposium, pp. 343356.
Durif, C., Guibert, A., Elie, P., 2009. Morphological discrimination of the silvering stages
of the European eel. In: Casselman, J.M., Cairns, D.K. (Eds.), Eels at the Edge:
Science, Status, and Conservation Concerns, vol. 58. American Fisheries Society
Symposium, pp. 103111.
Durif, C., Skiftesvik, A.B., 2019. Sluttrapport Forskningsfangst Etter Ål. Rapport fra
havforskningen, Issue, 2018.
Durif, C.M.F., Elie, P., 2008. Predicting downstream migration of silver eels in a large
river catchment based on commercial shery data [Article]. Fish. Manag. Ecol. 15
(2), 127137.
Kristensen, M.L., Olsen, E.M., Moland, E., Knutsen, H., Gronkjaer, P., Koed, A., Kallo, K.,
Aarestrup, K., 2021. Disparate movement behavior and feeding ecology in sympatric
ecotypes of Atlantic cod. Ecol. Evol. 11 (16), 1147711490.
Kroglund, F., Høgberget, R., Haraldstad, T., Økland, F., Thorstad, E.B., Teien, H.C.,
Rosseland, B.O., Salbu, B., Nilsen, T.O., Stefansson, S.O., Guttrup, J., 2011. Storelva
Smolt 2008: Påvirkes Laksesmolt Av Aluminium I Brakkvann?
Telemetriundersøkelser Storelva I Holt. Aust-Agder, 2008.
Larsen, A., Flaten, G.A.F., Sandaa, R.A., Castberg, T., Thyrhaug, R., Erga, S.R.,
Jacquet, S., Bratbak, G., 2004. Spring phytoplankton bloom dynamics in Norwegian
coastal waters: microbial community succession and diversity. Limnol. Oceanogr. 49
(1), 180190.
Reeve, C., Rowsey, L.E., Speers-Roesch, B., 2022. Inactivity and the passive slowing
effect of cold on resting metabolism as the primary drivers of energy savings in
overwintering shes. J. Exp. Biol. 225 (8)
Sadler, K., 1979. Effects of temperature on the growth and survival of the European eel,
Anguilla-Anguilla L. J. Fish. Biol. 15 (4), 499507.
Shuter, B.J., Finstad, A.G., Helland, I.P., Zweimuller, I., Holker, F., 2012. The role of
winter phenology in shaping the ecology of freshwater sh and their sensitivities to
climate change. Aquat. Sci. 74 (4), 637657.
M. Rohtla et al.
Estuarine, Coastal and Shelf Science 276 (2022) 108016
Shuter, B.J., Post, J.R., 1990. Climate, population viability, and the zoogeography of
temperate shes. Trans. Am. Fish. Soc. 119 (2), 314336.
Speers-Roesch, B., Norin, T., Driedzic, W.R., 2018. The benet of being still: energy
savings during winter dormancy in sh come from inactivity and the cold, not from
metabolic rate depression. Proc. Biol. Sci. 285, 20181593
rspb.2018.1593, 1886.
Team, R.C., 2021. R: a language and environment for statistical computing. https://
Tesch, F.W., 2003. The Eel, fth ed. ed. Blackwell Publishing.
Thibault, I., Dodson, J.J., Caron, F., 2007. Yellow-stage American eel movements
determined by microtagging and acoustic telemetry in the St Jean River watershed,
e, Quebec, Canada. J. Fish. Biol. 71 (4), 10951112. https://doi:10.1111/j.109
Thorstad, E.B., Okland, F., Westerberg, H., Aarestrup, K., Metcalfe, J.D., 2013.
Evaluation of surgical implantation of electronic tags in European eel and effects of
different suture materials. Mar. Freshw. Res. 64 (4), 324331.
Tjomsland, T., Kroglund, F., 2010. Modellering Av Strøm Og Saltholdighet I
Sandnesfjorden Ved Risør, 6049-2010.
Tomie, J.P.N., Cairns, D.K., Hobbs, R.S., Desjardins, M., Fletcher, G.L., Courtenay, S.C.,
2017. American eel (Anguilla rostrata) substrate selection for daytime refuge and
winter thermal sanctuary. Mar. Freshw. Res. 68 (1), 95105.
Ultsch, G.R., 1989. Ecology and physiology of hibernation and overwintering among
fresh-water shes, turtles and snakes [Article]. Biol. Rev. 64 (4), 435516. https://
van Deurs, M., Christensen, A., Frisk, C., Mosegaard, H., 2010. Overwintering strategy of
sandeel ecotypes from an energy/predation trade-off perspective [Article]. Mar.
Ecol. Prog. Ser. 416, 201U215.
Villegas-Rios, D., Freitas, C., Moland, E., Thorbjornsen, S.H., Olsen, E.M., 2020. Inferring
individual fate from aquatic acoustic telemetry data. Methods Ecol. Evol. 11 (10),
Walsh, P.J., Foster, G.D., Moon, T.W., 1983. The effects of temperature on metabolism of
the American eel Anguilla rostrata (LeSueur): compensation in the summer and
torpor in the winter. Physiol. Zool. 56, 532540.
Westerberg, H., Sj¨
oberg, N., 2015. Overwintering dormancy behaviour of the European
eel (Anguilla anguillaL.) in a large lake. Ecol. Freshw. Fish 24 (4), 532543. https://
Winslade, P., 1974. Behavioral-studies on lesser sandeel ammodytes-marinus (raitt) .3.
Effect of temperature on activity and environmental-control of annual cycle of
activity. J. Fish. Biol. 6 (5), 587599.
Zuur, A.F., Ieno, E.N., Elphick, C.S., 2010. A protocol for data exploration to avoid
common statistical problems. Methods Ecol. Evol. 1 (1), 314.
M. Rohtla et al.
ResearchGate has not been able to resolve any citations for this publication.
Full-text available
Winter dormancy is a seasonal survival strategy common among temperate ectotherms, characterized by inactivity, fasting, and low metabolic rates. Previous reports of metabolic rate depression (MRD) in winter-dormant ectotherms, including many fishes, may result from confounding influences of temperature-dependent variation in activity on metabolic rate measurements. We hypothesize that, as demonstrated recently in the winter-dormant cunner (Tautogolabrus adspersus), inactivity and the passive physicochemical (Arrhenius) effect of cold on standard metabolic rate (SMR) are the common primary mechanisms underlying the low metabolic rates among winter-dormant fishes. Using automated video tracking, we investigated threshold temperatures for winter dormancy onset (major reductions in activity, increased sheltering, and fasting) in four phylogenetically-diverse teleost species reported to be winter dormant: cunner, pumpkinseed sunfish (Lepomis gibbosus), American eel (Anguilla rostrata), and mummichog (Fundulus heteroclitus). All species showed large activity and feeding reductions, but the magnitude of change and dormancy threshold temperature was species-specific. We propose that a continuum of overwintering responses exists among fishes from dormant to lethargic to active. The relationship between activity and metabolic rate was then measured using video-recorded automated respirometry during acute cooling and following cold acclimation in pumpkinseed, mummichog, and eel. In all species, activity and metabolic rate were strongly correlated at all temperatures, and cooling caused reduced activity and metabolic rate. When variation in activity was controlled for across temperatures spanning the dormancy thresholds, the thermal sensitivity of metabolic rate including SMR indicated the predominance of passive physicochemical influences (mean Q10<3.5), rather than active MRD. Activity reductions and physicochemical slowing of metabolism due to cold appear to be the primary energy saving mechanisms in overwintering fishes.
Full-text available
Coexistence of ecotypes, genetically divergent population units, is a widespread phenomenon, potentially affecting ecosystem functioning and local food web stability. In coastal Skagerrak, Atlantic cod (Gadus morhua) occur as two such coexisting ecotypes. We applied a combination of acoustic telemetry, genotyping, and stable isotope analysis to 72 individuals to investigate movement ecology and food niche of putative local “Fjord” and putative oceanic “North Sea” ecotypes—thus named based on previous molecular studies. Genotyping and individual origin assignment suggested 41 individuals were Fjord and 31 were North Sea ecotypes. Both ecotypes were found throughout the fjord. Seven percent of Fjord ecotype individuals left the study system during the study while 42% of North Sea individuals left, potentially homing to natal spawning grounds. Home range sizes were similar for the two ecotypes but highly variable among individuals. Fjord ecotype cod had significantly higher δ13C and δ15N stable isotope values than North Sea ecotype cod, suggesting they exploited different food niches. The results suggest coexisting ecotypes may possess innate differences in feeding and movement ecologies and may thus fill different functional roles in marine ecosystems. This highlights the importance of conserving interconnected populations to ensure stable ecosystem functioning and food web structures. We investigated the behavior and trophic ecology of sympatric ecotypes of Atlantic cod. Differences in both behavior and trophic ecology were found between the ecotypes. This suggests coexisting ecotypes may possess innate differences in feeding and movement ecologies and may fill different functional roles in marine ecosystems, thus highlighting the importance of conserving interconnected populations to ensure stable ecosystem functioning and food web structures.
Full-text available
Acoustic telemetry has become a popular means of obtaining individual behavioural data from a wide array of species in marine and freshwater systems. Fate information is crucial to understand important aspects of population dynamics such as mortality, predation or dispersal rates. Here we present a method to infer individual fate from acoustic telemetry arrays of receivers with overlapping detection ranges. Our method depends exclusively on information on animal movements and the characteristics and configuration of the telemetry equipment. By answering a limited number of simple questions, our method identifies six different fates: tagging mortality, natural mortality, fishing mortality, predation, dispersal and survival. Applying the method to a cod telemetry dataset, we were able to determine fate for 97% of the individuals. We validate the results using several external sources of information, such as recaptures from fishers and control fish with known fate. The method is readily applicable to a wide array of species with minimal adjustments, expanding the range of hypotheses that can be tested using telemetry data.
Full-text available
Striped Bass in Canadian waters occur at the most northern extent of the species range. At these latitudes, overwintering represents a significant portion of annual activities (Nov- May). During the winter period, Striped Bass generally occupy inland waterways, aggregate densely, feed little, and conduct minimal movements rendering them vulnerable to fisheries exploitation, pollution and habitat alterations. Despite the importance of winter survival to population viability, the winter period remains an understudied aspect of Striped Bass ecology throughout the species range. Here we review what is known as well as the knowledge gaps regarding the overwintering ecology for Striped Bass in Canadian waters. We discuss: all reported Canadian Striped Bass overwintering locations and their characteristics; environmental conditions that may be required or Striped Bass winter survival; environmental cues that may cause Striped Bass to enter overwintering habitats in late fall and depart in early spring; possible threats to overwintering habitat; and we propose numerous research questions necessary for improving Striped Bass conservation and management both within Canada and across the species range.
Full-text available
We addressed hypotheses that anguillid eels use mud as a substrate refuge only in the absence of substrate cavities, and that the winter distribution of eels in coastal bay and estuarine habitat is limited to waters warmer than the freezing point of fish tissue (,À0.78C). In the seasonally ice-covered southern Gulf of St Lawrence, Canada, locations of summer fyke and winter spear fisheries indicate that American eels (Anguilla rostrata) are widely distributed in both summer and winter in shallow soft-bottomed bay and estuarine habitat. Captive eels in fresh water preferred mud substrates during summer, during pre-winter cooling and during post-winter warming periods, but in winter chose mud and cobble substrates at approximately equal frequencies. Plasma antifreeze was not detected in blood sampled from eels speared in mud under winter ice. Winter bottom water temperatures in an eel wintering site were below the approximate freezing point of fish tissue 29.9% of the time. Mud of eel wintering grounds is warmer than overlying water and appears to serve as a thermal sanctuary that allows eels to safely overwinter under ice-covered waters. American eels in the southern Gulf of St Lawrence spend ,67% of their annual cycle within the substrate.
Full-text available
The identification of five stages for female and two stages for male eels Anguilla anguilla using multivariate analysis was carried out on a large sample of individuals collected at six different locations in France. Stages corresponded to a growth phase (stages I and II), a pre-migrant phase (III) and two migrating phases (IV and V). It is likely that an important period of growth triggered silvering through the production of growth hormone (GH) in stage III eels. In migrating eels gonad development, gonadotropin hormone (GTH-II) production and increase of eye surface were similar at all sites. Differences among locations were found in gut regression and pectoral Fin length. As variability for these increased with the size of the watershed and values were highest for the most downstream locations, fin length and gut regression may indicate the time since an eel started its migration. (c) 2005 The Fisheries Society of the British Isles.
Full-text available
At acclimation temperatures between 10 and 20 C (typical of native summer temperatures), American eels feed regularly and display the compensatory adjustments in oxygen consumption rates which are characteristic of many ectothermic species (i.e., ). However, eels acclimated to temperatures typical of winter months (i.e., ≤ 5 C) for periods in excess of 5 wk cease feeding and display a dramatic decrease in oxygen consumption rates ( between 5- and 10 C-acclimated eels = 4.10) indicative of a torpid metabolic state. Studies of oxygen consumption rates after acute transfer of eels from higher to lower temperatures indicate that (1) compensation to 10 and 15 C requires a minimum of 16-20 h, and (2) that respiratory rates typical of the torpid state occur within the first 1-2-h posttransfer. The low temperature torpid state is accompanied by no significant increases in enzyme activities above levels of 15 C-acclimated fed eels, and several examples of decreases were noted. Although a 6-mo fast at 10 or 15 C results in substantial decreases in lipid stores, enzyme activities, and hepatosomatic index, eels fasted at 5 C have metabolic reserves and enzyme activities nearly identical with those of the 15 C-acclimated fed eels. Possible molecular mechanisms of torpor and the ecological implications of torpor and fasting are discussed.
Winter dormancy is used by many animals to survive the cold and food-poor high-latitude winter. Metabolic rate depression, an active downregulation of resting cellular energy turnover and thus standard (resting) metabolic rate (SMR), is a unifying strategy underlying the persistence of organisms in such energy-limited environments, including hibernating endotherms. However, controversy exists about its involvement in winter-dormant aquatic ectotherms. To address this debate, we conducted simultaneous, multi-day measurements of whole-animal oxygen consumption rate (a proxy of metabolic rate) and spontaneous movement in a model winter-dormant marine fish, the cunner (Tautogolabrus adspersus). Winter dormancy in cunner involved a dampened diel rhythm of metabolic rate, such that a low and stable metabolic rate persisted throughout the 24 h day. Based on the thermal sensitivity (Q10) of SMR as well as correlations of metabolic rate and movement, the reductions in metabolic rate were not attributable to metabolic rate depression, but rather to reduced activity under the cold and darkness typical of the winter refuge among substrate. Previous reports of metabolic rate depression in cunner, and possibly other fish species, during winter dormancy were probably confounded by variation in activity. Unlike hibernating endotherms, and excepting the few fish species that overwinter in anoxic waters, winter dormancy in fishes, as exemplified by cunner, need not involve metabolic rate depression. Rather, energy savings come from inactivity combined with passive physico-chemical effects of the cold on SMR, demonstrating that thermal effects on activity can greatly influence temperature-metabolism relationships, and illustrating the benefit of simply being still in energy-limited environments.