Marine migration and habitat use
of anadromous brown trout Salmo trutta
Sindre Håvarstein Eldøy, Jan Grimsrud Davidsen, Eva Bonsak Thorstad, Fred Whoriskey, Kim
Aarestrup, Tor Fredrik Næsje, Lars Rønning, Aslak Darre Sjursen , Audun Håvard Rikardsen and
Jo Vegar Arnekleiv
S. H. Eldøy (email@example.com), J. G. Davidsen (firstname.lastname@example.org), L. Rønning
(email@example.com), A. D. Sjursen (firstname.lastname@example.org) and J. V. Arnekleiv
(email@example.com). NTNU University Museum, Norwegian University of Science and
Technology, NO-7491 Trondheim, Norway.
E. B. Thorstad (firstname.lastname@example.org) and T. F. Næsje (email@example.com). Norwegian
Institute for Nature Research, NO-7485 Trondheim, Norway.
F. Whoriskey (firstname.lastname@example.org). Ocean Tracking Network, Dalhousie University, Halifax, NS
B3H 4J1, Canada.
K. Aarestrup (email@example.com). Technical University of Denmark, National Institute of Aquatic
Resources, DK-8600 Silkeborg, Denmark.
A. H. Rikardsen (firstname.lastname@example.org). The Arctic University of Norway, NO-9037 Tromsø,
Corresponding author: Jan Grimsrud Davidsen, NTNU University Museum, Department of
Natural History, NO-7491 Trondheim; Norway. Tel.: +47 924 64314; email:
The biology and ecology of anadromous brown trout Salmo trutta at sea is poorly understood.
This study provided information on spatial and temporal distribution of sea trout in the ocean.
The behaviour of 115 individuals (veteran migrants, 270-700 mm) was tracked by using acoustic
telemetry in a fjord system during April-September in 2012-2013. Overall, fish spent 68% of
their marine residence time close to river mouths (< 4 km). Most fish registrations (75%) were in
near shore habitats, but pelagic areas were also used. The maximum migration distance of tagged
fish was categorized as short (< 4 km from river mouth, 40% of fish), medium (4-~13 km, 18%
of fish) or long (> ~13 km, 42% of fish). Long distance migrants had poorer body condition in
spring prior to migration, used pelagic areas more often and returned earlier to freshwater than
short and medium distance migrants. Marine residence time was 7-183 days, and was positively
correlated to body length and smolt age, but negatively correlated to the date of sea entry.
Key words: acoustic telemetry; life history strategy; migratory behaviour; sea trout;
The brown trout (Salmo trutta L.) is an iteroparous salmonid species with indigenous
populations in Europe, North Africa and western Asia (MacCrimmon et al. 1970). It has been
introduced by humans to all other continents except Antarctica (MacCrimmon and Marshall
1968). The brown trout is an opportunistic carnivore that with its large ecological variability has
adapted to and found suitable niches in a variety of habitat types (Klemetsen et al. 2003). Brown
trout often migrate to utilise the best suited habitat during different stages of its life cycle,
moving either within freshwater systems, or repeatedly between freshwater and marine habitats,
to ultimately increase their individual fitness (Jonsson and Jonsson 1993). By exploiting better
feeding habitats (i.e., the sea or a lake), migration can enable individuals to attain higher growth
rates, larger sizes at-age, and for females higher fecundities (Hendry et al. 2004), all of which
may provide fitness benefits. The costs related to migration may include physiological
adjustments, the allocation of energy for swimming, and increased probability of mortality, e.g.
owing to predation, parasitism and diseases during migration (Gross et al. 1988, Jonsson and
Brown trout populations in coastal rivers may consist of both anadromous (hereafter
referred to as sea trout) and resident individuals originating from the same parents (Jonsson and
Jonsson 1993). The mechanisms controlling whether an individual becomes resident or migratory
are yet to be fully understood (Acolas et al. 2012), but an individual’s tendency to migrate seems
partly genetically determined and partly caused by phenotypic plasticity (Jonsson and Jonsson
1993). Factors such as metabolic rate, growth rate, body size, energy reserves, sex and genetics
are thought to influence whether an individual adopts migratory or resident behaviour (Thorpe
1987, Forseth et al. 1999, Wysujack et al. 2009). The balance between migration and residency is
influenced by environmental factors such as food availability, fish density, and interspecific
competition in combination with inter-individual differences, presumably underpinned by
genetically determined reaction norms (Pulido 2011). Similar intrinsic and environmental factors
may also influence individual behavioural strategies during marine migrations, determining
whether to become a short or long distant migrant, and which feeding habitats to utilise.
However, little is known about the inter-individual variation of migration behaviour and
strategies in the marine environment, and of the factors that may influence this variation.
Previous studies of sea trout in the marine environment have revealed a large variation in
migration timing, residence periods (Jensen 1968, Jonsson 1985, Jensen and Rikardsen 2008,
Jensen et al. 2012), migration distance (Berg and Berg 1987, Jensen et al. 2014) and prey choice
(Knutsen et al. 2001, Rikardsen and Amundsen 2005, Rikardsen et al. 2007). In Europe, sea trout
can enter estuaries from fresh water during all months of the year (Went 1962, Jonsson and
Jonsson 2002, Jonsson and Jonsson 2009) and the marine residence-time may differ considerably
among individual fish. For instance in Irish rivers, marine residence time was found to vary
between 43 and 362 days (Piggins 1964). Migratory distances may also differ significantly. In
Russia, Chernitsky et al. (1995) suggested that some trout resided in the estuary of the River
Varsina, while others migrated to the open Barents Sea. Intra-population variation in marine
migration distance was also recorded in a Danish population, where 47% of the tagged sea trout
post-smolts remained close to their home river in a coastal fjord, and 53% migrated to the open
Kattegat Sea (del Villar-Guerra et al. 2013). The authors suggested that the variation in migration
distance was consistent with a continuum of partial migration, in which a decision-making point
existed after fjord entry on whether to stay in the fjord or migrate to the open sea. However, both
smolts and sea trout kelts (repeat spawning individuals) in a nearby fjord all migrated into the
Kattegat sea (Aarestrup et al. 2014, Aarestrup et al. Accepted), demonstrating a large life history
variability both within and among nearby populations.
During the last decades, the abundance of sea trout has declined markedly in many
regions (ICES 2013). As an example, the catches in Norwegian rivers have, except for the
northernmost areas, declined by 23–66% during the last two decades (Anon. 2011). Recent
findings from several other countries where sea trout occur indicate similar decreases, and for
some areas it is hypothesized that this results from reduced marine survival caused at least in part
by changes in food supply or increased parasite infestations related to fish farming (ICES 2013).
In sea trout populations, mortality in the freshwater phase, especially during the earliest
embryonic and post-emergence life stages, can have a population regulating effect, whereas
mortality in the marine phase is not regulatory, but has a population reducing effect (Milner et al.
2003, Jonsson and Jonsson 2011). Hence, it is not believed that there are compensatory
mechanisms for additional mortality in the marine phase, and elevated marine mortality rates can
result in a proportional reduction in the number of spawning adults. Because sea trout typically
are females (e.g. Knutsen et al. 2004, Jensen et al. 2012), additional marine mortality has an
accentuated potential to negatively affect population recruitment by reducing the egg supply. The
marine phase is therefore an important life stage of sea trout. However, their biology and ecology
in the sea is poorly understood (Drenner et al. 2012, ICES 2013), and to understand the causes for
the decrease in the abundance of sea trout in many regions, increased knowledge on the marine
life stage is fundamental. To identify which anthropogenic or natural factors impact sea trout, and
to what extent, it is essential to determine the habitats utilized by the sea trout at different times.
Migration distance is also important, as short distant migrants will mainly be impacted by local
factors close to a population’s river mouth, whereas long distance migrants may be impacted by
multiple factors acting along the migration routes and in the different feeding habitats.
Most previous marine tracking studies of sea trout have focused on post-smolt migration
behaviour (e.g. Moore et al. 1998, Thorstad et al. 2004), whereas only a few studies have covered
older life stages (Bendall et al. 2005, Jensen and Rikardsen 2008, 2012, Jensen et al. 2014,
Aarestrup et al. Accepted). The aim of the present study was to provide novel information on the
marine habitat utilization during the summer season for sea trout that had previously performed
one or more previous marine migrations, termed veteran migrants. Spatial and temporal
distributions of tagged fish were recorded throughout the summer using acoustic telemetry in a
marine fjord in Central Norway. Specifically, marine migration distance from the trout’s putative
home river mouth, marine residence time and utilisation of littoral versus pelagic habitat was
examined. In order to explain individual variation in marine residence time, and possible
differences between the short, medium and long distance migrants, information on individual
morphometric (body length, body condition, age) and life history characteristics (back calculated
smolt length, age at smolting, previous number of marine seasons) were analysed in relation to
the observed migration patterns.
MATERIALS AND METHODS
The study was performed in two interconnected fjords (Hemnfjord and Snillfjord) in Sør-
Trøndelag County, Central Norway. Together, the two fjords cover more than 60 km2 of sea
surface and have 65 km of shoreline (Fig. 1). The fjord system is connected to the open sea
through a 36 km long strait.
The Søa watercourse has a drainage basin of 113 km2 and a mean annual water discharge
of 13.9 m3 s-1. The freshwater section accessible to anadromous fish is 10.2 km long and includes
Lake Rovatnet (surface area 7.65 km2), which offers suitable overwintering habitat and
conditions for sea trout. River Søa drains from the lake to the sea in Hemnfjord.
The River Snilldalselva consists of two branches, Snilldalselva and Bergselva.
Snilldalselva has a drainage basin of 42.7 km2, mean annual water discharge of 1.4 m3 s-1 and a
4.8 km long section accessible to anadromous trout. Bergselva has a drainage basin of 69.3 km2,
mean annual water discharge of 2.1 m3 s-1 and an accessible stretch of 1.1 km. Both branches are
highly influenced by floods and have few deep pools, consequently they are considered to be
poor overwintering areas for sea trout.
Three temperature and salinity recorders (DST milli-CT, Star-Oddi Ltd, Iceland) were deployed
in the fjord system, the first 1 km from the mouth of the River Søa in the inner Hemnfjord (Array
H1, Fig. 1), a second 600 m from the river mouth of the River Snilldalselva in the inner Snillfjord
(Fig. 1), and the third at the middle receiver of the outermost array (Array H3, Fig. 1). They were
mounted at 1 m depth on the same moorings as the automatic acoustic receivers.
FISH CAPTURE AND TAGGING
Five groups of sea trout were captured and tagged with acoustic transmitters during 12 April
2012-12 May 2013 (Table 1). A total of 80 individuals were tagged in the Søa watercourse,
consisting of 30 fish tagged in the outlet of Lake Rovatnet during the spring of 2012 (HS12
tagging group), 21 fish tagged in Lake Rovatnet during autumn 2012 (HA12), and 29 fish tagged
in the river mouth of River Søa during the spring of 2013 (HS13). A total of 35 individuals were
tagged in River Snilldalselva, consisting of 20 fish tagged during autumn 2012 (SA12), and 15
fish tagged during spring 2013 (SS13). The fish were captured using 3-5 gillnets with 35-42 mm
mesh width. The nets were checked continuously, and captured fish were retrieved as soon as
vibrations/visual observations indicated a fish was entangled. This reduced fish stress and
injuries. The fish were taken out of the nets by cutting net mesh with scissors to prevent damage
to gills, skin and scales. Prior to tagging, the captured fish were kept up to two hours in a net cage
in a calm part of the river or shoreline.
The sea trout were implanted with individually coded acoustic transmitters. Study
partners contributed tags to the study, which resulted in using different models of tags having
different characteristics and capabilities depending on partner resources and research interests.
The different models had the same shape but differed in length and diameter which allowed
adaptation of tag size to the length of the fish (HS12: n = 15 model MP-9-long, natural length
(LN) 335 - 440 mm, n = 15 model MP-13, LN 350 - 600 mm; HA12: n = 10 model V9-2x, LN 270
- 380 mm, n = 11 model V13-1x, LN 370 - 700 mm; SA12: n = 5 model MP-9-long, LN 310 - 400
mm, n = 6 model MP-13, LN 340 - 650 mm, n = 9 model V13-1x, LN 340 - 440 mm; HS13: n =
29 model ADT-9-long, LN 330 - 580 mm, SS13; n = 15 model ADT-9-long, LN 320 - 460 mm).
Natural length of the fish was measured from the tip of the snout to the tip of the longer lobe of
the caudal fin, without compressing the lobes along the midline. Estimated battery life was 246
days (MP-9-Long), 267 days (ADT-9-long), 282 days (V9-2L), 525 days (MP-13) or 622 days
(V13-1L), respectively. Hence, 41 fish tagged in 2012 could also be tracked in 2013. Transmitter
models MP and ADT were produced by Thelmabiotel AS, Norway, and all V models by
VEMCO Inc., Canada. Tag size was chosen according to body length and condition of the fish to
minimise tag size relative to fish size. Tag mass in air relative to fish body mass was on average
1.46% (range 0.30-3.09%). The tag used for any individual fish was believed to be small enough
that it would not significantly affect behaviour or survival (e.g., Cooke et al. 2011).
Prior to tagging, the fish were anaesthetised with 2-phenoxy-ethanol (EC No 204-589-7;
SIGMA Chemical Co., USA; 0.5 ml l-1 water). A 1.5 - 2 cm incision was made in the body cavity
on the ventral surface anterior to the pelvic girdle. After the tag was inserted via the incision into
the body cavity, the incision was closed using two independent monofilament sutures
(RESORBA Wundversorgung GmbH & Co. KG, Germany; 5/0 Resolon). During the 3-5 min
surgery, the gills were gently irrigated. After surgery, the fish were placed in a holding tank for
recovery (3-5 min) before they were released in a calm part of the river or near the shoreline
close to the capture site.
TRACKING OF TAGGED FISH
The tagged fish were tracked using a total of 50 acoustic receivers (Vemco Inc., Canada, model
VR2W and VR2). Of these, 39 were deployed in the fjord system, while 11 were deployed in
different watercourses, including those where the fish were captured for tagging (Fig. 1). All
receivers deployed in the fjord were mounted on moorings 5 meters below the surface, and were
operative from 20.04.2012-04.12.2013. The receivers deployed in rivers were moored on 50 mm
iron pipes which were hammered into the riverbed. With exception of the four receivers in River
Søa between Lake Rovatnet and the Hemnfjord which were in operation during the whole study
period, all receivers in freshwater habitats were operative from 20.04 2012 – 02.12.2012 and
from 22.04.2013 – 04.12.2013.
Receivers recorded transmitter identification code (individual fish identity), detection date and
time for each signal received. Receiver range was tested at the middle receiver of array H1 (Fig.
1) on 22.08.2013 (calm, clear weather, high tide) and at the Hafsmo salmon farming site (Fig. 1)
on 03.12.2013 (calm, clear weather, slack tide) by deploying a transmitter (model ADT-9-long,
146 dB re 1uPa @1m) at 3 and 5 meters depth and at increasing distance from the receiver in
steps of 50 meters. The maximum receiver range was on both occasions 300 - 350 meters. The
transmitter model used in the range test was expected to have the shortest range of all transmitter
models used in the study, based on its technical specifications.
SCALE SAMPLE ANALYSIS
A small number of fish scales (5-7 scales) were sampled from the studied animals during the
tagging procedure. Information obtained from the scales on smolt length, age at smoltification,
age when studied, and numbers of previous seaward migrations were used in the analyses of the
migratory behaviour. Scale growth was assumed to be proportional to length growth (Dahl 1910,
Lea 1910, Závorka et al. 2014). The ages assigned by the research team to the experimental
animals were verified by sending a subsample of the scales for reading by personnel at the
Norwegian Institute for Nature Research and the Technical University of Denmark. Uncertain
values of age, length and age at smoltification and number of previous seaward migrations were
excluded from analyses.
The sea trout tagged in the river mouth of River Søa during spring 2013 (HS13) had
uncertain river of origin, due to presence nearby (500 m) of another watercourse housing sea
trout. This group of fish was therefore separated from the groups tagged in Lake Rovatnet when
analysing morphology and life history of the individuals by the watercourses of tagging.
The initial number of detections (registrations) logged onto all receivers used in the study was 5
147 075. Mean number of detections of the tagged individuals was 44 745 (SD = 91 294, range 0-
597 433). A total of 1 360 (0.03%) registrations with false IDs were excluded from the dataset.
Data from the two receivers in the outlet of the River Søa and the three innermost receivers in
Snillfjord were anticipated to contain higher frequencies of false detections due to concurrent
signals from high numbers of simultaneously occurring tagged fish. Concurrent signals (tag
collisions) can confound receiver detections and generate false ID codes. A data filter that
required at least two registrations from a tagged individual within a time span of 10 minutes was
applied to these receivers, which excluded 46 223 (0.90%) registrations from further analyses.
STATISTICAL ANALYSES AND COMPUTER SOFTWARE
After sorting and extracting data using Access 2013 and Excel 2013 (Microsoft Co., USA), the
statistical analyses were conducted using R version 2.15.3 (R Core Team 2014; www.r-
project.org). For one and two-way analysis of variance between two groups, Welch’s t-test were
conducted, assuming unequal variance. For analysis of variance between three or more groups,
Tukey’s ANOVA was conducted using the R-package Multcomp (Hothorn et al. 2008).
The hypothesis that numbers of days spent at sea depended on some combination of fish
age, body length, condition factor, previous numbers of time the fish had been to sea, time of sea
entry (Julian day number), maximum distance migrated away from the home river and smolt age
and length was tested using the R-package MuMIn (Barton´ 2015). In total, 576 models of
varying complexity were fitted for hypothesis testing. To avoid autocorrelation between body
length and condition factor, residual values (resvalbc) from the linear model
log(condition)~log(length) were used instead of the body condition per se. The global model
included age at tagging, length (LN), resvalbc, previous number of times the fish had been to sea,
time of sea entry, short, medium or long distance migratory strategy, back calculated smolt age
and smolt length and the interaction terms length*resvalBC, strategy*length and
strategy*resvalbc. The other 575 models were all nested models from the global model. The
approximating models were compared using Akaike information criterion (AIC) (Anderson et al.
2001). AIC ranks the candidate models to determine which model provides the best description
of the data with the fewest parameters. The hypothesis was tested for those 27 sea trout for which
data on all variables were available.
DEFINING SHORT, MEDIUM AND LONG DISTANCE MIGRANTS
The fish were categorised as short, medium or long distance migrants according to the maximum
distance at which they were detected from their release point during 1 April-1 October in either
2012 or 2013(see Fig. 1). Short distance migrants were only recorded at receivers up to 4 km
from the river mouth. Medium distance migrants were registered up to 10 km from the river
mouth for fish tagged in Søa watercourse, and up to 13 km for the fish tagged in River
Snilldalselva. Long distance migrants were registered at receivers more than 10 km from the river
mouth for fish tagged in Søa watercourse, and more than 13 km for fish tagged in River
Snilldalselva. The slight difference in the distances which defined migrant groups for the two
watercourses (10 vs. 13 km) was due to logistical concerns that resulted in different distances
between the receiver arrays in the two fjords. Fish that did not return to freshwater and were not
recorded by any receiver in the marine habitat after 1 July in either 2012 or 2013, were excluded
from the migration distance analysis, because they potentially were lost from the study before
they had reached their maximum dispersal. An exception was done for fish registered at receivers
more than 10 km (Søa watercourse) or 13 km (River Snilldalselva) from the river mouths, since
they already had been recorded as long distance migrants.
CALCULATING MARINE RESIDENCE TIME
The study area was divided into different zones based on geographic location (Fig. 1). Residence
time was only calculated for individuals returning to freshwater or for fish recorded in the fjord
after 1 October in 2012 or 2013. The calculation of residence time by tagged fish in different
fjord zones was carried out using the following criteria:
1. In the case of a transition to a zone further out in the fjord, the residence time in the next zone
started at the time of the last registration at a receiver in the previous zone.
2. In the case of transition to a zone further into the fjord, the residence time in the next zone
started at the time of the first registration at a receiver in the inner zone.
3. For transitions into freshwater, the freshwater residence started at the time of the last
registration at a river mouth receiver.
4. For transitions from freshwater to fjord zones, the fjord residence started at the first registration
at a river mouth receiver.
Receivers in river mouths were considered as part of the fjord. For the fish tagged in
2013, estimated marine residence times were considered as minimums since the fish were
captured in the river mouths, and it was possible that they had spent a preceding period in marine
habitat before they were tagged. Nine fish tagged in Lake Rovatnet in spring 2012 conducted sea
migrations during summer in both 2012 and 2013. These fish were only included in the statistical
analyses of marine residence during the first year, to avoid repeated measures concerns.
USE OF PELAGIC VS LITTORAL HABITATS
The receiver arrays that contained both pelagic and near shore receivers (array H1, H2, H3 and
S1, Fig. 1) were used to investigate the importance of littoral and pelagic habitats for the tagged
sea trout. Receivers deployed near the shore or in areas with shallow water (< 10 meters depth)
where the sea trout was likely to feed at or near the bottom or along cliff walls within the receiver
range, were defined as near shore receivers. Receivers deployed over deep water, without
coastline or shallow areas (< 25 meters depth) within the receiver range were defined as pelagic
receivers. The proportional numbers of littoral and pelagic registrations at the receiver arrays,
corrected for the proportion of littoral (8 receivers) versus pelagic (9 receivers), were investigated
for each fish for the period 1 April-1 October in 2012 or 2013. This was assumed to give a rough
estimate of relative preference of littoral and pelagic habitats. Potential differences between
littoral and pelagic habitats were tested with a Chi-square test.
From 1 May-1 October, marine water temperatures in the study area varied from 3.8 °C to 19.4
°C . The salinity levels during the same period were brackish in the outer areas (2012: mean = 28
‰, SD ± 1.8 ‰, 2013: mean = 21 ‰, SD ± 2.0 ‰), the inner Hemnfjord (2012: mean = 29 ‰,
SD ± 2.7 ‰, 2013: mean = 23 ‰, SD ± 7.6 ‰) and the inner Snillfjord (2012: mean = 26 ‰, SD
± 4.7 ‰, 2013: mean = 24 ‰, SD ± 4.8 ‰).
MORPHOLOGICAL CHARACTERISTICS OF TAGGED FISH
Among the study animals there was considerable variation both among individuals (Table 1) and
tagging groups (Table 2) regarding body size, body condition, age, back calculated smolt length,
age at smoltification and number of previous marine seasons.
The two groups of fish tagged in Lake Rovatnet (HS12 and HA12) had greater mean
smolt length, higher mean age at smoltification, higher mean age and a tendency towards having
spent more previous seasons at sea than the groups of fish tagged in River Snilldalselva (SA12
and SS13; Table 2). Similarly, the groups of fish tagged in Lake Rovatnet had higher mean smolt
lengths, ages at smoltification, and total age than the fish tagged in the mouth of River Søa
(HS13; Table 2).
Sea trout tagged in River Snilldalselva (SA12 and SS13) had lower mean natural length
and greater mean body condition than the group of fish tagged in the mouth of River Søa (HS13;
Table 2). The fish tagged in the mouth of the River Snilldalselva during the spring of 2013
(SS13) had a higher body condition at tagging than the fish tagged both in Lake Rovatnet in
spring 2012 (HS12,) and the mouth of River Søa in spring 2013 (HS13; Table 1 and 2).
Fish tagged in the River Snilldalselva during autumn 2012 (SA12) had shorter body
lengths at smoltification than the fish tagged in Lake Rovatnet in spring 2012 (HS12) and autumn
2012 (HA12, Table 2). Similarly, at smoltification fish tagged in Lake Rovatnet in autumn 2012
(HA12) had greater body length than individuals tagged in the mouth of River Søa (HS13) and in
the river mouth of River Snilldalselva (SS13) during spring 2013 (Table 2).
The group of fish tagged in Lake Rovatnet during autumn of 2012 (HA12) had greater
ages at smoltification than those tagged in the mouth of River Søa in spring 2013 (HS13), in
River Snilldalselva in autumn 2012 (SA12) and in the mouth of River Snilldalselva in spring
2013 (SS13; Table 2). The fish tagged in Lake Rovatnet in autumn 2012 (HA12) had greater total
age than the fish tagged in River Snilldalselva in spring 2013 (SS13, Table 2).
MORPHOLOGICAL CHARACTERISTICS OF SHORT, MEDIUM AND LONG DISTANCE
In total, 100 of the 115 tagged sea trout were recorded by the acoustic receivers in the fjord
system. Individual sea trout were tracked from 6-624 days. Based on the previously described
criteria, a total of 88 fish were categorized as either short, medium or long distance migrants
(Table 3). The proportions of short, medium and long distance migrants varied among the tagging
groups. The fish tagged in Lake Rovatnet in spring 2012 (HS12) consisted of 6 short (26%), 5
medium (22%) and 12 long distance migrants (52%). All sea trout tagged in Lake Rovatnet in
autumn 2012 (HA12) were long distance migrants (11 individuals, 100%). The fish tagged in the
river mouth of River Søa in spring 2013 (HS13) consisted of 19 short (70%), 4 medium (15%)
and 4 long distance migrants (15%). The sea trout tagged in River Snilldalselva in autumn 2012
(SA12) had 4 short (31%), 2 medium (15%) and 7 long distance migrants (54%), while the those
tagged in spring 2013 (SS13) consisted of 6 short (43%), 5 medium (36%) and 3 long distance
migrants (21%). The body lengths of the 15 individuals that were not recorded at any receivers
did not differ significantly from the rest of the individuals (t-test, n = 115, P = 0.22).
There was no difference in mean body length (LN) among short, medium and long
distance migrants (Table 3, ANOVA, n = 88, P = 0.20). However, most (n = 7) of the largest
individuals (≥ 450 mm, n = 12) conducted long distance migrations, while fewer large individuals
performed medium (n = 3) and short distance (n = 2) migrations. Among the smallest individuals
(≤ 350 mm, n = 18), there were equal proportions of short (n = 6), medium (n = 6) and long
distance (n = 6) migrants.
There was large inter-individual variation in mean body condition in spring (Table 3).
Long distance migrants had significantly (Tukey ANOVA) poorer body condition in spring prior
to the marine migration than short (n = 29, P = 0.013) and medium distance migrants (n = 33, P =
0.018). The body condition in spring of short and medium distance migrants did not differ (n =
44, P = 0.92).
Age, back calculated smolt length, age at smoltification and number of previous marine
seasons varied among the groups of short, medium and long distance migrants (Table 3). Long
distance migrants had larger smolt lengths than both short (Tukey ANOVA, n = 57, P = 0.023)
and medium distance migrants (n = 43, P = 0.013). The long distance migrants had a near
significant higher age at smoltification than short distance migrants (n = 50, P = 0.057), but were
similar in age to the medium distance migrants (n = 36, P = 0.104). Long distance migrants
tended to have had more previous marine seasons than the medium distance migrants (n = 38, P
= 0.057), but not more previous marine seasons than the short distance migrants (n = 44, P =
0.255). The long distance migrants were older than both the short (n = 41, P = 0.043) and the
medium distance migrants (n = 35, P = 0.032).
Among the nine fish tagged in Lake Rovatnet in spring 2012 that were followed through
their sea migration both during the summer 2012 and again in 2013, there were identical
numbers of short (n = 3), medium (n = 3) and long distance (n = 3) migrants during 2012. In
2013, one short distance migrant from 2012 performed a medium distance migration and one
medium distance migrant from 2012 performed a long distance migration. The seven other
individuals repeated the migration pattern from the year before. However, this change in
maximum migratory dispersal was not significant, but the sample size was low (Chi-squared; n =
9, P = 0.72).
MARINE RESIDENCE TIME DURING SUMMER
During 1 April-1 October (2012 and 2013), 51 of the 115 tagged sea trout were never registered
in the marine fjord, or after an initial period of detections on the marine receivers the detections
stopped and the fish were not recorded returning to freshwater. The reasons for loosing track of
the fish were in about half of the cases not known. However, 15 individuals were reported
captured and killed by anglers, 8 individuals tagged in the Lake Rovatnet were never recorded to
leave the lake, and 4 individuals migrated out of the study area and did not return. After the study
ended, two of the individuals that migrated out of the study area were recaptured by anglers 130
km southwest of their tagging location.
There was large inter-individual variation in the total residence time in marine habitats
during 1 April-1 October in 2012 and 2013 (Fig. 2). Among tagged fish tracked throughout these
periods, the mean marine residence time was 100 days (SD 52 days, range 7 - 183 days). The
largest variation was found within the fish tagged in Lake Rovatnet in spring 2012 (HS12) which
had a mean residence of 91 days (SD 59 days, range 7 - 171 days). The fish tagged in the outlet
of spawning streams of Lake Rovatnet during autumn 2012 (HA12) and tracked during summer
2013, had the lowest intragroup variation with a mean marine residence time of 53 days (SD 15
days, range 27 - 72 days). When comparing marine residence times of the different tagging
groups, the fish tagged in Lake Rovatnet in autumn 2012 (HA12) had shorter marine residence
times than the fish tagged in the mouth of River Søa in spring 2013 (HS13, Tukey ANOVA, n =
25, P = 0.049) and fish tagged in River Snilldalselva in autumn 2012 (SA12, n = 17, P = 0.0105).
The four best predictive models all indicated that the number of days spent at sea was
positively correlated to body length (LN) and smolt age, and negatively correlated to the Julian
day number of sea entry and migration distance (Table 3). The best model (r2 = 0.65, P < 0.001)
included age, body length (LN), smolt age, timing of sea entry and migration distance (Table 3).
Fish from all tagging groups utilized all areas of the fjord. However, the innermost parts
of the fjord, near the tagging location of the sea trout (zone 1 and 2, up to 4 km from the river
mouth) were found to be especially important areas for the tagged individuals, as they spent on
average 68% (SD 39%, range 0.002% - 100%) of their marine residence time in these areas (Fig.
3). Fish tagged in the Søa watercourse spent a significantly longer time in the innermost part of
Hemnfjord (zone 1, mean 71.1 days, SD 59.1 days, range 0.2 – 170.8 days) than in inner
Snillfjord (zone 2, mean 0.6 days, SD 2.0 days, range 0 – 12.1 days, Tukey ANOVA, n = 90, P <
0.001), central Snillfjord (zone 3, mean 1.87 days, SD 4.5 days, range 0 – 18.7, n = 90, P <
0.001), central Hemnfjord (zone 4, mean 11.9 days, SD 19.9 days, range 0 – 116.7 days, n = 90,
P < 0.001) and outer areas (zone 5, mean 6.4 days, SD 14.8 days, range 0 – 68.2 days, n = 90, P <
0.001). Fish tagged in the River Snilldalselva spent a longer time in the innermost part of
Snillfjord (zone 2, mean 92.6 days, SD 69.1 days, range 0.002 – 183.0 days) than in the inner
Hemnfjord (zone 1, mean 0.3 days, SD 1.1 days, range 0 – 5.0 days, Tukey ANOVA; n = 38, P <
0.001), central Snillfjord (zone 3, mean 10.2 days, SD 29.3 days, range 0 – 124.0 days, n = 38, P
< 0.001), central Hemnfjord (zone 4, mean 9.9 days, SD 28.5 days, range 0 – 121.0 days, n = 38,
P < 0.001) and outer areas (zone 5, mean 7.9 days, SD 25.1 days, range 0 – 101.0 days, n = 38, P
When comparing the residence time in the innermost parts of the fjords (zone 1 for fishes
tagged in the Søa watercourse and fjord zone 2 for fishes tagged in the River Snilldalselva), there
was no difference between fish tagged in the Søa watercourse and those tagged in River
Snilldalselva (Fig. 3, two-sided t-test, n = 64, P = 0.25). Nor were there differences between these
two groups in their residence times in the central parts of Snillfjord (zone 3, n = 64, P = 0.23),
central parts of Hemnfjord (zone 4, n = 64, P = 0.78) or the outer study area (zone 5, n = 64, P =
MARINE RESIDENCE TIME VS MIGRATION DISTANCE
Long distance migrants had, despite large inter-individual variation, shorter mean marine
residence time than both short (Tukey ANOVA, n = 50, P = 0.05) and medium distance migrants
(n = 38, P = 0.005, Table 4). There was no difference between 2012 and 2013 in the mean marine
residence time for long distance migrants (two-sided t-test, n = 24, P = 0.99).
Large inter-individual variation in the mean residence time in the different fjord zones
was observed (Fig. 3). For long distance migrants from both Rivers Søa and Snilldalselva, the
time spent in the inner fjord was significantly shorter than for the short distance migrants (Tukey
ANOVA, Søa: n = 37, P = 0.002; Snilldalselva: n = 13, P = 0.039). Similar differences were
evident between long and medium distance migrants from Søa (n = 26, P = 0.008) but not from
Snilldalselva (n = 12, P = 0.092).
LITTORAL VS PELAGIC HABITAT UTILIZATION
Overall, at the receiver arrays containing both near shore and pelagic receivers the tagged
fish had larger proportions of their registrations at receivers along the shoreline (mean 75%, SD
19%, range 37-100%) compared to receivers in the pelagic areas (mean 25%, SD 19%, range 0-
63%, Chi-squared; n = 73, P < 0.001) (Fig. 4). The fish had larger proportions of registrations at
receivers deployed near the shore than in pelagic areas at array H1 (near shore; mean 76%, range
35-100%, pelagic; mean 24%, range 0-65%; Chi-squared; n = 64, P < 0.001), array S1, (near
shore; mean 80%, range 41-100%, pelagic; mean 20%, range 0-59%, n = 29, P < 0.001), and
array H2 (near shore; mean 64%, range 0.04-100%, pelagic; mean 36%, range 0-96%; n = 23, P
< 0.001), but not in array H3 (near shore; mean 50%, range 0-100%, pelagic; mean 50%, range
0-100%; n = 27, P < 0.001).
Long distance migrants had higher proportions of pelagic registrations than medium
distance migrants (Fig. 5, Tukey ANOVA, n = 146, P = 0.020), and nearly significant higher
portions of pelagic registrations compared to short distance migrants (n = 146, P = 0.052, Fig. 4).
Short and medium distance migrants did not differ in their uses of pelagic and inshore areas (n =
146, P = 0.72).
MORPHOLOGICAL CHARACTERISTICS AND LIFE HISTORY
The sea trout differed in morphology and life history both within and between the watercourses.
Sea trout tagged in River Snilldalselva had better body condition than fish tagged in the Søa
watercourse, and individuals tagged in the mouth of River Snilldalselva in the spring of 2013 had
better body condition than the other groups of fish also tagged during the spring. Differences in
body condition in the spring might be influenced by differences in overwintering conditions and
whether an individual fish had spawned in the previous autumn (Jonsson and Jonsson 2011). For
the fish tagged in the river mouth in the spring of 2013, their area of residence prior to tagging is
not known, i.e. if they had been in the sea or fresh water. Marine residence during winter has
been reported for sea trout in both the southern and northern parts of Norway (Knutsen et al.
2004, Jensen and Rikardsen 2008, 2012), and Jonsson and Jonsson et al. (2009) found that sea
trout spending the winter at sea had better growth during the first two years after smoltification
compared to sea trout that returned to freshwater for overwintering.
Fish tagged in Lake Rovatnet did not differ in natural body length (LN) or body condition
from fish tagged in the River Snilldalselva, but were older and tended ( nearly statistically
significant)toward having experienced more previous marine seasons. Since we tagged all fish of
suitable minimum sizes (>27 cm) that we captured, this may indicate a systematic difference in
the ages of sea trout between the two sites. Furthermore, fish from Lake Rovatnet had a larger
back calculated mean smolt size and greater age at smoltification compared to fish tagged in the
River Snilldalselva. This was probably caused by environmental differences between the
watercourses. The parr in the Søa watercourse could reside in Lake Rovatnet, enabling them to
have better growth before smoltification. In contrast, the River Snilldalselva offers few deep
pools and there is no access to lakes. Hence, variable environmental conditions, constraints in
food supply or limited availability of appropriate shelter may cause the parr in this river to
smoltify at younger age than in the Søa watercourse. This is consistent with previous studies on
how the environment influences smoltification in partly migrating trout populations (Jonsson and
Jonsson 1993, Wysujack et al. 2009). The group of fish tagged in the river mouth of River Søa in
spring 2013 were smaller and younger at smoltification than the fish tagged in the Lake Rovatnet,
possibly because some of these fish originated from the neighbouring watercourse.
Large inter-individual variation in the migration distance was observed. Some individuals
remained in the innermost parts of the fjord, while others spent most of their marine residence
outside the study area. The proportions of short and long distance migrants varied greatly among
the groups of tagged fish. Fish captured at different locations and times of the year may have
been at different stages in their life history, which may have influenced their subsequent
migratory behaviour. Other causes for the variations observed in migratory strategies may have
been due to behavioural and/or genetic differences. Previous studies have also shown large
variation in migration distance among populations of anadromous sea trout (Jensen 1968,
Svärdson and Fagerström 1982, Berg and Berg 1987), which these authors attributed to
combinations of environmental and genetic factors (Klemetsen et al. 2003). del Villar-Guerra et
al. (2013) suggested that variables such as morphological characteristics, ontogeny, genetics and
life history might influence the sea trout’s marine behaviour and the extent of its marine
No difference was found in body length between short and long distance migrants, and
individuals of all size classes performed long distance migrations. By contrast, Jensen et al.
(2014) found that large individuals were more likely to conduct long distance marine migrations
than smaller individuals. They suggested that this could be caused by a higher abundance of
suitable fish prey for the larger individual’s further out in the fjord at their study site. Similarly,
Knutsen et al. (2001) found that small post-smolt sea trout fed inshore on shallow water prey
communities while larger sea trout were feeding further offshore on pelagic fish.
Fast growing sea trout change to a more piscivorous diet at a smaller size and younger age
than slower growing individuals (Klemetsen et al. 2003), which might explain why in this study
some smaller individuals conducted long distance migrations. Alternatively, the small long
distance migrants may have had similar feeding behaviour as the short distance migrants, but
dispersed further out in the fjord by chance, or due to competition with conspecifics in inshore
areas and the availability of suitable alternative habitat and conditions further away from the river
The long distance migrants had poorer body condition than short distance migrants at the
time of tagging, suggesting that individuals with a poorer body condition experienced a greater
need to maximize benefits from distant feeding opportunities. Wysujack et al. (2009) found that
poor body condition promoted migratory behaviour in brown trout parr. Similarly, Davidsen et al.
(2014) found that starved sea trout post-smolts migrated further out into a fjord compared to fully
fed individuals. However, Boel et al. (2014) found a different pattern in their study of migration
distances of brown trout in a freshwater system, were energy stores were positively correlated
with migration distances. An alternative hypothesis to account for the pattern observed in this
study may be that that fish with poor body condition were outcompeted from the preferred shore
habitats. Migratory strategies have previous been shown to be influenced by different needs for
food intake (Halttunen et al. 2013), and Damsgård et al. (1998) showed that starving fish may
undertake more risky behaviour than well fed individuals.
MARINE RESIDENCE DURING SUMMER
Large intragroup variation in marine residence time during the summer months was
observed. Individuals tagged in the Lake Rovatnet during spring 2012 had the largest intragroup
variation, while individuals tagged in Lake Rovatnet in autumn 2012 had the smallest variation.
Previous studies have revealed that marine residency varies both among and within populations,
with a range of factors influencing the duration of the marine residence of an individual, such as
age, maturity (Jonsson 1985) and environmental conditions in fresh water prior to the seawards
migration (Jensen and Rikardsen 2008). In the present study, the duration of the seaward
migration for 27 of the tagged fish was found to be positively correlated to the body length (LN)
and smolt age, but negatively correlated to the date of sea entry.
Our fish spent on average 68% of their marine residence time in the innermost parts of the
fjords, near the mouth of the river where they were tagged. Since all fish in the present study
were veteran migrants with one or more previous marine seasons, and since seawater tolerance
in salmonids is known to increase with body size (Hoar 1988, Ugedal et al. 1998), most
individuals in the present study probably had good osmoregulatory capabilities. Larsen et al.
(2008) suggested that local adaptation may cause differences in seawater tolerance among sea
trout populations. However, the innermost parts of both Snillfjord and Hemnfjord had levels of
salinities similar to the outer parts of the fjord system during the present study, further suggesting
that salinity likely did not affect the spatial distribution of the experimental fish in the fjords to
any great extent.
Long distance migrants, who were found to be older than both short and medium distance
migrants, surprisingly spent a shorter time at the sea than individuals moving shorter distances.
Previous studies have shown that older sea trout individuals generally return earlier from the
marine migration (Jonsson 1985), however, the reasons for this remain obscure.
LITTORAL VERSUS PELAGIC HABITAT UTILIZATION
The sea trout stayed more often in littoral than pelagic habitats, based on the observed higher
proportions of registrations of tagged fish on acoustic receivers in near shore compared to pelagic
areas. These results are consistent with findings by Jensen et al. (2014), who found that sea trout
in the Alta Fjord only spent 33% of their time in the pelagic habitat. The near shore habitat
utilization is also consistent with previous studies on sea trout feeding behaviour, which suggest
that the main prey (crustaceans, polychaetes, insects and fish) are found in near shore, shallow
areas (Pemberton 1976, Knutsen et al. 2001). However, the data also show that the pelagic zone
may be an important habitat for especially the long distance migrants, and pelagic feeders are in
other studies have been shown to feed almost exclusively on fish (Rikardsen and Amundsen
2005). The long distance migrants in this study spent a minimal portion of their total marine
residence time in the innermost areas of the fjord, compared to short and medium distance
migrants. Long distance migrants had greater proportions of pelagic registrations than medium
distance migrants, and tended (nearly statistically significant) to show greater proportions of
pelagic registrations compared to short distance migrants.
Overall, the data suggests that the long distance migrants had a higher degree of pelagic
feeding behaviour, that they were in lower condition at the start of the migration and that they
returned earlier than the medium and short distance migrants. It is likely that these fish found
more energy rich prey in the outer part of the fjord and therefore potentially gained weight faster
and therefore also returned earlier to freshwater as they had utilized their compensatory growth
potential. Energy rich pelagic fish species are often found to be a considerable part of the diet in
larger sea trout, with herring (Clupea harengus L. 1758) as a key prey species (Pemberton 1976,
Knutsen et al. 2001, Rikardsen and Amundsen 2005, Rikardsen et al. 2006).
In summary, this study showed that sea trout both within and between watercourses
draining to the same fjord system may differ in morphology, life history, migration behaviour and
marine habitat use. Such plasticity may reinforce population resilience in areas with dynamic
environmental conditions or during periods of climatic changes. Altered patterns of fish
migration have often documented as an effect of contemporary global climate change is (e.g.
Cotton 2003, Parmesan 2007, Visser et al. 2009). A better understanding of the underlying causes
of the different marine migratory strategies in sea trout is now needed in order to predict how
changes in the marine habitat and different anthropogenic impacts may influence brown trout
populations with anadromous individuals.
This study was financed or supported by contributions from the Hemne municipality, the County
Governor of Sør-Trøndelag, Sør-Trøndelag County Authority, the Norwegian Environment
Agency, the Lake Rovatnet landowners association, the TrønderEnergi AS, the Aqua Gen AS, ,
the Norwegian Institute for Nature Research, the DTU Aqua, the Arctic University of Norway,
and the NTNU University Museum. The crew of RV Gunnerus (Martin Georg Hansen, Ola
Magne Taftø, Hans Erlandsen, Vegard Pedersen Sollien, Paul Skarsvåg, Stein Hugo
Hemmingsen and Kristian Lian) are all thanked for their extensive help during the field work.
Marc Daverdin at the NTNU Museum is thanked for assistance with data analyses. The acoustic
receivers were provided by Dalhousie University’s Ocean Tracking Network,
www.oceantrackingnetwork.org. The experimental procedures were approved by the Norwegian
National Animal Research Authority.
Aarestrup, K., Baktoft, H., Koed, A., del Villar-Guerra, D., and Thorstad, E.B. 2014. Comparison
of the riverine and early marine migration behaviour and survival of wild and hatchery-
reared sea trout Salmo trutta smolt. Mar. Ecol. Prog. Ser. 496: 197-206.
Aarestrup, K., Baktoft, H., Thorstad, E.B., Svendsen, J.C., Höjesjö, J., and Koed, A. Accepted.
Survival and progression rates of sea trout kelts (Salmo trutta L.) during downstream
migration in freshwater and at sea. Mar. Ecol. Prog. Ser.
Acolas, M., Labonne, J., Baglinière, J., and Roussel, J. 2012. The role of body size versus growth
on the decision to migrate: a case study with Salmo trutta. Naturwissenschaften 99: 11-21.
Anderson, D.R., Link, W.A., Johnson, D.H., and Burnham, K.P. 2001. Suggestions for presenting
the results of data analyses. J. Wildl. Manage. 65: 373-378.
Anon. 2011. Status of Norwegian salmon populations in 2011. Norwegian Scientific Advisory
Committee for Atlantic Salmon Management, pp. 1-285.
Barton´, K. 2015. MuMIn - R package for model selection and multi-model inference.
Bendall, B., Moore, A., and Quayle, V. 2005. The post-spawning movements of migratory brown
trout Salmo trutta L. J. Fish Biol. 67: 809-822.
Berg, O.K., and Berg, M. 1987. Migrations of sea trout, Salmo trutta L., from the Vardnes river
in northern Norway. J. Fish Biol. 31: 113-121.
Boel, M., Aarestrup, K., Baktoft, H., Larsen, T., Madsen, S.S., Malte, H., Skov, C., Svendsen,
J.C., and Koed, A. 2014. The physiological basis of the migration continuum in brown
trout (Salmo trutta). Physiol. Biochem. Zool. 87: 334-345.
Chernitsky, A.G., Zabruskov, G.V., Ermolaev, V.V., and Shkurko, D.S. 1995. Life history of
trout, Salmo trutta L., in the Varsina River estuary, (The Barents Sea). Nord. J. Freshwat.
Res. 71: 183-189.
Cooke, S.J., Woodley, C.M., Eppard, M.B., Brown, R.S., and Nielsen, J.L. 2011. Advancing the
surgical implantation of electronic tags in fish: a gap analysis and research agenda based
on a review of trends in intracoelomic tagging effects studies. Rev. Fish Biol. Fish. 21:
Cotton, P.A. 2003. Avian migration phenology and global climate change. Proc. Natl. Acad. Sci.
USA 100: 12219-12222.
Dahl, K. 1910. Alder og vekst hos laks belyst ved studier av deres skjæl. p. 60. Centraltrykkeriet,
Damsgård, B., and Dill, L.M. 1998. Risk-taking behavior in weight-compensating coho salmon,
Oncorhynchus kisutch. Behav. Ecol. 9: 26-32.
Davidsen, J.G., Daverdin, M., Sjursen, A.D., Rønning, L., Arnekleiv, J.V., and Koksvik, J.I.
2014. Does reduced feeding prior to release improve the marine migration of hatchery
brown trout Salmo trutta L. smolts? J. Fish Biol. 85: 1992-2002.
del Villar-Guerra, D., Aarestrup, K., Skov, C., and Koed, A. 2013. Marine migrations in
anadromous brown trout (Salmo trutta). Fjord residency as a possible alternative in the
continuum of migration to the open sea. Ecol. Freshwat. Fish 23: 594-693.
Drenner, S.M., Clark, T.D., Whitney, C.K., Martins, E.G., Cooke, S.J., and Hinch, S.G. 2012. A
synthesis of tagging studies examining the behaviour and survival of anadromous
salmonids in marine environments. Plos One 7(3): e31311.
Forseth, T., Næsje, T., Jonsson, B., and Hårsaker, K. 1999. Juvenile migration in brown trout: a
consequence of energetic state. J. Anim. Ecol. 68: 783-793.
Gross, M.R., Coleman, R.M., and McDowall, R.M. 1988. Aquatic productivity and the evolution
of diadromous fish migration. Science 239: 1291-1293.
Halttunen, E., Jensen, J.L.A., Næsje, T.F., Davidsen, J.G., Thorstad, E.B., Chittenden, C.M.,
Hamel, S., Primicerio, R., and Rikardsen, A.H. 2013. State-dependent migratory timing of
postspawned Atlantic salmon (Salmo salar). Can. J. Fish. Aquat. Sci. 70: 1063-1071.
Hendry, A.P., Bohlin, T., Jonsson, B., and Berg, O.K. 2004. To sea or not to sea? Anadromy
versus non-anadromy in salmonids. In Evolution Illuminated: Salmon and Their
Relatives. Edited by A.P. Hendry and S.C. Stearns. Oxford University Press, New York,
NY. USA. pp. 92-125.
Hoar, W.S. 1988. The physiology of smolting salmonids. In Fish Physiology, Vol. XI B. Edited
by W.S. Hoar and D.J. Randall. Academic Press, New York. pp. 275-343.
Hothorn, T., Bretz, F., and Westfall, P. 2008. Simultaneous inference in general parametric
models. Biometr. J. 50: 346-363.
ICES. 2013. Report of the Workshop on Sea Trout (WKTRUTTA), 12–14 November 2013,
ICES Headquarters, Copenhagen, Denmark ICES CM 2013/SSGEF:15, p. 280.
Jensen, A.J., Finstad, B., Fiske, P., Hvidsten, N.A., Rikardsen, A.H., and Saksgård, L. 2012.
Timing of smolt migration in sympatric populations of Atlantic salmon (Salmo salar),
brown trout (Salmo trutta), and Arctic charr (Salvelinus alpinus). Can. J. Fish. Aquat. Sci.
Jensen, J.L.A., and Rikardsen, A.H. 2008. Do northern riverine anadromous Arctic charr
Salvelinus alpinus and sea trout Salmo trutta overwinter in estuarine and marine waters?
J. Fish Biol. 73: 1810–1818.
Jensen, J.L.A., and Rikardsen, A.H. 2012. Archival tags reveal that Arctic charr Salvelinus
alpinus and brown trout Salmo trutta can use estuarine and marine waters during winter.
J. Fish Biol. 81: 735-749.
Jensen, J.L.A., Rikardsen, A.H., Thorstad, E.B., Suhr, A.H., Davidsen, J.G., and Primicerio, R.
2014. Water temperatures influence the marine area use of Salvelinus alpinus and Salmo
trutta. J. Fish Biol. 84: 1640–1653.
Jensen, K.W. 1968. Seatrout (Salmo trutta L.) of the River Istra, Western Norway. Rep. Inst.
Fresh. Res. Drottningholm 48: 1187-1213.
Jonsson, B. 1985. Life history patterns of freshwater resident and sea-run migrant brown trout in
Norway. Trans. Am. Fish. Soc. 114: 182-194.
Jonsson, B., and Jonsson, N. 1993. Partial migration: niche shift versus sexual maturation in
fishes. Rev. Fish Biol. Fish. 3: 348-365.
Jonsson, B., and Jonsson, N. 2009. Migratory timing, marine survival and growth of anadromous
brown trout Salmo trutta in the River Imsa, Norway. J. Fish Biol. 74: 621-638.
Jonsson, B., and Jonsson, N. 2011. Ecology of Atlantic salmon and brown trout: habitat as a
template for life histories. p. 680. Springer Science+Buisness Media B.V., London.
Jonsson, N., and Jonsson, B. 2002. Migration of anadromous brown trout (Salmo trutta) in a
Norwegian river. Freshwat. Biol. 47: 1391-1401.
Klemetsen, A., Amundsen, P.-A., Dempson, J.B., Jonsson, B., Jonsson, N., O´Connell, M.F., and
Mortensen, E. 2003. Atlantic salmon Salmo salar L., brown trout Salmo trutta L., and
Arctic charr Salvelinus alpinus (L.): a review of aspects of their life histories. Ecol.
Freshwat. Fish 12: 1-59.
Knutsen, J.A., Knutsen, H., Gjøsæter, J., and Jonsson, B. 2001. Food of anadromous brown trout
at sea. J. Fish Biol. 59: 533-543.
Knutsen, J.A., Knutsen, H., Olsen, E.M., and Jonsson, B. 2004. Marine feeding of anadromous
brown trout (Salmo trutta L.) during winter. J. Fish Biol. 64: 89-99.
Larsen, P.F., Nielsen, P.F., Koed, A., Thomsen, D.S., Olsvik, P.A., and Loeschcke, V. 2008.
Interpopulation differences in expression of candidate genes for salinity tolerance in
winter migrating anadromous brown trout (Salmo trutta L.). BMC Genet. 9: 1-9.
Lea, E. 1910. On the methods used in herring investigations. Publs. Circonst. Cons. Perm. Int
Explor. Mer. 53: 7-25.
MacCrimmon, H.R., and Marshall, T. 1968. World distribution of brown trout, Salmo trutta. J.
Fish Res. Board. Can. 25: 2527-2548.
MacCrimmon, H.R., Marshall, T.L., and Gots, B.L. 1970. World distribution of brown trout,
Salmo trutta: further observations. J. Fish Res. Board. Can. 4: 811-818.
Milner, N.J., Elliott, J.M., Armstrong, J.D., Gardiner, R., Welton, J.S., and Ladle, M. 2003. The
natural control of salmon and trout populations in streams. Fish. Res. 62: 111-125.
Moore, A., Ives, M., Scott, M., and Bamber, S. 1998. The migratory behaviour of wild sea trout
(Salmo trutta L.) smolts in the estuary of the River Conwy, North Wales. Aquaculture
Parmesan, C. 2007. Influences of species, latitudes and methodologies on estimates of
phenological response to global warming. Glob. Chang. Biol. 13: 1860-1872.
Pemberton, R. 1976. Sea trout in North Argyll sea lochs. 2. Diet. J. Fish Biol. 9: 195-208.
Piggins, D.J. 1964. The results of tagging sea trout kelts during 1962 and 1963. Rep. Salmon Res.
Pulido, F. 2011. Evolutionary genetics of partial migration: the threshold model of migration
revis(it)ed. Oikos 120: 1776-1783.
Rikardsen, A.H., and Amundsen, P.A. 2005. Pelagic marine feeding of Arctic charr and sea trout.
J. Fish Biol. 66: 1163-1166.
Rikardsen, A.H., Amundsen, P.A., Knudsen, R., and Sandring, S. 2006. Seasonal marine feeding
and body condition of sea trout (Salmo trutta) at its northern distribution. ICES J. Mar.
Sci. 63: 466-475.
Rikardsen, A.H., Dempson, J.B., Amundsen, P.-A., Bjørn, P.A., Finstad, B., and Jensen, A.J.
2007. Temporal variability in marine feeding of sympatric Arctic charr and sea trout. J.
Fish Biol. 70: 837-847.
Svärdson, G., and Fagerström, Å. 1982. Adaptive differences in the long-distance migration of
some trout (Salmo trutta L.) stocks. Rep. Inst. Freshw. Res., Drottningholm 60: 51-80.
Thorpe, J.E. 1987. Smolting versus residency: developmental conflicts in salmonids. In Common
Strategies of anadromous and aatadromous Fishes. Edited by M.J. Dadswell, R.J. Klauda,
C.M. Moffitt, R.L. Saunders, R.A. Rulifson and J.E. Cooper. American Fisheries Society
Symposium, Bethesda. pp. 244-252.
Thorstad, E.B., Økland, F., Finstad, B., Sivertsgård, R., Bjørn, P.A., and McKinley, R.S. 2004.
Migration speeds and orientation of Atlantic salmon and sea trout post-smolts in a
Norwegian fjord system. Environ. Biol. Fishes 71: 305-311.
Ugedal, O., Finstad, B., Damsgård, B., and Mortensen, A. 1998. Seawater tolerance and
downstream migration in hatchery-reared and wild brown trout. Aquaculture 168: 395-
Visser, M.E., Perdeck, A.C., Van Balen, J.H., and Both, C. 2009. Climate change leads to
decreasing bird migration distances. Glob. Chang. Biol. 15: 1859-1865.
Went, A.E.J. 1962. Irish sea trout, a review of investigations to date. Sci. Proc. R. Dublin Soc.
Wysujack, K., Greenberg, L.A., Bergman, E., and Olsson, I.C. 2009. The role of the environment
in partial migration: food availability affects the adoption of a migratory tactic in brown
trout Salmo trutta. Ecol. Freshwat. Fish 18: 25-59.
Závorka, L., Slavík, O., and Horký, P. 2014. Validation of scale-reading estimates of age and
growth in a brown trout Salmo trutta population. Biologia 69: 691-695.
Figure 1: Locations of automatic receivers (red pentagon = marine, green circle = freshwater)
and temperature and salinity data loggers (T/S) in the study area. Area zones (Z1-Z5) and outer
boundaries for definition of short (S) and medium (M) migration distance (fish from the Søa
watercourse = green lines, fish from the Snilldalselva River = red lines). Arrays across the fjord
included both near shore and pelagic receivers (H1, H2, H3 and S1).
Figure 2: Total residence time (days) in the marine environment during 1 April – 1 October 2012
or 2013 for tagging groups HS12 (tagged in Lake Rovatnet in spring 2012), HA12 (tagged in
Lake Rovatnet in autumn 2012), HS13 (tagged in river mouth of River Søa in spring 2013), SA12
(tagged in River Snilldalselva in autumn 2012), and SS13 (tagged in the river mouth of River
Snilldalselva in spring 2013). The box-and-whisker plots show median values (black lines), the
interquartile ranges (boxes) and the 5th and 95th percentiles (whiskers).
Figure 3: Residence time in the different fjord zones of short, medium and long distance
migrants during 1 April-1 October. The different fjord zones are indicated in Fig. 1. The box-and-
whisker plots show median values (black lines), the interquartile ranges (boxes) and the 5th and
95th percentiles (whiskers). Circles indicate outliers.
Figure 4: Proportions of individuals’ registrations at near shore (white) and pelagic (grey)
receivers at array H1, H2, H3 and S1 during 1 April-1 October. The box-and-whisker plots show
median values (black lines), the interquartile ranges (boxes) and the 5th and 95th percentiles
(whiskers). Circles indicate outliers.
Figure 5: Proportions of pelagic registrations at receiver arrays (H1, H2, H3 and S1) for short,
medium and long distance migrants during 1 April-1 October. The box-and-whisker plots show
median values (black lines), the interquartile ranges (boxes) and the 5th and 95th percentiles
(whiskers). Circles indicate outliers.
Table 1: Tagging groups, tagging location, number of individuals, natural body length (LN), body
mass, body condition, age, back calculated smolt length, age at smoltification and number of
previous marine seasons prior to tagging of fish in the different groups.
Tagging group HS12 HA12 SA12 SS13 HS13
Tagging date 12-14 April 2012
Capture and tagging site Søa (Lake Rovatnet) Søa (Lake
n 30 21 20 15 29
Natural length (mm) Mean ± SD 396 ± 61 412 ± 121 392 ± 75 381 ± 53 417 ± 55
Range 335-600 270-700 310-650 275-460 330-580
Body mass (g) Mean ± SD 586 ± 287 866 ± 908 581 ± 419 620 ± 286 713 ± 337
Range 330-1600 210-3660 310-2180 220-1210 300-1970
Fulton's K Mean ± SD 0.90 ± 0.12 0.95 ± 0.12 0.89 ± 0.09 1.05 ± 0.10 0.89 ± 0.10
Range 0.74-1.22 0.77-1.30 0.73-1.07 0.87-1.33 0.75-1.07
Scale reading estimates
Smolt length (mm)
Mean ± SD 166 ± 42 182 ± 52 132 ± 30 140 ± 34 137 ± 32
Range 105-270 112-276 98-197 102-236 96-210
n (cover) 22 (73%) 14 (67%) 18 (90%) 12 (80%) 22 (76%)
Age at smoltification (years)
Mean ± SD 2.63 ± 0.72 3.00 ± 0.74 2.35 ± 0.61 2.27 ± 0.65 2.19 ± 0.40
Range 2-4 2-4 2-4 2-4 2-3
n (cover) 16 (53%) 12 (57%) 17 (85%) 11 (73%) 21 (72%)
Previous marine seasons
Mean ± SD 3.39 ± 1.24 3.92 ± 2.36 3.43 ± 1.02 2.40 ± 0.52 3.06 ± 0.68
Range 2-7 2-10 2-6 2-3 2-4
n (cover) 18 (60%) 13 (62%) 14 (70%) 10 (67%) 16 (55%)
Mean ± SD 5.69 ± 1.65 6.73 ± 2.37 5.85 ± 1.28 4.78 ± 0.83 5.20 ± 0.77
Range 4-10 4-13 5-9 4-6 4-7
n (cover) 13 (43%) 15 (71%) 13 (65%) 9 (60%) 15 (52%)
Table 2: Differences in morphology and life history among fish from watercourses and tagging groups (HS12; Lake Rovatnet in
spring 2012, HA12; Lake Rovatnet in autumn 2012, SA12; River Snilldalselva in autumn 2012, SS13; Mouth of River Snilldalselva in
spring 2013, HS13; Mouth of River Søa in spring 2013). Significant P-values are shown in bold, non-significant Tukey ANOVA
values are excluded.
Morphological and life history
characteristic Alternative hypothesis (H1) Statistical test n P
fish tagged in Lake
Rovatnet and fish
tagged in River
Body length HS12 and HA12 <> SA12 and SS13 t-test 86 0.321
Body condition HS12 and HA12 <> SA12 and SS13 t-test 86 0.127
Smolt length HS12 and HA12 ≤ SA12 and SS13 t-test 66 < 0.001
Age at smoltification HS12 and HA12 ≤ SA12 and SS13 t-test 56 0.007
Previous marine seasons HS12 and HA12 ≤ SA12 and SS13 t-test 55 0.055
Age HS12 and HA12 ≤ SA12 and SS13 t-test 50 0.042
fish tagged in Lake
Rovatnet and fish
tagged in mouth of
Length HS12 and HA12 <> HS13 t-test 80 0.422
Body condition HS12 and HA12 <> HS13 t-test 80 0.258
Smolt length HS12 and HA12 ≤ HS13 t-test 58 < 0.001
Age at smoltification HS12 and HA12 ≤ HS13 t-test 49 < 0.001
Previous marine seasons HS12 and HA12 <> HS13 t-test 47 0.136
Age HS12 and HA12 ≤ HS13 t-test 43 0.012
fish tagged in River
tagged in mouth of
Body length SA12 and SS13 ≥ HS13 t-test 64 0.025
Body condition SA12 and SS13 ≤ HS13 t-test 64 0.014
Smolt lenght SA12 and SS13 <> HS13 t-test 52 0.817
Age at smoltification SA12 and SS13 <> HS13 t-test 49 0.372
Previous marine seasons SA12 and SS13 <> HS13 t-test 40 0.813
Age SA12 and SS13 <> HS13 t-test 37 0.528
groups of tagging
Body length HS12 <> HA12 <> SA12 <> SS13 <>
HS13 ANOVA 115 0.78
Body condition SS13 ≤ HS12 Tukey ANOVA 45 0.014
Body condition SS13 ≤ HS13 Tukey ANOVA 43 0.009
Smolt lengtht HS12 ≤ SA12 Tukey ANOVA 40 0.05
Smolt length HA12 ≤ SA12 Tukey ANOVA 32 0.004
Smolt length HA12 ≤ HS13 Tukey ANOVA 36 0.008
Smolt length HA12 ≤ SS13 Tukey ANOVA 26 0.044
Age at smoltification HA12 ≤ HS13 Tukey ANOVA 33 0.004
Age at smoltification HA12 ≤ SA12 Tukey ANOVA 29 0.049
Age at smoltification HA12 ≤ SS13 Tukey ANOVA 23 0.044
Previous marine seasons HS12 <> HA12 <> SA12 <> SS13 <>
HS13 ANOVA 71 0.098
Age HA12 ≤ SS13 Tukey ANOVA 24 0.032
Table 3: Model selection for estimating the determinants of the duration of the marine residence
time. The models estimate the relative contributions to the duration of the marine residence time
from the parameters age (A), body length (L), number of previous marine seasons (P), residual
values (resvalbc) from the linear model log(condition)~log(length) (R), Julian day of sea entry
(SE), smolt age (SA), smolt length (SL), and maximum distance migrated away from the home
river (S). AIC is the score based on Akaike information criterion. AIC weights represent the
relative likelihood of the model. The table displays the four best fitting of the total of 576 tested
Table 4: Natural body length (LN), Fulton’s body condition, age, back calculated smolt length,
age at smoltification, number of previous marine seasons and total marine residence time during
summer of short, medium and long distance migrants.
N (%) 35 (40%) 16 (18%) 37 (42%) 88 (100%)
Natural body length (mm)
Fulton’s body condition
Scale reading estimates
Smolt length (mm)
Mean 137 127 166 148
SD ± 35 ± 37 ± 44 ± 44
n 26 12 31 69
Mean 2.24 2.18 2.68 2.41
SD ± 0.52 ± 0.60 ± 0.80 ± 0.69
n 25 11 25 61
Mean 3.00 2.64 3.52 3.18
SD ± 0.61 ± 0.67 ± 1.34 ± 1.09
n 17 11 27 55
Mean 5.13 4.90 6.12 5.57
SD ± 0.72 ± 0.74 ± 1.62 ± 1.35
n 16 10 25 51