Environmental biology of fishes, tập 95, số 1, 2012

Environ Biol Fish (2012) 93:305–318
DOI 10.1007/s10641-011-9914-z

Short-and long term niche segregation and individual
specialization of brown trout (Salmo trutta)
in species poor Faroese lakes
Jakob Brodersen & Hilmar J. Malmquist & Frank Landkildehus &
Torben L. Lauridsen & Susanne L. Amsinck & Rikke Bjerring &
Martin Søndergaard & Liselotte S. Johansson & Kirsten S. Christoffersen &
Erik Jeppesen

Received: 24 January 2011 / Accepted: 14 August 2011 / Published online: 16 September 2011
# Springer Science+Business Media B.V. 2011

Abstract Trophic niche divergence is considered to
be a major process by which species coexistence is
facilitated. When studying niche segregation in lake
ecosystems, we tend to view the niche on a onedimensional pelagic-littoral axis. In reality, however,
the niche use may be more complex and individual
fidelity to a niche may be variable both between and
within populations. In order to study this complexity,
J. Brodersen : F. Landkildehus : T. L. Lauridsen :
S. L. Amsinck : R. Bjerring : M. Søndergaard :
L. S. Johansson : E. Jeppesen
Department of Bioscience, Aarhus University,
Vejlsøvej 25,
DK-8600 Silkeborg, Denmark
J. Brodersen
Department of Biology/Aquatic Ecology, Lund University,
223 62 Lund, Sweden

H. J. Malmquist
Natural History Museum of Kópavogur,
Hamraborg 6a,
IS-200 Kópavogur, Iceland
K. S. Christoffersen
Freshwater Biological Laboratory,
University of Copenhagen,
Helsingørsgade 51,
DK-3400 Hillerød, Denmark

relative simple systems with few species are needed.
In this paper, we study how competitor presence
affects the resource use of brown trout (Salmo trutta)
in 11 species-poor Faroese lakes by comparing
relative abundance, stable isotope ratios and diet in
multiple habitats. In the presence of three-spined
sticklebacks (Gasterosteus aculeatus), a higher proportion of the trout population was found in the
E. Jeppesen
Greenland Climate Research Centre (GCRC),
Greenland Institute of Natural Resources,
Nuuk, Greenland

E. Jeppesen
The Sino-Danish Center for Education
and Research (SDC),
Beijing, China

Present Address:
J. Brodersen (*)
Department of Fish Ecology and Evolution, EAWAG Swiss

Federal Institute of Aquatic Science and Technology,
Center of Ecology, Evolution and Biochemistry,
Seestrasse 79,
CH-6047 Kastanienbaum, Switzerland
e-mail:


306

pelagic habitat, and trout in general relied on a more
pelagic diet base as compared to trout living in allopatry
or in sympatry with Arctic charr (Salvelinus alpinus).
Diet analyses revealed, however, that niche-segregation
may be more complex than described on a onedimensional pelagic-littoral axis. Trout from both
littoral and offshore benthic habitats had in the
presence of sticklebacks a less benthic diet as
compared to trout living in allopatry or in sympatry
with charr. Furthermore, we found individual habitat
specialization between littoral/benthic and pelagic trout
in deep lakes. Hence, our findings indicate that for
trout populations interspecific competition can drive
shifts in both habitat and niche use, but at the same
time they illustrate the complexity of the ecological
niche in freshwater ecosystems.
Keywords Niche complexity . Stable isotopes . Trout .
Stickleback . Aquatic ecology . Faroe Islands

Introduction
A central, but much debated (e.g. Hubbell 2001;
Chase and Leibold 2003) concept in ecological theory

is the ‘ecological niche’. Hutchinson (1957; 1959)
originally defined the ecological niche as a hypervolume in an n-dimensional space with environmental
variables as axes. However, empirical measurements
of all potential dimensions will probably never be
accomplished for any species occurring in a natural
ecosystem (Chase and Leibold 2003). Ecologists are
therefore challenged as they have to reduce the
number of potential axes of resource specialization
to a single or a few measurable axes. In lake
ecosystems, the niche is often measured on a twodimensional scale with limnetic/pelagic and benthic/
littoral organisms as end points (e.g. Schluter and
McPhail 1992; Svanbäck and Persson 2009), which
conveniently can be determined by two end-member
stable isotope analyses (Post 2002). However, when
treating the niche as a one-dimensional variable, we
may trade-off the ability to measure the complexity of
reality for convenience. Also we might end up
measuring habitat use rather than niche. These
concepts of habitat and niche are highly entangled,
which is likely due to the confusion of whether niche
refers to aspects of environment or species (Chase and
Leibold 2003). There are in that regard a large

Environ Biol Fish (2012) 93:305–318

biological difference between how a species exploits
a habitat and in which habitat a species forage.
However, it is not only important to recognize that
each species has a certain niche, but also to
acknowledge that each individual in a population

may vary in its niche use, both compared to other
individuals, but also temporally. The importance of
individual phenotypic variation is generally recognized as the raw material on which evolution acts.
Individual flexibility may enable adaptation to current
conditions in the changing environment, and the sum
of individual adaptations will shape the response of
populations to variations in the environment, for
instance changed competition or predation pressure.
Competition may particularly affect niche use. In
sympatry, ecologically similar species are expected to
diverge in habitat use and/or diet, whereas in
allopatry, species are expected to converge in their
use of the same primary resources (MacArthur and
Levins 1967; Schoener 1982; Tilman 1987).
In fishes, partitioning of resources by ecologically
similar species has been well documented, in particular among Arctic charr (Salvelinus alpinus (L.)) and
brown trout (Salmo trutta L.) in temperate lakes
(Langeland et al. 1991; Jansen et al. 2002, Klemetsen
et al. 2003; Jonsson et al. 2008). In sympatry, charr
and trout populations usually utilize distinct habitat
and prey resources. Generally, charr feed on zooplankton in offshore habitats, while trout utilize the
littoral zone and feed on benthic invertebrates and
surface arthropods (Langeland et al. 1991; Klemetsen
et al. 2003; Jonsson et al. 2008). In allopatry, charr,
but not trout, usually alter their use of resources and
exploit the littoral zone to a greater extent. Therefore,
trout are usually regarded the competitively superior,
and shifts in charr habitat use are ascribed to trout
forcing charr to use alternative resources (e.g.
Klemetsen et al. 2003). While this pattern is well

described in the literature, less is known about
resource use by brown trout living in sympatry with
fish species other than Arctic charr. An emerging
question is whether more specialized littoral species
may drive trout into a more pelagic resource use.
Undertaking field studies on competition and
behavioral adaptations is difficult because the observed behavior is the sum of complex interactions,
where each consumer displays dietary overlap with
several other species (Tilman 1987; Hansson 1995).
Therefore, species-poor ecosystems serve as valuable


Environ Biol Fish (2012) 93:305–318

sites for the study of behavioral interactions and niche
segregation. On the Faroe Islands, situated in the MidAtlantic, a total of seven freshwater fish species occur,
but a maximum of four coexist in a single lake
(Jeppesen et al. 2002a; unpubl. data). Most Faroese
lakes host brown trout only, but in some lakes other
fish species co-occur, usually only three-spined
stickleback (Gasterosteus aculeatus L.).
A previous study of four Faroese lakes revealed that
in the one lake with Arctic charr, Lake Leynavatn,
brown trout relied more on benthic food than in the lakes
without charr (Malmquist et al. 2002). Moreover, the
density, somatic growth and condition factor of trout
were lowest in Leynavatn (Malmquist et al. 2002).
Stable isotope investigations supported the suggestion
that interspecific competition between trout and charr
was important in Leynavatn, with trout diet appearing

to be of a more littoral origin (Jeppesen et al. 2002b).
This study also indicated that the presence of threespined sticklebacks in other lakes may drive trout into
a more pelagic feeding mode (Jeppesen et al. 2002b).
However, in order to verify this theory, more comprehensive studies with more lakes are needed.
In this study we investigated the differences in
habitat use and diet of brown trout in eleven
species-poor Faroese lakes with notably different
fish communities. Our aim was to determine if the
habitat use and diet of trout were affected by fish
community structure and other environmental variables such as lake depth, area, and nutrient status.
We also examined the role of body size of trout in
relation to diet and habitat as another metric of
competitive interactions. We expected that the
fundamental niche as described by diet and habitat
of brown trout in lakes with only trout would differ
from the realized niche described by diet and
habitat use in lakes with presence of potential fish
competitors, i.e. Arctic charr and three-spined
sticklebacks.

Materials and methods
Study area
All sampling was carried out during July–August
2000 in 11 lakes located on the five largest Faroese
islands (Fig. 1). The lakes included a wide range of
areas (0.6–356 ha) and depths (max depth: 0.7–52 m)

307

(Table 1). Total phosphorous concentrations varied,

with the highest nutrient levels occurring in Lake
Vatnsnes which is used for rearing Atlantic salmon
(Salmo salar L.) in cages. Brown trout is the most
widespread freshwater fish species on the Faroe
Islands and occurred in all the study lakes. Threespined stickleback (hereafter sticklebacks) were
found in five and European eel (Anguilla anguilla
L.) in four of the study lakes. Arctic charr, European
flounder (Platichthys flesus (L.)) and Atlantic salmon occurred only in one study lake (salmon due to
fish farming). As shown in Table 1, the lakes varied
substantially in a number of physical, chemical and
biological parameters, all of which can be hypothesized to influence trout ecology. Compared to analysis of
very similar lake types, this will further enable us to
evaluate the relative importance of competitor presence/
absence compared to other environmental factors. For
further information on the characteristics of Faroese
lakes, see Landkildehus et al. (2002).
Fish sampling
Fish were caught in multi-mesh sized gill nets (type
NORDIC: 14 different mesh sizes ranging from 6.2575 mm (Appelberg et al. 1995)), placed overnight.
“Littoral nets” were set parallel to the shore at a depth
of 1.5 m, offshore nets in the middle of the lake along
the major length axis of the lake- either at the bottom,
“offshore benthic net” and in lakes with a maximum
depths >8 m also in the mid pelagic (in half the depth
of the epilimnion), “pelagic nets” (Jeppesen et al.
2001). The number of nets ranged between 4 and 9
per lake depending on area and depth, except in
Mjavavatn, a small lake where only one mid-lake
offshore benthic net was used. Catch per unit effort of
trout (CPUE: #trout net−1 h−1) was used as a measure

for relative fish density, both for each habitat within a
lake and as an average for all nets per lake. Each fish
was measured (fork length) to the nearest mm and
weighed to the nearest 0.1 g. Based on individual fish
weights, we further calculated trout biomass per unit
of effort (BPUE).
Brown trout stomachs were removed after capture
and stored individually in 96% ethanol. Brown trout,
Arctic charr and sticklebacks were frozen individually
for stable isotope analysis. Although sticklebacks
were often caught in gill nets, we additionally
sampled the lakes with fyke nets and shoreline netting


308

Environ Biol Fish (2012) 93:305–318
7°30'

7°00'

0

6°30'

20 km
62°20'

62°20'

Streymoy
Saksunarvatn

Esteroy
Leynavatn
Mjauvøtn

Fjallavatn

Toftavatn

Vagar
Sørvagsvatn

62°00'

62°00'

Sandsvatn
Grothusvatn

Sandoy

61°40'
61°40'

Suderoy

Bessavatn
Vatnsnes

Mjavavatn

7°30'

7°00'

6°30'

Fig. 1 Geographical location of the eleven Faroese study lakes

to validate presence/absence of sticklebacks. In
Sandsvatn, Sørvágsvatn and Toftavatn, sticklebacks
sampled with fyke nets were additionally used for
stable isotope analyses.

Stomach content analysis
Stomach contents of 165 trout (7.5–39.5 cm) were
enumerated and identified to the lowest taxonomic


Environ Biol Fish (2012) 93:305–318

309

Table 1 Characteristics of the 11 Faroese study lakes.
Chlorophyll data are from Amsinck et al. (2006). A=lake area
(ha); Ptot =total phosphorus (μg l−1), Zm =maximum depth (m);
S=Secchi depth (m), Chl a=chlorophyll a (μg l−1); Zoo=
crustacean zooplankton (# l−1). CPUE data refer to average

number caught per net per hour. Fish fauna: T: Brown trout; TS:
Three-spined stickleback; C: Arctic Charr; E: Eel; S: Atlantic
Salmon and F: Flounder. *In Sørvágsvatn, no sticklebacks were
caught in gill nets, some individuals being caught in fyke nets

Lake

A

Zm

Ptot

S

Chl a

Zoo

CPUET

CPUETS

CPUEC

CPUES

CPUEF

Fish fauna

Mjávavatn

0.6

0.8

19

0.8

1.81

0.0

0.50

0

0

0

0

T

Bessavatn

5.4

2.0

30

2.0

1.98

54.0

0.22

0

0

0

0

T

Saksunarvatn

8.1

16.0

6

8.8

1.13

53.2

0.35

0

0

0

0

T&E

Mjáuvøtn

3.1

5.7

15.2

4.3

1.76

175.5

1.83

0

0

0

0

T

Leynavatn

18.0

32.5

3.4

10.0

1.23

21.7

0.51

0

1.06

0

0

T&C

Vatnsnes

14.7

9.5

76

1.7

25.17

42.5

0.22

0

0

0.04

0

T&S

Sandsvatn

79.7

2.4

43

2.4

1.05

85.6

0.38

0.02

0

0

1.02

T, TS, F & E

Gróthúsvatn

13.4

0.7

35

0.7

1.03

22.8

0.69

0.01

0

0

0

T, TS & E

Sørvágsvatn

356.0

52.0

5.2

12.5

0.72

5.2

0.36

0*

0

0

0

T & TS

Toftavatn

52.2

17.5

10.8

5.8

0.98

25.6

0.21

0.03

0

0

0

T, TS & E

Fjallavatn

101.9

46.6

3

14.0

0.46

4.8

0.56

0.02

0

0

0

T & TS

level possible. For each lake a maximum of eight fish,
representing all sizes if possible, was selected
randomly from the fish caught in each of the three
habitats sampled: the littoral, the pelagic and the midlake offshore benthic nets defined above.
Stomach contents were identified and counted
using a dissection microscope. Food items were
identified to group (Hirudinea, Hydracarina, Turbellaria and Diptera other than mentioned below),
order (Coleoptera, Heteroptera, Trichoptera and
Ostracoda, Copepoda), family (Tipulidae), subfamily (Chironominae, Orthocladinae, Tanypodinae),
genus (Gammarus, Lymnaea, Pisidium, Daphnia,
Alona, Eurycercus, Acroperus, Bosmina, Holopedium and Chydorus) or species level (Gasterosterus
aculeatus). Insects were furthermore categorized as
larvae, pupae or adults.

Stable isotope analysis
To compare the relative contribution of pelagic and
littoral components to the diet, and to determine the
trophic level of the diet of trout and stickleback, we
performed analyses of the stable carbon and nitrogen
isotope content. Approximately 5–8 mg (wet weight)
of white dorsal muscle was extracted from each fish.
During sorting, all samples were kept at room
temperature for as short a time as possible, after
which they were frozen again before being lyophi-

lized. Lyophilized samples were homogenized and
packed into tin capsules (4×6 mm). The samples were
analyzed for δ13C and δ15N isotopes using a PDZ
Europa ANCA-GSL elemental analyzer interfaced to
a PDZ Europa 20–20 isotope ratio mass spectrometer
at the UC Davis Stable Isotope Facility, USA. Since
lipid corrected values showed little deviation from
original values (Δδ13C; average=0.10; SD=0.31) due
to relatively low C:N-ratios (average=3.46; SD=
0.31), indicating a relatively low lipid content (Post
et al. 2007), we used original δ13C-values rather than
lipid corrected ones.
Average δ13C-values for littoral invertebrates (chironomidae, Lymnaea sp., Eurycercus sp., Gammarus sp.,
Haliplus spp., Helopdella spp., Hydracarina spp.,
Limnephilidae, Phryganeidae sp., Polycentropus,
Tipulidae) and periphyton scraped of stones collected
in the littoral zone were used as littoral δ13C-baseline.
Since large mussels were not found, and collected
Pisidium sp. turned out to yield unrealistic baselines,

we used zooplankton as the pelagic baseline. The relative
contribution of pelagic resources in diet was calculated
from the 2-end-member-mixing model (Post 2002):

where ά is the proportion of carbon obtained from the
base of food web 1 and δ13Csc is the stable isotope ratio
of the secondary consumer. Our two-end-members were


310

the bases of the pelagic and littoral food webs (base1
and base2, respectively) ά was constrained to be
between 0 and 1. Since δ15N values were solely used
for within-lake comparisons, baseline values were not
needed for δ15N.
Environmental variables
Depth integrated samples of zooplankton were taken at a
mid-lake station with a modified Patalas sampler (3.3 L).
Zooplankton was identified under a stereo microscope
and counted in the lab (for more details, see Amsinck et
al. 2006). Total phosphorous was measured as described
by Jensen et al. (2002) and chlorophyll a was
determined according to Christoffersen et al. (2002).
These samples were taken simultaneously with the fish
samples.
Data treatment and statistical analysis
To test for possible effects of environmental variables
(lake area, Zmax, Chl a, Ptot zooplankton density and
presence of charr and stickleback) on the density

(CPUE) and biomass (BPUE) of trout, we used
multiple analysis of variance (MANCOVA, Wilks
Lambda).
For statistical analysis, food items were divided
into the following groups: zooplankton (Daphnia,
Bosmina, Holopedium, Chydorus and Copepoda),
benthic cladocerans (Eurycercus, Alona and Acroperus), chironomid larvae (larvae of Chironomidae),
insect pupae (pupae of Chironomidae, Trichoptera,
and other flies), emerged insects (emerged insects of
Chironomidae, Trichoptera, Tipulidae, other flies and
terrestrial insects), and benthic macroinvertebrates
(Hirudinea, Pisidium, Lymnaea, Hydracarina, Gammarus, Trichoptera larvae, Coleoptera larvae and
adults, Turbellaria and Heteroptera). This resulted in
six different diet groups. For each group we calculated
the proportion of the total diet as the number of food
items in the given group divided by the total number of
food items in the stomach. The effect of stickleback
presence on trout diet was tested with a nested
MANCOVA (Wilk’s lambda; lake nested under the
presence/absence of sticklebacks) on arcsine transformed data with the individual fish length as a
covariate. Differences in ά-values between lake types
were tested with a nested ANOVA. Lakes were in both
cases treated as fixed rather than random factors, since

Environ Biol Fish (2012) 93:305–318

stickleback presence was not randomly assigned to
the different lakes, but is expected to be fixed for
each lake, i.e. a lake with sticklebacks is assumed
to host sticklebacks at every revisit (see discussion

in Bennington and Thayne 1994 and Domenici et
al. 2008). This complies with common practice when
analyzing effects of predator presence in different
lakes (e.g. Reznick and Endler 1982; Reznick 1989;
Leips and Travis 1999; Kelly et al. 2000; Jennions
and Telford 2002; Langerhans et al. 2004; Domenici
et al. 2008).
To test for the relative importance of environmental
variables, including presence/absence of stickleback
competitors, on the stomach composition of trout, we
used ordination analysis. Canonical Correspondence
Analysis (CCA) was performed due to a high gradient
length of axis 1 in Detrended Canonical Analysis
(3.029 standard units). Seven environmental variables
(lake area, maximum depth, total phosphorous,
chlorophyll a content, zooplankton density, density
of trout (catch per unit effort) and presence-absence
data on sticklebacks) were included in the CCA. To
explore the relative importance of the 7 environmental
variables these were run as sole environmental variables
in CCA analyses. The larger the ratio between the
eigenvalue of CCA axis 1 (λ1) and CCA axis 2 (λ2) the
more variation explained by the single environmental
variable. ‘Species’ data were arcsine-transformed stomach content data, i.e. the proportion of each diet group
(based on the average values of stomach content for
each lake), whereas environmental variables were log
transformed with the exception of presence-absence of
sticklebacks. ‘Species’ occurring only in one lake were
excluded from the analysis. For the CCA, the benthic
macroinvertebrate group was split into individual taxa

(see above).
Sticklebacks may act both as competitors and as
prey for trout. Since piscivorous trout are known to
prey more heavily on sticklebacks than on trout, fish
will likely be a more frequent diet item in the
presence of sticklebacks as a consequence of prey
availability rather than competition. Thus, when fish
occurred in the diet of trout they were excluded from
the analysis to avoid a false positive result of
competition. Furthermore, we excluded the largest
trout caught (> 40 cm; N=5) from the diet and
isotopic analyses, since they were likely to be obligate
piscivores. These trout had little or nothing in their
stomachs.


Environ Biol Fish (2012) 93:305–318

311

Results
Fish abundance, habitat distribution and population
demography
Catch per unit effort of trout (CPUE; Fig. 2) ranged
between 0.02 trout hour−1 net−1 (Vatnsnes, pelagial
nets) and 2.00 trout hour−1 net−1 (Mjauvötn, littoral
nets). In all lakes but Leynavatn and Sandsvatn,
brown trout was the most abundant fish species in the
catches of all three habitats sampled. Arctic charr
occurred only in Leynavatn where it dominated the

catch in the offshore benthic and pelagic habitats.
European flounder was found only in Sandsvatn,
where it dominated the catch in both habitats
sampled (littoral and offshore benthic). Threespined sticklebacks were caught in five of the 11
lakes sampled (Table 1). In the six deepest lakes,
where pelagic nets were used, the number of trout
caught in pelagic nets relative to the total number of
trout caught was significantly higher in lakes with
sticklebacks (10.7–45.6%) than in lakes without

sticklebacks (2.9–3.8%) (Mann–Whitney U-test,
Z=−1.96; p=0.05). Neither the density (CPUE) nor
the biomass (BPUE) of trout in the lakes was related
to any of the environmental variables, including
presence of competitors (MANCOVA; Wilks Lambda;
p>0.05 for all). The size of trout varied highly
significantly between the lakes (Kruskal-Wallis test;
χ2 =154.6; p<0.001; Fig. 3); the difference was,
however, not related to any of the environmental
variables, including presence of other species.
Long term resource use: Stable isotope analyses
Trout in lakes with sticklebacks were found to have
significantly higher ά-values, indicating a relatively
higher dependence on pelagic resources than trout
from lakes without sticklebacks (nested ANOVA; F=
50.4; p<0.001; Fig. 4). The ά-values were generally
habitat-dependent (Two-way ANOVA, lake random
factor; F=7.40; p<0.001), with pelagic trout having a
TS absent

TS present
Sa Ch Fl

TS absent

e

50

Fl
Littoral
Benthic
Pelagic

2.0

1.5

Length (cm)

2.5

e b,c,d d,e
ns
*
***

a a,b,c e c,d,e a,b b,c,d b,c,d
ns ** ns ns *** ** ns

40
*
*
*

30
20

Fjallavatn

Toftavatn

Sørvágsvatn

Gróthúsvatn

Sandsvatn

Leynavatn

Vatnsnes

Mjáuvøtn

Saksunarvatn

Bessavatn

Mjávavatn

0

Fig. 2 Catch per unit effort (CPUE: # trout net−1 h−1) of brown
trout in three different habitats, i.e. littoral, pelagic and offshore
benthic, in eleven Faroese lakes with contrasting fish assemblages (TS: three-spined stickleback present; Sa: Atlantic salmon
present; Ch: Arctic charr present; Fl: flounder present)

Fjallavatn

Toftavatn

Sørvágsvatn

Gróthúsvatn

Sandsvatn

Vatnsnes

Mjáuvøtn

Leynavatn

0.5

Saksunarvatn

0
1.0

Bessavatn

10
Mjávavatn

CPUE (no. net–1 hour –1)

TS present
Sa Ch

Fig. 3 Length distribution of brown trout from 11 Faroese
lakes with contrasting fish assemblages (TS: three-spined
stickleback present; Sa: Atlantic salmon present; Ch: Arctic
charr present; Fl: flounder present). Grey bars denote fish
caught in the littoral zone, open bars denote fish caught
offshore in pelagic or benthic nets (see text). Boxes represent
the interquartile range (containing 50% of values), lines across
boxes indicate medians, and whiskers extend to the highest and
lowest values, excluding statistical outliers (circles) and
extreme values (vertical asterisks). Letters above bars show
results of a non-habitat specific post hoc test for between-lake
differences and asterisks indicate significant size difference
between habitats


312

Environ Biol Fish (2012) 93:305–318

ά-values, than charr (ANOVA; F=68.7; p<0.001),

whereas there was no difference between trout and
salmon in the one lake containing salmon (ANOVA;
F=0.151; p=0.70).
The level of δ15N-value in Sandsvatn differed
significantly between sticklebacks (average±SD; 4.4±
0.42) and trout (4.0 ± 0.54) relative to length
(ANCOVA; F=15.59; p=0.001), sticklebacks having
a higher δ15N-value compared to small trout. In
Sørvágsvatn, there was no effect of length and no
significant difference between δ15N-value of trout
(5.1±1.2) and sticklebacks (5.0±0.88) (ANCOVA;
F=0.142; p=0.709), and in Toftavatn, sticklebacks
(5.4±0.34) had a significantly higher δ15N-value than
trout (3.7 ±3.6) (ANCOVA; F = 22.54; p< 0.001).
Hence, the δ15N-values of sticklebacks were in none
of the lakes lower than the δ15N-values recorded for
similar sized trout.

more pelagic isotopic signature (higher ά-values)
compared to both littoral and off-shore benthic caught
trout (Tukey post hoc test; p<0.001 for pelagic
against both littoral and off-shore benthic trout).
Littoral and off-shore benthic trout were not found
to have different isotopic signatures (Tukey post hoc
test; p=0.146). This pattern was consistent to all deep
lakes, i.e. lakes where pelagic nets were used, with
pelagic caught trout on average having a more pelagic
signal than both littoral and off-shore benthic caught
trout.
In all three lakes with a sufficient number of

sticklebacks caught to allow statistical analyses,
trout had significantly more pelagic isotopic signals, i.e. higher ά-values, than sticklebacks
(ANOVA; F>6.1 for all; p<0.027 for all; Fig. 4).
In the one lake containing charr, trout had a
significantly more littoral isotopic signal, i.e. lower
Fig. 4 Relative contribution of pelagic resources
(ά) calculated from relative
content of δ13C (see text) in
trout, sticklebacks, charr and
salmon from eleven Faroese
lakes with contrasting fish
community composition

Sa Ch

Fl

Sandsvatn

1.0

TS present

Leynavatn

TS absent

0.6

0.4

0.2

Trout

Salmon

Charr

Stickleback

Fjallavatn

Toftavatn

Sørvágsvatn

Gróthúsvatn

Vatnsnes

Saksunarvatn

Mjáuvøtn

Bessavatn

0

Mjávavatn

Proportion pelagic diet

0.8


Environ Biol Fish (2012) 93:305–318

313

a
100

Fig. 5 Relative proportion by number of food items in the„
stomach of brown trout from three different habitats of eleven
Faroese lakes. Taxa were grouped into zooplankton (Daphnia,
Bosmina, Holopedium, Chydorus and Copepoda), benthic cladocerans (Eurycercus, Alona and Acroperus), chironomid larvae
(larvae of Chironomidae), insect pupae (pupae of Chironomidae,
Trichoptera, and other flies) and emerged insects (emerged insects
of Chironomidae, Trichoptera, Tipulidea, other flies and terrestrial
insects), and other benthic macroinvertebrates (Hirudinea, Pisidium,
Lymnaea, Hydracarina, Gammarus, Trichoptera larvae, Coleoptera
larvae and adult Turbellaria, and Heteroptera)

Fl

Littoral (%)

80
60

40
20
0
100

b

Benthic (%)

80
60
40
20
0
100

c
80

Pelagic (%)

60
40

Macro zoobenthos
Benthic cladocerans
Chironomid larvae

Fjallavatn

Toftavatn

Sørvágsvatn

Gróthúsvatn

Leynavatn

Saksunarvatn

Mjáuvøtn

0

Bessavatn

20

Mjávavatn

We found substantial variation in the stomach contents
of trout between lakes and within lakes (Fig. 5). The
presence of sticklebacks significantly affected the diet
of trout caught in the littoral zone (MANCOVA (lake
nested under the presence or absence of sticklebacks);
Wilks’ lambda=7.982; p<0.001). In particular, there
were significant differences in the proportion of
emerged insects and benthic macroinvertebrates in the
trout diet, the former being higher and the latter lower
in the presence of sticklebacks. Although only marginally significant, also the share of benthic cladocerans

and insect pupae in the stomach of trout tended to
differ depending on stickleback presence (Table 2).
However, individual fish length, used as a co-variate in
the analysis, was not found to have any effect on trout
diet (Wilks’ lambda=0.626; p=0.709).
The diet of trout in the offshore benthic habitat was
significantly affected by the presence of sticklebacks
(MANCOVA (lake nested under the presence or absence
of sticklebacks); Wilks’ lambda=4.248; p=0.002), but
not by individual fish length (MANCOVA; Wilks’
lambda=0.807; p=0.132). Tests of between subject
effects (Table 2) show that benthic trout from
stickleback-containing lakes fed significantly more on
emerged insects and benthic cladocerans and less on
chironomid larvae and benthic macroinvertebrates.
In both the littoral and the offshore benthic
habitats, there were significant lake effects on trout
diet (MANCOVA; littoral: Wilks’ lambda=3.259; p<
0.001; benthic: Wilks lambda=6.259; p<0.001). This
difference was evident for all diet groups (p<0.05 for
all; Table 2). An effect of stickleback presence on the
diet of pelagic trout could not be traced, since few
pelagic trout were caught in lakes without sticklebacks.
Excluding the effect of habitats, the multivariate
CCA showed that lake area and the presence of

TS present
Sa Ch

Sandsvatn

TS absent

Vatnsnes

Short term resource use: direct diet observations

Insect pupae
Emerged insects
Zooplankton

sticklebacks explained most of the between-lake
variation in the stomach contents of trout (Table 3).
These variables were, however, significantly correlated (Mann–Whitney U-test; p=0.018), which is also
seen from the ordination including all 7 environmental variables (Fig. 6). The ordination (Fig. 6) showed


314

Environ Biol Fish (2012) 93:305–318

Table 2 Presence/absence of stickleback (TS presence) and
lake effects on diet numerical composition of brown trout in the
littoral and the offshore benthic zones. Both TS presence and
lake effects were tested with nested MANCOVA (lake nested

under stickleback presence/absence with fish length as a covariate, see text). Values in the table refer to subsequent
‘between-subjects effects’ tests. Length is not included in the
tables as it was not significant in any of the habitats

Diet group

Type III ssq

d.f.

mean square

F

p

Zooplankton

.305

1

.305

2.420

.124

Chironomid larvae

.160

1

.160

2.077

.154

Benthic cladocerans

.373

1

.373

3.941

.051

Insect pupae

.157

1

.157

3.085

.084

Emerged insects

1.344

1

1.344

20.120

< .001

Benth. macroinvert.

2.244

1

2.244

21.027

< .001

Zooplankton

3.659

9

.407

3.223

.003

Chironomid larvae

1.439

9

.160

2.070

.044

Littoral
TS presence

Lake effects (TS presence)

Benthic cladocerans

3.509

9

.390

4.115

< .001

Insect pupae

1.063

9

.118

2.314

.025

Emerged insects

1.245

9

.138

2.071

.044

Benth. macroinvert.

8.796

9

.977

9.157

< .001

Zooplankton

.130

1

.130

1.094

.301

Chironomid larvae

.417

1

.417

5.976

.018

Benthic
TS presence

Lake effects (TS presence)

Benthic cladocerans

.647

1

.647

7.616

.008

Insect pupae

.003

1

.003

.050

.824

Emerged insects

.052

1

.052

6.955

.011

Benth. macroinvert.

.707

1

.707

6.543

.014

Zooplankton

11.519

6

1.920

16.121

< .001

Chironomid larvae

2.128

6

.355

5.081

< .001

Benthic cladocerans

3.346

6

.558

6.568

< .001

Insect pupae

1.283

6

.214

3.085

.012

Emerged insects

.195

6

.033

4.384

.001

Benth. macroinvert.

5.405

6

.901

8.333

< .001

that stickleback presence was related to relatively
high proportions of zooplankton and insects (pupae
and emerged) as well as Pisidium and Hydracarina in
the trout diet, whereas food items such as Heteroptera,
Table 3 Results
from CCA’s including only
one variable at the time.
Variables sorted by
decreasing importance
for the between-lake variation. ‘% explained’ denotes
the total percentage
explained of the
between-lake variation (no
covariation included)

Environmental variable

Hirudinea, Gammarus and Trichoptera were mainly
found in trout stomachs from stickleback-free lakes.
In the charr-containing lake, zooplankton did not

occur in the trout stomach contents.
CCA 11/CCA 12

11

% explained

F-ratio

P-value
0.039

Lake area

0.629

0.302

21.0

2.391

Stickleback presence

0.533

0.276

19.2

2.136

0.047

Zmax

0.444

0.198

13.8

1.436

0.198

Ptot

0.433

0.228

15.9

1.696

0.143

Trout CPUE

0.370

0.199

13.8

1.450

0.233

Zooplankton biomass

0.326

0.181

12.6

1.292

0.246

Chlorophyll a

0.283

0.147

10.2

1.022

0.382


Environ Biol Fish (2012) 93:305–318

315

Fig. 6 Canonical correspondence analysis (CCA) plot of
brown trout stomach content
(arcsine transformed relative
proportions) and 7 environmental variables (abbreviations see Table 1). CCA
axis 1 and 2 explain 41%
and 23% of the variation,
respectively. Species
variables: triangles, environmental variables: arrows,
lakes: circles: lakes without
stickleback, squares: lakes
with stickleback

Discussion
The interactions between species drive individuals to
change their use of resources, leading to an apparent
altered niche use of populations (Bolnick et al. 2007).
Jeppesen and co-workers (2002b) suggested that
competition from three-spined sticklebacks could
modify the relative use of pelagic and littoral
resources in trout. In this study, we confirm this
suggestion with results both from relative distribution,

stable isotope analyses and direct diet observations.
We also found evidence that trout in some lakes,
besides the general effect of competitors on the
populations relative pelagic/littoral resource use,
individually specialized in habitat usage. While the
competitors appeared to affect trout at population
level, the individual specialization in habitat use
seems to be dependent on habitat availability, i.e. the
availability of the pelagic habitat.
However, our results also suggest that by treating
the niche as a one-dimensional pelagic-littoral variable, we might miss the larger picture. The observed
diet differences show that both littoral and off-shore
benthic trout in the presence of sticklebacks would

feed more on emerged insects and less on benthic
macroinvertebrates, and in off-shore habitats also
less on chironomid larvae. Such a difference in diet,
being short or long term, would require different
hunting modes and microhabitat use, with emerged
insects caught on the surface and benthic macroinvertebrates and chironomid larvae caught on the
bottom substrate, even though the different prey
types, at least in the littoral habitat, may be
relatively close to each other. Such a specialization
would be difficult to trace in the isotopic signature,
since most of the emerged insects in the diet were
of aquatic origin and hence belonged to either the
chironomid larvae or benthic macroinvertebrate diet
group at an earlier life stage.
Although foraging in the littoral habitat may,
particularly in shallow lakes, be more profitable than

in offshore habitats, for example due to greater
abundance of large, energetically rewarding food
items, this may not apply to all lake types. Especially
in deep lakes, the relative importance of the pelagic
zone in terms of food availability may be greater than
in the littoral zone (e.g. Gasith 1991; Jeppesen et al.
1997; Horppila et al. 2000; Vadeboncoeur et al. 2002).


316

Interestingly, however, it appears that the relative
dependency of the pelagic habitat is more related to
the presence of competitors than to physical lake
characteristics. Specifically, our multivariate analysis
(Fig. 6) suggests that trout in lakes with sticklebacks
utilize more open-water food items, such as zooplankton, emerged insects and insect pupae as compared
to lakes without sticklebacks, where they feed
more on bottom-associated food items, such as
benthic cladocerans and benthic macroinvertebrates. Since sticklebacks were generally found in
the larger, albeit not deeper, lakes, it was in
principle not possible from the CCA analysis to
determine whether the difference in diet between
lakes was due to competitor presence or lake area.
Whereas lake depth and/or volume are known to
have a number of effects on lake ecosystems, such
as periphyton vs. phytoplankton contribution to
primary production (Vadeboncoeur et al. 2003;
Vadeboncoeur et al. 2008), zooplankton community
structure (Amsinck et al. 2006), fish population

structure (Riget et al. 2000), fish assemblages
(Mehner et al. 2007), top-down control (Jeppesen
et al. 2003), ecological effects of lake area appear to
be rather weak (e.g. Søndergaard et al. 2005). Since
neither lake depth nor lake volume were related to
stickleback presence or trout diet, we see it as more
likely that the observed diet differences between
lakes are related to competitor presence rather than
to lake area. Assuming that charr is a more effective
zooplanktivore than trout, due to its evolved behavioral and morphological traits (e.g. Malmquist 1992;
Forseth et al. 2003; Klemetsen et al. 2003), charr
may be a strong competitor in large, deep lakes
where the pelagic niche is prominent, as in deep
Leynavatn. Indirect evidence for this is also provided
by large, deep Lake Takvatn, Norway, where an
increase in the density of brown trout was attributed
to competitive release followed by a decrease in the
density of Arctic charr (Klemetsen et al. 2002;
Amundsen et al. 2007).
As with many other species, two individual trout
will never exploit their resources in an exactly similar
fashion. This may be due to stochastic differences in
encounter rate with different prey types or it may be
due to individual specialization. By analyzing direct
diet observations, it is rarely possible to distinguish
between random differences in diet and actual
individual specialization, even when comparing fish

Environ Biol Fish (2012) 93:305–318

from different habitats, as the stomach content of any
individual will be dependent on the habitat in which it
most recently foraged, which will often be the same
as the habitat in which the fish was caught. However,
stable isotope analysis provides a longer time average
of littoral/pelagic dependency, which reduces the
influence of stochastic prey encounter, and can
therefore be used to describe individual specialization
more accurately. It is noteworthy that pelagic-caught
trout, in the deepest Faroese lakes, where we also
conducted sampling with pelagic gill nets, differed in
isotopic signal from trout caught both in the littoral
and the off-shore benthic area, especially since offshore benthic invertebrates often have a more pelagic
signal than littoral benthic invertebrates (e.g. Hershey
et al. 2006), as has been observed in the Faroese lakes
(E. Jeppesen and J. Brodersen unpubl. data). If
individual fish showed fidelity to all habitats, we
would have expected to see pelagic and off-shore
benthic trout to differ in isotopic signature from
littoral trout. However, it appears that intrapopulation specialization occurs along a pelagicbenthic gradient, where the latter includes both the
littoral and the off-shore benthic habitats.
In six out of the seven deepest Faroese lakes brown
trout tended to be smaller in the littoral habitat than in
the off-shore benthic and pelagic habitats. Such a
pattern has been observed for trout in other relatively
deep lakes (e.g. Klemetsen et al. 2003; Saksgård and
Hesthagen 2004) and may be explained by the greater
structural complexity of the littoral zone, offering
better refuge to small fish (Werner and Gilliam 1984).
In general, brown trout often display ontogenetic diet

shifts, switching from relatively small invertebrate
food items to large items such as sticklebacks and
other fish species (L’Abée-Lund et al. 1992, 2002;
Kahilainen and Lehtonen 2002; Klemetsen et al.
2003). However, the habitat specific diet of the
Faroese trout was not related to size, suggesting that
ontogenetic changes occur in habitat choice, but not
in habitat specific diet.
In conclusion, brown trout in Faroese lakes should
be regarded as a generalist species (MacArthur and
Levins 1967), potentially facing competition from
more specialized pelagic Arctic charr and benthic
feeding three-spined sticklebacks. Whereas generalist
species are often found to struggle under competition
from multiple directions (e.g. Werner 1977), this is
not the case for Faroese trout, since in no lakes does it


Environ Biol Fish (2012) 93:305–318

face competition from both sides. Besides the population effects of competition on population niche use
characteristics, we have shown that individual trout
specialize in pelagic/benthic habitat use whenever the
lake characteristics allow this. Lastly, our study
illustrates the complexity of niche use in freshwater
lakes and the potential shortcomings of applying a
one-dimensional axis for describing niche use.
Acknowledgements We are grateful to Jane Stougaard Pedersen,
Karina Jensen and Lissa Skov Hansen for identification of
zooplankton samples. Special thanks go to Kirsten Landkildehus

Thomsen for chemical analysis and Anne Mette Poulsen for
manuscript editing. We also wish to thank Juana Jacobsen and
Kathe Møgelvang for graphical layout. The project was partly
funded by the Carlsberg Foundation, The Nordic Arctic Research
Programme 1999–2003 and The Danish North Atlantic
Research Programme. The study was also supported by
the EU projects EUROLIMPACS (www.eurolimpacs.ucl.ac.
uk) and WISER (www.wiser.eu), REFRESH, CRES, Greenland Research Centre and by Galathea 3. Furthermore, we
thank Andy Jones, Geraldine Thiere and Karin Olsson for
valuable comments on the manuscript.

References
Amsinck SL, Strzelczak A, Bjerring R, Landkildehus F,
Lauridsen TL, Christoffersen K, Jeppesen E (2006) Lake
depth rather than fish planktivory determines cladoceran
community structure in Faroese lakes—evidence from
contemporary data and sediments. Freshw Biol 51:2124–
2142
Amundsen P-A, Knudsen R, Andersen K (2007) Intraspecific
competition and density dependence of food consumption
and growth in Arctic charr. J Anim Ecol 76:149–158
Appelberg M, Berger H-M, Hesthagen T, Kleiven E, Kurkilahti
M, Raitaniemi J, Rask M (1995) Development and
intercalibration of methods in Nordic freshwater fish
monitoring. Water, Air, & Soil Pollution 85:401–406
Bennington CC, Thayne WV (1994) Use and misuse of mixedmodel analysis of variance in ecological studies. Ecology
75:717–722
Bolnick DI, Svanbäck R, Araújo MS, Persson L (2007)
Comparative support for the niche variation hypothesis
that generalized populations also are more heterogeneous.

Proc Natl Acad Sci USA 104:10075–10079
Chase JM, Leibold MA (2003) Ecological niches: linking
classical and contemporary approaches. University of
Chicago Press, Chicago, USA
Christoffersen K, Pålsson C, Kritzberg E, Granéli W (2002)
Abundance and biomass of microbial communities in
relation to phyto- and bacterioplankton production in five
Faroese lakes. Ann Soc Sci Færoensis Suppl 36:59–69
Domenici P, Turesson H, Brodersen J, Brönmark C (2008)
Predator-induced morphology enhances escape locomotion in crucian carp. P Roy Soc Lond Ser B 275:195–201

317
Forseth T, Ugedal O, Jonsson B, Fleming IA (2003) Selection
on Arctic charr generated by competition from Brown
trout. Oikos 101:467–478
Gasith A (1991) Can littoral resources influence ecosystem
processes in large, deep lakes? Verh Int Ver Limnol
24:1073–1076
Hansson S (1995) Effects of exploitative food competition on
food niche dynamics—a simulation analysis. Ecol Modell
77:167–187
Hershey AE, Beaty S, Fortino K, Kelly S, Keyse M, Luecke C,
O'Brien WJ, Whalen SC (2006) Stable Isotope Signatures
of Benthic Invertebrates in Arctic Lakes Indicate Limited
Coupling to Pelagic Production. Limnol Oceanogr
51:177–188
Horppila J, Ruuhijärvi J, Rask M, Karppinen C, Nyberg K,
Olin M (2000) Seasonal changes in the diets and relative
abundances of perch and roach in the littoral and pelagic
zones of a large lake. J Fish Biol 56:51–72

Hubbell SP (2001) The Unified Neutral Theory of Species
Abundance and Diversity. Princeton University Press,
Princeton, NJ
Hutchinson GE (1957) Concluding remarks. Cold Springs
Harbor Symp Quant Biol 22:415–427
Hutchinson GE (1959) Homage to Santa Rosalia or why are
there so many kinds of animals. Am Nat 93:145–159
Jansen PA, Slettvold H, Finstad AG, Langeland AL (2002)
Niche segregation between Arctic char (Salvelinus alpinus) and Brown trout (Salmo trutta): an experimental
study of mechanisms. Can J Fish Aquat Sci 59:6–11
Jennions MD, Telford SR (2002) Life-history phenotypes in
populations of Brachyrhaphis episcopi (Poeciliidae) with
different predator communities. Oecologia 132:44–50
Jensen JP, Landkildehus F, Søndergaard M, Jeppesen E,
Christoffersen K (2002) Water chemistry of five lakes at
the Faroe Islands. Ann Soc Sci Færoensis Suppl 36:34–38
Jeppesen E, Jensen JP, Søndergaard M, Lauridsen TL, Pedersen
LJ, Jensen L (1997) Top-down control in freshwater lakes:
the role of nutrient state, submerged macrophytes and
water depth. Hydrobiologia 342(343):151–164
Jeppesen E, Christoffersen K, Landkildehus F, Lauridsen T,
Amsinck S (2001) Fish and crustaceans in northeast Greenland lakes with special emphasis on interactions between
Arctic charr (Salvelinus alpinus), Lepidurus arcticus and
benthic chydorids. Hydrobiologia 442:329–337
Jeppesen E, Christoffersen K, Malmquist HJ, Faafeng B,
Hansson L-A (2002a) Ecology of five Faroese lakes:
summary and synthesis. Ann Soc Sci Færoensis Suppl
36:126–139
Jeppesen E, Landkildehus F, Lauridsen TL, Jensen JP, Bjerring
R, Søndergaard M, Amsinck SL (2002b) Food web

interactions in five Faroese lakes tracked by stable
isotopes. Ann Soc Sci Færoensis Suppl 36:114–125
Jeppesen E, Jensen JP, Jensen C, Faafeng B, Hessen DO,
Søndergaard M, Lauridsen T, Brettum P, Christoffersen K
(2003) The impact of nutrient state and lake depth on topdown control in the pelagic zone of lakes: a study of 466
lakes from the temperate zone to the arctic. Ecosystems:
313–325
Jonsson B, Jonsson N, Hindar K, Northcote TG, Engen S
(2008) Asymmetric competition drives lake use of coexisting salmonids. Oecologia 157:553–560


318
Kahilainen K, Lehtonen H (2002) Brown trout (Salmo trutta
L.) and Arctic charr (Salvelinus alpinus (L.)) as predators
on three sympatric whitefish (Coregonus lavaretus (L.))
forms in the subarctic Lake Muddusjarvi. Ecol Freshwat
Fish 11:158–167
Kelly CD, Godin JGJ, Abdallah G (2000) Geographical
variation in the male intromittent organ of the Trinidadian
guppy (Poecilia reticulata). Can J Zool 78:1674–1680
Klemetsen A, Amundsen P-A, Grotnes PE, Knudsen R,
Kristoffersen R, Svenning M-A (2002) Takvatn through
20 years: long-term effects of an experimental mass
removal of Arctic charr, Salvelinus alpinus, from a
subarctic lake. Environ Biol Fish 64:39–47
Klemetsen A, Amundsen P-A, Dempson JB, Jonsson B,
Jonsson N, O’Connel MF, 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

L’Abée-Lund JH, Langeland A, Sægrov H (1992) (1992)
Piscivory by Brown trout Salmo trutta L. and Arctic charr
Salvelinus alpinus (L.) in Norwegian lakes. J Fish Biol
41:91–101
L’Abée-Lund JH, Aass P, Sægrov H (2002) Long-term variation
in piscivory in a Brown trout population: effect of changes
in available prey organisms. Ecol Freshwat Fish 11:260–
269
Landkildehus F, Jeppesen E, Jensen JP, Dali S (2002) Lake
descriptions. Ann Soc Sci Færoensis Suppl. 36 28–33.
Langeland A, L’Abée-Lund JH, Jonsson B, Jonsson N (1991)
Resource partitioning and niche shift in Arctic charr,
Salvelinus alpinus and Brown trout, Salmo trutta. J Anim
Ecol 60:895–912
Langerhans RB, Layman CA, Shokrollahi M, DeWitt T (2004)
Predator-driven phenotypic diversification in Gambusia
affinis. Evolution 58:2305–2318
Leips J, Travis J (1999) The comparative expression of life
history traits and its relationship to the numerical
dynamics of four populations of the least killifish. J Anim
Ecol 68:595–616
MacArthur RH, Levins R (1967) The limiting similarity,
convergence and divergence of coexisting species. Am
Nat 101:377–385
Malmquist HJ (1992) Phenotype-specific feeding behaviour of
two Arctic charr Salvelinus alpinus morphs. Oecologia
3:354–361
Malmquist HJ, Ingimarsson F, Jóhannsdóttir EE, Gíslason D,
Snorrason SS (2002) Biology of Arctic charr (Salvelinus
alpinus) and Brown trout (Salmo trutta) in four Faroese

lakes. Ann Soc Sci Færoensis Suppl 36:94–103
Mehner T, Holmgren K, Lauridsen TL, Jeppesen E, Diekmann
M (2007) Lake depth and geographical position modify

Environ Biol Fish (2012) 93:305–318
lake fish assemblages of the European ‘Central Plain’
ecoregion. Freshwat Biol 52:2285–2297
Post DM (2002) Using stable isotopes to estimate trophic
position: models, methods, and assumptions. Ecology
83:703–718
Post DM, Layman CA, Arrington DA, Takimoto G, Quattrochi
J, Montaña CG (2007) Getting to the fat of the matter:
models, methods and assumptions for dealing with lipids
in stable isotope analyses. Oecologia 152:179–189
Reznick DN (1989) Life-history evolution in guppies: 2.
Repeatability of field observations and the effects of
season on life histories. Evolution 43:1285–1297
Reznick DN, Endler JA (1982) The impact of predation on life
history evolution in Trinidadian guppies (Poecilia reticulata). Evolution 36:160–177
Riget F, Jeppesen E, Landkildehus F, Lauridsen TL, GeertzHansen P, Christoffersen K, Sparholt H (2000) Landlocked
Arctic charr (Salvelinus alpinus) population structure and
lake morphometry in Greenland – is there a connection?
Polar Biol 23:550–558
Saksgård R, Hesthagen T (2004) A 14-year study of habitat use
and diet of Brown trout (Salmo trutta) and Arctic charr
(Salvelinus alpinus) in Lake Atnsjoen, a subalpine
Norwegian lake. Hydrobiologia 521:187–199
Schluter D, McPhail JD (1992) Ecological character displacement and speciation in stickleback. Am Nat 140:85–108
Schoener TW (1982) The controversy over interspecific
competition. Am Sci 70:586–595

Søndergaard M, Jeppesen E, Jensen JP (2005) Pond or lake: does it
make any difference? Arch Hydrobiol 162:143–165
Svanbäck R, Persson L (2009) Population density fluctuations
change the selection gradient in Eurasian perch. Am Nat
173:507–516
Tilman D (1987) The importance of the mechanisms of
interspecific competition. Am Nat 129:769–774
Vadeboncoeur Y, Vander Zanden MJ, Lodge DM (2002) Putting
the lake back together: Reintegrating benthic pathways
into lake food web models. Bioscience 52:44–54
Vadeboncoeur Y, Jeppesen E, Vander Zanden MJ, Schierup HH, Christoffersen K, Lodge DM (2003) From Greenland to
green lakes: Cultural eutrophication and the loss of benthic
pathways in lakes. Limnol Oceanogr 48:1408–1418
Vadeboncoeur Y, Peterson G, Vander Zanden MJ, Kalff J (2008)
Benthic algal production across lake size gradients:
interactions among morphometry, nutrients, and light.
Ecology 89:2542–2552
Werner EE (1977) Species packing and niche complementarity
in three sunfishes. Am Nat 111:553–578
Werner EE, Gilliam JF (1984) The ontogenetic niche and
species interactions in size-structured populations. Annu
Rev Ecol Syst 15:393–425


Environ Biol Fish (2012) 93:319–331
DOI 10.1007/s10641-011-9916-x

Temporal variation in chum salmon, Oncorhynchus keta,
diets in the central Bering Sea in summer and early autumn
Osamu Sakai & Orio Yamamura &

Yasunori Sakurai & Tomonori Azumaya

Received: 9 February 2010 / Accepted: 14 August 2011
# Springer Science+Business Media B.V. 2011

Abstract Seasonal, ontogenetic, and diel variations
in the diets of chum salmon, Oncorhynchus keta,
were examined by analyzing the stomach contents of
1398 fish (300–755 mm fork length) collected in the
Bering Sea during summer and early autumn of 2002.
Whereas mesozooplankton, including euphausiids,
hyperiids, and gastropods, constituted the greatest
portion of the stomach contents during the summer,
forage fishes (Stenobrachius leucopsarus and Atka
mackerel, Pleurogrammus monopterygius) were the
most important items during early autumn. Although
no apparent diel trend was found in feeding intensity,
distinct diel differences in prey composition were
observed. Chum salmon caught in the morning
contained Stenobrachius leucopsarus, whereas those
O. Sakai (*)
National Research Institute of Far Seas Fisheries,
Fisheries Research Agency,
5-7-1 Orido, Shimizu, Shizuoka,
Shizuoka 424-8633, Japan
e-mail:
Y. Sakurai
Graduate School of Fisheries Sciences,
Hokkaido University,
3-1-1 Minato-cho, Hakodate,

Hokkaido 041-8611, Japan
O. Yamamura : T. Azumaya
Hokkaido National Fisheries Research Institute,
Fisheries Research Agency,
116, Katsurakoi, Kushiro,
Hokkaido 085-0802, Japan

caught in the afternoon had mainly fed on euphausiids. Thus, chum salmon diets change temporally
because of changes in prey availability that result
from differences in the annual life cycles and diurnal
vertical migrations of prey species.
Keywords Bering Sea . Chum salmon . Feeding
habits . Oncorhynchus keta . Seasonal diet shift

Introduction
Chum salmon (Oncorhynchus keta) is a species of
Pacific salmon that ranges widely in latitude, from the
southeast coast of Japan (Kyushu Island) and northwest coast of the U.S. (central Oregon) in the south to
the shores of the Arctic Ocean in Russia and Canada in
the north (Salo 1991). Chum salmon fry quickly
migrate to the ocean and grow rapidly, capitalizing on
oceanic productivity during their extensive feeding
migrations. Most chum salmon migrate into the Bering
Sea during the summer growth period (Seeb et al.
2004). Other species of Pacific salmon, including pink
(O. gorbuscha), sockeye (O. nerka), coho (O. kisutch),
and chinook (O. tshawytscha), are also distributed in
the Bering Sea during the summer (Manzer et al.
1965), and thus the Bering Sea is recognized as an
important foraging area for Pacific salmon during their

oceanic life stages (Brodeur et al. 1999).
Previous studies have indicated that chum salmon
are primarily planktivorous in the Bering Sea (e.g.,


320

Kanno and Hamai 1971; Sobolevskii and Senchenko
1996; Tadokoro et al. 1996), but their diet is the most
diverse among Pacific salmon, consisting of small
fishes, squids, crustaceans, and gelatinous organisms
(Azuma 1995; Davis et al. 1998). Of these, gelatinous
organisms are unique to chum salmon diets, compared
to the other Oncorhynchus species. Most previous
studies of chum salmon, however, were conducted in
early summer (June and July). Few studies have
examined other seasons in the Bering Sea, with the
exception of studies conducted in Russian or US
waters. According to these coastal studies, chum
salmon tends to consume high energy content prey
(e.g. fish, euphausiids) in autumn (Moss et al. 2009)
and low energy content prey (e.g. gelatinous organisms) in winter (Sobolevskii et al. 1994; Sobolevskii
and Senchenko 1996). These previous studies primarily targeted the first year of marine life, which
provided valuable information to understand the
seasonal aspects of chum salmon feeding habits
before the first wintering. On the other hand, seasonal
feeding ecology in the second and subsequent years
of marine life remains unclear. In the Bering Sea, age
2 to 4 chum salmon appeared to be affected by
density-dependent effect (Seo et al. 2011), thus the

information of feeding ecology during these stages is
a vital component to understand the growth control
mechanisms. Moreover, presence or absence of
ontogenetic change in the feeding ecology is also
important information to understand the intra-specific
competition under the limited caring capacity. Against
this background, therefore, we targeted second and
subsequent years of chum salmon feeding habits with
their ontogenetic changes in the present study.
At higher latitudes, lower species diversity and
higher amplitudes in seasonal biomass fluctuations
Fig. 1 Sampling stations in
the central Bering Sea
where chum salmon were
collected during summer
(white circles) and early
autumn (black diamonds) in
2002

Environ Biol Fish (2012) 93:319–331

are expected in zooplankton (Rudjakov et al. 1995;
Macpherson 2002; Woodd-Walker et al. 2002).
Chum salmon are non-selective feeders, typically
feeding “opportunistically” (Pearcy et al. 1988).
Therefore, chum salmon may have seasonal fluctuations in their use of different prey items, reflecting
local prey availability. In addition, chum salmon
inhabit surface layers throughout the day (Ogura and
Ishida 1995), whereas many of their prey organisms,
such as zooplankton and micronekton, undergo diel

vertical migrations in this region (Schabetsberger et
al. 2000). Therefore, diurnal changes in chum
salmon diet are also expected, but studies examining
diurnal variations in the oceanic diets of chum
salmon are lacking. In the present study, we
examined the feeding habits of chum salmon in the
Bering Sea during summer and early autumn to
identify not only seasonal and ontogenetic variations, but also diel changes.

Materials and methods
Sample collection
Surface trawl samples were collected from the RV
Kaiyo-maru during two periods in 2002: 29 June–15
July (summer) and 16 August–18 September (early
autumn). The same area was surveyed during both
seasons (Fig. 1). A total of 24 and 38 stations were
sampled during summer and early autumn, respectively. The mouth opening of the surface trawl was
approximately 60 m in both height and width. The net
was attached to a canvas kite, which suspended the
net from the sea surface, so that the trawl sampled the
upper 60 m of the water column. The duration of each


Environ Biol Fish (2012) 93:319–331

tow was 1 h at a ship speed of 5 knots (9.3 km/hr).
Tows were made twice a day, once in the morning
(08:38–11:54) and again in the afternoon (14:24–
19:30). Up to 30 chum salmon individuals were
selected randomly from each tow. Each fish was

measured and weighed to the nearest 1 mm in fork
length (FL) and 1 g in body weight (BW). Sex was
determined, and gonad weight (GW) was measured to
the nearest 1 g. Stomachs were excised and preserved
in a 10% formalin-seawater solution.
Hydrographic observations were also conducted at
each station using either a Seabird SBE-9plus or a
SBE-19 conductivity-temperature-depth sensor
(CTD), deployed to a depth of 2000 m or close to
the sea bottom. Statistical significances of the
seasonal difference for the sea surface temperature
and number of fish caught were determined using
two-sample t-test.
Stomach content analysis
In the laboratory, stomachs were cut open in a Petri
dish filled with water, and the contents were removed
from the stomach and pylorus. Stomach contents were
placed on filter paper and excess water was removed
by suction. The contents were then sorted to the
lowest possible taxonomic group under a stereoscopic
microscope. When an unusually intact prey item, such
as a fish with complete scales, was found, it was
regarded as being ingested in the net during the tow
and was excluded from further analysis. The digestion
stage of each prey item was categorized according to
the following criteria (Pearcy et al. 1979):
Stage 1: well preserved, cuticle is practically
remaining;
Stage 2: partially digested, bodily form is
maintained;

Stage 3: substantially digested, bodily form is
collapsed;
Stage 4: well-digested, hard parts and digested
matter remain.
When fresh or lightly digested prey items were
found in a stomach, body lengths (BL) were measured
to the nearest 1 mm. For the gravimetrically analysis
of stomach contents, variation in the amount of liquid
causes the sources of error (Hyslop 1980). With the
aim to remove the digestive fluid, surface water, and

321

moisture content from prey items, dry weight is
preferred to wet weight as a unit of measurement
(Suzuki 1993). In this study, thus, each prey item was
dried at 52°C in a drying oven for 24 h and in a
desiccator for 36–48 h, and then weighed to the
nearest 1 mg.
The importance of individual prey taxa was
represented by the frequency of occurrence (F) and
by dry weight composition (DW); DW calculations
included the weight of digested matter. The data were
divided into 16 groups (sc1–sc16) according to chum
salmon body size (≤400 mm, 401–500 mm, 501–
600 mm, >600 mm FL groupings), season (summer
and early autumn), and diel period (morning and
afternoon). In this study, Redundancy analysis (RDA)
was used to investigate the effects for the feeding
habits of chum salmon. RDA is one of ordination

techniques and an extension of multiple linear
regression (Makarenkov and Legendre 2002). RDA
is able to express how much of the variance in one set
of variables can be explained by the others. As the
example of RDA for the fish biology, there are some
studies to investigate the effects of environmental
variables (e.g. depth, temperature, salinity etc.) toward
the fish abundance and/or distribution (RodríguezCabello et al. 2007; Van de Putte et al. 2010). In our
case, the effects of chum salmon body size (four FL
groups), season (summer and early autumn), and diel
period (morning and afternoon) toward the DW of
chum salmon stomach contents were investigated by
RDA. The significance of each effect was tested using a
permutation test with up to 999 random permutations.
Diel change in feeding intensity was assessed
using the stomach contents index (SCI) according to
Yamamura et al. (2002):


SCW
SCI ¼
 103 ;
BW
where SCW is the dry weight (g) of the total stomach
contents ingested by an individual. Diel feeding
changes were also examined by comparing the
digestive stage composition of prey. These stomach
content data were condensed into 10 sub-periods
during the daylight period by comparing sampling
time with the time of sunrise and sunset at each

sampling station. An overall difference of SCI between
the time groups was determined by Kruskal-Wallis test.
All of the statistical analyses were conducted using the
program R (version 2.8.1).


322

Results
Catch pattern and overall diet
Sea surface temperatures (temperature at 5 m depth) at
each station increased from summer (4.7–8.7°C) to early
autumn (7.3–10.9°C) (Two-sample t-test: t=−10.2, df=
60, P<0.001). A total of 10,175 chum salmon were
caught throughout both seasons. The number of fish
caught per haul (catch per unit effort, CPUE) of chum
salmon also increased from summer (0–551 individuals)
to early autumn (22–670 individuals) (Welch’s twosample t-test, log(x+1)-transformed prior to analysis: t=
−3.92, df=33.1, P<0.001). A total of 1,398 stomachs
were examined; no empty stomachs were encountered.
Chum salmon diets varied widely, consisting of over
30 genera and species (Appendix 1). Overall, the most
important prey were nekton (39% DW, 31% F), which
predominantly included myctophids, juvenile walleye
pollock (Theragra chalcogramma), and juvenile Atka
mackerel (Pleurogrammus monopterygius). The greatest proportion of the myctophids in the stomach
contents consisted of Stenobrachius leucopsarus. The
body sizes of walleye pollock and Atka mackerel in
stomach contents were 24–78 mm SL and 48–144 mm
FL, respectively. Mesozooplankton prey were ingested

frequently. Of these, euphausiids were the most
important gravimetrically, representing 23% DW, which
was dominated by Thysanoessa longipes. Gelatinous
organisms (9% DW, 52% F) and pelagic gastropods
(6% DW, 25% F) were also important, and the most
common taxa were Beroe sp. and Limacina helicina,
respectively. Hyperiid amphipods, dominated by Themisto pacifica, occurred most frequently (72% F), but
their gravimetric contribution was <4%. Digested
matter constituted a substantial portion of the stomach
contents (15% DW) and was often represented by a
grayish substance, suggesting that it contained gelatinous organisms, but the advanced digestive stage
prevented us from sorting it into specific taxa.
Dietary differences by season, diel period, and fish
body size
The diets of chum salmon varied with season, diel
period, and fish size (Fig. 2). During summer, the
stomach contents of all chum salmon size groups
mainly consisted of mesozooplankton, primarily
including euphausiids, gastropods, and gelatinous

Environ Biol Fish (2012) 93:319–331

organisms. Euphausiids were especially important in
afternoon diets of individuals that were >400 mm FL
(45–60% DW). The diets of the smallest body size
class (≤400 mm FL) differed from those of the largest
body size class; hyperiid amphipods and fish larvae
were also important (>20% DW) in the smallest size
class in morning and afternoon stomachs, respectively. During early autumn, diets in all size
classes shifted to include nekton, which primarily

included Stenobrachius myctophids and juvenile
Atka mackerel. Myctophids were equally important
in all size classes in the morning (about >40% DW),
whereas Atka mackerel were more important for
larger individuals in the afternoon (501–600 mm FL
fish: 31% DW, >600 mm FL fish: 83% DW). Juvenile
walleye pollock was also a characteristic prey during
early autumn, tending to be found in stomachs from
smaller body size classes, but their contribution was
limited (<9% DW). Euphausiids were frequently
ingested during early autumn as well, but their size
composition differed substantially from those ingested
during summer (Fig. 3). Although the >20 mm
euphausiid size class made up 58% of the total DW
of Thysanoessa ingested during summer, the medium
size class (10–20 mm total length: TL) became
exclusively important during early autumn.
The results of the redundancy analysis (RDA)
definitely showed the above-described differences in
chum salmon stomach contents between both seasons
and diel periods. RDA explained 63.1% of the
constrained variance. The first two dimensions, which
are displayed in the ordination graph (Fig. 4), showed
56.1% of the variance. Axis 1 and axis 2 of the
ordination graph roughly explain the effects of season
and diel period, respectively. Each prey category
score in the ordination graph shows a trend for higher
DW composition in a particular prey category. Thus,
higher DW compositions of mesozooplankton (e.g.,
euphausiids, gastropods, copepods) corresponded to

summer stomach contents. Euphausiids in particular
were interpreted as a characteristic prey in summerafternoon stomach contents (sc6, sc7, and sc8). On
the other hand, higher DW composition of nekton (e.g.,
myctophids, Atka mackerel, walleye pollock) corresponded to early autumn stomach contents. Effects of
diel period were conspicuous in early autumn. Morning
and afternoon stomachs tended to correspond to a higher
DW composition of myctophids (sc9, sc10, sc11, and
sc12) and Atka mackerel (sc15 and sc16), respectively.


Environ Biol Fish (2012) 93:319–331

323

Fig. 2 Comparison of percentages in dry weight (DW) of chum salmon diets between morning and afternoon samples for each season

The significance of these seasons and diel period effects
was indicated by the permutation test (Table 1). Differences in chum salmon diets among fish body size
categories were observed in each season and diel
period. However, they did not explain the major trends

in the main prey categories; the permutation test did
not find this effect to be significant.
SCI values differed significantly with time of day
during both seasons (Kruskal-Wallis test, P<0.001).
However, the difference was equivocal, showing no
clear diel tendency (Fig. 5). Prey with less advanced
digestion (stages 1 and 2) occurred throughout the
day with no apparent diel pattern (Fig. 6).

Discussion

Fig. 3 Weight composition of different size classes of
Thysanoessa occurring in the stomachs of chum salmon;
summer (white bar) and early autumn (gray bar)

In the present study, chum salmon diets varied
markedly with season and time of day. These differences would be attributed to the availability of prey
items. Seasonal prey shifts from mesozooplankton to
nekton and diurnal prey differences apparently
reflected seasonal and diel changes in the overlap
between predator and prey distributions. This overlap
would be explained by seasonal fluctuations in
biomass and diel vertical migrations of prey species


324

Environ Biol Fish (2012) 93:319–331

Fig. 4 RDA ordination graph showing the first two axes. The
eigenvalues are 587.08 (axis 1) and 215.39 (axis 2). The presented
axes display 56.1% of the total variance within the data. Prey items
are labeled with the following codes; Po1=Polychaeta, Sag=
Sagittoidea, Gas=Gastropoda, Cep=Cephalopoda, Cop=Copepoda, Mys=Mysidacea, Gam=Gammaridea, Hyp=Hyperiidea,
Eup=Euphausiacea, Dec=Decapod crustacea, Gel=Gelatinous
prey, MYCT= Myctophidae, WALLEYE= Walleye pollock,

ATKA=Atka mackerel, OtherFish=Other nektons, Fishlarv=Fish
larvae, Others=Other items, Dig=Digested matter. The “sc” prefix

indicates the stomach content group; sc1-4=summer/morning
stomachs, sc5-8=summer/afternoon stomachs, sc9-12=early
autumn/morning stomachs, sc13-16=early autumn/afternoon
stomachs. In each season and diel period, codes are numbered in
ascending order according to fish size (≤400 mm FL, 401–500 mm
FL, 501–600 mm FL, and >600 mm FL)

in the environment. Considering the seasonal changes
in particular, some mesozooplankton species that were
observed in summer stomachs were rarely consumed
in early autumn (e.g., gastropods Limacina helicina and
copepods Neocalanus cristatus, N. plumchrus/flemingeri). These species are known to reach a biomass peak
in spring-summer, after which they disappear from the
epipelagic layer in the subarctic North Pacific (Tsuda et
al. 1999; Kobari and Ikeda 2000; Cooney et al. 2001).
Similar fluctuations have also been reported in the

Bering Sea (Rudjakov et al. 1995). Thus lower
gravimetric importance of these mesozooplankton in
early autumn stomachs was construed as a reflection of
a seasonal biomass reduction. Euphausiid Thysanoessa
spp., mainly T. longipes, had the highest gravimetric
importance in summer. In early autumn, reductions in
both their contribution and body size were observed,
with larger-sized individuals, >20 mm TL, being more
important during the summer. Previous studies have
shown that T. longipes has a life span of 3 years, and
that individuals attain a maximum size of 24–28 mm TL
by the summer of their third year (Iguchi and Ikeda
2004). Thus, it is likely that the T. longipes individuals

consumed by chum salmon in the summer were
approaching maximum size, and the individuals probably would have died in early autumn when the newly
recruited cohort (smaller-sized Thysanoessa) became
available as chum salmon forage. Seasonal biomass
reductions in the environment would reduce the
availability of mesozooplankton prey items in early

Table 1 Results of a permutation test examining the significance of constraints in RDA. The significance level shows a
marginal effect, as obtained with a permutation test under the
reduced model with up to 999 random permutations
Variable

df

Eigenvalue

F

P

Season

1

Diel time period

1

486.24

9.2308

0.005

221.47

4.2043

Fish body size

3

0.010

196.26

1.2419

0.290


Environ Biol Fish (2012) 93:319–331

325

Fig. 5 Mean stomach
content index (±1 S.D.) by
time of fish collection
relative to sunrise, noon,
and sunset during summer

(upper) and early autumn
(lower). Sample sizes area
listed above each data point

autumn, and these bottom-up processes would force
chum salmon to switch from mesozooplankton to
alternative prey.
The optimum prey selected by chum salmon in
early autumn, as an alternative to mesozooplankton,
were nekton: juvenile Atka mackerel, walleye pollock, and myctophids, especially Stenobrachius leucopsarus. Increasing numbers of both Juvenile Atka
mackerel and walleye pollock were observed in trawl
catches in early autumn relative to summer; catch per
haul for walleye pollock and Atka mackerel increased
from 1±0.5 (mean±S.E.) to 523±232 and from 3646±
1996 to 5981±963, respectively (Azumaya et al.
unpubl. data1). Although no previous study has
reported seasonal variations in the abundances of these
forage fishes, their autumnal increases evidently led to
increased consumption by chum salmon. On the other
hand, the density of S. leucopsarus in the study area
remained fairly stable between summer and autumn
(Tanimata et al. 2008). This implies that factors other
than a seasonal biomass change in S. leucopsarus
promoted consumption by chum salmon in early
autumn. One possibility is that the increasing nighttime
period, from 7 h in the summer to 12 h in early
autumn, led to increased encounter rates between S.
1

/>202003/717(Japan).pdf

leucopsarus and chum salmon. Stenobrachius leucopsarus undergo extensive diel vertical migrations,
ascending to the epipelagic layer during the nighttime
(Pearcy et al. 1977; Watanabe et al. 1999), and chum
salmon typically inhabit surface waters throughout
both the day and night (Ogura and Ishida 1995).
Chum salmon may have only had access to S.
leucopsarus during the night, when S. leucopsarus
move into surface waters. Thus the seasonal increase in
nighttime length led to increased opportunities for
encountering and ingesting S. leucopsarus. The comparison of diets between morning and afternoon also
supports the hypothesis that chum salmon ingest
myctophids (S. leucopsarus) during the night;
myctophids were more common in the stomachs of
chum salmon collected in the morning compared to
individuals collected in the afternoon. Although the
evacuation rate of fish from chum salmon stomachs
has not been examined, 69% evacuation of shrimp
was found after 5 h at 13°C (Arai et al. 2003). The
sea surface temperature in the present study area was
<13°C in both seasons (4.7–8.7°C and 7.3–10.9°C
for summer and autumn, respectively), indicating
that the duration of gastric evacuation for myctophids was probably longer than observed in the
laboratory study. This suggests that stomach contents
sampled in the morning and afternoon reflect feeding
during the preceding night and morning, respectively.


326

Environ Biol Fish (2012) 93:319–331

Fig. 6 Digestive stage
composition for prey by
time of collection relative to
sunrise, noon, and sunset
during summer (upper) and
early autumn (lower)

Therefore the prevalence of myctophids in morning
samples reflects ingestion during the night, when their
vertical distribution overlapped with chum salmon.
Although we did not collect samples after sunset, we
found no apparent diurnal trends in SCI or the digestive
stage composition. This implies that chum salmon feed
intermittently throughout the daytime, as found in
previous studies (Azuma 1992; Davis et al. 2000).
These previous studies reported that the day-night
change in the feeding intensity of chum salmon was
not substantial compared to other Oncorhynchus species
(Pearcy et al. 1984; Azuma 1992), suggesting that chum
salmon were able to find food at all times of the day
(Pearcy et al. 1984; Davis et al. 2000). This suggests
that chum salmon have a flexible feeding behavior,
allowing efficient ingestion of available prey in the
environment wherever and whenever prey is encountered, such as zooplankton during summer and S.
leucopsarus during autumn and/or at night, as noted
by Pearcy et al. (1984).
Previous studies reported that gelatinous organisms, such as salps and ctenophores, represent large
proportions of the diets of chum salmon (Pearcy et al.

1988; Kaeriyama et al. 2004). However, in the present
study, gelatinous organisms represented fairly limited
proportions (11% DW and 5% DW in summer and
early autumn, respectively). This difference was due
to procedural differences for examining stomach
contents. While we used DW composition, previous
studies used wet weight composition, which would lead

to overestimation of the contribution of gelatinous
organisms containing more water (Davis et al. 1998).
Size-dependent predator-prey relationships, in
which prey size increases as predators grow, is a
common phenomenon found in fish feeding studies
(Scharf et al. 2000). Ontogenetic shifts in prey items
were not clear in the present study; differences in diet
composition with fish body size varied depending on
the season and diel period. In particular, in summer,
smaller fish (≤400 mm FL) tended to ingest hyperiids
and fish larvae more frequently compared to larger fish.
Larger prey (e.g., juvenile Atka mackerel) were found in
the stomachs of larger fish (>500 mm FL) in early
autumn afternoons. On the other hand, myctophids,
which are also larger prey items, were found in
stomachs from all size classes in early autumn
mornings. Myctophids were ingested in both seasons,
but they were rarely found in the stomachs of smaller
fish in summer. These findings suggest that the prey size
range of chum salmon is wide and is less dependent on
predator size than seasonal and diel prey availability.
Previous studies rarely reported “piscivorous” chum

salmon, possibly because the studies were mainly
conducted in spring-summer (e.g., Azuma 1995;
Tadokoro et al. 1996; Davis et al. 2000) and/or focused
on juvenile fish (Sobolevskii and Senchenko 1996). In
spring-summer, the feeding habits of chum salmon
would reflect the abundance of available mesozooplankton in the environment, and their “planktivorous” nature
was reported accordingly. In addition, smaller body


Environ Biol Fish (2012) 93:319–331

327

habits with changes in the availability of prey items in
the environment. The present study found that chum
salmon changed their dominant prey items seasonally,
which provides essential information for modeling
their energy intake from prey organisms throughout
the year. The knowledge of seasonal change of
feeding habits would improve our understanding of
the relation between the ocean environment and chum
salmon, and which will contribute to consider the
chum salmon growth mechanism during their oceanic
phase through the further study as the application to
modeling approach.

sizes might limit the ingestion of larger sized prey items,
such as nekton. Our results are limited in only one year
samples, thus, yearly fluctuation could not be argued. As
the suggestion by Tadokoro et al. (1996), there is odd/

even year fluctuation of the pink salmon abundance in
the summer Bering Sea, and chum salmon feeding
habits is possibly associated with it. Our finding
corresponds to the feeding habits in even year, therefore
further investigation especially for odd year would be
valuable to understand the inter-specific competition
from the aspect of seasonal shift of prey and predators.
Recently, ecosystem model based approach was
applied to understand the ocean growth of Pacific
salmon (e.g. Aydin et al 2005; Kishi et al. 2010).
Feeding behavior of salmon is the most fundamental
component of these ecosystem models, and from the
standpoint of model construction and configuration, it
is necessary to evaluate temporal variations in feeding

Acknowledgments We thank the officers and crew of the
RV Kaiyo-maru. We are grateful to A. Tsuda and S. Urawa
for their cooperation in sample collection. We also thank R.
Brodeur, N. Davis, K. Uchikawa, and N. Tanimata for their
advice. This study was financially supported by the Japan
Fisheries Agency.

Appendix 1
Table 2 Diets of chum salmon as percent dry weight composition (DW) and frequency of occurrence (F). DW and mean body length
(BL; mm) for different size classes are also shown
Taxon

Total
DW

POLYCHAETA (total)

F

≤400 mm

400–500 mm

500–600 mm

>600 mm

DW

DW

DW

DW

BL

BL

0.02

1.65

0.01

1

0.01

–

0.02

–

0.02

–

<0.01

0.64

–

–

0.01

–

<0.01

–

0.15

16.09

0.52

0.15

16.31

0.52

4.72

25.11

2.3

Clione limacina

0.36

10.52

<0.01

–

0.19

6.9

0.86

–

0.08

Limacina helicina

4.29

23.03

2.29

1.86

2.28

2.97

7.08

3.02

6.78

4.48

Gastropoda (unidentified)

0.07

0.29

–

–

<0.01

–

0.23

–

–

–

CEPHALOPODA (total)

2.42

15.09

Berryteuthis anonychus

0.04

0.14

–

–

0.07

–

–

–

–

Gonatus middendorffi

0.63

4.58

0.79

G. onyx

0.25

0.07

–

–

–

Gonatidae (unidentified)

0.77

8.08

0.2

7

0.89

0.74

2.72

3.66

–

<0.01

0.93

<0.01

1

<0.01

Tomopteris sp.
SAGITTOIDEA (total)
Sagitta sp.
GASTROPODA (total)

Cephalopoda (unidentified)
OSTRACODA (total)
Ostracoda (unidentified)

<0.01

0.18
22.54

0.89

29.69

2.37

Neocalanus spp.

0.14

8.73

0.06

N. cristatus

0.71

16.02

2.21

N. plumchrus / flemingeri

0.01

5.01

<0.01

0.18

25.47

0.94

0.34
<0.01

20.35
–
13.16
–
2.7

0.1

–

0.01

–
–

2.36
–

0.2

17.07

0.27

9.47

0.84

92.6

–

–

0.55

12.3

1.24

0.45

–

<0.01
0.29

0.86

38.45
–

<0.01
–

0.97
3.78

–

6.87

<0.01

0.78
3.88

–

2.03

<0.01
–

0.04

–
<0.01
0.01

8.17

2.25
14.47

<0.01

0.04

2.47

4.64

COPEPODA (total)

0.02

BL

0.02

Rhynchonereella angelini

0.03

BL

<0.01

–

0.07
3.77

7.81

0.63

7.25

0.66

7.48

–

0.02

2.95

0.01

–

<0.01
0.05
<0.01

4.83
7.36
–


328

Environ Biol Fish (2012) 93:319–331

Table 2 (continued)
Taxon

Total
DW

F

≤400 mm

400–500 mm

500–600 mm

>600 mm

DW

DW

DW

DW

BL

BL

BL

BL

0.02

0.72

<0.01

–

0.03

–

<0.01

–

Eucalanus bungii

<0.01

1.14

<0.01

–

<0.01

–

<0.01

–

–

–

Paraeuchaeta elongata

<0.01

0.5

<0.01

5.05

<0.01

–

<0.01

–

–

–

Candacia columbiae

<0.01

0.07

Copepoda (unidentified)

Calanidae (unidentified)

–

–
–

–

–

–

–

–

<0.01

–

–

–

–

<0.01

–

<0.01

0.29

0.07

MYSIDACEA (total)

<0.01

0.36

–

Gnathophausia zoea

<0.01

0.14

–

–

0.01

–

–

–

–

–

Mysidacea (unidentified)

<0.01

0.21

–

–

<0.01

–

–

–

–

–

0.02

1.65

0.02

GAMMARIDEA (total)

0.02

1.14

0.02

HYPERIIDEA (total)

Gammaridea (unidentified)

3.98

71.53

8.8

Themisto pacifica

3.66

67.6

8.44

Hyperia medusarum

0.07

13.23

0.08

Hyperoche medusarum

<0.01

1.79

<0.01

Hyperiidae (unidentified)

<0.01

0.5

0.24

31.26

Phronima atlantica

<0.01

P. sedentaria
Hyperiidea (unidentified)
EUPHAUSIACEA (total)

Primno abyssalis

<0.01

<0.01

–

0.01

0.02
–

0.02

18.3

0.01

19.9

2.24
5.33

–

–

1.39

1.85

5.7

1.3

0.07

13.6

0.08

13.2

0.07

11.2

4.85

<0.01

5

<0.01

6.4

<0.01

9.4

–

–

<0.01

–

<0.01

0.29

7.11

0.14

–

<0.01

0.57

<0.01

0.43

11.3

4.48

–

0.01

4.8
4.77

–

–

–

–

8.29

0.31

–

<0.01

9.8

<0.01

–

–

–

–

–

<0.01

–

<0.01

–

–

–

–

–

<0.01

–

<0.01

–

–

–

23.16

42.35

18.66

Thysanoessa longipes (unmeasured)

8.2

9.59

3.44

–

8.53

–

8.49

–

9.47

–

T. longipes (≤10 mm)

0.07

0.72

0.34

9.33

0.03

9.8

0.1

9.35

–

–

T. longipes (10–20 mm)

2.26

7.51

0.52

11.9

3.26

14.37

1.28

13.2

1.93

16.91

T. longipes (20–30 mm)

3.04

1.65

–

–

0.27

21.45

3.06

22.9

15.59

23.13

T. inspinata

<0.01

0.29

9

0.01

10.45

<0.01

10.4

–

–

T. inermis

<0.01

0.14

–

–

<0.01

<0.01

–

–

–

T. spinifera

0.05

0.07

0.61

–

–

–

–

–

–

–

Thysanoessa spp. (unmeasured)

1.91

6.8

1.09

–

2.74

–

1.49

–

0.27

–

Thysanoessa spp. (≤10 mm)

0.4

2.07

0.37

8.01

0.63

8.18

0.2

8.5

–

–

11.23

11.16

9.05

11.22

Thysanoessa spp. (10–20 mm)

5.9

Thysanoessa spp. (20–30 mm)

<0.01

0.07

Euphausiidae (unidentified)
APPENDICULATA (total)

Oikopleuridae (unidentified)
GELATINOUS ORGANISMS (total)

13.3

<0.01

–

–
–

1.32

8.44

1.07

<0.01

0.07

–

<0.01

0.07

–

7.36

51.5

18.16

<0.01

–

0.24

25.86

10.93

6.37

–

0.01

–

1.32

3.3

<0.01

3.46

11

–

–

1.94

2.8

–

<0.01
–

1.6

27.29

–

–

6.29

<0.01

–

–

–

0.03

–

–
–

–

9.12

–

9.96

Beroe sp.

0.64

1.57

–

–

0.15

–

0.42

–

3.46

–

Hydrozoa (unidentified)

0.13

2

0.09

–

0.14

–

0.14

–

0.06

–

Unid. Gelatinous organisms

6.59

50.5

3.37

–

5.99

–

8.56

–

6.45

–

DECAPOD CRUSTACEA (total)

0.94

33.33

0.55

Benthesicymidae (unidentified)

<0.01

0.07

Hippolytidae (Zoea)

0.04

0.29

Phyllocarida (Zoea)

<0.01

0.14

Paguridae (Zoea)

<0.01

0.36

–

0.94

1.44

0.05

–

<0.01

–

–

–

–

–

–

0.02

–

0.09

–

–

–

–

–

<0.01

–

–

–

–

–

–

–

<0.01

–

–

–

–

–

<0.01