271

ARTICLE

Can. J. Fish. Aquat. Sci. Downloaded from www.nrcresearchpress.com by Thompson Rivers University on 07/22/13
For personal use only.

The complexity of cormorants: stable isotopes reveal multiple prey
sources and feeding site switching
Aleksandra Bugajski, Matthew W. Reudink, Jennifer L. Doucette, Samantha E. Franks, Björn Wissel, and Christopher M. Somers

Abstract: Conflict between cormorants (Phalacrocorax spp.) and humans over fisheries is currently one of the most widespread
wildlife management issues in the world. Cormorant impact assessments typically assume a single source of prey near the
breeding colony. However, cormorants can fly long distances (>20 km), resulting in fish removal from multiple areas. Knowledge
of the source of cormorant prey is critical for fisheries impact assessments. Carbon and nitrogen stable isotopes values from fish
in double-crested cormorant (Phalacrocorax auritus) regurgitations revealed several prey sources for breeding birds in northcentral Saskatchewan, Canada. Cormorants also switched feeding locations on a daily and seasonal basis. Foraging patterns
inferred by stable isotope analysis closely paralleled observations of relative cormorant densities on different lakes. Up to 80% of
prey came from areas well removed from the breeding colony lake (≤30 km). Cormorant foraging ecology can be complex and
may be driven by factors independent of breeding colony site selection. The approach of estimating biomass removal from
breeding colony lakes as the guideline for management decisions may be invalid in many locations.
Résumé : Les conflits entre cormorans (Phalacrocorax spp.) et humains relatifs à la pêche constituent un des problèmes de gestion
de la faune les plus répandus à l'échelle planétaire. Les évaluations des impacts des cormorans partent généralement de
l'hypothèse d'une seule source de proies à proximité de la colonie de nidification. Les cormorans peuvent toutefois voler sur de
longues distances (>20 km) et ainsi prélever du poisson de plusieurs secteurs. La connaissance des sources de proies des
cormorans est essentielle à l'évaluation des impacts sur les pêches. Les concentrations d'isotopes stables du carbone et de l'azote
de poissons dans les régurgitations de cormorans à aigrettes (Phalacrocorax auritus) révèlent l'existence de plusieurs sources de
proies pour des oiseaux nidifiant dans le centre nord de la Saskatchewan (Canada). Ces cormorans changent également de sites
d'alimentation sur des bases quotidienne et saisonnière. Les patrons de quête de nourriture tirés de l'analyse des isotopes stables
concordent étroitement avec les observations sur la densité relative de cormorans dans différents lacs. Jusqu'à 80 % des proies
proviennent de secteurs à bonne distance du lac de la colonie de nidification (≤30 km). L'écologie de la quête de nourriture
des cormorans peut s'avérer complexe et dépendre de facteurs non reliés au choix du site de la colonie de nidification. L'approche
reposant sur l'estimation du retrait de biomasse de lacs de colonie de nidification pour orienter les décisions de gestion pourrait
s'avérer inadéquate pour de nombreux sites. [Traduit par la Rédaction]

Introduction
Conflicts between piscivorous birds and humans over fisheries
and aquaculture resources are common and widespread, especially those involving cormorants (Phalacrocorax spp.; Doucette
et al. 2011). A predominantly piscivorous diet, reputation for efficient foraging, and increasing populations of these birds have
raised concerns over their impact on fisheries in a variety of locations worldwide (Hatch 1995; Frederiksen et al. 2001). In general,
cormorants are perceived as major competitors with humans over
economically valuable fisheries resources. As a result, cormorant
population control programs occur across much of the Northern
Hemisphere, where millions of dollars are spent on cormorant
management annually (e.g., Shwiff et al. 2009). The extent of the
conflict depends on estimates of cormorant impacts on fisheries,
which differ dramatically among studies (reviewed by Doucette
et al. 2011). The circumstances linking cormorant foraging and
economically important fish loss remain uncertain and in need of
further study.
Bioenergetics modeling is widely used to estimate cormorant
consumption of fish (reviewed by Ridgway 2010). Central place

foraging theory suggests that colonial birds should forage within
the immediate vicinity of breeding colonies to decrease energy
expended (Ashmole 1963; Gorke and Brandl 1986; Elliott et al.
2009). Thus, many assessments assume that all fish biomass removal by cormorants is from their nesting lake (e.g., Rudstam
et al. 2004) or a particular area near the breeding colony (e.g.,
Diana et al. 2006). However, cormorants can use feeding sites that
are distant from breeding colonies (Anderson et al. 2004). Several
studies have shown that cormorants routinely forage within
20 km of breeding colonies (e.g., Custer and Bunck 1992; Platteeuw
and Van Eerden 1995; Kotzerka et al. 2011), and maximum flight
distances up to 30 km are commonly recorded (Hobson et al. 1989;
Grémillet et al. 1995; Stapanian et al. 2002). Cormorants may thereby
routinely have multiple prey sources, spreading their consumption
over several fish populations. However, this feature of their foraging
and feeding ecology is seldom considered during fisheries impact
assessments.
Studies of cormorant foraging and feeding behaviour, including site selection, have been notoriously difficult. Radio telemetry
generally allows detailed examination of only a few marked individuals from colonies containing hundreds to thousands of breed-

Received 23 May 2012. Accepted 7 November 2012.
Paper handled by Associate Editor Gary Sprules.
A. Bugajski, J.L. Doucette, B. Wissel, and C.M. Somers. Department of Biology, University of Regina, Regina, SK S4S 0A2, Canada.
M.W. Reudink. Department of Biological Sciences, Thompson Rivers University, Kamloops, BC V2C 0C8, Canada.
S.E. Franks.* Centre for Wildlife Ecology, Department of Biological Sciences, Simon Fraser University, Burnaby, BC V5A 1S6, Canada.
Corresponding author: Christopher M. Somers (e-mail: chris.somers@uregina.ca).
*Present address: 212-2150 East Hastings Street, Vancouver, BC V5L 0A5, Canada.

Can. J. Fish. Aquat. Sci. 70: 271–279 (2013) dx.doi.org/10.1139/cjfas-2012-0237

Published at www.nrcresearchpress.com/cjfas on 9 November 2012.

272

Can. J. Fish. Aquat. Sci. Vol. 70, 2013

Can. J. Fish. Aquat. Sci. Downloaded from www.nrcresearchpress.com by Thompson Rivers University on 07/22/13
For personal use only.

Fig. 1. Map of the study area (inset = North America; study area near Lac La Ronge is shown with a star) showing potential foraging locations
for cormorants breeding on Egg Lake. Nemeiben Lake (Nem.) and Lac La Ronge (LLR) were each split into three areas because of habitat
heterogeneity. Locations of gill nets to sample fish from known sites are shown (triangles). Line transects used to survey for foraging
cormorants are also shown (lines with circles).

ing pairs (e.g., Custer and Bunck 1992; Coleman et al. 2005; Seefelt
and Gillingham 2006), resulting in potential issues that apply to
individual patterns being extrapolated colony-wide. In addition, telemetry studies only provide information on the location of birds,
and when they are actively foraging (Coleman and Richmond 2007),
but they do not indicate when cormorants are successfully obtaining
prey in a given area. Thus, there is a need for alternative methods of
determining where cormorants are feeding to refine biomass removal estimates when assessing fisheries conflicts.
Stable isotope analysis makes it possible to track movements of
animals based on what they consume, and this technique does not
rely on tagged individuals (Rubenstein and Hobson 2004). Stable
isotopes offer a very useful tool for cormorant feeding studies
(Hobson 2009; Doucette et al. 2011), but they have not been applied
extensively for diet source tracking of cormorants (for exceptions
see Bearhop et al. 1999; Ofukany et al. 2012). Isotopic characterization of cormorant tissues directly may not be useful for prey
source tracking if cormorants are foraging in multiple locations,
as these tissues would reveal an integrated signature from multiple sources. However, the prey items themselves (fish) may vary
substantially in stable isotopes values based on their provenance
(e.g., Overman and Parrish 2001; Murchie and Power 2004; Barks
et al. 2010). Thus, if fish from multiple feeding locations are isotopically distinct, it may be possible to trace prey collected by cormorants back to their area of origin (Rubenstein and Hobson
2004). To our knowledge, no previous studies have taken advantage of stable isotopes values of prey to make inferences about
cormorant foraging and feeding ecology.
The purpose of this study was to understand double-crested
cormorant (Phalacrocorax auritus; hereafter, cormorant) foraging
and feeding ecology in an environment where they have several
lakes and foraging areas to choose from. We compared stable
isotopes values for carbon (!13C) and nitrogen (!15N) in fish collected from known locations to fish regurgitated by cormorants in
breeding colonies. Our main objectives were to (i) evaluate
whether the variance in stable isotopes values over a small spatial

scale within and among close bodies of water was sufficient for
determining sources of cormorant prey and (ii) determine
whether cormorants feed primarily from one or several sources of
fish during the breeding season.

Materials and methods
Study area
We conducted our study in the La Ronge area of north-central
Saskatchewan, Canada (55°09=N, 105°12=W; Fig. 1). We focused on
the largest breeding colonies in the area, which are located on Egg
Lake (Fig. 1): colony A in 2009 (55°07=14.88==N, 105°33=17.05==W;
!4000 breeding pairs) and colony B in 2010 (55°07=21.05==N,
105°30=20.11==W; !800 breeding pairs). Egg, Bigstone, Nemeiben,
and Head lakes and Lac La Ronge are suspected major foraging
sites for breeding cormorants in the study area (Fig. 1). These sites
were identified by local residents and also fall within a 20 km
foraging radius of these cormorant colonies (Custer and Bunck
1992; Neuman et al. 1997). Egg (10 690 ha; mean depth 3 m) and
Bigstone (1980 ha; mean depth 1.5 m) lakes are situated in the
Boreal Plain ecozone and are characterized as shallow and mesotrophic (Cushing 1964). Nemeiben Lake (15 483 ha) is oligotrophic and situated in the Boreal Shield ecozone. It has three
distinct basins: north (mean depth 30 m), central (mean depth
7 m), and outlet (mean depth 3 m; Duffy 2008). Head Lake (932 ha;
mean depth 3 m) is a small outlet connected to the northwest side
of Nemeiben Lake. Lac La Ronge (130 700 ha) also has three basins:
north, central, and south. North Lac La Ronge is in the Boreal
Shield ecozone and is characterized by deep (≥25 m), clear waters;
south Lac La Ronge is in the Boreal Plain ecozone and has relatively shallow (≤10 m), productive waters; central Lac La Ronge is
the transition zone between the two other areas (Duffy 2007).
Lac La Ronge and Nemeiben Lake have been used for commercial, sport, and subsistence fishing for decades, and their fisheries
have declined in quality since the 1960s (Gloutney and Chen 1992;
Wallace 1993). Of particular interest is the Lac La Ronge fishery,
Published by NRC Research Press

Bugajski et al.

Can. J. Fish. Aquat. Sci. Downloaded from www.nrcresearchpress.com by Thompson Rivers University on 07/22/13
For personal use only.

which has continued to decline since 1992 (Duffy 2007), coincident
with the discovery of established cormorant breeding colonies of
unknown age on nearby Egg Lake. Stakeholders are currently
concerned that cormorants may cause further decline or inhibit
recovery of fish populations.
Cormorant diet and selection of a representative prey
species
We collected regurgitated stomach contents from nestling cormorants on Egg Lake in 2009 (n = 147) and 2010 (n = 701). Yellow
perch (Perca flavescens; hereafter perch) made up 30% and 55% of
the total fish biomass consumed by nestlings in 2009 and 2010,
respectively (data not shown). In 2009, perch ranged in size from
45 to 190 mm (mean total length (TL) = 92.5 ± 29.1 mm); in 2010, TL
ranged from 36 to 196 mm (mean TL = 78.9 ± 21.5 mm). Based on
their importance to cormorants, we chose perch for all stable
isotopes analyses.
Fish from known locations
We sampled fish from sites within four lakes that were identified as potential cormorant foraging locations (Fig. 1). Bottom-set,
multipanel gill nets (mesh sizes: 1.3, 2.5, 3.8, 5.1, 6.4, and 7.6 cm)
were set at depths ranging from 2 to 10 m overnight for 12–16 h.
Gill nets were set at six sites in 2009 from 4 to 13 July: Egg Lake,
Bigstone Lake, north and central Lac La Ronge, and central and
eastern Nemeiben Lake. In 2010, gill nets were set from 23 June to
10 July at the same sites, and also south Lac La Ronge, west Nemeiben Lake, and Head Lake (Fig. 1). Gill nets were also set on
1 August 2010, to determine any temporal variation in stable isotope values of perch and to increase sample sizes for Egg and
Bigstone lakes. No change in !15N values and only a small shift in
!13C (mean difference = 0.9‰–2.1‰) to lower values was observed
(data not shown). All captured perch were measured (TL) and frozen.
We randomly chose up to 30 gill-netted perch (TL = 70–130 mm) from
each potential cormorant foraging area for stable isotope analysis
(n = 122 in 2009; n = 146 in 2010).
Fish from cormorant diet
We collected spontaneously regurgitated perch in the 70–130 mm
size range from cormorant nestlings in colony A on 18 June and
5 July 2009 and expanded in 2010 by collecting regurgitated fish in
colony B from 14 June to 22 July, on 10 separate days. Boluses were
collected haphazardly throughout the colony to represent as
many breeding pairs as possible. We analyzed a total of 186 prey
items in 2009: 18 items from 18 June and 168 items from 5 July. In
2010, we analyzed up to 50 fish on each of 10 sampling days for a
total of 391 individuals.
Stable isotopes analysis
We removed dorsal muscle tissue from perch and rinsed it in
deionized water. Muscle samples were dried at 48 °C for ≥48 h and
ground into a fine powder. All samples were weighed (0.5–1.0 mg)
into tin capsules and analysed using a Finnigan Delta Plus isotope
ratio mass spectrometer at the Environmental Quality Analysis
Laboratory (Faculty of Science, University of Regina, Regina, Saskatchewan). Carbon and nitrogen stable isotopes values are expressed in ! notation (‰) relative to international standards,
PeeDee belemnite (C) and atmospheric air (N2; Peterson and Fry
1987). Lipids were not extracted from fish tissue; analyses of a
subset of 2009 samples showed a mean C:N ratio of 3.6 ± 0.1 (n =
53), and arithmetic lipid normalization (Post et al. 2007) showed
little difference (0.3‰ ± 0.1‰; n = 53) between !13Cuntreated and
!13Cnormalized values. Based on this analysis, and no a priori need to
standardize by lake (we wanted to maximize differences by site),
we did not correct for lipids in this study. Fish tissue samples were

273

analyzed along with two in-house laboratory standards: bovine
liver (!13C = −21.6‰, !15N = 7.5‰) and wheat (!13C = −24.5‰, !15N =
3.5‰), both of which had standard deviation values of ±0.2‰
within runs. We also analyzed replicate samples of fish tissue
from the same individuals (as per Bond and Hobson 2012), which
had mean differences of 0.01‰ ± 0.2‰ for C samples and 0.02‰ ±
0.1‰ for N2 samples.
Relative cormorant density
To relate cormorant foraging behaviour to patterns suggested
from stable isotope analysis of regurgitated fish, and to characterize general cormorant foraging patterns in our study area, we
surveyed cormorants in 2010 using line transects in potential foraging areas (Fig. 1). Transects were 600 m wide; observations were
calibrated prior to surveying using spherical buoys placed 300 m
from the boat (Gould and Forsell 1989). The number of transects
per site and the total distance and area covered were as follows:
Egg Lake: two transects, 7.3 km, 4.4 km2; Bigstone Lake: one transect, 4.3 km, 2.6 km2; Nemeiben Lake: four transects, 11.8 km,
7.1 km2; North Lac La Ronge: four transects, 9.2 km, 5.5 km2;
Central Lac La Ronge: seven transects, 23.0 km, 13.8 km2; South
Lac La Ronge: two transects, 10.1 km, 6.0 km2. Line transects were
driven at 15 km·h−1 when visibility was >5 km, wave height
was <1 m, and wind speed was <20 km·h−1. Two observers counted
cormorants inside the transect boundaries. Transects on all lakes
were surveyed 23 times between 8 June and 24 August 2010. We
also recorded the location of cormorant foraging flocks (≥10 individuals) observed during surveys and other boat travel from 9 June
to 22 August 2010. The date, flock size, and approximate water
depth were recorded; we used a geographic information system
(GIS) to examine spatial and temporal patterns in cormorant foraging sites (ArcGIS Desktop10, ESRI, Redlands, California, USA).
Statistical analyses
We used a maximum likelihood approach to assign perch to
their most likely area of origin based on stable isotope values
(Royle and Rubenstein 2004). Areas of origin were determined
based on geographic location (e.g., different lakes); however, delineation of origin boundaries was complicated by large bodies of
water, in which different locations within a single lake could act
as different sources (areas of origin), with corresponding differences in isotopic signatures. Thus, we created several models for
our analyses by pooling and removing sites (see online Supplementary Tables S1–S61). Our approach used a two-step process,
where we first used linear discriminant analysis (LDA) with leaveone-out cross-validation to test isotopic distinctness of gill-netted
perch among sites (Gillanders and Kingsford 2000) and to determine the best model for each year based on classification success.
The leave-one-out method removes an individual perch of known
origin from the analysis and reclassifies it based on the discriminant function of the remaining perch (n − 1). Following LDA classification, we used a multivariate (!13C, !15N) normal probability
density function (MDF) to determine the likelihood that regurgitated perch originated from a given site based on the values for
both isotopes (see supplementary online material for other models1). The LDA–MDF approach yields a single estimate of provenance for each unknown perch.
Second, to account for analytical error in isotope measurements and generate a measure of assignment confidence for our
top models, we used a Bayesian error-based resampling simulation approach. This technique enables cross-validation of perch
from known sampling locations (gill-netted perch) and the assignment of unknown origin (regurgitated) perch based on the best
models from LDA–MDF assignments (Wunder et al. 2005; Wunder
and Norris 2008; Franks et al. 2012). Because fish population den-

1Supplementary data are available with the article through the journal Web site at http://nrcresearchpress.com/doi/suppl/10.1139/cjfas-2012-0237.

Published by NRC Research Press

sity varies across space, we incorporated relative abundance
(catch per unit effort of yellow perch based on gill netting) as the
prior probability of origin (as in Franks et al. 2012). Briefly, for
each isotope value for each perch, we randomly sampled 100 values from a normal distribution with a mean equal to the isotope
value for that tissue and a standard deviation derived from repeated measures of lab standards; this produced 100 new isotopic
data sets for all individuals. We randomly chose one data set to
define the parameters (mean, variance–covariance) of the multivariate probability density functions for the sites of origin. We
took two approaches for choosing the data set to define the parameters: (i) using all known-origin samples and (ii) using only
samples correctly assigned during the cross-validation simulation
(see below). Both approaches provided similar results; we present
data using all known-origin samples because it better represents
the extent of isotopic variation in the populations. For each individual in each data set, we determined the probability of assignment to a potential site of origin, which produced 100 assignment
outcomes for each perch. Finally, we repeated the previous step
100 times, each time using a new data set to define the parameters
of the probability density function. This produced 10 000 assignment outcomes for each individual. This approach also allowed us
to assess our confidence in assignments. For example, an individual that was assigned to a given region 9000 times out of 10 000
simulations was considered to have a 90% confidence of assignment to that region. Regurgitated perch with assignment confidence levels ≥50% and 80% were used for determining the
proportion of prey assigned to each location; here we present only
data based on the 80% confidence threshold. The proportion of
regurgitated perch assigned to each location was assumed to reflect the whole diet of cormorants. These values were chosen to
represent low and high confidence thresholds in the probability
of assignment (as in Franks et al. 2012). MDF assignments were
analysed by collection date as well as chick rearing stage (early =
chicks in nests; late = chicks in crèches).
To further examine our level of confidence in assignments of
unknown-origin perch, we examined the ability of our isotopic
data to correctly assign known origin (gill-netted) perch back to
their source location. We used the resampling simulation approach described above with an exclusion criterion (i.e., leaveone-out cross-validation), whereby we calculated the parameters
(mean, variance–covariance) of the probability density function,
but excluded the individual to be assigned from the data set. We
then examined the likelihood of that individual originating from
each of the potential source regions, as above. The LDAs and MDFs
were conducted in the statistical software package R 2.12.1, using
the rrcov (Todorov and Filzmoser 2009), mvtnorm, and mvnmle
packages (Genz et al. 2011).
We used a generalized linear model to compare cormorant densities among sites; the number of transects and transect length
did not significantly influence density (data not shown). We performed a negative binomial regression on cormorant density
based on site and chick rearing period. Density was used to standardize cormorant abundance among lakes. To eliminate noninteger values, density was multiplied by 100 and the decimal
points were removed. Transect count data are reported as geometric means (± standard error); statistical tests of generalized linear
models were performed in PASW Statistics 18 (SPSS Inc., IBM,
Armonk, New York, USA).

Results
Fish from known locations
Gill-netted perch showed a wide range of stable isotope values
and varying degrees of overlap in !13C and !15N among sites. Carbon stable isotope values spanned a total of 11.8‰ (−30.2‰ to
−18.5‰) in 2009 and 14.6‰ (−33.1‰ to −18.5‰) in 2010 (Fig. 2).
Nitrogen stable isotope values spanned a total of 6.3‰ (7.1‰–

Can. J. Fish. Aquat. Sci. Vol. 70, 2013

Fig. 2. Carbon and nitrogen stable isotopes values of gill-netted
yellow perch (TL = 70–130 mm) in (a) 2009 and (b) 2010. All gillnetted fish were collected from suspected cormorant foraging
locations using bottom-set gill nets and represent known locations.
In 2010, all Lac La Ronge zones were combined.
(a)

14.0

12.0

10.0

8.0

δ15N (‰)

Can. J. Fish. Aquat. Sci. Downloaded from www.nrcresearchpress.com by Thompson Rivers University on 07/22/13
For personal use only.

274

(b)

6.0
-32.0

-27.0

-22.0

-17.0

-22.0

-17.0

14.0

12.0

10.0

8.0

6.0
-32.0

-27.0
δ13C (‰)

13.3‰) in 2009 and 6.4‰ (5.9‰–12.2‰) in 2010 (Fig. 2). The best
model for 2009 assumed distinct perch populations for Egg and
Bigstone lakes, two populations in Lac La Ronge (north and central), and two populations from Nemeiben Lake (central and east).
Eighty-one percent (99/122) of gill-netted perch were correctly assigned back to their lake of origin (Table 1; Table S11). Of these, 89%
(88/99) were assigned with greater than 80% confidence, and 99%
(98/99) were assigned with greater than 50% confidence. Assignment probabilities were high for most sites; the highest correctly
assigned rate was 100% (6/6) for perch from eastern Nemeiben
Lake, while the lowest rate was 67% (14/21) for perch from central
Lac La Ronge. The range of correct assignment rates for the remaining lakes was between 73% and 93% (Table 1).
The best model for 2010 assumed distinct perch populations for
Egg Lake, Bigstone Lake, Head Lake, Lac La Ronge, and three distinct populations for Nemeiben Lake (west, central, and east).
Sixty-seven percent (98/146) of gill-netted perch were correctly
assigned back to their lake of origin (Table 1; Table S21). Of these,
94% (92/98) were assigned with greater than 80% confidence, and
100% (98/98) were assigned with greater than 50% confidence. Correct assignment rates were low for Egg and central Nemeiben
lakes at 44% (7/16) and 0% (0/17), respectively. Conversely, the remaining sites had high assignment rates ranging from 74% to
100% (Table 1).
Published by NRC Research Press

Bugajski et al.

275

Table 1. Cross-validation resampling simulation of gill-netted yellow perch (TL = 70–130 mm) from known locations, in 2009 using an MDF
model with two Lac La Ronge (LLR) and Nemeiben Lake populations and in 2010 using an MDF model with one LLR and three Nemeiben
Lake populations.

Can. J. Fish. Aquat. Sci. Downloaded from www.nrcresearchpress.com by Thompson Rivers University on 07/22/13
For personal use only.

Predicted sites
Actual sites

Bigstone

Egg

Central
Nemeiben

East
Nemeiben

North
LLR

West
Nemeiben

Central
LLR

Head

LLR

2009
Bigstone (n = 29)
Egg (n = 30)
Central Nemeiben (n = 14)
Eastern Nemeiben (n = 6)
North LLR (n = 22)
Central LLR (n = 21)

24 (0.83)
1 (0.03)
0 (0.00)
0 (0.00)
0 (0.00)
0 (0.00)

1 (0.03)
28 (0.93)
2 (0.14)
0 (0.00)
2 (0.09)
2 (0.10)

0 (0.00)
0 (0.00)
11 (0.79)
0 (0.00)
3 (0.14)
3 (0.14)

4 (0.14)
0 (0.00)
0 (0.00)
6 (1.00)
0 (0.00)
0 (0.00)

0 (0.00)
1 (0.03)
1 (0.07)
0 (0.00)
16 (0.73)
2 (0.10)

—
—
—
—
—
—

0 (0.00)
0 (0.00)
0 (0.00)
0 (0.00)
1 (0.05)
14 (0.67)

—
—
—
—
—
—

—
—
—
—
—
—

2010
Bigstone (n = 20)
Egg (n = 16)
Central Nemeiben (n = 17)
East Nemeiben (n = 13)
West Nemeiben (n = 24)
Head (n = 9)
LLR (n = 47)

16 (0.80)
0 (0.00)
4 (0.24)
1 (0.08)
0 (0.00)
0 (0.00)
0 (0.00)

1 (0.05)
7 (0.44)
3 (0.18)
0 (0.00)
0 (0.00)
0 (0.00)
6 (0.13)

0 (0.00)
1 (0.06)
0 (0.00)
0 (0.00)
1 (0.04)
0 (0.00)
2 (0.04)

2 (0.10)
0 (0.00)
0 (0.00)
11 (0.85)
0 (0.00)
0 (0.00)
0 (0.00)

—
—
—
—
—
—
—

0 (0.00)
0 (0.00)
8 (0.47)
0 (0.00)
20 (0.83)
0 (0.00)
4 (0.09)

—
—
—
—
—
—
—

1 (0.05)
0 (0.00)
0 (0.00)
1 (0.08)
0 (0.00)
9 (1.00)
0 (0.00)

0 (0.00)
8 (0.50)
2 (0.12)
0 (0.00)
3 (0.13)
0 (0.00)
35 (0.74)

Note: Each perch was assigned to the site with the greatest number of assignments out of 10 000 simulations. In each simulation, perch were assigned to the site
with the highest probability of origin. Values on the diagonal represent the number (and proportion) of correctly assigned individuals.

Fish from cormorant diet
Yellow perch regurgitated by cormorant nestlings had similar
ranges in !13C and !15N values to those gill netted from potential
foraging sites. Carbon stable isotope values spanned a total of
9.7‰ (−30.9‰ to −21.2‰) in 2009 and 14.2‰ (−31.6‰ to −17.2‰) in
2010 (Fig. 3). Nitrogen stable isotope values spanned a total of
5.5‰ (7.5‰–13.0‰) in 2009 and 7.4‰ (6.4‰–13.8‰) in 2010 (Fig. 3).
Based on MDF analyses, confidence in assignment of regurgitated
perch to their site of origin in 2009 was ≥80% for 94% (174/186) of
individuals and ≥50% for 100% (186/186) of individuals, whereas in
2010, confidence was ≥80% for 85% (332/391) of individuals and
≥50% for 99% (389/391) of individuals. We present the following
data using perch with assignments ≥80%. In 2009, 75% of perch
were assigned to Bigstone Lake, followed by 18% to Egg Lake, 4% to
north Lac La Ronge, 2% to central Nemeiben Lake, 1% to central Lac
La Ronge, and 0% to eastern Nemeiben Lake. In contrast, in 2010,
24% percent of perch were assigned to western Nemeiben Lake,
followed by 20% to Lac La Ronge, 17% to eastern Nemeiben Lake,
14% to central Nemeiben Lake, 10% to Bigstone Lake, 8% to Egg
Lake, and 7% to Head Lake.
In both study years the proportions of perch assigned to each
site changed by date (Fig. 4; Tables S3, S41). For example, in 2009,
13% (2/15) and 81% (128/159) of regurgitated perch were assigned to
Bigstone Lake on 18 June and 5 July, respectively (Fig. 4). Similar
shifts were also observed in 2010: 88% (28/32) of perch collected on
1 July were assigned to western Nemeiben Lake, whereas 9% (4/43)
of perch were assigned to the same site when collected on 4 July.
The chick rearing stage also affected assignment location. During
early chick rearing (n = 281), 62% of regurgitated perch were assigned to all Nemeiben Lake sites, with all other sites having <15% of assignments. In contrast, 20% of perch regurgitated
during late chick rearing (n = 51) were assigned to Nemeiben Lake,
31% to Egg Lake, and 49% to Lac La Ronge. Analyses based on
confidence in assignments ≥50% showed similar results (see also
Tables S5, S61).
Relative cormorant density
Variance in relative cormorant density was explained by site
("2 = 219.8, df = 5, p < 0.001) and chick rearing stage ("2 = 16.2, df =
1, p < 0.001) in a negative binomial regression model. There was a
significant interaction between chick rearing stage and site ("2 =
150.6, df = 5, p < 0.001). This interaction was driven largely by a
reduction in cormorant density on Nemeiben Lake during the late

chick rearing stage, with corresponding increases on Egg Lake
and south Lac La Ronge (Fig. 5). Seasonal switching between foraging areas by cormorants is further supported by observations of
foraging flocks (Fig. 6). We detected 66 cormorant flocks ranging
from 10 to 2500 individuals (385.4 ± 56.3) in water depths of 1.2–
13.1 m (4.5 ± 0.3 m). In the early chick rearing stage, 38% (13/34) of
flocks were observed on Nemeiben Lake and 44% (15/34) on Lac La
Ronge. During the late chick rearing stage, 66% (21/32) of flocks
were observed on Egg Lake.

Discussion
Stable isotope values from prey and relative densities by site
suggest that breeding cormorants foraged widely over much of
the La Ronge area, including different lakes and regions within
lakes. Cormorants also focused much of their feeding at sites that
were well removed from the breeding colony lake and switched
locations regularly; feeding site changes were common to many
individuals within colonies. Therefore, cormorants are not foraging solely on the lake containing their breeding colony, or even on
lakes closest to their breeding colony. Acentric colony locations
on the perimeter of the foraging range have been observed in
other colonial waterbirds and can lead to increased flight distances to foraging areas (Wittenberger and Dollinger 1984). Correspondingly, cormorants nesting on Egg Lake were collecting
prey fish from areas over 20 km away from their breeding colony
in different habitats and ecozones. Use of these distant areas was
a major part of the foraging and feeding ecology of cormorants
(>80% of fish in 2010), indicating a relatively weak spatial connection between the location of the breeding colonies and foraging
sites. This insight into the feeding ecology of cormorants may
profoundly affect assessment of potential fisheries impacts in our
study area. Cormorants are spreading their fish biomass removal
over a variety of sites, perhaps mitigating negative effects on fish
in any given body of water. Conversely, multiple prey populations
may be negatively affected by cormorant consumption, rather
than just one.
Multiple foraging lakes and habitats may be a more common
feature of cormorant ecology than previously thought. Our findings are supported by other studies on great cormorants in Europe, where the birds did not consistently forage on the lake they
nested on. For example, great cormorants nesting on Lake Ymsen,
Sweden, spent 20% of the time foraging at other nearby lakes, 7
Published by NRC Research Press

276

Can. J. Fish. Aquat. Sci. Vol. 70, 2013

Fig. 3. Carbon and nitrogen stable isotopes values of regurgitated
yellow perch (TL = 70–130 mm) in (a) 2009 and (b) 2010. Regurgitated
yellow perch are separated by collection date and chick rearing
period (early = 17 June – 15 July; late = 16 July – 31 August) for 2009
and 2010, respectively. The approximate outline (rectangles) of
stable isotopes values for yellow perch collected from known
locations is shown for reference.

Density (birds·km-2)

14.0
Central
LLR

North
LLR

10.0
East
Nemeiben

Egg
8.0

20

Early chick rearing

18

Late chick rearing

16
14
12
10
8
6
4
2
0

Bigstone

(b)

22

18 June
5 July

12.0

δ15N (‰)

Can. J. Fish. Aquat. Sci. Downloaded from www.nrcresearchpress.com by Thompson Rivers University on 07/22/13
For personal use only.

(a)

Fig. 5. Geometric mean (±SE) cormorant density at each site in the
Lac La Ronge (LLR) study area during the early (17 June – 15 July;
open bars) and late (16 July – 31 August; filled bars) chick rearing
periods in 2010. Data were obtained from line transect surveys.

Bigstone

6.0
-32.00

-27.00

-22.00

14.0

-17.00

Early chick rearing
West
Nemeiben

Late chick rearing
Lac La
Ronge

12.0

10.0
East
Nemeiben
8.0

Egg
Bigstone

6.0
-32.00

Head Lake
-27.00

-22.00

-17.00

δ13C (‰)

Fig. 4. The proportion of regurgitated yellow perch that were
assigned to each site by collection date in 2009 and 2010.

Egg

Nemeiben North LLR Central LLR South LLR
Site

and 11 km away (Engström 2001). Similarly, great cormorants nesting on Lake Selent, Germany, foraged 50% of the time on that lake
and the other 50% of the time in two different marine areas located approximately 30 km away (Grémillet et al. 1995). In an
extreme case, great cormorants commuted 60 km from a breeding colony on Sheep Island in Northern Ireland, to feed in Lough
Neagh (Warke et al. 1994; Warke and Day 1995). The distances
travelled by cormorants in our study are comparable to those
documented for cormorants in other areas. Thus, cormorants
might commonly move large distances, 10–60 km away from their
breeding colony. To our knowledge there are currently no published North American studies on double-crested cormorants that
specifically address multiple foraging sites, despite evidence that
cormorant movement may be a common phenomenon. For example, Eisenhower and Parrish (2009) reported that foraging cormorants in a basin of Lake Champlain declined in July of 3 years,
indicating that they switched to other foraging areas. It is critical
that cormorant foraging ecology, specifically where they consume
prey, and how sites are selected, be more completely understood
to inform management decisions.
Despite temporal changes in the sources of prey, cormorants
were still consuming mainly small yellow perch (TL = 70–130 mm).
It is currently unclear why cormorants switched locations to capture the same prey; however, prey availability likely played a role
in their decisions (Warke et al. 1994; Warke and Day 1995;
Eisenhower and Parrish 2009). Cormorant prey capture rate has
been shown to relate more to prey density than to visual acuity
(Hao 2008), and prey availability has a greater influence on daily
food intake than water temperature and dive depth (Grémillet
and Wilson 1999). This suggests that cormorants forage in locations that provide the most efficiently available prey. Our observations suggest a similar explanation: gill nets in pelagic zones on
Egg Lake in 2010 yielded 0.34 perch·net−1·h−1 in June when cormorants were foraging elsewhere, but catch per unit effort increased
54-fold in August when Egg Lake became a more important feeding site. Thus, cormorants may have foraged elsewhere when
perch were not readily available in Egg Lake, perhaps because the
perch were occupying unsuitable aquatic habitats. Recent research has shown that fish activity levels and movement may
affect gill-net capture rates (Rennie et al. 2012), so seasonal differences in these parameters may also have influenced our perch
catch per unit effort during the two sampling periods on Egg Lake.
It is therefore not clear whether gill net capture data are a suitable
Published by NRC Research Press

Bugajski et al.

277

Can. J. Fish. Aquat. Sci. Downloaded from www.nrcresearchpress.com by Thompson Rivers University on 07/22/13
For personal use only.

Fig. 6. Locations of cormorant foraging flocks (≥10 individuals) observed in the Lac La Ronge area in 2010. Flocks were divided into early and
late chick rearing periods. Flock sizes ranged from 10 to 2500 individuals.

proxy for cormorant prey availability. In addition, cormorants
may have also been selecting foraging sites and prey to suit rearing of nestlings, which may benefit from particular size ranges
and species of prey. Future research examining the variables that
affect cormorant decisions regarding foraging sites would be very
useful for understanding which fish populations they are likely to
interact with.
Our findings demonstrate the utility of stable isotopes for tracking the provenance of cormorant prey. Relative cormorant density in 2010 showed trends in foraging locations in accordance
with stable isotope analysis. The MDF assigned 62% of regurgitated perch to Nemeiben Lake during the early chick rearing stage
and 20% to the same lake in the late chick rearing stage. Relative
cormorant densities and the locations of foraging flocks in 2010
showed a parallel shift, supporting our conclusions. Our analyses
of stable isotopes indicate high confidence in our assignment
of regurgitated perch; however, we recognize that the crossvalidation success of gill-netted perch of known origin
was <100%. Thus, assignments of regurgitated perch have some
degree of uncertainty. Overall, assignment rates based on our
stable isotopes data were similar to those that distinguished fish
populations using multiple (four or five) trace elements (Thorrold
et al. 1998; Gillanders and Kingsford 2000; Brazner et al. 2004).
Successful assignment rates in these studies ranged from 63% to
76%, with individual sites ranging from 50% to 100% (Thorrold
et al. 1998; Gillanders and Kingsford 2000; Brazner et al. 2004).
Considering that our study was done on a smaller spatial scale
with only two markers (!13C and !15N), the maximum likelihood
approach performed well and is an approach that should be considered for additional studies of prey source tracking in colonial
birds. For example, Hobson et al. (2012) showed marked spatial
variation in baseline and fish stable isotope values within Lake
Winnipeg (Manitoba, Canada), which supports our findings and
may make source tracking possible on that lake as well. In
addition, the general notion of using stable isotopes to determine the provenance of resources may be transferable to a wide

variety of other central place foragers beyond birds, provided
that they exist in landscapes with isotopically distinct patches
(Cook and Dawes-Gromadzki 2005).
It is important to consider that the sites sampled in this study
were assumed to be the only ones used by breeding cormorants
from Egg Lake. It is possible that cormorants fed at other locations
where prey had stable isotopes values similar to those sampled
from our “known” sites. Thus, there is some uncertainty involved
with assigning prey to specific locations. However, it is clear that
multiple isotopic populations were involved. Stable isotopes values can only vary so much over a small spatial scale (e.g., Overman
and Parrish 2001; Barks et al. 2010), so our technique likely has
limitations in terms of the number of sites that can be sampled
and the subsequent resolution of the MDF. A multivariate approach using several biogeochemical markers may greatly increase precision in detecting small spatial scale variation among
fish populations (Thorrold et al. 1998; Gillanders and Kingsford
2000; Brazner et al. 2004). Future studies should examine variation among sites and evaluate resolving power before large spatial
scale application of a stable isotope or chemical marker approach
(see Hobson et al. 2012).
Bioenergetics modeling indicates that cormorants can consume
large amounts of fish (Grémillet et al. 2003; Ridgway 2010). A
single double-crested cormorant weighing 2.0 kg will eat !500 g
fish·day−1, based on consuming 25% of its body mass daily (Dunn
1975; Glahn and Brugger 1995). A variety of studies have used such
per-capita consumption rates to estimate biomass removal by cormorants during the breeding season and have assumed that prey
originates near the colony (e.g., Johnson et al. 2002; Rudstam et al.
2004; Diana et al. 2006). For example, a small colony of 730 adult
double-crested cormorants was estimated by Rudstam et al. (2004)
to consume 47 300 kg of fish, whereas Johnson et al. (2002) estimated that a colony of 16 820 adults consumed 2.1 million kg of
fish during one breeding season. The approach of estimating fish
biomass consumed by cormorants solely from areas near breeding or roosting areas can be inaccurate for assessing impacts on
Published by NRC Research Press

Can. J. Fish. Aquat. Sci. Downloaded from www.nrcresearchpress.com by Thompson Rivers University on 07/22/13
For personal use only.

278

fish populations. Our findings show that cormorants obtained
prey from multiple distant (!30 km) locations, which negates the
approach of simply assuming one prey source. In addition, simple
applications of bioenergetic models present other serious problems for making management decisions. For example, fish biomass removal estimates by cormorants do not directly represent
the fish biomass that would have otherwise been recruited into
cohorts used by fisheries. Accumulating evidence, including our
findings presented here, suggests that biomass removal estimates
are only one small part of assessing cormorant impacts on fisheries resources and have limited usefulness.
Management implications
The conflict between cormorants and humans in the La Ronge
area, and other locations worldwide, stems from the fact that
some cormorant prey come from lakes with economically important fisheries that have declined in production over time. In our
case, the majority of cormorant prey (75%) in 2010 was obtained
from lakes that are heavily used by humans for fishing and recreational activities (Nemeiben Lake 55%, Lac La Ronge 20%) and have
experienced declining fisheries quality. These feeding areas are
well removed from the breeding colony location on Egg Lake;
thus, it would be a major error to assume that cormorants primarily interact with fish populations on Egg Lake in the assessment of
potential fisheries impacts. Egg Lake has essentially no sport fishery and only a small commercial harvest of lake whitefish, so it
would be easy to erroneously conclude that there is little cormorant conflict. However, the possible influence of cormorant predation on sport fish populations in Lac La Ronge and Nemeiben
Lake must be considered in any management planning exercise
for our study area. We recommend at minimum that a much more
detailed assessment of the species, size class, and age of cormorant prey fish be combined with a site-adjusted biomass removal
model that accounts for multiple feeding sites and their relative
importance to cormorants. This approach would enable fisheries
managers to evaluate the potential importance of cormorants as a
source of fish mortality compared with other factors at each site.
For example, adult cormorants in 2010 consumed approximately
104 064 kg of fish (based on 542 g·bird−1·day−1 and 800 breeding
pairs with a 120-day residence period; Ridgway 2010); however,
only 57 235 and 20 813 kg of this total are predicted to be drawn
from Nemeiben Lake and Lac La Ronge, respectively, based on our
study. This raises two key points regarding management: (i) multiple feeding sites spread out cormorant biomass removal, reducing cormorant impact at individual sites, and (ii) the biomass
removal may or may not be a significant fisheries management
concern for a given body of water. Lac La Ronge, the site of primary concern in our study area, is a large lake (130 700 ha) in
which 20 813 kg of fish biomass, comprising mostly yellow perch,
cisco, and white suckers, is likely irrelevant to sport and commercial fisheries. Thus, we would not recommend any cormorant
management action based on the Lac La Ronge fishery. The other
sites, especially Nemeiben Lake, would need to be carefully assessed in a similar vein. As a final point, it is interesting that
cormorants in our study area are choosing to nest on a lake that is
isolated (Egg Lake; no road access) and has very little human use,
but they travel long distances to feed on lakes in which food webs
have been heavily disrupted by humans (Duffy 2007). It is highly
probable that human disruption of food webs creates conditions
more favourable for cormorants (e.g., by reducing abundance of
top predator fish; Hobson et al. 1989; Gloutney and Chen 1992;
Post et al. 2002), and thereby leads to the current situation. We
suggest that effective long-term management of intact aquatic
food webs is a key feature of preventing cormorant conflict with
humans over fisheries resources.

Can. J. Fish. Aquat. Sci. Vol. 70, 2013

Acknowledgements
Support was provided by the Saskatchewan Ministry of Environment's Fish and Wildlife Development Fund, the Natural Sciences
and Engineering Research Council of Canada (NSERC), the Canada
Research Chair (CRC) program, the Canada Foundation for Innovation (CFI), and the Science and Innovation Fund of the Province
of Saskatchewan. P. Barks, N. Hoggarth, A. White, L. Heisler, and
M. Duffy provided project assistance. We thank B. Hall, three
anonymous reviewers, and the editors-in-chief for helpful comments on this manuscript. This research was approved by the
University of Regina President's Committee on Animal Care (AUP
No. 09-06). Collection permits were provided by the Saskatchewan
Ministry of Environment.

References
Anderson, C.D., Roby, D.D., and Collis, K. 2004. Foraging patterns of male and
female double-crested cormorants nesting in the Columbia River estuary.
Can. J. Zool. 82(4): 541–554. doi:10.1139/z04-019.
Ashmole, N.P. 1963. The regulation of numbers of tropical oceanic birds. Ibis,
103b(3): 458–473. doi:10.1111/j.1474-919X.1963.tb06766.x.
Barks, P.M., Doucette, J.L., and Somers, C.M. 2010. Lack of angling-sized yellow
perch in a Canadian boreal lake: potential influences of growth rate, diet,
and predation by double-crested cormorants. Trans. Am. Fish.Soc. 139(4):
1029–1040. doi:10.1577/T09-174.1.
Bearhop, S., Thompson, D.R., Waldron, S., Russell, I.C., Alexander, G., and
Furness, R.W. 1999. Stable isotopes indicate the extent of freshwater feeding
by cormorants Phalacrocorax carbo shot at inland fisheries in England. J. Appl.
Ecol. 36(1): 75–84. doi:10.1046/j.1365-2664.1999.00378.x.
Bond, A.L., and Hobson, K.A. 2012. Reporting stable-isotope ratios in ecology:
Recommended terminology, guidelines and best practices. Waterbirds, 35(2):
324–331. doi:10.1675/063.035.0213.
Brazner, J.C., Campana, S.E., and Tanner, D.K. 2004. Habitat fingerprints for Lake
Superior coastal wetlands derived from elemental analysis of yellow perch
otoliths. Trans. Am. Fish. Soc. 133(3): 692–704. doi:10.1577/T03-091.1.
Coleman, J.T.H., and Richmond, M.E. 2007. Daily foraging patterns of adult
double-crested cormorants during the breeding season. Waterbirds, 30(2):
189–198. doi:10.1675/1524-4695(2007)30[189:DFPOAD]2.0.CO;2.
Coleman, J.T.H., Richmond, M.E., Rudstam, L.G., and Mattison, P.M. 2005. Foraging location and site fidelity of the double-crested cormorant on Oneida Lake,
New York. Waterbirds, 28(4): 498–510. doi:10.1675/1524-4695(2005)28[498:
FLASFO]2.0.CO;2.
Cook, G.D., and Dawes-Gromadzki, T.Z. 2005. Stable isotope signatures and landscape functioning in banded vegetation in arid-central Australia. Landsc.
Ecol. 20(6): 649–660. doi:10.1007/s10980-005-0069-1.
Cushing, C.E., Jr. 1964. Plankton and water chemistry in the Montreal River
lake-stream system, Saskatchewan. Ecology, 45(2): 306–313. doi:10.2307/
1933843.
Custer, T.W., and Bunck, C. 1992. Feeding flights of breeding double-crested
cormorants at two Wisconsin colonies. J. Field. Ornithol. 63(2): 203–211.
Diana, J.S., Maruca, S., and Low, B. 2006. Do increasing cormorant populations
threaten sportfishes in the Great Lakes? A case study in Lake Huron. J. Great
Lakes Res. 32(2): 306–320. doi:10.3394/0380-1330(2006)32[306:DICPTS]2.0.CO;2.
Doucette, J.L., Wissel, B., and Somers, C.M. 2011. Cormorant-fisheries conflicts:
stable isotopes reveal a consistent niche for avian piscivores in diverse food
webs. Ecol. Appl. 21(8): 2987–3001. doi:10.1890/10-2384.1.
Duffy, M. 2007. Lac La Ronge endorsement surveys 1997–2005. Fish and Wildlife
Technical Report 2007-3, Saskatchewan Ministry of Environment, Regina,
Saskatchewan.
Duffy, M. 2008. Assessment of Nemeiben Lake fish stocks, 2003. Fish and Wildlife Technical Report 2008-2, Saskatchewan Ministry of Environment, Regina, Saskatchewan.
Dunn, E.H. 1975. Caloric intake of nestling double-crested cormorants. Auk,
92(3): 553–565. doi:10.2307/4084609.
Eisenhower, M.D., and Parrish, D.L. 2009. Double-crested cormorant and fish
interactions in a shallow basin of Lake Champlain. Waterbirds, 32(3): 388–
399. doi:10.1675/063.032.0303.
Elliott, K.H., Woo, K.J., Gaston, A.J., Benvenuti, S., Dall'Antonia, L., and
Davoren, G.K. 2009. Central-place foraging in an Arctic seabird provides evidence for Storer-Ashmole's halo. Auk, 126(3): 613–625. doi:10.1525/auk.2009.
08245.
Engström, H. 2001. Long term effects of cormorant predation on fish communities and fishery in a freshwater lake. Ecography, 24(2): 127–138. doi:10.1034/j.
1600-0587.2001.240203.x.
Franks, S.E., Norris, D.R., Kyser, T.K., Fernández, G., Schwarz, B., Carmona, R.,
Colwell, M.A., Sandoval, J.C., Dondua, A., Gates, H.R., Haase, B., Hodkinson, D.J.,
Jiménez, A., Lanctot, R.B., Ortego, B., Sandercock, B.K., Sanders, F.,
Takekawa, J.Y., Warnock, N., Ydenberg, R.C., and Lank, D.B. 2012. Range-wide
patterns of migratory connectivity in the western sandpiper Calidris mauri.
J. Avian Biol. 43(2): 155–167. doi:10.1111/j.1600-048X.2012.05573.x.
Published by NRC Research Press

Can. J. Fish. Aquat. Sci. Downloaded from www.nrcresearchpress.com by Thompson Rivers University on 07/22/13
For personal use only.

Bugajski et al.
Frederiksen, M., Lebreton, J.D., and Bregnballe, T. 2001. The interplay between
culling and density-dependence in the great cormorant: a modelling approach. J. Appl. Ecol. 38(3): 617–627. doi:10.1046/j.1365-2664.2001.00620.x.
Genz, A., Bretz, F., Miwa, T., Mi, X., Leisch, F., Scheipl, F., and Hothorn, T. 2011.
mvtnorm: Multivariate normal and t distributions [online.] Available from
http://CRAN.R-project.org/package=mvtnorm [accessed 1 September 2012].
Gillanders, B.M., and Kingsford, M.J. 2000. Elemental fingerprints of otoliths of
fish may distinguish estuarine ‘nursery’ habitats. Mar. Ecol. Prog. Ser. 201:
273–286. doi:10.3354/meps201273.
Glahn, J.F., and Brugger, K.E. 1995. The impact of double-crested cormorants on
the Mississippi Delta catfish industry: a bioenergetics model. Colon. Waterbird, 18(Spec. Publ. 1): 168–175.
Gloutney, B.L., and Chen, M.Y. 1992. Status of the Lac La Ronge fishery 1989 and
1990. Fisheries Technical Report 92-2, Saskatchewan Ministry of Environment, Regina, Saskatchewan.
Gorke, M., and Brandl, R. 1986. How to live in colonies: spatial foraging strategies
of the black-headed gull. Oecologia, 70(2): 288–290. doi:10.1007/BF00379253.
Gould, P.J., and Forsell, D.J. 1989. Techniques for shipboard surveys of marine
birds. Fish and Wildlife Technical Report 25, United States Department of the
Interior Fish and Wildlife Service, Washington, D.C.
Grémillet, D., and Wilson, R.P. 1999. A life in the fast lane: energetics and
foraging strategies of the great cormorant. Behav. Ecol. 10(5): 516–524. doi:
10.1093/beheco/10.5.516.
Grémillet, D., Schmid, D., and Culik, B. 1995. Energy requirements of breeding
great cormorants Phalacrocorax carbo sinensis. Mar. Ecol. Prog. Ser. 121: 1–9.
doi:10.3354/meps121001.
Grémillet, D., Wright, G., Lauder, A., Carss, D.N., and Wanless, S. 2003. Modelling the daily food requirements of wintering great cormorants: a bioenergetics tool for wildlife management. J. Appl. Ecol. 40(2): 266–277. doi:10.1046/
j.1365-2664.2003.00806.x.
Hao, W. 2008. Spatial analysis of the impact of dredging induced increased water
turbidity on the visual acuity of great cormorants in Poyang Lake, China.
M.Sc. thesis, School of Resources and Environmental Sciences, Wuhan University, Wuhan, Hubei, China.
Hatch, J.J. 1995. Changing populations of double-crested cormorants. Colon.
Waterbird, 18(Spec. Publ. 1): 8–24.
Hobson, K.A. 2009. Trophic interactions between cormorants and fisheries: Towards a more quantitative approach using stable isotopes. Waterbirds, 32(4):
481–490. doi:10.1675/063.032.0401.
Hobson, K.A., Knapton, R.W., and Lysack, W. 1989. Population, diet and reproductive success of double-crested cormorants breeding on Lake Winnipegosis, Manitoba, in 1987. Colon. Waterbird, 12(2): 191–197.
Hobson, K.A., Ofukany, A., Soto, D.X., and Wassenaar, L.I. 2012. An isotopic
baseline (!13C, !15N) for fishes of Lake Winnipeg: Implications for investigating impacts of eutrophication and invasive species. J. Great Lakes Res.
38(Suppl. 3): 58–65. doi:10.1016/j.jglr.2010.11.008.
Johnson, J.H., Ross, R.M., and McCullough, R.D. 2002. Little Galloo Island,
Lake Ontario: a review of nine years of double-crested cormorant diet and
fish consumption information. J. Great Lakes Res. 28(2): 182–192. doi:10.1016/
S0380-1330(02)70575-3.
Kotzerka, J., Hatch, S.A., and Garthe, S. 2011. Evidence for foraging-site fidelity
and individual foraging behavior of pelagic cormorants rearing chicks in the
Gulf of Alaska. Condor, 113(1): 80–88. doi:10.1525/cond.2011.090158.
Murchie, K.J., and Power, M. 2004. Growth- and feeding-related isotopic dilution
and enrichment patterns in young-of-the-year yellow perch (Perca flavescens).
Freshw. Biol. 49(1): 41–54. doi:10.1046/j.1365-2426.2003.01163.x.
Neuman, J., Pearl, D.L., Ewins, P.J., Black, R., Weseloh, D.V., Pike, M., and
Karwowski, K. 1997. Spatial and temporal variation in the diet of doublecrested cormorants (Phalacrocorax auritus) breeding on the lower Great Lakes
in the early 1990s. Can. J. Fish. Aquat. Sci. 54(7): 1569–1584. doi:10.1139/f97-064.
Ofukany, A.F.A., Hobson, K.A., and Wassenaar, L.I. 2012. Connecting breeding
and wintering habitats of migratory piscivorous birds: implications for tracking contaminants (Hg) using multiple stable isotopes. Environ. Sci. Technol.
46(6): 3263–3272. doi:10.1021/es204135s. PMID:22385064.
Overman, N.C., and Parrish, D.L. 2001. Stable isotope composition of walleye: 15N
accumulation with age and area-specific differences in !13C. Can. J. Fish.
Aquat. Sci. 58(6): 1253–1260. doi:10.1139/f01-072.

279
Peterson, B.J., and Fry, B. 1987. Stable isotopes in ecosystem studies. Annu. Rev.
Ecol. Syst. 18(1): 293–320. doi:10.1146/annurev.es.18.110187.001453.
Platteeuw, M., and Van Eerden, M.R. 1995. Time and energy constraints of fishing
behaviour in breeding cormorants Phalacrocorax carbo sinensis at Lake Ijsselmeer, The Netherlands. Ardea, 83(1): 223–234.
Post, D.M., Layman, C.A., Arrington, D.A., Takimoto, G., Quattrochi, J., and
Montaña, C.G. 2007. Getting to the fat of the matter: models, methods and
assumptions for dealing with lipids in stable isotope analyses. Oecologia,
152(1): 179–189. doi:10.1007/s00442-006-0630-x. PMID:17225157.
Post, J.R., Sullivan, M., Cox, S., Lester, N.P., Walters, C.J., Parkinson, E.A.,
Paul, A.J., Jackson, L., and Shuter, B.J. 2002. Canada's recreational fisheries:
the invisible collapse? Fisheries, 27(1): 6–17. doi:10.1577/1548-8446(2002)027<
0006:CRF>2.0.CO;2.
Rennie, M.D., Johnson, T.B., and Sprules, W.G. 2012. Energy acquisition and
allocation patterns of lake whitefish (Coregonus clupeaformis) are modified
when dreissenids are present. Can. J. Fish. Aquat. Sci. 69(1): 41–59. doi:10.1139/
f2011-126.
Ridgway, M.S. 2010. A review of estimates of daily energy expenditure and food
intake in cormorants (Phalacrocorax spp.). J. Great Lakes Res. 36(1): 93–99.
doi:10.1016/j.jglr.2009.10.002.
Royle, J.A., and Rubenstein, D.R. 2004. The role of species abundance in determining breeding origins of migratory birds with stable isotopes. Ecol. Appl.
14(6): 1780–1788. doi:10.1890/04-0175.
Rubenstein, D.R., and Hobson, K.A. 2004. From birds to butterflies: animal movement patterns and stable isotopes. Trends Ecol. Evol. 19(5): 256–263. doi:10.
1016/j.tree.2004.03.017. PMID:16701265.
Rudstam, L.G., VanDeValk, A.J., Adams, C.M., Coleman , J.T.H., Forney, J.L., and
Richmond, M.E. 2004.Cormorant predation and the population dynamics of
walleye and yellow perch in Oneida Lake. Ecol. Appl. 14(1): 149–163. doi:10.
1890/03-5010.
Seefelt, N.E., and Gillingham, J.C. 2006. Foraging locations of double-crested
cormorants in the Beaver Archipelago of northern Lake Michigan: potential
for impacts on smallmouth bass. Waterbirds, 29(4): 473–480. doi:10.1675/15244695(2006)29[473:FLODCI]2.0.CO;2.
Shwiff, S., Kirkpatrick, K., and DeVault, T. 2009. The economic impact of doublecrested cormorants to central New York. New York Wildlife Services Report,
National Wildlife Research Center, Fort Collins, Colorado.
Stapanian, M.A., Bur, M.T., Tyson, J.T., Seamans, T.W., and Blackwell, B.F. 2002.
Foraging locations of double-crested cormorants on western Lake Erie: site
characteristics and spatial associations with prey fish densities. J. Great Lakes
Res. 28(2): 155–171. doi:10.1016/S0380-1330(02)70573-X.
Thorrold, S.R., Jones, C.M., Swart, P.K., and Targett, T.E. 1998. Accurate classification of juvenile weakfish Cynoscion regalis to estuarine nursery areas based
on chemical signatures in otoliths. Mar. Ecol. Prog. Ser. 173: 253–265. doi:10.
3354/meps173253.
Todorov, P., and Filzmoser, P. 2009. An object-oriented framework for robust
multivariate analysis. J. Stat. Softw. 32(3): 1–47.
Wallace, R.G. 1993. Nemeiben Lake: walleye, pike, and trout fisheries, 1978–1992.
Fisheries Technical Report 93-5, Saskatchewan Ministry of Environment, Regina, Saskatchewan.
Warke, G.M.A., and Day, K.R. 1995. Changes in abundance of cyprinid and percid
prey affect rate of predation by cormorants Phalacrocorax carbo carbo on
salmon Salmo salar smolt in Northern Ireland. Ardea, 83(1): 157–166.
Warke, G.M.A., Day, K.R., Greer, J.E., and Davidson, R.D. 1994. Cormorant (Phalacrocorax carbo [L]) populations and patterns of abundance at breeding and
feeding sites in Northern Ireland, with particular reference to Lough Neagh.
Hydrobiologia, 279 –280(1): 91–100. doi:10.1007/BF00027844.
Wittenberger, J.F., and Dollinger, M.B. 1984. The effect of acentric colony location on the energetics of avian coloniality. Am. Nat. 124(2): 189–204. doi:10.
1086/284263.
Wunder, M.B., and Norris, D.R. 2008. Improved estimates of certainty in stableisotope-based methods for tracking migratory animals. Ecol. Appl. 18(2):
549–559. doi:10.1890/07-0058.1. PMID:18488615.
Wunder, M.B., Kester, C.L., Knopf, F.L., and Rye, R.O. 2005. A test of geographic
assignment using isotope tracers in feathers of known origin. Oecologia,
144(4): 607–617. doi:10.1007/s00442-005-0071-y. PMID:15891836.

Published by NRC Research Press