food web structure of a coastal arctic marine ecosystem ... · berland sound from atlantic waters...

MARINE ECOLOGY PROGRESS SERIESMar Ecol Prog Ser

Vol. 482: 17–28, 2013doi: 10.3354/meps10278

Published May 22

INTRODUCTION

Over the next few decades, Arctic marine ecosys-tems will continue to face increased stress from amultitude of climate change scenarios (Wassmann etal. 2011). The predicted effects of these stressorsrange in scale from changes in the physiology of indi-vidual animals (e.g. decreased health) to overarchingecosystem-level effects (e.g. due to decreased icecover), and have the potential to affect the structure,

functioning and stability (in terms of variability, resil-ience and persistence) of Arctic food webs. It is,therefore, essential for ecologists to identify the foodweb structures that confer stability to these imperiledecosystems.

The mechanisms governing the stability of foodwebs has been of interest to ecologists for decades(MacArthur 1955), and structures linked to increasedfood web stability are known to arise within the com-plex feeding interactions that make up food webs

© Inter-Research and Her Majesty the Queen in Right ofCanada 2013 · www.int-res.com

*Email: [email protected]

Food web structure of a coastal Arctic marine ecosystem and implications for stability

Bailey C. McMeans1,*, Neil Rooney2, Michael T. Arts3, Aaron T. Fisk1

1Great Lakes Institute for Environmental Research, University of Windsor, 401 Sunset Ave., Windsor, Ontario N9B 3P4, Canada 2Department of Integrative Biology, University of Guelph, Guelph, Ontario N1G 2W1, Canada

3National Water Research Institute, Environment Canada, 867 Lakeshore Road, PO Box 5050, Burlington, Ontario L7R 4A6, Canada

ABSTRACT: There is little doubt that Arctic ecosystems will continue to face unprecedentedchange in the coming decades. The identification of food web structures that confer stability tothese systems is, therefore, a priority. Here, we use stable isotopes and fatty acids to resolve thefood web structure of a seasonally ice-covered fjord (Cumberland Sound, Baffin Island, Canada)sampled in late summer. We show that the food web is structured such that upper trophic levelscouple separate energy channels (based on phytoplankton or macroalgae), a previously docu-mented food web structure that has been linked with stability in temperate ecosystems, but neverestablished in a seasonally dynamic, ice-covered ecosystem. Herbivorous zooplankton (e.g.Calanus hyperboreus) relied exclusively on phytoplankton, whereas herbivorous benthos usedeither phytodetritus (e.g. Hiatella arctica) or macroalgae (e.g. Tectura testudinalis), supporting theexistence of separate energy channels. Upper trophic level fishes and marine mammals reliedmore heavily on phytoplankton- than macroalgal-derived carbon (58 to 100% reliance on phyto-plankton), but 6 out of 8 species sampled derived energy from both carbon sources. Since benthicinvertebrate predators used both phytodetrital- and macrolgal-based resources, the coupling ofseparate energy channels was also iterated within the benthos. The temporally pulsed nature ofphytoplankton production, characteristic of Arctic seas, indicates that Arctic consumers also act ascouplers of resources in time because phytoplankton- and detrital-based carbon would likelyreach upper trophic levels earlier and later in the season, respectively. Potential changes in therelative production of macroalgae and phytoplankton under climate change scenarios couldimpact the stability-promoting food web structure reported here.

KEY WORDS: Resource coupling · Ecological patterns · Arctic marine ecology · Food web · Stableisotopes · Fatty acids · Macroalgae · Climate warming

Resale or republication not permitted without written consent of the publisher

Mar Ecol Prog Ser 482: 17–28, 2013

(e.g. Polis et al. 1996). More recent work combiningfood web theory with empirical observations hasrevealed that a seemingly common structure, thecoupling of different energy channels in space bymobile, upper trophic level predators, imparts stabil-ity to food webs (Rooney et al. 2006). These energychannels are based on different basal resources andarise because lower trophic levels often deriveenergy predominantly from singular carbon sources(e.g. phytoplankton or detritus). Upper trophic levelorganisms act to couple different energy channels asthey move through spatial environments, consumingmultiple prey types (McCann & Rooney 2009). To -gether, these factors impart a ‘hump-shape’ structureto food webs when graphed on axes of percentreliance on 1 of 2 carbon sources versus trophic posi-tion (Fig. 1; Rooney et al. 2006), which is hypothe-sized to be iterative across spatial scales and withinand between ecosystems (McCann & Rooney 2009).Importantly, biomass turnover rates of taxonomicgroups and the strength of interactions between con-sumers and resources are markedly different be -tween the energy channels, resulting in weak andstrong energy channels. Strong energy channels arethose characterized by high production rates andstrong trophic interactions, whereas weak energychannels are based on a resource with low produc-tion and constituted by weak trophic interactions(Rooney et al. 2006). The coupling behavior of mobiletop predators has the potential to generate asynchro-nous dynamics of prey species between energy chan-nels, producing a less variable resource base forpredators. The weak channel also competes with the

strong channel, muting some of the energy flow thatwould go through the strong channel in the absenceof the weak channel. This muting also enhances sta-bility. Thus far, coupling of energy channels by uppertrophic levels has been identified in terrestrial, mar-ine and freshwater food webs from latitudes withtemperate climates (Rooney et al. 2006, Dolson et al.2009), but it is unknown if food webs from latitudesthat experience high seasonal variability, and whichare driven by brief but large pulses of primary pro-duction (Renaud et al. 2011), exhibit such structure.

Climate warming effects are hypothesized to alterthe relative contribution of various primary produc-ers to overall food web production in the Arctic(Wassmann et al. 2011, Weslawski et al. 2011).Pelagic primary production in Arctic seas currentlytakes place during a 1 to 4 mo window, when intenseice algae and phytoplankton blooms fuel pelagicenergy channels (Søreide et al. 2006) and, when notcompletely grazed, sink to provide labile phytodetri-tus that is rapidly exploited by benthos (Lovvorn etal. 2005). Contrary to the situation in temperate seas,Arctic consumers must contend with >10 mo, in someareas, of sea-ice and snow-cover that markedlyreduces or entirely halts primary production (Wes-lawski et al. 1991). Based on the large magnitude andefficient transfer of energy from phytoplankton toupper trophic levels during productive periods (Falk-Petersen et al. 2007), one may expect that summerArctic food webs may deviate from theoretical pre-dictions (Fig. 1; Rooney et al. 2006), and instead bedriven predominantly by only one carbon source, i.e.phytoplankton. An additional primary productionsource in some coastal Arctic Seas, however, ismacroalgae, although few studies have assessedwhether Arctic consumers rely heavily on macroalgalcarbon (Dunton & Schell 1987, Hobson et al. 1995).Given the differential biomass turnover rate ofmacroalgae compared to phytoplankton, macroalgaecould provide the basis for an important weak energychannel that could supplement the strong energychannel based on phytoplankton. Such a structure, ifapparent, could be important for the stability ofcoastal Arctic food webs. More broadly, identifyingwhether or not the coupling of different energy chan-nels by upper trophic levels occurs in Arctic foodwebs would lend insight into the ubiquity of suchstructures among ecosystems.

In the present study, we used stable nitrogen(δ15N), carbon isotopes (δ13C) and fatty acids to delin-eate the structure of a coastal, Arctic food web and toanswer the following 2 questions: (1) is a coastal Arc-tic food web structured such that lower trophic levels

18

Fig. 1. Conceptual model showing hypothesized food webstructure arising from lower trophic levels (black and whitesymbols) feeding predominantly within 1 of 2 energy chan-nels and upper trophic levels (grey symbols) acting asresource couplers by using prey from both energy channels

McMeans et al: Resource coupling by Arctic consumers

use distinct energy channels, based on phytoplank-ton and macroalgae, which are coupled by uppertrophic levels? (2) If so, what is the relative contribu-tion of different resources to upper trophic levels?

MATERIALS AND METHODS

Study site and species sampling

Sampling was conducted within or just outside themouth of Pangnirtung fjord in Cumberland Sound,Baffin Island (see Fig. S1 in the Supplement fora map of sampling locations; available at www.int-res. com/articles/suppl/m482p017_supp.pdf). Pang-nirtung fjord is characterized by wide (up to 600 m)intertidal flats, expansive growths of rockweed Fucusdistichus, a maximum depth of 150 m and ice coverfrom November until June or July (see Fig. S2 in the Supplement for photographs). Parts of CumberlandSound, including the southern portion of Pangnir-tung fjord sampled here, are immediately below theArctic Circle (by ~30 km), but we retain the conven-tion of previous authors by defining CumberlandSound as part of the Canadian Arctic (e.g. Lee 1973).Cumberland Sound waters experience colder tem-peratures and a longer duration of ice-cover thansimilar latitudes in Europe due to the influence of theGulf Stream on the latter. Both Arctic (i.e. BaffinIsland Current) and Atlantic (i.e. Greenland Current)water masses (Aitken & Gilbert 1989) influenceCumberland Sound. The fauna is subsequently ofboth Arctic and Atlantic Ocean origin (Aitken &Gilbert 1989), which is characteristic of other Arcticfjords (e.g. Kongsfjorden, Svalbard; Hop et al. 2006).Throughout the Canadian Arctic, marine benthicalgae are predominantly of Atlantic origin (Lee 1973).Carmack & Wassmann (2006) classify CumberlandSound, and the remainder of Canadian Archipelagoshorelines, as ‘outflow shelves’ in their review ofpan-Arctic shelf types.

Particulate organic matter (POM), rockweed and21 benthic and pelagic invertebrate and vertebratespecies were sampled for stable isotopes and/or fattyacids during summer in August 2007, 2008 and 2009and classified into functional groups (see Table 1 forspecies sampling year and functional group classifi-cations) based on previously reported habitat anddiet attributes (Table S1 in the Supplement). The fol-lowing species were sampled only for stable isotopes:POM, jellyfish Agnatha digitale, periwinkle Littorinasp., arrow worm Sagitta sp. and herring Clupeaharengus. Similar sized individuals of each species

were collected, with the exception of adult sculpinMyoxocephalus scorpius, of which small (24 cm) size classes were sampled andtreated separately due to size-related diet variabilityin this species (Cardinale 2000).

POM was sampled via a 10 µm plankton net(Wildlife Supply Company®) from 50 m to the sur-face and rockweed was sampled either by hand or byPonar grab sampler. Each rockweed sample con-sisted of the distal tip of one leaf from one plant. Ben-thos were sampled via dip net, except for scallopsChlamys islandica which were collected in water30 to 40 m deep using a dredge. Zooplankton were captured by towing a plankton net (243 µm mesh;Wildlife Supply Company®) at the surface and byperforming vertical hauls down to ~50 m. Pelagic fishwere sampled via dip nets and gill nets and sculpinwere captured using baited fishing line. The remain-ing fishes were collected using bottom long lines (50hooks, ~200 m long). Marine mammals were cap-tured during Inuit subsistence hunting.

We sampled representative species for each func-tional group listed in Table 1. Abundance and bio-mass data do not exist for Cumberland Sound spe-cies, but we collected the most commonly observedspecies during our sampling efforts. The speciessampled included those unique to sub-Arctic andArctic Seas (e.g. copepod Calanus hyperboreus;ringed seal Pusa hispida). Species that enter Cum-berland Sound from Atlantic waters (herring andcapelin Mallotus villosus) or from the surroundingrivers (arctic char Salvelinus alpinus) were also sam-pled because they are members of the summer foodweb. The obvious species missing from our study ispolar cod Boreogadus saida which were not presentin our samples.

Multiple individuals (2 to 10) of each zooplankter,polychaete worm and amphipod Gammarus oceani-cus were pooled for stable isotope and fatty acidssamples to obtain sufficient material for analysis.Similar tissues were sampled for both stable isotopesand fatty acids, except that dorsal muscle and blub-ber (inner layer) were sampled from marine mam-mals for these analyses, respectively. All sampleswere placed into cyrovials and immediately frozen at−20°C (stable isotopes) and at −80°C (fatty acids),and kept at these temperatures, until analysis.

Stable isotope and fatty acid analysis

Lipid extracted samples were analyzed for stableisotopes as previously described in McMeans et al.

19

http://www.int-res.com/articles/suppl/m482p017_supp.pdfhttp://www.int-res.com/articles/suppl/m482p017_supp.pdf

Mar Ecol Prog Ser 482: 17–28, 201320

Fu

nct

ion

al g

rou

pS

pec

ies

Com

mon

nam

eF

ig. 2

Dat

eT

issu

eδ1

3 Cδ1

5 NT

Pα

Pri

mar

y P

arti

cula

te o

rgan

ic m

atte

rP

OM

na

2008

Wh

ole

−22

.18.

24p

rod

uce

rsF

ucu

s d

isti

chu

sR

ock

wee

dn

a20

08L

eaf

−14

.9 ±

1.1

76.

40 ±

1.0

820

09−

15.1

3 ±

1.8

07.

49 ±

1.2

9

Zoo

pla

nk

ton

Her

biv

ores

Cal

anu

s h

yper

bor

eus

Cop

epod

1a20

08W

hol

e−

20.3

7 ±

0.1

49.

88 ±

0.5

32.

0 ±

0.2

98 ±

320

09W

hol

e−

20.8

4 ±

0.4

99.

90 ±

0.3

42.

0 ±

0.1

96 ±

6M

ysis

ocu

lata

bM

ysid

220

09W

hol

e−

20.7

9 ±

0.1

09.

02 ±

0.1

31.

7 ±

0.1

95 ±

2O

mn

ivor

esA

gla

nth

a d

igit

ale

Jell

yfis

h3

2008

Wh

ole

−20

.46

± 0

.26

10.1

2 ±

0.3

92.

1 ±

0.1

99 ±

1M

yoxo

cep

hal

us

scor

piu

s(l

arva

e)L

arva

l sc

ulp

in4

2009

Wh

ole

−20

.54

± 0

.44

10.5

4 ±

0.2

72.

2 ±

0.1

95 ±

7S

tich

aeu

s p

un

ctat

us

(lar

vae)

Lar

val

arct

ic s

han

ny

520

09W

hol

e−

21.0

9 ±

0.2

210

.63

± 0

.34

2.2

± 0

.110

0 ±

1C

arn

ivor

eS

agit

tasp

.A

rrow

wor

m6

2008

Wh

ole

−19

.25

± 0

.47

14.1

± 0

.01

3.2

± 0

.194

± 9

Ben

thos

Her

biv

ores

Ch

lam

ys i

slan

dic

a(F

F)

Sca

llop

7a20

08M

usc

le−

18.6

3 ±

0.6

59.

65 ±

0.1

61.

9 ±

0.1

59 ±

15

2009

Mu

scle

−18

.79

± 0

.20

8.89

± 0

.38

1.7

± 0

.153

± 3

Hia

tell

a ar

ctic

a(F

F)

Cla

m8

2008

Man

tle

−19

.52

± 0

.23

8.14

± 0

.43

1.5

± 0

.171

± 5

Gam

mar

us

ocea

nic

us

(G)

Am

ph

ipod

920

08W

hol

e−

18.5

5 ±

0.2

99.

48 ±

0.7

11.

9 ±

0.2

57 ±

5L

itto

rin

asp

. (G

)P

eriw

ink

le10

2009

Wh

ole

−16

.13

± 0

.62

9.39

± 0

.50

1.8

± 0

.15

± 8

Tec

tura

tes

tud

inal

is(G

)L

imp

et11

2008

Foo

t−

15.9

6 ±

0.4

69.

87 ±

0.1

92.

0 ±

0.1

4 ±

5C

arn

ivor

esB

ucc

inu

m c

yan

eum

Wh

elk

1220

08F

oot

−17

.27

± 0

.63

12.9

2 ±

0.5

82.

9 ±

0.2

46 ±

12

Nu

dib

ran

chc

Nu

dib

ran

ch13

2008

Wh

ole

−16

.81

± 0

.23

13.5

4 ±

0.1

83.

1 ±

0.1

39 ±

6P

olyc

hae

ted

Pol

ych

aete

1420

08W

hol

e−

16.1

8 ±

0.8

011

.21

± 0

.93

2.4

± 0

.313

± 1

1

Fis

hes

&

Car

niv

ores

Som

nio

sus

mic

roce

ph

alu

sG

reen

lan

d s

har

k21

2008

Mu

scle

−17

.74

± 0

.67

17.0

7 ±

0.6

84.

6 ±

0.3

86 ±

14

mar

ine

Am

bly

raja

hyp

erb

orea

Arc

tic

skat

e18

2007

Mu

scle

−18

.37

16.7

93.

910

0m

amm

als

Clu

pea

har

eng

us

Her

rin

g15

2007

Mu

scle

−19

.85

13.4

2.9

100

Mal

lotu

s vi

llos

us

Cap

elin

1620

08M

usc

le−

19.3

7 ±

0.1

513

.51

± 0

.29

3.1

± 0

.196

± 3

Sal

veli

nu

s al

pin

us

Arc

tic

char

1720

08M

usc

le−

18.6

1 ±

0.2

314

.51

± 0

.64

3.4

± 0

.285

± 8

Myo

xoce

ph

alu

s sc

orp

ius

(<24

cm

)S

hor

thor

n

1920

08M

usc

le−

17.9

9 ±

0.5

815

.07

± 0

.39

3.5

± 0

.174

± 1

2(>

24 c

m)

Scu

lpin

2020

08M

usc

le−

16.7

9 ±

1.0

717

.17

± 1

.14

4.1

± 0

.358

± 1

8P

hoc

a g

roen

lan

dic

aH

arp

sea

l22

2008

Mu

scle

−17

.95

± 0

.13

14.6

6 ±

0.5

53.

9 ±

0.3

80 ±

6P

usa

his

pid

aR

ing

sea

l23

2008

Mu

scle

4.1

± 0

.391

± 1

0

aM

ean

of

2008

an

d 2

009

show

n i

n F

ig. 2

bB

enth

ic-a

ssoc

iate

d b

ut

cap

ture

d i

n z

oop

lan

kto

n n

ets

c Un

kn

own

sp

ecie

s dU

nk

now

n s

pec

ies

bu

t p

red

ator

y b

ased

on

lar

ge

jaw

s

Tab

le 1

. S

pec

ies

sam

ple

d f

rom

Cu

mb

erla

nd

Sou

nd

du

rin

g s

um

mer

(A

ug

ust

). Y

ear

of s

amp

lin

g,

tiss

ue

anal

yzed

an

d a

ssoc

iate

d c

odes

for

Fig

. 2

are

pro

vid

ed.

Ben

thic

her

biv

ores

are

eit

her

fil

ter-

feed

ers

(FF

) or

gra

zers

(G

). V

alu

es o

f δ1

3 C, δ1

5 N (

bot

h ‰

), t

rop

hic

pos

itio

ns

(TP

) an

d p

erce

nt

reli

ance

on

ph

ytop

lan

kto

n v

ersu

s m

acro

alg

ae

(α)

are

mea

n ±

SD

(sp

ecie

s w

ith

out

SD

had

n =

1)


(2009). Acid washing was performed for amphipodand periwinkle (details provided in the Supplement).Stable isotopes are expressed as delta δ values whereδ X = 1000[(Rsample × Rstandard−1) – 1], X = 15N or 13C andR = the ratio of 15N:14N or 13C:12C. Replicate analysesof NIST (National Institute of Standards and Technol-ogy) standard bovine muscle (NIST 8414, N = 159)and internal lab standard (tilapia muscle, N = 159)yielded a precision (i.e. 1 SD) of 0.13 and 0.20‰ forδ15N and 0.07 and 0.08‰ for δ13C, respectively.

Fatty acid methyl esters were generated from thetotal lipid extract (see McMeans et al. 2012 for de-tailed analytical methods) and separated on a HewlettPackard 6890 gas chromatograph (GC) (splitless in-jection, column = Supelco [SP-2560 column]). Fattyacids were identified using a 37 component fatty acidstandard (Supelco 47885-U) and are reported as pro-portions (i.e. % of total identified fatty acids).

Data analyses

Applying stable isotopes to accurately estimatetrophic positions (TP) and carbon sources (α, in thepresent study, relative reliance on phytoplanktonversus macroalgae) requires identifying appropriatebaselines (Post 2002). Values of δ13C are often higherin macroalgae versus phytoplankton and have beenused as baselines in isotope mixing models (Fredrik-sen 2003, Nilsen et al. 2008). However, obtaining aphytoplankton sample free of other carbon sourcesand detritus is difficult (Iken et al. 2010) and usingPOM or offshore phytoplankton values as baselinesin mixing models can result in over-estimation ofconsumer reliance on macroalgae (Miller & Page2012). Primary consumers, on the other hand, aretemporarily and spatially less variable in their δ13Cthan primary producers (Vander Zanden et al. 1998,Iken et al. 2010) and should also be good integratorsof spatial and among-species isotopic variabilitywithin available phytoplankton and macroalgae. We,therefore, used the δ15N and δ13C of a primary con-sumer (i.e. assumed to occupy TP = 2) of phytoplank-ton (copepod Calanus hyperboreus; Stevens et al.2004, Søreide et al. 2008) and macroalgae (limpetTectura testudinalis; Fredriksen 2003) as baselinesfor calculations of TP and α. Since outputs from 1-and 2-source TP models (see Post 2002) were similar,the following 1-source TP model was applied for allindividuals:

(1)

Copepod mean δ15N (from the same year as thegiven consumer was sampled) was used as theδ15Nbaseline and 3.4‰ was used as the diet-tissue dis-crimination factor (i.e. Δ15N; Post 2002). CalculatingGreenland shark Somniosus microcephalus TP canbe problematic due to, for example, uncertainty overΔ15N (Hussey et al. 2012). The Greenland shark’s TPwas, therefore, calculated using: (1) 2.3‰ for theΔ15N (derived for large sharks; Hussey et al. 2010),and (2) the mean δ15N of capelin as the δ15Nbaseline(capelin TP = 3.1, Table 1). Percent reliance on phyto-plankton versus macroalgae (α) was calculated foreach individual using a 2-source mixing equation(Post 2002) modified to incorporate enrichment ofconsumer 13C at each trophic step (i.e. Δ13C) abovethe primary consumer baselines as follows:

(2)

Δ13C was set at 0.8‰ (Dunton & Schell 1987,Fredriksen 2003), TPbaseline is 2 and TPconsumer is theresult of the 1-source TP model for each individual(Eq. 1). By using primary consumers, instead of pri-mary producers, as baselines for isotope models, weassumed that the stable isotope signatures of cope-pod and limpet represented an integrated signatureof available phytoplankton and macroalgae in thearea. Since all of our sampling was conducted withinPangnirtung fjord (with the exception of arctic skateand herring, Fig. S1), we consider this an acceptableassumption for elucidating the structure of thiscoastal food web.

The data were first assessed for normality (q-qplots) and homoscedacity (Levene’s tests). Sincerockweed, copepod and scallop were sampled dur-ing multiple years for stable isotopes (2008 and 2009,Table 1), 2-way ANOVAs (factors = year and species)were used to identify inter-annual differences in δ13Cand δ15N for copepod and scallop. Separate Welch’s ttests were used for this purpose in rockweed due tohigher stable isotope variability in this versus othersampled species (Table 1). ANOVA and Tukey’s posthoc tests were used to compare δ13C, δ15N and TPsamong functional groups (i.e. zooplankton versusbenthos versus fish and marine mammals) and δ13C-derived values of α (i.e. % reliance on phytoplank-ton) were compared among these groups via aKruskal-Wallis non-parametric ANOVA followed byMann-Whitney tests (with Bonferroni corrections).

Different classes of macroalgae differ in their fattyacid profiles, but in general, macroalgae have lower

TP TPN N

consumer baselineconsumer base= + δ δ

15 15– lline15NΔ

αδ

=( )13C C TP TPconsumer 13 consumer baseline– –Δ ×[[ ]–

–

δδ δ

13

13 13

C

C Climpet

copepod limpet

×100

21


C22 polyunsaturated fatty acids (PUFA) relative tophytoplankton (Graeve et al. 2002). The brown macro-algae, Phaeophyta, also have high proportions of n-6PUFA (Graeve et al. 2002) relative to phytoplankton,which have high 16:1n-7 and n-3 PUFA (Stevens etal. 2004). No biomass data exist for CumberlandSound macroalgae, but rockweed is clearly abundanton the shores of Pangnirtung fjord (Fig. S2). We,therefore, assume that rockweeds provide the largestsource of fresh and detrital macroalgae to consumers.Fatty acid profiles of Cumberland Sound rockweedversus published values for POM from the CanadianArctic (Table S2 in the Supplement) agreed with pre-vious reports that, in general, brown macroalgaehave high proportions of certain fatty acids including18:2n-6 and 20:4n-6 (Graeve et al. 2002) and phyto-plankton have high 16:1n-7, 22:6n-3 and n-3:n-6(Stevens et al. 2004). To identify similarities in theproportions of these 5 fatty acid trophic markersamong Cumberland Sound species, non-metric mul-tidimensional scaling (NMDS, dimensions = 2,Euclidean distances) was performed on: (1) lowertrophic levels: rockweed, zooplankton, benthos and(2) upper trophic levels: fish and marine mammals.All analyses were performed in R (R DevelopmentCore Team 2010) and the significance level was set at0.01.

RESULTS

No differences existed in δ13C or δ15N between2008 and 2009 for rockweed (Welch’s t test, p > 0.05),copepod or scallop (2-way ANOVA, p > 0.05, Table 1provides all values of δ15N and δ13C). The among-functional group comparison of δ15N and δ15N-derived TPs (Table 1) revealed that zooplankton =benthos < fish and marine mammals (Table 2). Val-ues of δ13C decreased significantly from benthos >fish and marine mammals > zooplankton, and calcu-lated values of α exhibited the opposite trend, in -creasing from benthos < fish and marine mammals <zooplankton (Fig. 2, Table 2). Thus, coupling ofmacroalgae and phytoplankton energy channels byupper trophic levels was apparent because calcu-lated values of α for fish and marine mammals (i.e. 58to 100) fell in between, although overlapped with,that of benthos (4 to 71) and zooplankton (94 to 100)(Fig. 2). Herbivorous, omnivorous and carnivorouszooplankton relied entirely on pelagic carbon (Fig. 2,α > 95), whereas benthos exhibited a wider range ofresource use, from heavy reliance on macroalgae inperiwinkle (α = 5 ± 8, mean ± SD) and a carnivorous

polychaete (unknown species, α = 13 ± 11), to the useof both macroalgae and phytoplankton by scallop,clam Hiatella arctica and amphipods (Fig. 2, α = 53 ±3 to 71 ± 5). The carnivorous whelk Buccinum cya-neum and nudibranch (unknown species) appearedto couple the 2 aforementioned groups (Fig. 2, α = 46± 12 and 39 ± 6, respectively).

The NMDS performed on lower trophic level spe-cies’ fatty acid proportions supported δ13C-derived αvalues because all zooplankton separated away fromrockweed due to higher proportions of phytoplank-ton trophic markers (16:1n-7, 22:6n-3, n-3:n-6), andlower proportions of the macroalgae trophic markers18:2n-6 and 20:4n-6 (Fig. 3A, fatty acid proportionsprovided in Table S2). Clam and amphipod (high16:1n-7) and scallop (from both 2008 and 2009, high22:6n-3, Fig. 3A) fatty acids supported some phyto-plankton consumption, as indicated by α values (53to 71), although closer proximity to rockweed on theNMDS plot indicates greater reliance on this carbonsource by amphipod versus the clam and scallop(Fig. 3A). The remainder of benthos clustered moreclosely to rockweed due to high proportions of 18:2n-6 and 20:4n-6 (Fig. 3A), supporting consumption of

22

Fig. 2. Percent (%) reliance on phytoplankton (α) and trophicpositions of benthos (black symbols), zooplankton (white)and fish and marine mammal consumers (grey) from Cum-berland Sound (mean ± SD, see Table 1 for species codes). Values of α are the results of a 2-source δ13C mixing modelwith primary consumers of phytoplankton and macroalgaeas endpoints. Trophic positions were calculated using a one-

source, δ15N-based model


macroalgae as indicated by α values. High phyto-plankton fatty acid trophic markers in copepod (e.g.high 16:1n-7) and macroalgae markers in limpet (e.g.high 20:4n-6) support their respective consumptionof these carbon sources and their use as the phyto-plankton and macroalgae baselines, respectively, incalculations of α.

Fish and marine mammals separated from eachother on the NMDS plot of upper trophic levels due todifferent relative contributions of phytoplanktonfatty acids (Fig. 3B), likely due to reliance on phyto-plankton as indicated by α values, and species -specific fatty acid requirements. Unlike the otherfatty acid trophic markers, mean proportions of18:2n-6 varied little among fish and marine mammalspecies (1.1 to 1.8%, Table S2) and, therefore, maynot be useful as a trophic marker in upper trophiclevels (which agrees with results of Hall et al. 2006).Proportions of 20:4n-6 (Table S2), on the other hand,increased from a mean ± SD of 0.4 ± 0.1% in thepelagic capelin (α = 96 ± 3) to 1.5 ± 0.2% in the ben-thic-pelagic Greenland shark (α = 86 ± 14) to 4.4 ±1.5% in large individuals (i.e. >24 cm) of the benthicsculpin (α = 58 ± 18), which coincides with α valuesthat indicate low, medium and high reliance onmacroalgae-derived carbon in these species, respec-tively. Higher 20:4n-6 in the arctic skate (3.7%) thanthe pelagic capelin could indicate a greater relianceon macroalgae-derived carbon in arctic skate thanindicated by its α value of 100 (Fig. 2). Based on com-bined stable isotope and fatty acid data, all fish andmarine mammal species sampled acted as couplers ofphytoplankton- and rockweed-based resources tosome extent except for capelin (i.e. α = 96 ± 3, highproportions of phytoplankton fatty acids) and per-haps herring (i.e. α = 100, no fatty acid data).

DISCUSSION

Results from the present study indicate that thecoastal Cumberland Sound food web is structuredsuch that separate energy channels based on differ-ent basal resources (phytoplankton and macroalgae)were coupled by upper trophic levels, which agreeswith previously reported food web structures (Rooneyet al. 2006, McCann & Rooney 2009) and further indi-

23

Metric Functional Group Test statistic df p

Vertical food web structure δ15N Zooplankton = benthos < fish & mammals F = 255.01 2,150


cates that such structures arise even in seasonallyvariable ecosystems. The almost exclusive use ofphytoplankton by zooplankton, and macroalgae bythe benthic herbivores, limpet and periwinkle, sup-ported the existence of energy channels in Cumber-land Sound. These results are consistent with previ-ous findings that phytoplankton is the dominantenergy source to Arctic zooplankton during the sum-mer months (Søreide et al. 2006), and that benthicherbivore grazers exert strong control over, and,therefore, effectively exploit macroalgae (includingFucus spp.) in both intertidal and subtidal marinehabitats (Poore et al. 2012).

For the fishes and marine mammals, phytoplanktonwas an important energy source based on high val-ues of α (i.e. δ13C-derived % reliance on phytoplank-ton) and high proportions of phytoplankton fatty acidtrophic markers, which supports the notion that pro-duction in upper trophic levels is predominantlyphytoplankton-driven in many Arctic seas (Hobsonet al. 1995, Renaud et al. 2011). However, macro-algae also played a role in fueling the food webbecause some macroaglae-derived carbon was uti-lized by almost all fishes and marine mammals sam-pled (i.e. most species had mean α < 100). Values of αalso varied within-species based on standard devia-tions (Table 1), indicating intra- as well as inter -specific variability in the extent of resource coupling.The ability of individuals and species to differentiallyfeed on multiple prey types is advantageous for con-sumers inhabiting temporally variable ecosystemslike the Arctic, and based on recent advances in foodweb theory (McCann & Rooney 2009), such couplingof different resources is also likely to be important forArctic food web stability.

Higher δ13C in the bivalves (clam and scallop) andamphipod relative to zooplankton suggests that thephytoplankton consumed by these benthos waslikely to be in the form of detritus (sedimenting POMtypically becomes 13C enriched; Lovvorn et al. 2005).Thus, a consumer in Cumberland Sound that preyson herbivorous zooplankton and amphipods, likeArctic char (Table S1), could obtain carbon that orig-inated from 2 different phytoplankton pools: phyto-plankton (consumed by zooplankton) and phytode-tritus (consumed by amphipods), as well as fromrockweed (consumed by amphipods). Arctic benthosare known to use a wide range of resources, whichlikely explains their ability to maintain high biomasseven in areas of low autochthonous production(Feder et al. 2011). Within the framework of food webtheory, however, the ability of benthos to use multi-ple resources could also function to increase the

number of basal resource types that ultimately sup-port production in upper trophic levels, a potentiallyimportant yet unrecognized role in Arctic food webs.

Both stable isotope and fatty acid analysis sug-gested the existence of distinct energy channelswithin the benthos, based on macroalgal and phy-todetrital carbon, that were coupled by the benthicpredators whelk and nudibranch. Our results, there-fore, provide empirical evidence to support the sug-gestion that the coupling of different energy chan-nels is repeated at various scales (McCann & Rooney2009). Other mobile, benthic omnivores (e.g. crabsand shrimps; Feder et al. 2011) would also beexpected to couple distinct energy channels withinbenthic food webs.

Ice and pelagic algae can have similar fatty acids(Søreide et al. 2008) and we, therefore, cannot dis-count the possibility that ice algae, in addition tophytoplankton, also contributed to high levels of n-3and 16:1n-7 fatty acids observed in the suspensionfeeding bivalves and amphipods. However, we areconfident based on low 20:4n-6 and high 16:1n-7,22:6n-3 and n-3:n-6 that macroalgae was not themajor source of energy to these species. Other poten-tial energy sources not sampled here include alloch-thonous terrestrial carbon, which can be important toconsumers in some Arctic seas (e.g. Beaufort Sea;Dunton et al. 2006). However, the POM sample fromCumberland Sound (−22.13‰) clearly had a marineδ13C signature (terrestrial carbon −27 to −31‰, mar-ine carbon −22 to −25‰; Dunton et al. 2006), and thesmall δ13C range separating copepods and bivalvesof ~2‰ is consistent with tight benthic-pelagic cou-pling and low reliance on depleted, terrestrial carbonby Cumberland Sound consumers (similar to resultsfrom the Chukchi Sea; Dunton et al. 2006).

Both stable isotopes and fatty acids supportedreliance on rockweed by limpet and phytoplanktonby copepod. Previous studies have also concludedthat limpets obtain 100% of their carbon from macro-algae (i.e. Helcion pellucidum; Fredriksen 2003) andthat the copepod Calanus hyperboreus relies heavilyon phytoplankton (Søreide et al. 2006). Copepod δ15Nhas also been previously used to baseline TP calcula-tions (Hobson et al. 2002, Hedeholm et al. 2012). Thecalculated TP of species in the present study, basedon copepod δ15N, and calculated values of α, usingcopepod and limpet δ13C as baselines, agreed withputative diet information (Table S1). For example,capelin are known consumers of zooplankton (Scott& Scott 1988) that feed at TP = 2.8 to 3.1 in Greenlandwaters (Hedeholm et al. 2012). Calculated TP of 3.1and 96% reliance on phytoplankton-derived carbon

24


for capelin from the present study (Table 1), as wellas higher proportion of phytoplankton than macro-algae fatty acid trophic markers (e.g. 20:4n-6 =0.4%), support this previous diet information. Otherfishes, such as shorthorn sculpin, are known to con-sume both pelagic prey, like small herring, as well asbenthic-associated invertebrates like Mysis (Cardi-nale 2000), crab and other shorthorn sculpin (B.McMeans unpubl. data), which supports their calcu-lated TP and α of 3.5 and 74% for small sculpin and4.1 and 58% in large sculpin, respectively. There-fore, the assignment of species to functional groupsbased on literature data was supported by our stableisotope and fatty acid analysis. Selecting differentspecies within each functional group should not alterour conclusions because previous studies, whichinclude different Arctic zooplankton and benthicspecies than those sampled here, found that zoo-plankton rely heavily on phytoplankton productionand that benthos use a wider range of carbon sourcesthan zooplankton (Tamelander et al. 2008). Althoughnot sampled in the present study, polar cod are likelyto have acted as resource couplers because they aregeneralist consumers of both pelagic and benthicinvertebrates, with diet compositions likely reflectinglocal prey availability (Renaud et al. 2012). Addi-tional work is required to explore this idea. Further,the range of carbon sources used by benthos sampledin the present study, based on δ13C ranging from−15.96 to −19.52‰; is within previously reportedranges for benthos from other Arctic fjords, includingKongsfjorden (−16.9 to −19.6‰, Renaud et al. 2011)and from offshore ecosystems like the Chukchi Sea(−15.89 to −22.15‰; Iken et al. 2010). Based on thesefindings, we hypothesize that the coupling of differ-ent energy channels by upper trophic levels is a com-mon feature of Arctic food webs, including coastalfjords and the open ocean, albeit potentially based ondifferent basal resources than analyzed here (e.g.phytoplankton versus detritus or terrestrial carboninstead of macroalgae).

Another potential factor that could affect ourresults is that mobile, upper trophic levels such as, forexample, Greenland sharks, could have been feed-ing in areas outside of Cumberland Sound. Stableisotope ratios are known to vary over small and largescales in Arctic seas, although δ13C is less spatiallyvariable than δ15N (Hansen et al. 2012). The factthat limpet δ13C in Cumberland Sound was similarto what was found in the Norwegian sea (−16.16‰;Fredriksen 2003) and copepod δ13C from the presentstudy was in agreement with what has been foundfor conspecifics from the high Arctic (i.e. −20.4‰;

Hobson & Welch 1992), lends confidence to the sup-position that feeding in macroalgal and phytoplank-ton food chains outside of Cumberland Sound wouldbe reflected in consumer stable isotope profiles. Fur-ther, fatty acid data are generally more variableamong- than within-species, such that geographicalvariability in prey fatty acids is likely low (Thiemannet al. 2007). We, therefore, assume that the use ofmacroalgal-based prey in areas outside of Cumber-land Sound would be reflected in predator fatty acidprofiles.

The pulsed nature of phytoplankton growth, whichis a characteristic of Arctic seas (Weslawski et al.1991), would impart a temporal aspect to the foodweb structure reported here. In the pelagic energychannel, phytoplankton and secondary zooplanktonproduction is tightly coupled (Rysgaard et al. 1999),indicating a strong phytoplankton-herbivorous zoo-plankton interaction typical of a strong energy chan-nel (Rooney et al. 2006). Thus, energy (e.g. in theform of lipids) accumulated by herbivorous copepodslike Calanus hyperboreus and C. glacials is rapidlyand efficiently transferred to upper trophic levelsearly in the productive season (Fig. 4; Falk-Petersenet al. 2007). In contrast, secondary benthic pro -duction is not always coupled with phytoplanktonproduction (Link et al. 2011), indicating a weak inter-action between producer and benthic consumer.Instead, there appears to be a lag time between foodinput to the sea floor and increased benthic biomass(Link et al. 2011), which suggests that carbon routedthrough the benthic channel (as phytodetritus ormacroalgae) would not reach upper trophic levelsuntil later in the summer or fall (Fig. 4). One of themajor elements through which energy channels cou-pled in space confer stability to food webs is the top-down induced asynchrony in resource abundancebetween energy channels (Rooney et al. 2006,McCann & Rooney 2009). Here, we propose that Arc-tic food webs (and presumably other food webs inhighly seasonally environments) can be structuredsuch that consumers couple resources that are alsocompartmentalized in time (Fig. 4; McCann et al.2005), and that the mechanism generating asyn-chrony between benthic and pelagic energy chan-nels is the bottom-up effect of pulsed phytoplanktongrowth, which is a different (but not mutually exclu-sive) mechanism to the top-down mechanism pro-posed by Rooney et al. (2006).

Climate warming could negatively impact the foodweb structure reported here through both bottom-up(i.e. removal of resource heterogeneity) and/or top-down mechanisms (i.e. removal of resource coupling

25


in space or time by upper trophic levels). Resourceheterogeneity could decline, for example, throughdecreased benthic biomass. Declining benthic bio-mass is predicted to arise under certain climatechange scenarios from the effects of decreased ben-thic-pelagic coupling and increased sedimentationand habitat homogenization (Wassmann et al. 2011,Weslawski et al. 2011). Alternatively, macroalgaebiomass could increase with decreased ice cover dueto decreased ice scouring (Weslawski et al. 2011),and could become an increasingly important re -source for benthos in the face of decreased pelagic-benthic coupling. From a top-down perspective, re -ductions in sea ice and warmer water temperatureshave already resulted in the once benthic-dominatedcommunity of the Bering Sea shifting towards domi-nance by pelagic fish (Grebmeier et al. 2006). In -creasing contribution of pelagic consumers to Arcticfood webs by newly arriving species like capelin andherring, that rely entirely on pelagic carbon (i.e.through consumption of zooplankton, present study),could serve to decouple Arctic food webs. Anticipat-ing food web level shifts due to climate change is dif-ficult, and the outcome for Cumberland Sound willdepend on the relative strength of these mechanisms.

Since not all Arctic areas support macroalgae growth,including the open ocean, and are more influencedby, for example, terrestrial carbon (e.g. Beaufort Sea;Dunton et al. 2006), or ice algae during ice break up(Tamelander et al. 2008), further insight into climatechange affects on a pan-Arctic scale would be gainedfrom comparing structures of other Arctic foodwebs to the results presented here for CumberlandSound.

In summary, our results show that energy channelsbased on phytoplankton and macroaglae exist andare coupled by upper trophic levels in a coastal, sea-sonally ice-covered fjord in late summer. In a broadsense, our study affirms that heterogeneity in basalresources and feeding of upper trophic level con-sumers (within and between resource channels) arecommon structures, which exist even in food websthat experience high seasonality. We provide evi-dence that resource coupling was iterative within theCumberland Sound food web and suggest that uppertrophic level consumers were coupling resources inspace as well as time due to the pulsed nature ofphytoplankton growth in Arctic seas. Our study pro-vides testable hypotheses that food webs from otherArctic ecosystem types (e.g. open ocean) or during

26

Fig. 4. Conceptual model illustrating the transfer of energy through strong (phytoplankton) and weak (macroalgae and phy-todetritus) energy channels in a hypothetical Arctic marine food web. Estimates of annual phytoplankton productivity andsedimentation of phytoplankton reaching the benthos as phytodetritus are from Frobisher Bay, Baffin Island (Atkinson &Wacasey 1987). Thicker arrows arising from phytoplankton and phytodetritus to consumers reflect the larger amount ofenergy moving through these channels versus macroalgae, based on previous Arctic ecosystem energy flow models (e.g. inte-grated over the entire Lancaster Sound area, phytoplankton and macroalgae production was 55 and 1.7 g C m−2 yr−1,

respectively, Welch et al. 1992)


different times of the year (e.g. during ice break up)will also exhibit distinct energy channels that arecoupled by upper trophic levels. These energy chan-nels could be based on different resources than thosesampled here, including ice algae or terrestrial car-bon. Further work is also required to explore howinter- and intra-specific variability in the extent ofresource coupling in predator populations influencesfood web structure and stability. Results from ourstudy demonstrate that food web theory provides auseful framework with which to interpret the poten-tial effects of environmental change on food webstructure and stability. From a food web perspective,our results suggest that it is not changes in biomassor species composition, per se, but the removal ofheterogeneity in resource use among, and, perhapswithin species that is the biggest threat to the stabil-ity of Arctic food webs.

Acknowledgements. The authors thank K. Iken and K.McCann for constructive comments; N. Metuq, A. Metuq, K.Coghill, R. Currie, J. Brush and J. Olin for help with samplecollections; and J. Chao, A. Hussey, M. Rudy and S. Wol-faardt for help with sample analysis and 3 anonymousreviewers for constructive comments. This study was fundedby a grant from the Government of Canada Program forInternational Polar Year 2007/2008 (A.T.F and M.T.A, grantIPY C144) and Environment Canada (M.T.A).

LITERATURE CITED

Aitken AE, Gilbert R (1989) Holocene nearshore environ-ments and sea-level history in Pangnirtung fjord, BaffinIsland, NWT, Canada. Arct Alp Res 21: 34−44

Atkinson EG, & Wacasey (1987) Sedimentation in arcticCanada: particulate organic carbon flux to a shallowmarine benthic community in Frobisher Bay. Polar Biol 8: 3−7

Cardinale M (2000) Ontogenetic diet shifts of bull rout,Myoxocephalus scorpius (L.), in the south western BalticSea. J Appl Ichthyology 16: 231−239

Carmack E, Wassmann P (2006) Food webs and physical-biological coupling on pan-Arctic shelves: unifying con-cepts and comprehensive perspectives. Prog Oceanogr71: 446−477

Dolson R, McCann K, Rooney N, Ridgway M (2009) Lakemorphometry predicts the degree of habitat coupling bya mobile predator. Oikos 118: 1230−1238

Dunton KH, Schell DM (1987) Dependence of consumers onmacroalgal (Laminaria solidungula) carbon in an arctickelp community: 13 C evidence. Mar Biol 93: 615−625

Dunton KH, Weingartner T, Carmack EC (2006) Thenearshore western Beaufort Sea ecosystem: circulationand importance of terrestrial carbon in arctic coastal foodwebs. Prog Oceanogr 71: 362−378

Falk-Petersen S, Pavlov V, Timofeev S, Sargent JR (2007)Climate variability and possible effects on arctic foodchains: the role of Calanus. In: Ørbæk JB, Kallenborn R,Tombre I, Hegseth EN, Falk-Petersen S, Hoel AH (eds)

Arctic alpine ecosystems and people in a changing envi-ronment. Springer-Verlag, Heidelberg, p 147−166

Feder HM, Iken K, Blanchard AL, Jewett SC, Schonberg S(2011) Benthic food web structure in the southeasternChukchi Sea: an assessment using δ13C and δ15 N analy-ses. Polar Biol 34: 521−532

Fredriksen S (2003) Food web studies in a Norwegian kelpforest based on stable isotope (δ13C and δ15N) analysis.Mar Ecol Prog Ser 260: 71−81

Graeve M, Kattner G, Wiencke C, Karsten U (2002) Fattyacid composition of Arctic and Antarctic macroalgae: indicators for phylogenetic and trophic relationships.Mar Ecol Prog Ser 231: 67−74

Grebmeier JM, Overland JE, Moore SE, Farley EV and oth-ers (2006) A major ecosystem shift in the northern BeringSea. Science 311: 1461−1464

Hall D, Lee SY, Meziane T (2006) Fatty acids as trophic trac-ers in an experimental estuarine food chain: tracer trans-fer. J Exp Mar Biol Ecol 336: 42−53

Hedeholm R, Gronkaer P, Rysgaard S (2012) Feeding ecol-ogy of capelin (Mallotus villosus Muller) in West Green-land waters. Polar Biol 35: 1533−1543

Hobson KA, Welch HE (1992) Determination of trophic rela-tionships within a high Arctic marine food web usingδ13C and δ15N analysis. Mar Ecol Prog Ser 84: 9−18

Hobson KA, Ambrose WG Jr, Renaud PE (1995) Sources ofprimary production, benthic-pelagic coupling, andtrophic relationships within the Northeast WaterPolynya: insights from delta13C and delta15N analysis.Mar Ecol Prog Ser 128: 1−10

Hobson KA, Fisk AT, Karnovsky N, Holst M, Gagnon JM,Fortier M (2002) A stable isotope (δ13C, δ15N) model forthe North Water food web: implications for evaluatingtrophodynamics and the flow of energy and contami-nants. Deep-Sea Res II 49: 5131−5150

Holst Hansen J, Berg Hedeholm R, Sunksen K, Tang Chris-tensen J, Gronkjaer P (2012) Spatial variability of carbon(δ13C) and nitrogen (δ15N) stable isotope ratios in an Arc-tic marine food web. Mar Ecol Prog Ser 467: 47−59

Hop H, Falk-Petersen S, Svendsen H, Kwasniewski S,Pavlov V, Pavlova O, Søreide J (2006) Physical and bio-logical characteristics of the pelagic system across FramStrait to Konsfjorden. Prog Oceanogr 71:182–231

Hussey NE, Brush J, McCarthy ID, Fisk AT (2010) 15N and13C diet−tissue discrimination factors for large sharksunder semi-controlled conditions. Comp Biochem Phys-iol A 155: 445−453

Hussey N, MacNeil MA, Olin J, McMeans BC, Kinney MJ,Chapman D, Fisk AT (2012) Stable isotopes and elasmo-branchs: tissue types, methods, applications and assump-tions. J Fish Biol 80: 1449−1484

Iken K, Bluhm B, Dunton K (2010) Benthic food-web struc-ture under differing water mass properties in the south-ern Chukchi Sea. Deep-Sea Res II 57: 71−85

Lee R (1973) General ecology of the Canadian Arctic benthicmarine algae. Arctic 26: 32−43

Link H, Archambault P, Tamelander T, Renaud PE, Piepen-burg D (2011) Spring-to-summer changes and regionalvariability of benthic processes in the western CanadianArctic. Polar Biol 34: 2025−2038

Lovvorn JR, Cooper LW, Brooks ML, De Ruyck CC, BumpJK, Grebmeier JM (2005) Organic matter pathways tozooplankton and benthos under pack ice in late winterand open water in late summer in the north-centralBering Sea. Mar Ecol Prog Ser 291: 135−150

27

http://dx.doi.org/10.3354/meps291135http://dx.doi.org/10.1007/s00300-011-1046-6http://dx.doi.org/10.1016/j.dsr2.2009.08.007http://dx.doi.org/10.1111/j.1095-8649.2012.03251.xhttp://dx.doi.org/10.1016/j.cbpa.2009.09.023http://dx.doi.org/10.3354/meps09945http://dx.doi.org/10.1016/S0967-0645(02)00182-0http://dx.doi.org/10.3354/meps128001http://dx.doi.org/10.3354/meps084009http://dx.doi.org/10.1007/s00300-012-1193-4http://dx.doi.org/10.1016/j.jembe.2006.04.004http://dx.doi.org/10.1126/science.1121365http://dx.doi.org/10.3354/meps231067http://dx.doi.org/10.3354/meps260071http://dx.doi.org/10.1007/s00300-010-0906-9http://dx.doi.org/10.1016/j.pocean.2006.09.011http://dx.doi.org/10.1007/BF00392799http://dx.doi.org/10.1111/j.1600-0706.2009.17351.xhttp://dx.doi.org/10.1016/j.pocean.2006.10.004http://dx.doi.org/10.1046/j.1439-0426.2000.00231.xhttp://dx.doi.org/10.2307/1551515


MacArthur R (1955) Fluctuations of animal populationsand a measure of community stability. Ecology 36: 533−536

McCann KS, Rooney N (2009) The more food webs change,the more they stay the same. Philos Trans R Soc Lond. B364: 1789−1801

McCann KS, Rasmussen JB, Umbanhowar J, Humphries M(2005) The role of space, time, and variability in food webdynamics. In: de Ruiter P, Wolters V, Moore JC (eds)Dynamic food webs: multispecies assemblages, ecosys-tem development, and environmental change. AcademicPress, Boston, MA, p 56−70

McMeans BC, Olin JA, Benz GW (2009) Stable-isotope com-parisons between embryos and mothers of a placenta -trophic shark species. J Fish Biol 75: 2464−2474

McMeans BC, Arts MT, Rush S, Fisk AT (2012) Seasonal pat-terns in fatty acids of Calanus hyperboreus (Copepoda,Calanoida) from Cumberland Sound, Baffin Island,Nunavut. Mar Biol 159: 1095−1105

Miller RJ, Page HM (2012) Kelp as a trophic resource formarine suspension feeders: a review of isotope-basedevidence. Mar Biol 159: 1391−1402

Nilsen MNM, Pedersen TPT, Nilssen EMNEM, FredriksenSFS (2008) Trophic studies in a high-latitude fjord eco-system-a comparison of stable isotope analyses (13C and15N) and trophic-level estimates from a mass-balancemodel. Can J Fish Aquat Sci 65: 2791−2806

Polis GA, Holt RD, Menge BA, Winemiller KO (1996) Time,space, and life history: influences on food webs. Foodwebs: integration of patterns and dynamics. Chapmanand Hall, New York, NY, p 435−460

Poore AGB, Campbell AH, Coleman RA, Edgar GJ and oth-ers (2012) Global patterns in the impact of marine herbi-vores on benthic primary producers. Ecol Lett 15: 912−922

Post DM (2002) Using stable isotopes to estimate trophicposition: models, methods, and assumptions. Ecology 83: 703−718

R_Development_Core_Team (2010) R: A language and envi-ronment for statistical computing. R Foundation for Sta-tistical Computing, Vienna. www.R-project.org

Renaud PE, Tessmann M, Evenset A, Christensen GN (2011)Benthic food-web structure of an Arctic fjord (Kongsfjor-den, Svalbard). Mar Biol Res 7: 13−26

Renaud PE, Berge J, Varpe Ø, Lønne OJ, Nahrgang J, Otte-sen C, Hallanger I (2012) Is the poleward expansion byAtlantic cod and haddock threatening native polar cod,Boreogadus saida? Polar Biol 35: 401−412

Rooney N, McCann K, Gellner G, Moore JC (2006) Struc-tural asymmetry and the stability of diverse food webs.Nature 442: 265−269

Rysgaard S, Nielsen TG, Hansen BW (1999) Seasonal varia-tion in nutrients, pelagic primary production and grazingin a high-Arctic coastal marine ecosystem, Young Sound,Northeast Greenland. Mar Ecol Prog Ser 179: 13−25

Scott WB, Scott MG (1988) Atlantic fishes of Canada. Uni-versity of Toronto Press, Toronto

Søreide JE, Hop H, Carroll ML, Falk-Petersen S, HegsethEN (2006) Seasonal food web structures and sympagic-pelagic coupling in the European Arctic revealed by sta-ble isotopes and a two-source food web model. ProgOceanogr 71: 59−87

Søreide JE, Falk-Petersen S, Hegseth EN, Hop H, CarrollML, Hobson KA, Blachowiak-Samolyk K (2008) Seasonalfeeding strategies of Calanus in the high-Arctic Svalbardregion. Deep-Sea Res II 55: 2225−2244

Stevens CJ, Deibel D, Parrish CC (2004) Copepod omnivoryin the North Water Polynya (Baffin Bay) during autumn: spatial patterns in lipid composition. Deep-Sea Res I51: 1637−1658

Tamelander T, Reigstad M, Hop H, Carroll ML, Wassmann P(2008) Pelagic and sympagic contribution of organic mat-ter to zooplankton and vertical export in the Barents Seamarginal ice zone. Deep-Sea Res II 55: 2330−2339

Thiemann GW, Iverson SJ, Stirling I (2007) Variability in theblubber fatty acid composition of ringed seals (Phocahispida) across the Canadian Arctic. Mar Mamm Sci 23: 241−261

Vander Zanden MJ, Hulshof M, Ridgway MS, Rasmussen JB(1998) Application of stable isotope techniques to trophicstudies of age-0 smallmouth bass. Trans Am Fish Soc127: 729−739

Wassmann P, Duarte CM, Agustí S, Sejr MK (2011) Foot-prints of climate change in the Arctic marine ecosystem.Glob Change Biol 17: 1235−1249

Welch HE, Bergmann MA, Siferd TD, Martin KA, Curtis MF,Crawford RE, Conover RJ, Hop H (1992) Energy flowthrough the marine ecosystem of the Lancaster Soundregion, Arcric Canada. Arctic 45: 343−357

Weslawski J, Kwasniewski S, Wiktor J (1991) Winter in aSvalbard Fiord ecosystem. Arctic 44: 115−123

Węslawski JM, Kendall MA, Wlodarska-Kowalczuk M, IkenK, Kędra M, Legezynska J, Sejr MK (2011) Climatechange effects on Arctic fjord and coastal macrobenthicdiversity—observations and predictions. Mar Biodiv 41: 71−85

28

Editorial responsibility: Katherine Richardson,Copenhagen, Denmark

Submitted: August 3, 2012; Accepted: January 21, 2013Proofs received from author(s): April 26, 2013

http://dx.doi.org/10.1007/s12526-010-0073-9http://dx.doi.org/10.1111/j.1365-2486.2010.02311.xhttp://dx.doi.org/10.1577/1548-8659(1998)127%3C0729%3AAOSITT%3E2.0.CO%3B2http://dx.doi.org/10.1111/j.1748-7692.2007.00101.xhttp://dx.doi.org/10.1016/j.dsr2.2008.05.019http://dx.doi.org/10.1016/j.dsr2.2008.05.024http://dx.doi.org/10.1016/j.pocean.2006.06.001http://dx.doi.org/10.3354/meps179013http://dx.doi.org/10.1038/nature04887http://dx.doi.org/10.1007/s00300-011-1085-zhttp://dx.doi.org/10.1080/17451001003671597http://dx.doi.org/10.1890/0012-9658(2002)083[0703%3AUSITET]2.0.CO%3B2http://dx.doi.org/10.1139/F08-180http://dx.doi.org/10.1007/s00227-012-1889-6http://dx.doi.org/10.1111/j.1095-8649.2009.02402.xhttp://dx.doi.org/10.2307/1929601

cite28: cite5: cite14: cite3: cite27: cite13: cite1: cite26: cite39: cite12: cite40: cite38: cite11: cite24: cite37: cite8: cite23: cite36: cite6: cite22: cite35: cite4: cite21: cite34: cite2: cite20: cite33: cite18: cite17: cite31: cite16: cite9: cite29: cite7: cite30: cite15:

food web structure of a coastal arctic marine ecosystem ... · berland sound from atlantic waters...

Documents