Carbon flows in the benthic food web at the deep-sea

pointdepressedMécanique

22 févr. 2014 (il y a 3 années et 1 mois)

91 vue(s)

Carbon flows in
the benthic food web at the
deep
-
sea
1

observatory

HAUSGARTEN

(Fram Strait)

2


3

Dick van Oevelen
1,*
, Melanie Bergmann
2
, Karline Soetaert
1
, Eduard Bauerfeind
2
, Christiane
4

Hasemann
2
, Michael Klages
2
, Ingo Schewe
2
, Thomas Soltwedel
2
, N
ataliya
E.

Budaeva
3


5


6


7

1

Centre for Estuarine and Marine Ecology, Netherlands Institute of Ecology (NIOO
-
KNAW), PO

B
ox

8

140, 4400 AC Yerseke, The Netherlands

9

2

Alfred Wegener Institute for Polar and Marine Research, Am Handelshafen 12, D
-
27570
10

Bremerhaven, Germany

11

3

P
.P. Shirshov Institute of Oceanology, Russian Academy of Sciences, Nakhimovsky Pr., 36, 117997
12

Moscow, Russia

13


14

*

Corresponding author:
d.vanoevelen@nioo.knaw.nl

15

16

ABSTRACT

17

The
HAUSGARTEN

observatory is located in
the eastern
Fram Strait (Arctic Ocean) and used as
18

long
-
term monitoring site to follow changes in the Arctic benthic ecosystem. Linear inverse modelling
19

was
applied

to decipher carbon flows among the compartments of the benthic f
ood web at the central
20

HAUSGARTEN

station (
2
5
00
m) based on an empirical data set consisting of data on biomass,
21

prokaryote

production, total carbon deposition and community respiration. The model resolved
99

22

carbon flows among 4 abiotic and 10 biotic comp
artments, ranging from
prokaryotes

up to
mega
fauna
.
23

Total carbon input was
3.78
±
0.31

mmol C m
-
2

d
-
1
,
which is a comparatively small fraction of total
24

primary production

in the area
. The community respiration of
3.26
±
0.
20

mmol C m
-
2

d
-
1

is dominated
25

by
prokaryotes

(
93
%) and has lower contributions from surface
-
deposit feeding
macro
-

(
1.7
%) and
26

suspension feeding
m
ega
fauna

(
1.9
%),
w
hereas contribution
s

from nematode and other macro
-

and
27

megabenthic

compartments

w
ere

limited to <
1
%. The high
prokaryotic

contribution to carbon
28

processing suggests that functioning of the benthic food web at the central
HAUSGARTEN

station is
29

comparable to
those of
abyssal plain
sediments
that are
characterised by

strong energy limitation.
30

Faunal diet compositions suggest tha
t labile detritus is important for deposit
-
feeding nematodes

(24
%

31

of their diet
) and surface
-
deposit feeding macro
fauna

(~4
4
%)
,

but that semi
-
labile detritus
is
more
32

important in the diets of deposit
-
feeding macro
-

and
mega
fauna
. Dependency

indices
on thes
e food
33

sources
were also calculated as these integrate direct (i.e. direct grazing and predator


prey
34

interactions) and indirect (i.e. longer loops in the food web) pathways in the food web.
Projected sea
-
35

ice retreats for the Arctic Ocean typically antici
pate a decrease in the labile detritus flux to the already
36

food
-
limited benthic food web
.

The dependency indices
indicate that faunal compartment
s

depend
37

similarly on
labile and semi
-
labile detritus, which
suggests

that the benthic
biota may be
more
38

sensit
ive to changes in labile detritus inputs

than when assessed from diet composition alone.
39

Species
-
specific responses to different types of labile detritus

inputs
, e.g. pelagic algae versus
40

sympagic

algae, however
, are

presently unknown and are needed to assess the vulnerability of
41

individual components of the benthic food web.

42


43

Keywords:
Food web



Modelling



Sediment



Benthos



Arctic Ocean



Carbon processing

44


45

1.

Introduction

46

The Earth is warming rapidly due to anthro
pogenic inputs of CO
2

into the atmosphere

(IPCC,
47

2007)
. While research is mainly directed at the terrestrial consequences of global warming the
48

changes in the deep oceans, especially those in the vulnerable Polar regions receive less attention.
49

Climate change is expe
cted to affect Arctic marine ecosystems in various direct and indirect ways.
50

One direct effect is that seawater temperatures will rise and this will directly affect organisms
51

physiology

(Pörtner

et al.
, 2001)
. However, observed temperature changes in the deep Arctic ocean
52

are
still limited to <0.01
°
C y
-
1

(Glover

et al.
, 2010)
.

A more profound and faster impact is to be
53

expected through a
n
indirect mechanism: the retreat of the ice
-
edge and the continuous loss of multi
-
54

year ice will lead to a decreased flux of fast
-
sinking
sympagic

algae and fauna
(Hop

et al.
, 2006)
.
The
55

dominant primary producers in the upper water column may therefore shift from
sympagic

algae to
56

pelagic phytoplankton, which may be retained in the twilight zone

(Buesseler

et al.
, 2007)
.

This change
57

could shift
an

ecosystem
characterized by s
trong benthic
-
pelagic coupling to
one characterized by a
58

water column


dominated food web

(Grebmeier

et al.
, 2006; Hop

et al.
, 2006)
.

59

It may not be easy to de
tect changes in the quantity and composition of primary producers in
60

the upper water column directly because
algal
blooms and ice

cover are errati
c and difficult to sample
61

at

appropriate temporal resolution

(Bauerfeind

et al.
, 2009; Forest

et al.
, 2010)
.
The benthic
62

ecosystem, which depends directly on phytodetritus produced in the euphotic zone and which
63

integrates patterns in the overlying productivity over longer time periods, may yield a more consistent
64

signal.

65

In this context, the Alfred Wegner Institute
for Polar and Marine Research

(
Germany
)

66

established the deep
-
sea observatory
HAUSGARTEN

west of Svalbard
(Soltwedel

et al.
, 2005)

to
67

provide a long
-
term monitor of changes in the Arctic benthic ecosystem (Fig. 1).
The observatory

68

comprises nine sampling

stations along a bathymetric transect (1000


5500 m). A latitudinal transect
69

crosses the bathymetric transect at the central
HAUSGARTEN

station

(2500 m)
, which serves as an
70

experimental area for
biological
long
-
term experiments
(Gallucci

et al.
, 2008; Kanzog

et al.
, 2009)
.

71

Repeated sampling and deployments of moorings and long
-
term landers

has been conducted on an
72

annual basis since 1999 and has yielded a unique time
-
series dataset on
mega
-
,
macro
-

and

73

meio
benthic
,
prokaryotic
, biogeochemical and geological properties as well as on hydrography and
74

sedimentation patterns

(Bauerfeind

et al.
, 2009; Bergmann

et al.
, 2009; Hoste

et al.
, 2007)
.
This time
-
75

series has revealed decreases in the proportions of fresh phytodetrita
l matter at the seafloor and
in the
76

concentration of
sediment
-
bound organic matter
in the period 2001


2005

(Soltwedel

et al.
, 2005)
.

77

C
hanges in the quality and quantity

of detrital input can affect the structure of the benthic food
78

web profoundly

(Billett

et al.
, 2010; Ruhl

et al.
, 2008; Smith

et al.
, 2009)
.

Indeed, Hoste
et al.

(2007)

79

showed a decline in the microbial biomass of sediments and changes in
nematode

community
80

structur
e

at HAUSGARTEN
. These changes
,

however
,

operate in a food web context, in which biota
81

are linked through consumption and predation processes. Data sets on benthic food webs are typically
82

restricted to biomass estimates of large functional groups and occasional rate measurement
(Soetaert
83

and Van Oevelen,

2009)
, rendering knowledge based on field measurements alone insufficient to
84

derive a coherent picture of carbon flows in these systems. Recent advances in the use of so
-
called
85


inverse modelling


techniques, however, enable us not only to quantify foo
d web flows based on
86

limited data sets, but also to
assess

the uncertainty associated with this quantification
(Van Oevelen

e
t
87

al.
, 2010)
. These techniques allow us to analyse even complex deep
-
sea food webs quantitatively
88

(Van Oevelen

et al.
, 2009)
. The basic advantage is that site
-
specific field data on carbon processing
89

and carbon biomass are combined with more uncertain
data from the literature to collectively constrain
90

the magnitudes of the food web flows.

91

In this paper
,

we combine the comprehensive set of available empirical data to quantify the
92

carbon flows in the benthic food web of the central
HAUSGARTEN

station

(250
0 m)
. Detrital input to
93

the food web is divided into three classes of lability to do justice to the heterogeneity of natural detritus
94

and assess differences in diet contributions of these different detritus classes. Moreover, we determine
95

partitioning of r
espiration and secondary production to
identify

which food web compartments are
96

important pathways in the benthic food web. Trophic levels of the faunal compartments are calculated
97

and compared with trophic level position based on
δ
15
N isotope data
(Bergmann

et al.
, 2009)
, to verify
98

the resulting food web structure. Finally, dependency indices of biotic compartments on the basal
99

detritus a
nd
prokaryotic

resources

are calculated
. Dependency indices quantify the dependence of a
100

biotic compartment on other compartments
via
direct (i.e.
consumption
) and indirect (
transfer via
101

longer pathways)

interactions

(Ulanowicz, 2004)
. The model resul
ts will be used to speculate on
102

changes that can be anticipated in the benthic food web under a scenario of receding
sea
ice.

103

2.

Material and

methods

104

2.1
Data collection

105

An overview of the field data and references that were used in the food web model is give
n in
106

Table 1

and 2, with
a brief summary of the sampling methodology given here. Most samples were
107

taken during expedition ARK XIX/3c (July

August 2003)

with

the German research ice breaker
108

Polarstern

at the central HAUSGARTEN station (2500 m water depth).

109

The deposition of particulate organic carbon (POC) represents an important input parameter of
110

the inverse model, since it determines the total carbon processing by the benthic food web. Long
-
term
111

deployments of
deep
sediment trap
s

provide important constraints on the POC input.
The s
edimenting

112

particles

were

sampled

by

modified

automatic

Kiel sed
iment

traps

(see Bauerfeind

et al.
, 2009 for
113

details)
.
The

sediment

traps

were

installed

in

bottom
-
tethered

moorings

at

different

depths, but here
114

only the data from the deepest sediment trap (170 mab)

are considered
.
The

traps

were

programmed

115

to

collect

at 15

day intervals
. POC input data
from
the productive spring
-
summer season

are used in
116

the model
, since this depositional flux is most relevant for the benthic food web compartments that
117

were sampled in July 2003.
The

collector

cups

were

filled

with

sterile

water,

adjust
ed
to

a salinity of
40

118

and
poisoned

with

mercury

chloride

(0.14%

final

solution) and kept refrigerated till further processing
119

a
fter

recovery
.
Sub
-
samples

were analyzed for
,

amongst other

parameters,

particulate

organic

carbon

120

(see Bauerfeind

et al.
, 2009 for details)
. POC
deposition

showed little variation

during March


May
121

(range
0.7
7 to

1.44 mmol C m
-
2

d
-
1
),
but rose substantially to 3.99 mmol C m
-
2

d
-
1

in June (Fig. 2).
122

Because of this

range
in

deposition rates
, it was decided to include the

full range as input for the
123

model (Table 2).

124

Sediment samples were taken by a multiple corer and the
top

5 cm of the se
diment
analyzed
125

at

1
-
cm intervals
for organic carbon content, pigment concentration,
prokaryotic

biomass, hydrolytic
126

activity, meio
-

and
macrofaunal
biomass. Sediment porosity was estimated by measuring the weight
127

loss of wet sediment samples dried at 60 °
C

(average of
0.60 for the top 5 cm
)
. Total organic carbon
128

content was determined as the ash
-
free dry weight after combustion and converted to total organic
129

carbon in the sediment using sediment porosity and assuming a density of 2.5 g cm
-
3

for the sedimen
t
130

fraction.

P
articulate

proteins

were

analyz
ed
photometrically

following
0.5 N NaOH

extraction

(Greiser
131

and Faubel, 1988)
. Chloroplastic pigments were extracted and the chlorophyll
a

content was
132

determined with a fluorome
ter.
Prokaryotic

cell volume was determined with the Porton grid
133

(Grossmann and Reichardt, 1991)

after staining with acridine orange and converted to
prokaryotic

134

biomass using a conversion factor of 3.0×10
−13

g C μm
−3

(Borsheim

et al.
, 1990)
.
Prokaryotic

135

enzymatic turnover rates were measured as an indicator of the potential hydrolytic activi
ty of
136

prokaryotes

using fluorescein
-
di
-
acetate as fluorogenic substrate
(Köster

et al.
, 1991)
, hydrolysis rates
137

were converted to carbon units assuming that one mole fluorescein is equi
valent to four moles of
138

carbon (i.e. 2 acetate molecules). Sediment samples for nematode enumeration were sieved through
139

a 1 mm sieve and nematodes that were retained on a 32
μ
m were extracted
by
Ludox centrifugation

140

(Hoste

et al.
, 2007)
. Macrofaunal density estimates were
based

on

box
-
cor
e samples

(Budaeva

et al.
,
141

2008)
. Megafaunal density estimates were acquired by analysis of still images of the seafloor taken by
142

a towed camera system during RV
Polarstern
expedition ARK XVIII/1

in 2002
(Soltwedel

et al.
, 2009)
.
143

Total oxygen uptake by the benthic community was determined from the decrease i
n ox
ygen
144

concentration

in sediment cores that were incubated
in situ

during the
RV
Polarstern

cruise ARK XVI
/
2

145

(2002) and
RV
M
aria
S
.

Merian

expedition

2
-
4 (2006)

(Winkler titration

were done on
-
board
)

146

(Soltwedel, unpub
lished
).

147

2.2
Food web model

148

The food web model was
set

up

as
a
linear inverse model (LIM). The term
linear

refers to the
149

food web model being described as a
linear

function of the flows,
inverse
means that the food web
150

flows are derived from observed data. The
model

itself is the top
ology of the food web, which is
151

determined
a priori

by
delineating the

compartments and connecting them with flows.

152

Several reviews on linear i
nverse
modelling

ha
ve

been recently published and contain simple
153

models to exemplify the setup and solution of b
enthic food web LIMs

(Soetaert and Van Oevelen,
154

2009; Van Oevelen

et al.
, 2010)
.

Here, we therefore limit our methodological discussion on li
near
155

inverse models.
A LIM contains a mass balance for each food web compartment and a set of
156

quantitative data constraints. A LIM is captured by two matrix equations:

157

Equality equation:










(1)

158

Inequality equation:










(2)

159

in which vector


contains the un
known flows. E
ach row in the equality equation (1) imposes a strict
160

constraint: a linear combination of the flows must match the corresponding value in vector

.

The
161

inequality equation (2) imposes lower and upper bounds on flows or on linear combinations

of flows. A
162

default set of inequalities is the condition



,

which ensures that flows have directions that are
163

consistent with the imposed food web topology.

164

For the HAUSGARTEN station, t
he compartments
of the benthic food web
were defined as: labile
165

d
etritus

(lDet)
, semi
-
labile detritus

(sDet)
, refractory detritus

(rDet)
, dissolved organic carbon

(DOC)
,
166

prokaryotes

(
Pro
)
,
deposit
-
feeding
nematodes (NemDF)
, predatory+omnivore
nematodes (NemPO)
,
167

surface
deposit
-
feeding
macrofauna

(MacSDF)
, deposit
-
feeding
macrofauna

(MacDF)
, suspension
-
168

feeding
macrofauna

(MacSF)
, predatory+scavenging
macrofauna

(MacPS)
, deposit
-
feeding
169

megafauna

(MegDF),

suspension
-
feeding
megafauna
(MegSF)
and predatory+scavenging
megafauna

170

(MegPS)
.

171

Carbon stocks
were available
fo
r all

compartments
,
except
DOC (Table 1).

Labile detritus was defined
172

as all carbon associated with chlorophyll
a
. Chlorophyll
a

concentrations were summed in the top 5 cm
173

and were converted to carbon units by assuming a carbon to chlorophyll
a

ratio of 40

that is typical for
174

diatoms
(Allen

et al.
, 2005)
. Semi
-
labile detritus
was defined as the carbon equivalents
of
particulate
175

proteins

(converted to carbon equivalents by the conversion factor 0.49, Pusce
ddu

et al.
, 2010)

in the
176

top 5 cm
(Hoste

et al.
, 2007)

minus the labile detritus stock. Refractory detritus was defined as the
177

total OC stock in the top 5 cm of the sediment minus the labile and semi
-
labile detritus stocks.
178

Prokaryotic

carbon stocks were infe
rred from cell volumes (see above).
The biomass of nematodes
(>
179

85% of the meiobenthic

community, Hoste

et al.
, 2007)

was
partitioned among feeding modes based
180

on the
following nematode
feeding types
(Wieser, 1953)
:

deposit
-
feeding nematodes (Wieser type 1A,
181

1B an
d 2A) and predatory+omnivore nematodes (Wieser type 2B). Macrobenthic and megabenthic
182

species were divided into feeding types using specialized literature, natural abundance stable isotope
183

values
(Bergmann

et al.
, 2009)

and expert judgement.

184

Carbon
inputs
in
to the food web are deposition and/or feeding
o
n

suspended labile (lDet_w),
185

semi
-
labile (sDet_w) and refractory detritus (rDet_w).
Car
bon outputs
from the food web are
186

respiration to dissolved inorganic carbon (DIC), burial of rDet, DOC efflux to the water column and
187

export by the macro
-

and megafaunal compartments (e.g. consumption by fish).

188

Within the food web, the labile and semi
-
labi
le detritus pools in the sediment can be
189

hydrolysed
to DOC, or are grazed upon by
nematodes (NemDF and NemPS)
and MacSDF, MacDF,
190

MacPS, MegSDF, MegDF and MegPS.
Refractory detritus is only
hydrolysed
to DOC
. The

DOC is
191

taken up by
prokaryotes

or effluxes t
o the water column. Predatory feeding links are primarily defined
192

based on size class;
prokaryotes

are consumed by
the
nematode

and non
-
suspension
-
feeding macro
-

193

and
mega
faunal

compartments,
deposit
-
feeding nematodes are consumed by predatory nematodes,
194

both
nematode
compartments are consumed by non
-
suspension
-
feeding macro
-

and
mega
faunal

195

compartments, the
macro
faunal

compartments MacSDF, MacDF and MacSF are preyed upon by
196

predatory macro
-

an
d megafauna and predatory

macrofauna
is predated upon by
predatory
197

megafauna
.

198

Part of the sources
ingested
by the faunal compartments is not assimilated but instead
199

expelled as faeces
. T
he non
-
assimilated labile (e.g. labile detritus,
prokaryotic

and fauna
l
200

compartments) and semi
-
labile (semi
-
labile detritus) carbon
enter the
semi
-
labile and refractory
201

detritus, respectively. Respiration by faunal compartments is defined as the sum of maintenance
202

respiration (biomass
-
specific respiration) and growth respira
tion (overhead on new biomass
203

production).
Prokaryotic

mortality is defined as a flux to DOC and faunal mortality is defined as a flux
204

to labile detritus.

205

2.3
Data constraints

206

The range in POC fluxes, as measured with deep sediment traps, was included in the
207

inequality equation (Table 2).
There were

two measurements of sediment oxygen consumption
rates

208

and these were quite variable
,

and were therefore also

included in the inequality equation
(Table 2
)
.
209

Esterase activity reflect
s

potential hydrolysis rates rather than
in situ

hydrolysis rates
(Gumprecht

et
210

al.
, 1995)

and the measured hydrolysis rate was therefore imposed as upper bound

on total hydrolysis
211

(Table 2).

212

In addition to the site
-
specific data, a set of general constraints from the literature were
213

included in the inequality equation. These constraints were used to set bounds on

degradation rates of
214

the labile, semi
-
labile and refractory detritus pools, burial efficiency,
prokaryoti
c

growth efficiency, viral
-
215

induced
prokaryotic

lysis, release of DOC from the sediment,
grazing of prokaryotes by

nematodes,
216

assimilation efficiency of all faunal compartments, net growth efficiency of all faunal compartments,
217

production and mortality rate
s of all faunal compartments (Table 2).
The biomass
-
specific production
218

and mortality rates in combination with the biomass values of the faunal stocks constrain the total
219

carbon demand by the faunal compartments.
Since measurements of assimilation and

gro
wth
220

efficiencies of deep
-
sea benthos are very rare, an extensive literature review
(Van Oevelen

et al.
,
221

2006)

o
f temperate benthos was used as basis for these constraints. Assimilation efficiencies for semi
-
222

labile carbon were set to half the values of
the assimilation efficiencies of
labile carbon for the macro
-

223

and
mega
faunal

compartments. Faunal maintenance respi
ration was defined as 0.01 d
-
1

at 20°C
(see
224

references in Van Oevelen

et al.
, 2006)

and

is
corrected with a temperature
-
correction factor (Tlim)
225

based on the Q
10

formulation with a doubling of rates for every 10
°C

increase (Table 2).
The bottom
226

w
ater
temperature
s

at
HAUSGARTEN

w
ere ca.

-
0.8°C.

227

Both surface
-
deposit and deposit
-
feeding holothurian
s and
other
echinoderms ingest organic
228

matter with higher than ambient
chl
orophyll
a

and total

hydrolysable amino acid concentrations
229

(Ginger

et
al.
, 2001; Witbaard

et al.
, 2001)
,

al
though selectivity differs bet
ween feeding modes with
230

surface
-
depo
sit feeders typically exhibiting stronger selectivity than deposit feeders
(Wigham

et
al.
,
231

2003)
. Se
lectivity

between labile detritus and semi
-
labile detritus
for these organisms
was defined as
232

the ratio of
chl
orophyll

a

concentrations in
the
gut with respect to the ambient surface sediment. The
233

level of selectivity varies from 1
to

10 for deposit feeding holothurians at the Porcupine Abyssal Plain
234

to >500 for the surface
-
deposit
-
feeding holothurian
Amperima ro
sea

(Wigham

et al.
, 2003)
. Selectivity
235

at the Antarctic peninsula was le
ss eviden
t (selectivity of 2
to

7), possibly
because of
the

existence of
236

a food bank, but there was a clear separation between deposit and surface
-
deposit feede
rs
(Wigham

237

et al.
, 2008)
. T
herefore,
zero

to moderate
(
1
to

10
) selectivity

for

deposit feed
ers and strong selectivity
238

(50 to

100) for surface deposit feeders was assumed in the model (Table 2). Since no comparable
239

data are available for
macro
fauna
, similar selectivity ranges were defined for these communities
240

(Table 2). Finally, the predatory
nematodes and

macro
-

and
mega
faunal

compartments were assumed
241

to ingest a minimum of 75% through predatory feeding (Table 2).

242

2.4
Model solution

243

The complete food web model consists of
99

flows, 16 compartments and mass balances,
99

244

inequalities of



and
123

data inequalities. It is clear
that the

total number of flows in a food web
245

greatly
outnumbers the equations in the LIM (
99

16
).

As a result, a food web LIM is mathematically
246

under
-
determined, which implies that an infinitely large set of solutions fit
s

the matrix equations. Since
247

no un
ique solution can be found for

an

under
-
determined model, a recently developed likelihood
248

approach
was followed
(Van den Meersche

et al.
, 2009; Van Oevelen

et al.
, 2010)
. In short, a large
249

set of 50,000 solutions is sampled from the infini
tely large set of solutions. Each solution represents a
250

different food web configuration and is consistent with the matrix equations




and



.

T
he
251

mean and standard deviation for each food web flow is calculated from this set of sampled solution
s
252

and represents a central estimate (i.e. the mean) of the flow value an
d its associated uncertainty (i.e.
253

standard deviation)
(Van Oevelen

et al.
, 2010)
. This
will be noted as

mean ± standard deviation
.

254

Trophic levels of the biotic compartments and dependency indices were calculated for each solution in
255

th
e set of 50,000 solutions
using the R
-
packa
ge
NetIndices

(Kones

et al.
, 2009)
.
By running the model
256

50,000 times, the uncertainty in the empirical data (indicated by the flow ranges in Table 2) is
257

propagated onto an uncertainty estimate of the carbon flows as indicated by its standard deviation.
258

Convergence of the mean and standard d
eviation of the flows was checked visually to confirm that the
259

set of 50,000 model solutions was sufficiently large. Generally,
model
convergence
(within 10% of the
260

final mean and standard deviation for each flow value)
was achieved after
<
5
,000 solutions.

In the
261

calculation of trophic levels, t
he three detritus and dissolved organic carbon compartments were fixed
262

to a trophic level of one.

The model code
is
made available in the R
-
package
LIM

(Soetaert and Van
263

Oevelen, 2008)
.

264

2.5 Sensitivity analysis

265

The
data set included in the inverse
model
is inherently uncertain
. The
uncertainty
of
the
flux data and
266

rate parameters is included in the model by incorporating them as lower and upper bounds on their
267

values (Table 2).
In this way, this uncertainty propagates onto the final model solution as standar
d
268

deviation for each flow value (see d
escription of sampling methodology above)
.
The stock data
,

269

however
,

are also uncertain
, but this uncertainty
can
not be directly included by lower and upper
270

bounds
using this
sampling methodology. This is because
,

with a perturbation of the stock inputs, the
271

core equations




and




are no
t guaranteed to be
valid when
all
solution
s

are
averaged to

272

obtain the
final
model

solution
.
Henceforth, a sensitivity analysis was performed in which the stock
273

values w
ere perturbed one
-
by
-
one by increasing or decreasing
a

stocks value with 15% of its default
274

value. With the perturbed stock value

a new set of 500 solutions was sampled. The number of
500
275

was chosen
to save computing time
, while at the same time
it was lar
ge enough to reasonably
276

approach the

final
model solution
. The set of solutions was
subsequently averaged to obtain a
277

perturbed model solution.

These perturbed model solution
s

were compared with the default model
278

solution to assess the sensitivity of our m
odel results
f
or changes in the stock values.

279

3.

Results

280

A
complete overview

of
the mean and standard deviation for each food web
flow
is
given
in
281

the Appendix.

282

3.1
C
arbon flows

inferred by the inverse model

283

Total carbon input to
the
food web
wa
s
3.78
±
0.31

mmol C m
-
2

d
-
1

and is partitioned among
284

labile detritus deposition (
30
%), semi
-
labile detritus deposition (
3
1
%), refractory detritus deposition
285

(
32
%) and suspension feeding (
8
%). Total respiration is
3.26
±
0.
20
, burial is
0.3
2
±
0.08

and export from
286

the food

web is
0.0
2
±
0.
00
6

mmol C m
-
2

d
-
1
.
Total
r
espiration is dominated by
prokaryotes

(
93
%)

with
287

contributions

that are <
2
%

f
or each of

the faunal compartments

(Table 3)
. The
contribution
s

to total
288

respiration
by the individual compartments
are
well
-
constrained, given the small standard deviations

289

(Table
3).

290

Largest carbon flows in the
food web at the central
HAUSGARTEN

s
tation

is
the
deposition of
291

the three
classes of
detritus,
which
subsequent
ly dissolve

in
to DOC

that is taken up by
prokaryotes

292

and then respir
ed

by
prokaryotes

(Fig.
3
A).
All
carbon flows in this pathway are >
1

mmol C m
-
2

d
-
1
.
293

Prokaryotic

production is 1.84
±
0.12

mmol C m
-
2

d
-
1

and the
prokaryotic

growth efficiency is
0.3
8
±0.
0
3
.
294

M
uch

of the
prokaryotic

production
(
92
±
4
%)
undergoes

cell lysis
after viral infection
(Danovaro

et al.
,
295

2008)
,
and
this carbon
cycles back to DOC (
App
endix
). Other
important flows are carbon burial
296

(0.31
±
0.08

mmol C m
-
2

d
-
1
) and efflux of DOC from the sediment
(
0.19
±
0.
10
mmol C m
-
2

d
-
1
) (Fig. 3B).
297

Important faunal flows
(>0.
1
mmol C m
-
2

d
-
1
)
are uptake by
surface
-
deposit

feeding
macro
fauna

and
298

suspension
-
feeding
macro
-

and
mega
fauna

(Fig.
3
B).
Most carbon flows related to
faunal
299

compartments, however,
are between 0.005 and 0.05 mmol C m
-
2

d
-
1

(Fig.
3
C).

Finally,
export flows
300

and
carbon flows
associated

with

the predatory+omnivore nematodes
, pred
atory macro
fauna

and
301

mega
fauna

are typically <
3∙10
-
3

mmol C m
-
2

d
-
1

(Fig.

3
D).

302

Faunal

secondary production is highest for
macro
fauna

(0.
1
0
±0.
00
4

mmol C m
-
2

d
-
1
), followed
303

by
mega
fauna

(0.07±0.003

mmol C m
-
2

d
-
1
) and nematodes (0.0
4
±0.
00
4

mmol C m
-
2

d
-
1
)

(Fig. 4)
. The
304

fate of the
secondary production by the
non
-
predatory faunal compartments shows that
83
% of the
305

deposit
-
feeding nematode production is
grazed
, but only to a small extent (6%) by predatory
306

nematodes,

most production is predated upon by macro
-

(
40
%) and mega
fauna

(
38%) (Fig.
4
B)
.
The
307

maintenance costs are
relatively
higher

for
the macro
-

(22%)

and mega
fauna

(7
7
%)

compared to the
308

nematodes
, because

maintenance
costs are a

fixed fraction of the biomass
per day, whereas

309

biomass
-
specific production

rates decrease with faunal size (Table 2).
For macro
fauna
, a total of 56
%
310

is grazed by predatory macro
-

(18%) and megafauna (3
8
%) (Fig.
4
C)
.

Finally, for non
-
predatory
311

mega
fauna
, a similar proportion (
6
-
10
%)
of
the secondary production is grazed by predat
ory
312

mega
fauna
, lost through mortality and exported from the food web (Fig.
4
D).

313

The model
results
suggest

that f
aunal

diets are
typically
dominated by labile and semi
-
labile
314

detritus, with variable contributions among
the
compartments (Fig.
5
). Despite

the fact

that deposit
-
315

feeding nematodes form the principle carbon
source of predatory nematodes (>
80
%, Fig.
5
), this
316

represents
only 6
% of the

fate of

secondary production by deposit
-
feeding nematodes (Fig.
4
A). The
317

surface
-
deposit feeding
macrofauna

and
deposit
-
feeding macro
-

and
megafauna

derive carbon
mainly
318

from
three

principle

sources: labile detritus, semi
-
labile detritus and
prokaryotes
. Semi
-
labile detritus
319

dominates the diets of the deposit
-
feeding compartments (
52
±
9
%
for MacDF and
52
±
1
2
% for MegD
F),

320

whereas

prokaryotes

(
50
±
2
8
%
) are
of similar importance

for surface
-
deposit feeding
macrofauna

as
321

labile detritus (
44
±
28
%
) with a lower contribution from

semi
-
labile detritus (
4
±1%).

Diets of s
uspension

322

feeding
macro
-

and mega
fauna
diets are dominated by semi
-
labile detritus with a lower contribution of
323

labile detritus
(
67
±
23
%

and 33±
2
3%

for MacSF
,
respectively

and
59
±1
7
%

and 41±17%

for MegSF
,

324

respectively
)
(
Fig.
5
).
T
he diets
of
predatory macro
-

and megafauna

are diverse and

seem to be
325

similar among the two predatory compartments. Important contributions (>50%) are from the
326

macrofaunal compartments, most notably surface
-
deposit feeding mac
ro
fauna

(>24%), nematodes
327

(>11%), wi
t
h

labile (<5%) and semi
-
labile (<1
3
%) detritus repr
esenting a much lower contribution.

328

3.2
Trophic levels and dependencies on primary resources

329

The trophic
level
(
TL
)
of suspension f
eeding macro
-

and megafauna is fixed at

two (Fig.
6
),
330

because the two suspended detritus sources are presumed to have a fixed

trophic position of one (see
331

Material and Methods). The
T
L

of deposit feeding nematodes is fairly well
-
determined and is slightly
332

higher than 2 because of a small contribution of
prokaryotes

in their diet. The

TL

of deposit
-
feeding
333

macro
-

and megafauna is

similar and fairly well
-
d
etermined with lower and upper
quartiles
of 2.2 and
334

2.4 (Fig.
6
),
corresponding with
the similarity in their diet compositions (Fig.
5
). The
T
L

of surface
-
335

deposit feeding macrofauna
,
however
, is

much more uncertain and has lower and upper quartiles of
336

2.
3

and 2.
8

with a
median of 2.5 (Fig.
6
). This uncertainty is due to the
uncertain
diet contributions
337

described above of labile detritus (T
L

of 1) and
prokaryotes

(T
L

of 2). The
TL

of predatory+omn
ivore
338

nematodes is

well constrained between 2.8 and
2.9
,

because of
their
predominant feeding on deposit
-
339

feeding nematodes. The predatory macro
-

and megafauna have
similar and
highest
T
L

with lower and
340

upper quartiles between 2.
8



3.0
, respectively, but w
ith large excursions to lower and higher
values

341

for their trophic level

(Fig.
6
). The high uncertainty is a result of the uncertainty in the diet composition
342

and the diets of its preys, but overall the higher

TL
s

are expected for these predatory compartmen
ts.

343


The
direct and indirect
dependence on refractory detritus is lowest for all biotic compartments
344

in the food web (Fig.
7
), with lower and upper quartiles between 0.1
5

and 0.
77
. Dependence of
345

prokaryotes

is highest on sem
i
-
labile detritus (median of 1.5
) and
prokaryotes

(median 2.
5
) (Fig.
7
A).
346

Dependence on labile and semi
-
labile detritus is comparable for most biotic compartments with lower
347

quartiles between
0.9
8

and

1.
15

and upper quartiles between
2
.
1
8 and
2.
2
8

(Fig.
7
B
-
H). The level of
348

uncertainty is

high

for the

dependency values, particularly with respect to the upper levels of
349

dependency that can be >
8
, which is substantially higher than the median values (Fig.
7
).

350

Overall it is clear that the standard deviations are fairly limited for respiration
rates (Table 3)
351

and secondary production (see above) by the various compartments, which are measures of total
352

carbon processing. There is
,

however
,

a much higher variability in trophic levels (Fig.
6
) and
353

dependencies (Fig.
7
), which indicates that the unc
ertainty on the flows between the compartments is
354

substantially higher than the uncertainty on the total carbon processing by the compartments.

355

3.3 Sensitivity analysis

356


The
perturbations of the stock values with
±
15% of their default value in the
sensitivity analysis

357

gave following results: f
or
46
% of the flow values, the deviation in the perturbed solution was between
358

0 and 10% of the default flow value, for
3
7% this deviation was between 10 and 25% of the default
359

flow values, for 13% the deviatio
n was between 25 and 50%, for
4
% it was between 50 and 100% and
360

for <1% of the flow values a deviation of more than 100% of its default flow value was fou
nd. The
361

maximum deviation for a flow was 169%, which involved the export flow of surface
-
deposit feedi
ng
362

macro
fauna

and occurred under a reduced stock value of the predatory mega
fauna
. The reduced
363

stock of predatory mega
fauna

involved a reduction in its predation pressure on surface
-
deposit
364

feeding macro
fauna

due to which the export flow increased.

Overall, however, the sensitivity analysis
365

revealed that in the model results were generally insensitive to perturbations in the stock values.

366

4.

Discussion

367

Ecosystem dynamics in the Arctic Ocean are regulated by the strong seasonality in
the
light
368

and tempe
rature regime and the cover of sea
-
ice
(Grebmeier and Barry, 1991; Honjo

et al.
, 2010;
369

Wassmann

et al.
, 2006)
. The Arctic Ocean is surrounded by landmasses
with

extensive shallow
370

continental shelves

with strong benthic
-
pelagic
coupling
(Grebmeier and Barry, 1991)
, particularly in
371

the ice
-
edge and ice
-
fre
e regions such as the Bering Sea

(Grebmeier

et al.
, 2006)
, Barents Sea
(De
372

Laender

et al.
, 2010; Wassmann

et al.
, 2006)

and Chukchi Sea
(Moran

et al.
, 2005)
.
Th
is benthic
-
373

pelagic coupling on the shallow shelves

results in a comparativ
ely high fraction (>15%) of the primary
374

production being processed by the benthos
,

sustaining high levels of macrofaunal biomass
375

(Grebmeier

et al.
, 1988; Renaud

et al.
, 2007)
.
C
ompared to the shallow continental shelves, the deep
376

sediments of the Arctic Ocean are much less studied

(e.g. Bergmann

et al.
, 2009; Clough

et al.
, 1997;
377

Iken

et al.
, 2005; Kröncke, 1994; Vanreusel

et al.
, 2000; Wlodarska
-
Kowalczuk and Pearson, 2004)
,
378

especially at the
integration
level of the
whole food web.

379

Fram Strait is located between Spitsbergen and
Greenland and forms a deep (>2000 m) and
380

narrow connection between the Arctic Ocean and the Atlantic Ocean (Fig. 1). In this region, the
381

amount of organic matter processed in the sediment decreases
as

allometric functions of water depth
382

and primary product
ion

(Schluter

et al.
, 2000)
.
T
o detect and track the impact of large
-
scale
383

environmental changes in the transition zone between the northern North Atlantic and the central
384

Arctic Ocean
,

the long
-
term observatory
HAUSGARTEN

was established
(Budaeva

et al.
, 2008; Hoste

385

et al.
, 2007; Soltwedel

et al.
, 2009)
. Export fluxes from the euphotic zone at the
HAUSGARTEN

386

observatory are restricted to <10% of the primary production suggesting an efficient processing by the
387

pelagic food web
(Bauerfeind

et al.
, 2009)
.
Alt
hough information on biomass of various compartments
388

(Budaeva

et al.
, 2008; Hoste

et al.
, 2007; Soltwedel

et al.
, 2009)

and organic carbon deposition
389

(Bauerfeind

et al.
, 2009)

is well
-
documented, no inferences
have been

made on how the organic
390

carbon that arrives at the seafloor is processed within the benthic community. To
add
ress this
391

deficiency
, the available data were merged to a food web model using linear inverse methodology

392

(Soetaert and Van Oevelen, 2009)
. Since the resulting food web structure depends
heavily
on the
393

model assumptions and data quality, it is essential to start

with a critical
appraisal of

these.

394

4.1
Model assumptions

395

The inherent heterogeneity of sedimentary detritus
implies that various
detritus
fractions have
396

degradation rates
differing over

orders of magnitude
(Middelburg, 1989; Moore

et al.
, 2004; Westrich
397

and Berner, 1984)
. On
the
one hand, it is impossible to do justice to this continuum of degradation
398

rates

within a food web context, first
ly because

too many detritus compartment
s

would
need to be
399

defined and secondly
because
no data are available to constrain their dynamics. On the other hand,
400

mass
-
balance models of sediment food webs typically merge all dea
d organic matter into one
401

homogeneous “detritus” compartment
(e.g. Rowe

et al.
, 2008)
, which may

be a too crude
402

simplification
.
Here, three detritus compartments
were
defined
based on

empirical data

using a similar
403

approach as Van Oevelen
et al.

(2011)
.

Labile detritus was defined as all carbon
associated with
404

chlorophyll
a
. Chlorophyll
a

d
eposition is typically linked to the input of fresh phytodetritus
, both in
405

coastal
(Sun

et al.
, 1991)
, canyon
(Van Oevele
n

et al.
, 2011)

and abyssal plain
(Stephens

et al.
, 1997;
406

Witbaard

et al.
, 2000)

sediments.

Henceforth
,

the particulate organic carbon that is associated with
407

chlorophyll
a
, using a carbon : chlorophyll
a

ratio for living phytoplankton
(following Stephens

et al.
,
408

1997)
,

represents a natural choice to
define the
labile detritus compartment
.

In addition to a labile
409

detritus fraction, it is readily
established
that there is a refractory detritus pool in marine sediments

that
410

is
only

degradable by prokaryotes
(Benner

et al.
, 1986; Deming and Baross, 1993; Pfannkuche,
411

2005)
. Therefore
, total particulate organic carbon in the sediment minus the labile and semi
-
labile
412

detritus was defined refractory and
hence
not degradable by benthic fauna.

The

most ambiguous
413

detritus pool to define was s
emi
-
labile detritus
, which was here defined as the sum of extractable
414

p
articulate

proteins

in the top 5 cm. Amino acids are frequently used as indicator for the lability of
415

detritus
(Dauwe

et al.
, 1999; Kiriakoulakis

et al.
, 2001; Mayer

et al.
, 1995)

and termed are semi
-
labile
416

(Fabiano

et al.
, 2001)
, because they do not degrade in short time scales and proteins have
417

intermediate degradation rates in experimental decays studies
(Harvey

et al.
, 1995)
. Since th
e
418

extraction and hydrolyzation methods used for the characterization of proteins are harsher (e.g. low
419

pH, high temperatur
e
) than those
present

in digestive tract
s

(Mayer

et al.
, 1995)
, it is likely that this
420

definition represents an upper limit on

the semi
-
labile detritus stock. Va
n

Oevelen
et al.

(2011)

421

followed a similar approach,
al
though
they
defined
semi
-
labile detritus
as
the sum of lipi
ds,
422

carbohydrates and proteins
. However,
proteins comprise ~50% of this

total pool
.
To account for the
423

uncertainty in classifying the detritus classes, the lower and upper bounds on the degradation rates of
424

semi
-
labile detritus encompasses two orders of magnitude (Table 2). Also, o
ur s
ensitivity analysis

425

showed that the model

results presented are
in
sensitive to the stock values ± 15% (see Results).
426

Admittedly,
our separation into various detritus classes
is an operational definition
. However,
it
does
427

better justice to the

natural

detritus heterogeneity than considering simply

one

pool

and
it
is linked to
428

measurable quantities. Moreover, there is roughly an order of
magnitude
increase in
the different
429

detritus stocks

with decreasing lability

(Table 2), suggesting a reasonable coverage using these
430

metrics.

431

Some biotic compartmen
ts are missing from the food web topology (Fig.
3
). Micro
fauna

or
432

nanobenthos
(i.e. flagellates and ciliates) are not included
because of

a lack of biomass data
. T
his is a
433

problem in most deep
-
sea
studie
s such that

the role of
the nanofauna
in carbon cycling of deep
-
sea
434

food webs
remains

an open question
. However,

their limited biomass compared to
,

for example
435

Foraminifera
(Alongi, 1992)
,

may suggest a limited role in carbon processing. Meiofa
una were
436

represented by two nematode compartments (Fig.
3
). Hoste
et al.
(2007)

showed that nematodes
437

strongly dominated the metazoan
meiofauna

(85
-
99%) at
HAUSGARTEN
, such that the omission of
438

other metazoan
meiofauna

is probably not significant
, at least when it comes down to carbon
439

processing rates. Foraminifera, i.e. protozoan
meiofauna
,
had to be

omitted from
the

Hausgarten
food
440

web

model

because no biomass data were available
. Therefore
their role in
carbon processing
could
441

not be assessed

and this represents a shortcoming of the present model. Foraminifera
can have
442

comparable or higher biomass levels compared to the metazoan meiofauna in deep sediments
443

(Bernhard

et al.
, 2008; Gooday, 1986; Witte

et al.
, 2003)

and
have been shown to be important
444

contributors to short
-
term processing of fresh phytodetritus in deep sediments

(Moodley

et al.
, 2002
)
.
445

A

spe
cific study on the Foraminifera showed that their

contribution to total respiration
was
limited to
446

0.5


2.5% only
(Geslin

et al.
, 2010)

and the uptake of
13
C
-
phytodetritus by the larger (
>300

μ
m
)
447

Foraminifera is generally
limited
(Woulds

et al.
, 2007)
.
Moreover, if biomass of

Foraminifera
were
of
448

comparable magnitude as
the nematodes at
the
Hausgarten
, then their
role
will be
limited
considering
449

the limited role of nematodes in carbon processing
(
Fig.

3 and 4

and Table 3
).
The absence of
450

nanobenthos

and Foraminifera as specific compartments in essence implies that the
ir

role in carbon
451

processing is included in the prokaryotes
. This latter compartment

acts as a closure term
on
452

respiration,
because specific information of the prokaryotic respiration or

production
wa
s unavailable.
453

Overall
,

however, very detailed benthic biomass data were available ranging from
prokaryotes

to
454

megafauna
, that could be split among feeding types using taxonomic information and stable isotope
455

studies
(Bergmann

et al.
, 2009)
. Carbon flows could therefore be inferred at a high resolution,
456

especially
considering the fact that we are dealing with
a deep
-
sea food
web.

457

The

resulting standard deviations on the carbon flows are limited for
prokaryotic

and faunal
458

respiration rates and secondary production (see above), which are all measures of total carbon
459

processing. These carbon flows are predominantly constrained b
y biomass data and literature
460

constraints (Table 2). There is
,

however
,

higher variability in diet compositions (see ‘Results’), trophic
461

levels (Fig.
6
) and food dependencies (Fig.
7
), indicating that the uncertainty on the flows between the
462

compartments i
s substantially higher than the uncertainty on the total carbon processing by the
463

compartments (see also Appendix). This higher uncertainty on flows between compartments is
464

undoubtedly the result of limited data that are available to constrain these flows
,

a situation that is
465

typical for benthic food web reconstructions.

This may be improved by more detailed information on
466

diet composition using for example fatty acid composition data
(Iverson

et al.
, 2004; Meziane

et al.
,
467

1997)

or
trophic
level indicatios

using
δ
15
N values of specific amino acids

(Chikaraishi

et al.
, 2009)
.

468

4.2
Carbon budget

469

Respiration by the total community was estimated at
3.26
±
0.
20

mmol C m
-
2

d
-
1
.
This respiration
470

rate is
3 to 10 times
lower than the range of 10.3


35.6 mmol C m
-
2

d
-
1

that can be inferred from an
471

empirical relation based on
in
-
situ

sediment oxygen consumption rates from open slope sediments

472

(Andersson

et al.
, 2004)
,
al
though

substantially

higher than the 0.12


0
.25 mmol C m
-
2

d
-
1

measured
473

at similar depth at an Arctic continental slope of the Laptev Sea
(Boetius and Damm, 1998)
.
474

Bauerfeind
et al.

(2009)

report low export
fr
om the pelagic food web
(<10% of primary production) at
475

the central
HAUSGARTEN

station and explain this by effective recyclin
g within the pelagic community.
476

However,
short
-
term sedimentation

events
like ice
-
edge blooms or detached
sympagic

algae

may
477

cause
a
temporary
decoupling from the pelagic food web. The food web model resolves carbon fluxes
478

in spring/summer 2003, during which carbon export was comparatively large compared to other years
479

and was composed mostly of diatomaceous material
(Bauerfeind

et al.
, 2009)
.

This

decoupling in the
480

pelagic food web renders total respiration rates of the sediment comparatively low compared to other
481

open slopes.

482

Respiration was clearly
dominated by
prokaryotes

(
93
±0.6
%) with
c
ontributions
of less than
2
% by
483

different
faunal
compartments (
Table 3). Piepenburg
et al.

(1995)

conducted an extensive study in the
484

north
-
eastern

Barents Sea
,

in which sediment community oxygen consumption (SCOC) rates,
485

including micro
-
, meio and
macrofaunal

r
espiration, were amended with respiration rates of
486

megafauna

and fish. These authors also report a microbial dominance (57%) for slope sediments (200
487



550 m), with more limited contributions of meio
-

(7%), macro
-

(21%) and
megafauna

(16%).
488

Ambrose
et al.

(2001)

found contributions of up to 25% at shall
ow (<50 m) stations
,

with

respiration
489

rates
of >1 mmol C m
-
2

d
-
1

for epibenthic echinoderms. These faunal contributions and rates are
490

consistently higher than estimated
here
for the deeper
HAUSGARTEN

station (2500 m). This pattern
491

is also seen in other regions were faunal contributions can be as high as 50% at the shelf and shelf

492

break
sediments

c
ompared to continental slope and abyssal plain
environment
s
(Heip

et al.
, 2001;
493

Woulds

et al.
, 2009)
. The shift towards an increased microbial contribution to carbon processing
494

possibly relates to energy lim
itation at
great
er
depths, such that population densities of large
495

organisms simply become too low to remain reproductively viable
(Rex

et al.
, 2006)
. In all, the
496

respiration partitioning at the central
HAUSGARTEN

station is more comparable to abyssal plain food
497

webs that are under strong energy limitation, c
ompared to shallower sediments where the faunal
498

contribution is typically larger.

499

4.3
Faunal carbon flows and position in food web

500


The nematode
community
contributed
surprisingly
little to total respiration (<
1
%, Table 3)
,
501

especially when compared to a global estimate of around 7.5% for nematodes
(Soetaer
t

et al.
, 2009)
.
502

This limited contribution
of

nematodes is also found in two isotope pulse
-
chase experiments
503

conducted in Arctic sediments. Ingels
et al.

(2010)

infer
red

that <1% of the added organic matter
504

sources (
bacteria

and diatoms) at the central
HAUSGARTEN

station were processed by nematodes.
505

Urban
-
Malinga and Moens
(2006)

conducted
13
C
-
phytodetritus tracer experiments in two Arctic beach
506

sediments and
reported

meiofaunal processing
of

<5%. The diet composition of deposit
-
feeding
507

nematodes indicates that

a substantial 24% of their
carbon
requirements
is
derived

from
labile detritus
508

(Fig.
5
). This seems to
contradict

the findings of Ingels
et al.

(2010)
, who report a higher uptake of
13
C
-
509

labelled
bacteria

compared to
13
C
-
labelled diatoms. However,
the
total uptake of labelled
bacteria

was
510

limited to <6

10
-
5

mmol C m
-
2

d
-
1
, which is less than 0.1% of the carbon assimilation of 0.
06
±0.
00
7

511

mmol C m
-
2

d
-
1

inferred here. This indicates that although upta
ke
rates
of labelled
bacteria
w
ere

higher
512

than labelled diatoms, these labelled carbon sources were insignificant in the total carbon
513

requirements of the nematodes.
Whether this is related to the reduced food availability because of
514

reduced mixing of the carbon sources into the sediment or methodological bias because of freeze
-
515

drying of the carbon sources, as Ingels
et al.

(2010)

note, is presently unclear. The limited uptake of
516

ba
cterial

carbon
, however, is

in agreement with experimental results from Guilini
et al.

(2010)
, who
517

injected different
13
C enriched dissolved organic compounds into the top 5 cm of the sediment of a
518

shallower H
AUSGARTEN

station and traced the
13
C into
prokaryotes

and subsequently nematodes
519

(hence
forth the constraints in Table 2). We lack sufficient evidence to more precisely quantify labile
520

and semi
-
labile diet contributions for the nematode community (Fig.
5

and Fig.
7
B
-
C), but
modelling
521

results indicate a dominance of semi
-
labile detritus. Hoste

et al.

(2007)

found inter
-
annual c
hanges in
522

nematode biomass in the years 2001


2004 at the central
HAUSGARTEN

site, but these could not be
523

related directly to export fluxes from the water column. The authors indicated that the absence of this
524

relation may have been related to uncertainti
es in exact timing of export flux and the amount actually
525

arriving at the seafloor. Present results
,

however
,

indicate that this may also be due to a dependence
526

on a semi
-
labile pool that are much less dynamic given the degradation rate of
0.0018

d
-
1

527

corre
sponding to a half
-
life of
385

days.

Ruhl
et al.
(2008)

report response times
of
the benthic
528

community to a POC deposition event of 4


6 months for abundance and up to 10 months for
529

biomass.
It is not clear
,

however, to what extent the benthos were directly acquiring their carbon from
530

the freshly deposited c
arbon compared to the more stable semi
-
labile detritus pools.
Their figures also
531

show
that
benthic biomass
does not decrease monotonically with
POC

fluxes approach
ing

zero
,
532

indicating at least partial reliance on a
less fraction detritus pool
.

533


Only a
limited fraction of secondary production (~
6
%) of the deposit
-
feeding nematodes is
534

transferred to predatory nematodes, despite
the fact
that deposit
-
feeding nematodes represent
80
% of
535

the diet of predatory nematodes. This indicates that predation may not c
ontrol their biomass at this
536

stat
ion
. Gallucci
et al.
(2008)

used cage experiments at
HAUSGARTEN

to investigate the impact of
537

presence/absence of
megafauna

on the co
mmunity structure of nematodes. Total nematode densities
538

were higher within the cages, probably related to higher food abundance,
al
though the percentage of
539

predatory nematodes was low (~3%) and similar within and outside the cages.
S
imilar
ly

low
540

abundance
s

of predatory nematodes were reported
in sediments of
the Laptev Sea

(Vanaverbeke

et
541

al.
, 1997)
, suggesting that the ‘incomplete’
utili
s
ation
of secondary production within the nematode
542

sub
-
food web may be a more general phenomenon in Arctic sediments.

543

The non
-
predatory
macrofaunal

compartments account for half of the faunal secondary
544

production, which is efficiently transferred up the food web to the

macro
-

(18%) and megabenthic
545

(3
8
%) predators
.
The transfers also result in high contributions of non
-
predatory macro
fauna

in diets of

546

predatory macro
-

and mega
fauna
.

Only few studies are available
for comparison
. Rowe
et al.
(2
008)

547

determined predation rates by
megafauna

in
the

Gulf of Mexico, but
a priori

assumed that organisms
548

preferentially feed on larger prey items. Predation rates ranged from 2

10
-
4



0.03 mmol C m
-
2

d
-
1
,
549

depending on station depth. Total predation by the
megafauna

at the
HAUSGARTEN

station amounts
550

to
0.
05
5
±0.
00
7

mmol C m
-
2

d
-
1

and is comparable to the upper slope stations (500


1000 m)
of

the
551

Gulf of Mexico. Total
megafaunal

biomass at
HAUSGARTEN

(6 mmo
l C m
-
2
) is somewhat higher

552

compared

to these shallower stations (2.2


2.6 mmol C m
-
2
) despite the lower organic matter input at
553

HAUSGARTEN

(
3.78

vs. 4.2


9.5 mmol C m
-
2

d
-
1
). This
implie
s
that
the
Arctic HAUSGARTEN
554

megafauna

appear to

take more advan
tag
e

of the organic matter flux than
megafaunal assemblages
555

from the
tropical Gulf of Mexico.

556

The diet contribution of surface
-
deposit feeding
macro
fauna

seems to be dominated by
557

prokaryotes

(
50
±
2
8
%
), although there is substantial uncertainty in these diet reconstructions
as
558

evidenced by the large standard deviation
(see
also
‘Results’).

There is no experimental evidence on
559

the importance of
prokaryotes

in the diets of deposit
-
feeding macro
-

and mega
fauna
,
but

the model
560

results suggest 15

to 20
%. Constraints on deep
-
sea feeding strategies based on optimal foraging
561

theory imply that deposit feeders may use a strategy in which oxygen and ammonium is supplied to
562

prokaryotes

to facilitate (pre
-
)degradatio
n of detritus and the
prokaryotes

are subsequently grazed
563

(Jumars

et al.
, 1990)
.
Although this

theory has not been rigorously t
ested for benthic f
ood webs, our

564

results here show that
prokaryotes

may indeed contribute to carbon demands, but further experimental
565

work is required to substantiate these findings.

566

The trophic levels (TL) that are inferred from the model are not directly comparable with
th
ose

567

estimated at species
-
level

with
δ
15
N
measurements

(e.g. Bergmann

et al.
, 2009; Iken

et al.
, 2005; Iken

568

et al.
, 2001)
.
Here,
the trophic level is calculated for benthic
compartments

from
a large set of model
569

solutions that are feasible within the current data set. As such, these results can be interpreted as the
570

range of TLs that ar
e feasible within the different biotic compartments, wi
th two restrictions.

(
1)
T
he
571

ranges in the model results are based on
a
compartment
in which species are lumped
into functional
572

groups
and will therefore be more limited than those based on species
-
spe
cific
δ
15
N
,

in which more
573

extreme values can be found
.

(
2)
T
he
δ
15
N of detritus may increase during progressive degradation
574

resulting in a
δ
15
N difference between labile and semi
-
labile detritus

(Altabet, 1996)
. T
his
fractionation
575

effect is not included in the TL calculation, because
it

does not influences an organism’s TL.
Moreover,
576

the estimates of TL b
ased on
δ
15
N

may not be accurate if there is a large difference in
δ
15
N

of the
577

primary food sources.

578

Overall, the model results are consistent with the
δ
15
N results that Bergmann
et al.
(2009)

579

obtained at the
HAUSGARTEN

stations, in that deposit feeders mostly occupy
the second

and most
580

predators the third TL.

In addition, the model results show that the largest range of trophic levels is
581

found for the predatory compartments (
TL range
of
2.5


3.3, respectively
), which is qualitatively
582

consistent with the results from Bergmann
et al.
(2009)

but

the
ir

δ
15
N data indicate a larger range in
583

TLs of
up to
3.5.
The largest discrepancy is found for suspension feeders that are at TL of 2

in the
584

model, because in the model setup they are assumed to feed exclusively on suspended detritus (with
585

fixed TL of 1). Bergmann
et al.
(2009)
,

however, find a much larger variation in the suspension feeders
586

(range of 3 TLs), which they attribute to starvation effects, feeding selectivity and the uncertain
587

classification of sponges and anthozoans (that are potentially
ca
rnivorous or rely on microbial
588

farming
).

589

The spread in TLs also agrees with a study of
δ
15
N signatures of benthic fauna in the Canada
590

Basin
(Iken

et al.
, 2005)
, where benthic fauna are found mainly at a TL of 2


3. Moreover, they infer
591

that deposit feeders rely
to a large

extent
on less labile detritus because of the large difference

in δ
15
N
592

between benthic deposit feede
rs and fresh detritus (i.e. sympagic
algae and pelagic algae).
Our

model
593

results also indicate that semi
-
lab
ile detritus is an important (>5
0%) component o
f the diets of deposit
-
594

feeding macro
-

and
mega
fauna
.

595

Trophic
dependenc
y

quantif
ies

the dependence of a consumer on a resource through direct (i.e.
596

grazing) and indirect (i.e. longer loops in the food webs) pathways in the food web, thereby giving a
597

more complete view on trophic interactions than when looking at direct interactions only
(Ulanowicz,
598

2004)
. Overall, dependence
o
n

refractory detritus is low for all biotic compartments. The dependencies
599

that are inferred for the
other
basal resources show an interesting feature: although labile detritus is
600

general
ly

substantially less

impo
rtan
t

in the diet compositions
than semi
-
labile detritus
,

the
601

dependence on labile detritus is comparable or
even
slightly higher than dependence on semi
-
labile
602

detritus. This is due to the effect of combining all pathways, and not only the direct
interactions, into
603

the dependency indices.

As such, this may
indicate

that the benthic food web is more sensitive to
604

changes in labile detritus input as may be inferred from their diet compositions alone.

605

4.4
Speculations on
future conditions

606

Based on the
results of this food web reconstruction, we conclude that carbon
minerali
s
ation
at
607

the central
HAUSGARTEN

station (2500 m) is

strongly
dominated by
prokaryotes

with l
imited
608

contributions
of
the faunal compartments. The limited
vertical
export
of particulat
e organic matter

609

imposes energy limitation on the benthic food web such that
carbon processing resembles that of
an
610

abyssal plain food web. The Arctic Ocean is a region where major shifts in the ecosystem
are
611

expected
due to
cl
imate
c
hange
. Grebmeier
et al
.
(2006)

describe a shift in the Bering S
ea, where
612

high production of sympagic
algae favoured a high export to the benthos and resulted in a high
613

benthic production. This situation changed with a receding ice
-
edge to a
pelagic
dominated food web
614

with limited export fluxes and decreasing benthic production
.

615

How the benthic food web at

HAUSGARTEN will change under projected climate change will
616

mainly

depend on the changes in the pelagic food web structure and export of organic matter.
It could
617

be argued that ice
-
free conditions promote phytoplankton growth, as projected for some Arctic
regions
618

due to
temperature and light penetration increase
s

as a result of

shrinking sea ice

(Arrigo

et al.
, 2008;
619

Slagstad

et al.
, 2010)
. However, primary production may rise only slightly if increased thermal or
620

hal
ine stratification limit
s

mixing and upward nutrient transport
(Carmack

et al.
, 2006; Slagstad

et al.
,
621

2010)
. In addition, mesozooplankton abundance may increase in the Fram Strait
,

since Atlantic
622

species extend their range as more Atlantic water m
asses prevail and sea surface temperatures rise
623

(Hirche and Kosobokova, 2007)
. This would amplify the grazing pressure and lead t
o increased
624

retenti
on in the water column

(Carroll and Carroll, 2003)
.
T
he retreat of the ice edge and the
625

continuous loss of multi
-
year ice will lead to a lower flux of fast
-
sin
king
sympagic

algae and ice
-
related
626

POM

(Forest

et al.
, 2010; Hop

et al.
, 2006)
, which may affect megafaunal deposit feeders such as
627

holothuri
an
s

(Bergmann

et al.
, 2011)
. In sum, all this would lead to a decreas
ed carbon deposition at
628

the deep seafloor, which is already characterised by food limitation.

629

Based on the results from
our

model, we do not expect that shifts in the
overall functioning of the
630

benthic food web will occur rapidly, because semi
-
labile detri
tus
plays an significant role in the benthic
631

food web
. The semi
-
labile detritus i
s a stock with a comparatively
low degradation rate
(
0.0018
±
0.0007

632

d
-
1

and corresponding half
-
life of
382

days)

such that changes will be slower to observe than expected
633

from sole differences in detritus deposition rates.

The dependency indices
of the benthic fauna on
634

labile and semi
-
labile detritus were
, however,
of comparable magnitude such that
that the benthic
food
635

web
may be

more sensitive to changes in labile detritus input as may be inferred from their diet
636

compositions alone.

Unfortunately, t
here were not enough data to
allow better discrimination between
637

pelagic and
sympagic

plankton inputs
; if some species

specifically select
sympagic phyto
detritus, as
638

seen for some pelagic fauna
(Hop

et al.
, 2006)
, thes
e species will be especially vulnerable. It will be
639

important to study the species
-
specific feeding preferences in detail to assess the vulnerability of
640

individual components of the benthic food web.

641

5.

Acknowledgements

642

We thank the officers and crews of RV
s

Polarstern

and
M
aria
S
.

Merian

and the team of
the
643

remotely operated vehicle

‘‘Victor 6000’’ for their support.
We also acknowledge the work of our
644

technicians and student workers

in the laboratory and at
sea.

Three anonymous reviewers and Andy
645

Gooday

are thanked for constructive comments that considerably improved an earlier version of this
646

manuscript.
This research was supported by the HERMES project (contract

GOCE
-
CT
-
2005
-
511234),
647

funded by the European Commission’s Sixth Framework Programme under t
he priority “Sustainable
648

Development, Global Change and Ecosystems”, and HERMIONE project (grant agreement n°
649

226354") funded by the European Community's Seventh Framework Programme (FP7/2007
-
2013).

650

This is publication **** from the Netherlands Institute o
f Ecology (NIOO
-
KNAW), Yerseke and
651

publication awi
-
n**** of the Alfred Wegener Institute for Polar and Marine Research.

652

653

6.

References

654

Allen, J.T., Brown, L., Sanders, R., Moore, C.M., Mustard, A., Fielding, S., Lucas, M., Rixen, M.,
655

Savidge
, G., Henson, S., Mayor, D., 2005. Diatom carbon export enhanced by silicate upwelling in
656

the northeast Atlantic. Nature 437 (7059), 728
-
732.

657

Alongi, D.M., 1992. Bathymetric patterns of deep
-
sea benthic communities from bathyal to abyssal
658

depths in the wes
tern south Pacific (Solomon and Coral Seas). Deep
-
Sea Research Part a
-
659

Oceanographic Research Papers 39 (3
-
4A), 549
-
565.

660

Altabet, M.A., 1996. Nitrogen and carbon isotopic tracers of the source and transformation of particles
661

in the deep sea. In: Ittekkot, V
., Schäffer, P., Honjo, S., Depetris, P.J. (Eds.), Particle flux in the
662

ocean. Wiley, Chichester, pp. 155
-
184.

663

Ambrose, W.G., Clough, L.M., Tilney, P.R., Beer, L., 2001. Role of echinoderms in benthic
664

remineralization in the Chukchi Sea. Marine Biology 139

(5), 937
-
949.

665

Andersson, J.H., Wijsman, J.W.M., Herman, P.M.J., Middelburg, J.J., Soetaert, K., Heip, C., 2004.
666

Respiration patterns in the deep ocean. Geophysical Research Letters 31 (3), L03304, doi
667

03310.01029/02003gl018756.

668

Arrigo, K.R., van Dijken, G
., Pabi, S., 2008. Impact of a shrinking Arctic ice cover on marine primary
669

production. Geophys. Res. Lett. 35 (19), L19603.

670

Bauerfeind, E., Nothig, E.M., Beszczynska, A., Fahl, K., Kaleschke, L., Kreker, K., Klages, M.,
671

Soltwedel, T., Lorenzen, C.,
Wegner, J., 2009. Particle sedimentation patterns in the eastern Fram
672

Strait during 2000
-
2005: Results from the Arctic long
-
term observatory HAUSGARTEN. Deep
-
Sea
673

Research Part I
-
Oceanographic Research Papers 56 (9), 1471
-
1487.

674

Benner, R., Moran, M.A., Hods
on, R.E., 1986. Biogeochemical cycling of lignocellulosic carbon in
675

marine and freshwater ecosystems: Relative contributions of procaryotes and eucaryotes.
676

Limnology and Oceanography 31 (1), 89
-
100.

677

Bergmann, M., Dannheim, J., Bauerfeind, E., Klages, M., 2
009. Trophic relationships along a
678

bathymetric gradient at the deep
-
sea observatory HAUSGARTEN. Deep
-
Sea Research Part I
-
679

Oceanographic Research Papers 56 (3), 408
-
424.

680

Bergmann, M., Soltwedel, T., Klages, M., 2011. The interannual variability of megafaunal

681

assemblages in the Arctic deep sea: Preliminary results from the HAUSGARTEN observatory
682

(79°N). Deep Sea Research Part I: Oceanographic Research Papers In Press, Accepted
683

Manuscript.

684

Bernhard, J.M., Sen Gupta, B.K., Baguley, J.G., 2008. Benthic foraminife
ra living in Gulf of Mexico
685

bathyal and abyssal sediments: Community analysis and comparison to metazoan meiofaunal
686

biomass and density. Deep Sea Research Part II: Topical Studies in Oceanography 55 (24
-
26),
687

2617
-
2626.

688

Billett, D.S.M., Bett, B.J., Reid, W.
D.K., Boorman, B., Priede, I.G., 2010. Long
-
term change in the
689

abyssal NE Atlantic: The 'Amperima Event' revisited. Deep
-
Sea Research Part II
-
Topical Studies in
690

Oceanography 57 (15), 1406
-
1417.

691

Boetius, A., Damm, E., 1998. Benthic oxygen uptake, hydrolytic

potentials and microbial biomass at
692

the Arctic continental slope. Deep
-
Sea Research Part I
-
Oceanographic Research Papers 45 (2
-
3),
693

239
-
275.

694

Borsheim, K.Y., Bratbak, G., Heldal, M., 1990. Enumeration and biomass estimation of planktonic
695

bacteria and viruse
s by transmission electron
-
microscopy. Applied and Environmental Microbiology
696

56 (2), 352
-
356.

697

Budaeva, N.E., Mokievsky, V.O., Soltwedel, T., Gebruk, A.V., 2008. Horizontal distribution patterns in
698

Arctic deep
-
sea macrobenthic communities. Deep
-
Sea Researc
h Part I
-
Oceanographic Research
699

Papers 55 (9), 1167
-
1178.

700

Buesseler, K.O., Lamborg, C.H., Boyd, P.W., Lam, P.J., Trull, T.W., Bidigare, R.R., Bishop, J.K.B.,
701

Casciotti, K.L., Dehairs, F., Elskens, M., Honda, M., Karl, D.M., Siegel, D.A., Silver, M.W.,
702

Stei
nberg, D.K., Valdes, J., Van Mooy, B., Wilson, S., 2007. Revisiting carbon flux through the
703

ocean's twilight zone. Science 316 (5824), 567
-
570.

704

Carmack, E., Barber, D., Christensen, J., MacDonald, R., Rudels, B., Sakshaug, E., 2006. Climate
705

variability and

physical forcing of the food webs and the carbon budget on panarctic shelves.
706

Progress In Oceanography 71 (2
-
4), 145
-
181.

707

Carroll, M., Carroll, J., 2003. The Arctic Seas. In: Black, K., Shimmield, G. (Eds.), Biogeochemistry of
708

marine systems. Blackwell, O
xford, pp. 127
-
147.

709

Chikaraishi, Y., Ogawa, N.O., Kashiyama, Y., Takano, Y., Suga, H., Tomitani, A., Miyashita, H.,
710

Kitazato, H., Ohkouchi, N., 2009. Determination of aquatic food
-
web structure based on compound
-
711

specific nitrogen isotopic composition of am
ino acids. Limnology and Oceanography
-
Methods 7,
712

740
-
750.

713

Clough, L.M., Ambrose, W.G., Cochran, J.K., Barnes, C., Renaud, P.E., Aller, R.C., 1997. Infaunal
714

density, biomass and bioturbation in the sediments of the Arctic Ocean. Deep
-
Sea Research Part
715

II
-
To
pical Studies in Oceanography 44 (8), 1683
-
1704.

716

Danovaro, R., Dell'Anno, A., Corinaldesi, C., Magagnini, M., Noble, R., Tamburini, C., Weinbauer, M.,
717

2008. Major viral impact on the functioning of benthic deep
-
sea ecosystems. Nature 454 (7208),
718

1084
-
U1027
.

719

Dauwe, B., Middelburg, J.J., Herman, P.M.J., Heip, C.H.R., 1999. Linking diagenetic alteration of
720

amino acids and bulk organic matter reactivity. Limnology and Oceanography 44 (7), 1809
-
1814.

721

De Laender, F., Van Oevelen, D., Soetaert, K., Middelburg, J.J
., 2010. Carbon transfer in herbivore
-

722

and microbial loop
-
dominated pelagic food webs in the southern Barents Sea during spring and
723

summer. Marine Ecology
-
Progress Series 398, 93
-
107.

724

Deming, J.W., Baross, J.A., 1993. The early diagenesis of organic
matter: Bacterial activity. In: Engel,
725

M.H., Macko, S.A. (Eds.), Organic Geochemistry. Plenum Press, New York.

726

Fabiano, M., Pusceddu, A., Dell'Anno, A., Armeni, M., Vanucci, S., Lampitt, R.S., Wolff, G.A.,
727

Danovaro, R., 2001. Fluxes of phytopigments and la
bile organic matter to the deep ocean in the
728

NE Atlantic Ocean. Progress in Oceanography 50 (1
-
4), 89
-
104.

729

Forest, A., Wassmann, P., Slagstad, D., Bauerfeind, E., Nöthig, E.
-
M., Klages, M., 2010. Relationships
730

between primary production and vertical partic
le export at the Atlantic
-
Arctic boundary (Fram Strait,
731

HAUSGARTEN). Polar Biology
http://dx.doi.org/10.1007/s00300
-
010
-
0855
-
3
.

732

Gallucci, F., Fonseca, G., Soltwedel, T., 2008. Effects of megafauna

exclusion on nematode
733

assemblages at a deep
-
sea site. Deep
-
Sea Research Part I
-
Oceanographic Research Papers 55
734

(3), 332
-
349.

735

Geslin, E., Risgaard
-
Petersen, N., Lombard, F., Metzger, E., Langlet, D., Jorissen, F., 2010. Oxygen
736

respiration rates of benthic

foraminifera as measured with oxygen microsensors. Journal of
737

Experimental Marine Biology and Ecology doi:10.1016/j.jembe.2010.10.011.

738

Ginger, M.L., Billett, D.S.M., Mackenzie, K.L., Kiriakoulakis, K., Neto, R.R., Boardman, D.K., Santos,
739

V., Horsfall, I.M
., Wolff, G.A., 2001. Organic matter assimilation and selective feeding by
740

holothurians in the deep sea: some observations and comments. Progress in Oceanography 50 (1
-
741

4), 407
-
421.

742

Glover, A.G., Gooday, A.J., Baily, D.M., Billett, D.S.M., Chevaldonné, P.,
Colaco, A., Copley, J.,
743

Cuvelier, D., Desbruyères, D., Kalogeropoulou, V., Klages, M., Lampadariou, N., Lejeusne, C.,
744

Mestre, N.C., Paterson, G.L.J., Perez, T., Ruhl, H., Sarrazin, J., Soltwedel, T., Soto, E.H., Thatje,
745

S., Tselepides, A., Van Gaever, S.,
Vanreusel, A., 2010. Temporal Change in Deep
-
Sea Benthic
746

Ecosystems: A Review of the Evidence From Recent Time
-
Series Studies. Advances in Marine
747

Biology 85 (1
-
95).

748

Gooday, A.J., 1986. Meiofaunal foraminiferans from the bathyal Porcupine Seabight (northeas
t
749

Atlantic)
-

size structure, standing stock, taxonomic composition, species
-
diversity and vertical
-
750

distribution in the sediment. Deep
-
Sea Research 33 (10), 1345
-
1373.

751

Grebmeier, J.M., Barry, J.P., 1991. The influence of oceanographic processes on pelagic
-
benthic
752

coupling in polar regions: A benthic perspective. Journal of Marine Systems 2, 495
-
518.

753

Grebmeier, J.M., McRoy, C.P., Feder, H.M., 1988. Pelagic
-
benthic coupling on the shelf of the
754

northern Bering and Chukchi Seas. 1. Food
-
supply source and benthi
c biomass. Marine Ecology
-
755

Progress Series 48 (1), 57
-
67.

756

Grebmeier, J.M., Overland, J.E., Moore, S.E., Farley, E.V., Carmack, E.C., Cooper, L.W., Frey, K.E.,
757

Helle, J.H., McLaughlin, F.A., McNutt, S.L., 2006. A major ecosystem shift in the northern Bering
758

Sea. Science 311 (5766), 1461
-
1464.

759

Greiser, N., Faubel, A., 1988. Biotic factors. In: Higgens, R.P., Thiel, H. (Eds.), Introduction to the
760

study of meiofauna. Smithsonian Institution Press, Washington D.C., London, pp. 79
-
114.

761

Grossmann, S., Reichardt, W.
, 1991. Impact of
Arenicola marina
on bacteria in intertidal sediments.
762

Marine Ecology Progress Series 77 (1), 85
-
93.

763

Guilini, K., Van Oevelen, D., Soetaert, K., Middelburg, J.J., Vanreusel, A., 2010. Nutritional importance
764

of benthic bacteria for deep
-
sea

nematodes from the Arctic ice margin: Results of an isotope tracer
765

experiment. Limnology and Oceanography 55 (5), 1977
-
1989.

766

Gumprecht, R., Gerlach, H., Nehrkorn, A., 1995. FDA hydrolysis and resazurin reduction as a
767

measure of microbial activity in sedim
ents from the south
-
east Atlantic. Helgolander
768

Meeresuntersuchungen 49 (1
-
4), 189
-
199.

769

Harvey, H.R., Tuttle, J.H., Bell, J.T., 1995. Kinetics of Phytoplankton Decay During Simulated
770

Sedimentation
-

Changes in Biochemical
-
Composition and Microbial Activity
under Oxic and Anoxic
771

Conditions. Geochimica Et Cosmochimica Acta 59 (16), 3367
-
3377.

772

Heip, C.H.R., Duineveld, G., Flach, E., Graf, G., Helder, W., Herman, P.M.J., Lavaleye, M.,
773

Middelburg, J.J., Pfannkuche, O., Soetaert, K., Soltwedel, T., de Stigter, H.,

Thomsen, L.,
774

Vanaverbeke, J., de Wilde, P., 2001. The role of the benthic biota in sedimentary metabolism and
775

sediment
-
water exchange processes in the Goban Spur area (NE Atlantic). Deep
-
Sea Research
776

Part II
-
Topical Studies in Oceanography 48 (14
-
15), 322
3
-
3243.

777

Hirche, H.
-
J., Kosobokova, K., 2007. Distribution of
Calanus finmarchicus

in the northern North
778

Atlantic and Arctic Ocean
--
Expatriation and potential colonization. Deep
-
Sea Research II 54 (23
-
779

26), 2729
-
2747.

780

Honjo, S., Krishfield, R.A., Eglinton, T
.I., Manganini, S.J., Kemp, J.N., Doherty, K., Hwang, J., McKee,
781

T.K., Takizawa, T., 2010. Biological pump processes in the cryopelagic and hemipelagic Arctic
782

Ocean: Canada Basin and Chukchi Rise. Progress in Oceanography 85 (3
-
4), 137
-
170.

783

Hop, H.,
Falk
-
Petersen, S., Svendsen, H., Kwasniewski, S., Pavlov, V., Pavlova, O., Soreide, J.E.,
784

2006. Physical and biological characteristics of the pelagic system across Fram Strait to
785

Kongsfjorden. Progress in Oceanography 71 (2
-
4), 182
-
231.

786

Hoste, E., Vanhove
, S., Schewe, I., Soltwedel, T., Vanreusel, A., 2007. Spatial and temporal variations
787

in deep
-
sea meiofauna assemblages in the Marginal Ice Zone of the Arctic Ocean. Deep
-
Sea
788

Research Part I
-
Oceanographic Research Papers 54 (1), 109
-
129.

789

Iken, K., Bluhm, B
.A., Gradinger, R., 2005. Food web structure in the high Arctic Canada Basin:
790

evidence from delta C
-
13 and delta N
-
15 analysis. Polar Biology 28 (3), 238
-
249.

791

Iken, K., Brey, T., Wand, U., Voigt, J., Junghans, P., 2001. Food web structure of the benthic
792

co
mmunity at the Porcupine Abyssal Plain (NE Atlantic): a stable isotope analysis. Progress in
793

Oceanography 50 (1
-
4), 383
-
405.

794

Ingels, J., Van den Driessche, P., De Mesel, I., Vanhove, S., Moens, T., Vanreusel, A., 2010.
795

Preferred use of bacteria over phytop
lankton by deep
-
sea nematodes in polar regions. Marine
796

Ecology
-
Progress Series 406, 121
-
133.

797

IPCC, 2007. Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to
798

the Fourth Assessment Report of the Intergovernmental Panel on Clim
ate Change. In: Solomon,
799

S., Qin, D., Manning, M., Chen, Z., Marquis, M., Averyt, K.B., Tignor, M., Miller, H.L. (Eds.),
800

Cambridge, United Kingdom and New York, NY, USA.

801

Iverson, S.J., Field, C., Bowen, W.D., Blanchard, W., 2004. Quantitative fatty acid si
gnature analysis:
802

A new method of estimating predator diets. Ecological Monographs 74 (2), 211
-
235.

803

Jumars, P.A., Mayer, L.M., Deming, J.W., Baross, J.A., Wheatcroft, R.A., 1990. Deep
-
sea deposit
-
804

feeding strategies suggested by environmental and feeding co
nstraints. Philosophical Transactions
805

of the Royal Society of London Series A
-
Mathematical Physical and Engineering Sciences 331
806

(1616), 85
-
101.

807

Kanzog, C., Ramette, A., Queric, N.V., Klages, M., 2009. Response of benthic microbial communities
808

to chitin en
richment: an in situ study in the deep Arctic Ocean. Polar Biology 32 (1), 105
-
112.

809

Kiriakoulakis, K., Stutt, E., Rowland, S.J., Vangriesheim, A., Lampitt, R.S., Wolff, G.A., 2001. Controls
810

on the organic chemical composition of settling particles in the
Northeast Atlantic Ocean. Progress
811

in Oceanography 50 (1
-
4), 65
-
87.

812

Kones, J.K., Soetaert, K., van Oevelen, D., Owino, J.O., 2009. Are network indices robust indicators of
813

food web functioning? a Monte Carlo approach. Ecological Modelling 220, 370

382.

814

Kös
ter, M., Jensen, P., Meyer
-
Reil, L.A., 1991. Hydrolytic activity associated with organisms and
815

biogenic structures in deep
-
sea sediments. In: Chrost, R. (Ed.), Microbial Enzymes in Aquatic
816

Environments. Springer Verlag, Berlin, pp. 298
-
310.

817

Kröncke, I., 19
94. Macrobenthos composition, abundance and biomass in the Arctic Ocean along a
818

transect between Svalbard and the Makarov Basis. Polar Biology 14 (8), 519
-
529.

819

Mayer, L.M., Schick, L.L., Sawyer, T., Plante, C.J., Jumars, P.A., Self, R.L., 1995. Bioavailabl
e amino
-
820

acids in sediments
-

A biomimetic, kinetics
-
based approach. Limnology and Oceanography 40 (3),
821

511
-
520.

822

Meziane, T., Bodineau, L., Retiere, C., Thoumelin, G., 1997. The use of lipid markers to define
823

sources of organic matter in sediment and food w
eb of the intertidal salt
-
marsh
-
flat ecosystem of
824

Mont
-
Saint
-
Michel Bay, France. Journal of Sea Research 38 (1
-
2), 47
-
58.

825

Middelburg, J.J., 1989. A simple rate model for organic matter decomposition in marine sediments.
826

Geochimica Et Cosmochimica Acta 53 (
7), 1577
-
1581.

827

Moodley, L., Middelburg, J.J., Boschker, H.T.S., Duineveld, G.C.A., Pel, R., Herman, P.M.J., Heip,
828

C.H.R., 2002. Bacteria and Foraminifera: Key players in a short
-
term deep
-
sea benthic response to
829

phytodetritus. Marine Ecology Progress Serie
s 236, 23
-
29.

830

Moore, J.C., Berlow, E.L., Coleman, F.C., De Ruiter, P.C., Dong, Q., Hastings, A., Collins, N.,
831

McCann, K.S., Melville, K., Morin, P.J., Nadelhoffer, K., Rosemond, A.D., Post, D.M., Sabo, J.L.,
832

Scow, K.M., Vanni, M.J., Wall, D.H., 2004. Detri
tus, trophic dynamics and biodiversity. Ecology
833

Letters 7, 584
-
600.

834

Moran, S.B., Kelly, R.P., Hagstrom, K., Smith, J.N., Grebmeier, J.M., Cooper, L.W., Cota, G.F., Walsh,
835

J.J., Bates, N.R., Hansell, D.A., Maslowski, W., Nelson, R.P., Mulsow, S., 2005. Seas
onal changes
836

in POC export flux in the Chukchi Sea and implications for water column
-
benthic coupling in Arctic
837

shelves. Deep
-
Sea Research Part II
-
Topical Studies in Oceanography 52 (24
-
26), 3427
-
3451.

838

Pfannkuche, O., 2005. Allochtonous deep
-
sea benthic co
mmunities: functioning and forcing. In:
839

Kristensen, E., Haese, R.R., Kostka, J.E. (Eds.), Interactions between macro
-

and microorganisms
840

in marine sediments. American Geophysical Union, Washington DC, pp. 251
-
266.

841

Piepenburg, D., Blackburn, T.H., Vondorrie
n, C.F., Gutt, J., Hall, P.O.J., Hulth, S., Kendall, M.A.,
842

Opalinski, K.W., Rachor, E., Schmid, M.K., 1995. Partitioning of benthic community respiration in
843

the Arctic (northwestern Barents Sea). Marine Ecology
-
Progress Series 118 (1
-
3), 199
-
213.

844

Pörtner,
H.O., Berdal, B., Blust, R., Brix, O., Colosimo, A., De Wachter, B., Giuliani, A., Johansen, T.,
845

Fischer, T., Knust, R., Lannig, G., Naevdal, G., Nedenes, A., Nyhammer, G., Sartoris, F.J.,
846

Serendero, I., Sirabella, P., Thorkildsen, S., Zakhartsev, M., 2001
. Climate induced temperature
847

effects on growth performance, fecundity and recruitment in marine fish: developing a hypothesis
848

for cause and effect relationships in Atlantic cod (Gadus morhua) and common eelpout (Zoarces
849

viviparus). Continental Shelf Resea
rch 21 (18
-
19), 1975
-
1997.

850

Pusceddu, A., Bianchelli, S., Canals, M., Sanchez
-
Vidal, A., Durrieu De Madron, X., Heussner, S.,
851

Lykousis, V., de Stigter, H., Trincardi, F., Danovaro, R., 2010. Organic matter in sediments of
852

canyons and open slopes of the Port
uguese, Catalan, Southern Adriatic and Cretan Sea margins.
853

Deep Sea Research Part I: Oceanographic Research Papers 27, 441
-
457.

854

Renaud, P.E., Morata, N., Ambrose, W.G., Bowie, J.J., Chiuchiolo, A., 2007. Carbon cycling by
855

seafloor communities on the easter
n Beaufort Sea shelf. Journal of Experimental Marine Biology
856

and Ecology 349 (2), 248
-
260.

857

Rex, M.A., Etter, R.J., Morris, J.S., Crouse, J., McClain, C.R., Johnson, N.A., Stuart, C.T., Deming,
858

J.W., Thies, R., Avery, R., 2006. Global bathymetric patterns o
f standing stock and body size in the
859

deep
-
sea benthos. Marine Ecology
-
Progress Series 317, 1
-
8.

860

Rowe, G.T., Wei, C.L., Nunnally, C., Haedrich, R., Montagna, P., Baguley, J.G., Bernhard, J.M.,
861

Wicksten, M., Ammons, A., Briones, E.E., Soliman, Y., Deming, J
.W., 2008. Comparative biomass
862

structure and estimated carbon flow in food webs in the deep Gulf of Mexico. Deep
-
Sea Research
863

Part II
-
Topical Studies in Oceanography 55 (24
-
26), 2699
-
2711.

864

Ruhl, H.A., Ellena, J.A., Smith, K.L., 2008. Connections between cl
imate, food limitation, and carbon
865

cycling in abyssal sediment communities. Proceedings of the National Academy of Sciences of the
866

United States of America 105 (44), 17006
-
17011.

867

Schluter, M., Sauter, E.J., Schafer, A., Ritzrau, W., 2000. Spatial budget of

organic carbon flux to the
868

seafloor of the northern North Atlantic (60 degrees N
-
80 degrees N). Global Biogeochemical
869

Cycles 14 (1), 329
-
340.

870

Slagstad, D., Ellingsen, I., Wassmann, P., 2010. Primary and secondary production in a future Arctic
871

Ocean withou
t summer sea ice. Progress In Oceanography.

872

Smith, K.L., Ruhl, H.A., Bett, B.J., Billett, D.S.M., Lampitt, R.S., Kaufmann, R.S., 2009. Climate,
873

carbon cycling, and deep
-
ocean ecosystems. Proceedings of the National Academy of Sciences of
874

the United States
of America 106 (46), 19211
-
19218.

875

Soetaert, K., Franco, M., Lampadariou, N., Muthumbi, A., Steyaert, M., Vandepitte, L., vanden Berghe,
876

E., Vanaverbeke, J., 2009. Factors affecting nematode biomass, length and width from the shelf to
877

the deep sea. Marine E
cology
-
Progress Series 392, 123
-
132.

878

Soetaert, K., Van Oevelen, D., 2008. LIM: Linear Inverse Model examples and solution methods. R
879

package version 1.2.

880

Soetaert, K., Van Oevelen, D., 2009. Modeling food web interactions in benthic deep
-
sea ecosystems:
881

a
practical guide. Oceanography 22 (1), 130
-
145.

882

Soltwedel, T., Bauerfeind, E., Bergmann, M., Budaeva, N., Hoste, E., Jaeckisch, N., von Juterzenka,
883

K., Matthiessen, J., Mokievsky, V., Nöthig, E.M., Quéric, N.V., Sablotny, B., Sauter, E., Schewe, I.,
884

Urban
-
M
alinga, B., Wegner, J., Wlodarska
-
Kowalczuk, M., Klages, M., 2005. HAUSGARTEN:
885

Multidisciplinary investigations at a deep
-
sea, long
-
term observatory in the Arctic Ocean.
886

Oceanography 18 (3), 46
-
61.

887

Soltwedel, T., Jaeckisch, N., Ritter, N., Hasemann, C., Be
rgmann, M., Klages, M., 2009. Bathymetric
888

patterns of megafaunal assemblages from the arctic deep
-
sea observatory HAUSGARTEN. Deep
-
889

Sea Research Part I
-
Oceanographic Research Papers 56 (10), 1856
-
1872.

890

Stephens, M.P., Kadko, D.C., Smith, C.R., Latasa, M., 1
997. Chlorophyll
-
a and pheopigments as
891

tracers of labile organic carbon at the central equatorial Pacific seafloor. Geochimica Et
892

Cosmochimica Acta 61 (21), 4605
-
4619.

893

Sun, M.Y., Aller, R.C., Lee, C., 1991. Early diagenesis of chlorophyll
-
a

in Long Island
Sound
894

sediments: A measure of carbon flux and particle reworking. Journal of Marine Research, pp. 379
-
895

401.

896

Ulanowicz, R.E., 2004. Quantitative methods for ecological network analysis. Computational Biology
897

and Chemistry 28 (5
-
6), 321
-
339.

898

Urban
-
Malinga,
B., Moens, T., 2006. Fate of organic matter in Arctic intertidal sediments: Is utilisation
899

by meiofauna important? Journal of Sea Research 56 (3), 239
-
248.

900

Van den Meersche, K., Soetaert, K., Van Oevelen, D., 2009. xsample(): an R function for sampling
901

lin
ear inverse problems. Journal of Statistical Software 30 (1), 1
-
15.

902

Van Oevelen, D., Duineveld, G.C.A., Lavaleye, M.S.S., Mienis, F., Soetaert, K., Heip, C.H.R., 2009.
903

The cold
-
water coral community as hotspot of carbon cycling on continental margins: a fo
od web
904

analysis from Rockall Bank (northeast Atlantic). Limnology and Oceanography 54 (6), 1829

1844.

905

Van Oevelen, D., Soetaert, K., García, R., De Stigter, H., Cunha, M.R., Pusceddu, A., Danovaro, R.,
906

2011. Canyon conditions impact carbon flows in food we
bs of three sections of the Nazaré canyon.
907

Deep Sea Research Part II: Topical Studies in Oceanography.

908

Van Oevelen, D., Soetaert, K., Middelburg, J.J., Herman, P.M.J., Moodley, L., Hamels, I., Moens, T.,
909

Heip, C.H.R., 2006. Carbon flows through a benthic
food web: Integrating biomass, isotope and
910

tracer data. Journal of Marine Research 64 (3), 1
-
30.

911

Van Oevelen, D., Van den Meersche, K., Meysman, F., Soetaert, K., Middelburg, J.J., Vézina, A.F.,
912

2010. Quantitative reconstruction of food webs using linear i
nverse models. Ecosystems 13, 32

913

45.

914

Vanaverbeke, J., Arbizu, P.M., Dahms, H.U., Schminke, H.K., 1997. The Metazoan meiobenthos
915

along a depth gradient in the Arctic Laptev Sea with special attention to nematode communities.
916

Polar Biology 18 (6), 391
-
401.

917

V
anreusel, A., Clough, L., Jacobsen, K., Ambrose, W., Jivaluk, J., Ryheul, V., Herman, R., Vincx, M.,
918

2000. Meiobenthos of the central Arctic Ocean with special emphasis on the nematode community
919

structure. Deep
-
Sea Research Part I
-
Oceanographic Research Pa
pers 47 (10), 1855
-
1879.

920

Wassmann, P., Slagstad, D., Riser, C.W., Reigstad, M., 2006. Modelling the ecosystem dynamics of
921

the Barents Sea including the marginal ice zone II. Carbon flux and interannual variability. Journal
922

of Marine Systems 59 (1
-
2), 1
-
24.

923

Westrich, J.T., Berner, R.A., 1984. The role of sedimentary organic matter in bacterial sulfate
924

reduction: The
G

model tested. Limnology and Oceanography 29 (2), 236
-
249.

925

Wieser, W., 1953. Die Beziehung zwischen Mundhöhlengestalt, Ernährungsweise und Vork
ommen bei
926

freilebenden marinen Nematoden. Eine skologisen
-
morphologische studie. Arkiv für Zoologie 4,
927

439
-
484.

928

Wigham, B.D., Galley, E.A., Smith, C.R., Tyler, P.A., 2008. Inter
-
annual variability and potential for
929

selectivity in the diets of deep
-
water An
tarctic echinoderms. Deep
-
Sea Research Part II
-
Topical
930

Studies in Oceanography 55 (22
-
23), 2478
-
2490.

931

Wigham, B.D., Hudson, I.R., Billett, D.S.M., Wolff, G.A., 2003. Is long
-
term change in the abyssal
932

Northeast Atlantic driven by qualitative changes in exp
ort flux? Evidence from selective feeding in
933

deep
-
sea holothurians. Progress in Oceanography 59 (4), 409
-
441.

934

Witbaard, R., Duineveld, G.C.A., Kok, A., van der Weele, J., Berghuis, E.M., 2001. The response of
935

Oneirophanta mutabilis

(Holothuroidea) to the s
easonal deposition of phytopigments at the
936

porcupine Abyssal Plain in the Northeast Atlantic. Progress in Oceanography 50 (1
-
4), 423
-
441.

937

Witbaard, R., Duineveld, G.C.A., Van der Weele, J.A., Berghuis, E.M., Reyss, J.P., 2000. The benthic
938

response to the s
easonal deposition of phytopigments at the Porcupine Abyssal Plain in the North
939

East Atlantic. Journal of Sea Research 43 (1), 15
-
31.

940

Witte, U., Wenzhofer, F., Sommer, S., Boetius, A., Heinz, P., Aberle, N., Sand, M., Cremer, A.,
941

Abraham, W.R., Jorgensen,
B.B., Pfannkuche, O., 2003. In situ experimental evidence of the fate
942

of a phytodetritus pulse at the abyssal sea floor. Nature 424 (6950), 763
-
766.

943

Wlodarska
-
Kowalczuk, M., Pearson, T.H., 2004. Soft
-
bottom macrobenthic faunal associations and
944

factors affe
cting species distributions in an Arctic glacial fjord (Kongsfjord, Spitsbergen). Polar
945

Biology 27 (3), 155
-
167.

946

Woulds, C., Andersson, J.H., Cowie, G.L., Middelburg, J.J., Levin, L.A., 2009. The short
-
term fate of
947

organic carbon in marine sediments: Compa
ring the Pakistan margin to other regions. Deep
-
Sea
948

Research Part II
-
Topical Studies in Oceanography 56 (6
-
7), 393
-
402.

949

Woulds, C., Cowie, G.L., Levin, L.A., Andersson, J.H., Middelburg, J.J., Vandewiele, S., Lamont, P.A.,
950

Larkin, K.E., Gooday, A.J., Schum
acher, S., Whitcraft, C., Jeffreys, R.M., Schwartz, M., 2007.
951

Oxygen as a control on seafloor biological communities and their roles in sedimentary carbon
952

cycling. Limnology and Oceanography 52 (4), 1698
-
1709.

953


954


955