Origin and variability of downward biogeochemical fluxes on the Rhone continental margin (NW mediterranean)

*Corresponding author. Fax: #33-4-68-66-20-96.E-mail address: [email protected] (A. Monaco)1Present address: CEREGE, Europo( le de l'Arbois, BP 80, 13545 Aix en Provence Cedex 04, France.

Deep-Sea Research I 46 (1999) 1483}1511

Origin and variability of downwardbiogeochemical #uxes on the Rhone continental

margin (NW mediterranean)

A. Monaco*, X. Durrieu de Madron, O. Radakovitch1,S. Heussner, J. Carbonne

CNRS } Centre de Formation et de Recherche sur l+Environment Marine Universite& de Perpignan,66860 Perpignan Cedex, France

Received 27 January 1997; received in revised form 14 February 1998; accepted 19 January 1999

Abstract

A one year study of downward particle #uxes conducted in the northwestern MediterraneanSea is presented. Two mooring lines equipped with sediment traps and current meters weredeployed at around 1000 m depth on the northeastern continental slope of the Gulf of Lions,one inside the Grand-Rho( ne canyon and the other outside on the adjacent open slope. Meantotal mass #uxes increased slightly with trap depth inside the canyon, a feature quite typical of#uxes in continental margin environments. The near-bottom trap inside the canyon collectedmore material than its counterpart deployed at equivalent depth on the open slope, indicatinga preferential transport of material within the canyon. Major biogeochemical constituents(organic and inorganic carbon, opal, and siliciclastic residue) revealed a marked di!erence inparticle composition between the sub-surface (80 m) and deeper traps, suggesting the existenceof at least two sources of material. The two shallower traps showed a clear biological signal: #uxpeaks were related to periods of surface biological production, especially perceptible in summerand autumn. The particulate matter trapped at deeper levels in the canyon and on the openslope was characterized by a more stable composition with a major lithogenic contribution,originating from sedimentary material most probably resuspended on the upper- or mid-slope.The seasonal variability was dominated by the summer/winter alternation; the latter period wascharacterized by a weak strati"cation of the water column and an enhanced current variabilityfavoring vertical exchanges. The present results are compared with those obtained previously inthe Lacaze-Duthiers canyon on the southwestern side of the Gulf of Lions. The comparisonshows strong di!erences between the NE entrance and the SW exit of the gulf, with respect to

0967-0637/99/$ - see front matter ( 1999 Elsevier Science Ltd. All rights reserved.PII: S 0 9 6 7 - 0 6 3 7 ( 9 9 ) 0 0 0 1 4 - X

the general along-slope circulation of water masses, both in terms of intensity of particulate#uxes and transport processes. ( 1999 Elsevier Science Ltd. All rights reserved.

1. Introduction

The relevance of studies on continental margins to the problem of marine bi-ogeochemical cycling has been recognized since the creation of the JGOFS (JointGlobal Ocean Flux Study) program. More recently, the necessity of obtaining im-proved estimations of the global budget of matter and particularly of carbon oncontinental margins was emphasized (LOICZ, 1993).

The ECOMARGE program (ECOsystemes de MARGE continentale), part ofFrance-JGOFS, focused on the study of particle transfer and benthic ecosystemresponse on French continental margins (Monaco et al., 1990a). From 1985 to 1989two successive experiments were conducted in the Gulf of Lions (northwesternMediterranean). The "rst experiment (ECOMARGE-I) took place in 1985}1986 onthe southwestern slope, that is at the exit of the Gulf of Lions with respect to thegeneral circulation of water masses (Fig. 1). The results were published in a specialissue of Continental Shelf Research (1990, vol. 10, No. 9}11). Particulate #uxes, mea-sured over a 9 month period (July 1985}April 1986), showed clear di!erences betweenthe surface layer and the deeper water column, the latter being subject to strongadvective inputs from the shelf. Furthermore, the strong seasonal #uctuations of #uxeswere linked to climatic forcing (precipitation, river inputs) and, to a lesser extent, tobiological production (Monaco et al., 1990a). The second experiment (ECOMARGE-II) was performed during the years 1986}1988 in the region of the Grand-Rho( necanyon at the northeastern entrance of the Gulf of Lions (Fig. 1). This experiment wasbased on "ve synoptic, seasonal cruises that were intended to describe the hydrologi-cal, chemical and biological environment of the study site (e.g., Durrieu de Madron,1994; Henry et al., 1994), and on a one year deployment of moored instrumentation.

The present paper aims to describe the temporal qualitative and quantitativevariations of downward particle #uxes as measured by sediment traps. The discussionfocuses on the potential sources of particles and the comparison with the previousexperiment on the southwestern side of the gulf.

2. Study area

The water mass circulation in the Gulf of Lions is driven essentially by the cyclonicNorthern Current (Millot, 1990; Beckers et al., 1997). A shallow branch of this currentpenetrates over the shelf, whereas the main outer branch (0.5}1.8]106 m3 s~1) #owssouthwestward along the slope. This latter branch has an average core width of about30}50 km and a thickness of about 300}500 m. The current speed decreases from30}50 cm s~1 at the surface to a few cm s~1 at a few hundred meters depth. Deeperwaters are subject to its in#uence and follow isobaths in the same southwestward,along-slope direction (Millot, 1990; Durrieu de Madron et al., 1990).

1484 A. Monaco et al. / Deep-Sea Research I 46 (1999) 1483}1511

Fig. 1. Location and design of the sediment trap moorings of the ECOMARGE-II experiment in theGrand-Rho( ne Canyon (GRC site; 42350@72 N, 04349@07 E) and the adjacent open slope (OS site; 42345@37 N,04347@35 E) on the continental slope of the Gulf of Lions (Northwestern Mediterranean Sea). The locationof the "rst ECOMARGE experimental site in the Lacaze-Duthiers canyon is also indicated.

With a solid discharge of approximately 6.2]106 T yr~1 (Pont, 1997), the Rho( neriver constitutes the main source (&80%) of land-derived particulate matter in theGulf of Lions. Major #ows occur from March to May, usually with an annual #ood.On the shelf, suspended particulate matter (SPM) is transported primarily within thebottom nepheloid layer, which generally detaches from the bottom at the shelf-break.The numerous canyons that indent the shelf and the slope appear to be preferentialpathways for o!-shelf and downslope transfer of suspended and settling particles(Durrieu de Madron et al., 1990; Monaco et al., 1990b; Durrieu de Madron, 1994).

3. Material and methods

3.1. Field experiment

Two mooring lines equipped with sediment traps and current meters were deployedat around 950 m depth within the Grand-Rho( ne canyon and on the adjacent south-western open slope. This canyon runs across the Gulf of Lions continental slope o!

A. Monaco et al. / Deep-Sea Research I 46 (1999) 1483}1511 1485

the Grand-Rho( ne river mouth (Fig. 1). It is oriented N}S in the upper part and turnsto the SE in its lower part, below 1000 m depth. The canyon line was equipped withfour cylindroconical sediment traps (model PPS3 Technicap, 0.125 m2 collection area,six receiving cups; Heussner et al., 1990) at nominal depths of 80, 200, 600 and 900 m.The traps were coupled with Aanderaa rotor current meters (RCM 4 and 5) located5 m below the traps. The open slope line was equipped with only one sediment trap-current meter pair at a nominal depth of 900 m.

The experiment lasted one year, from January 1988 to January 1989, with threesuccessive deployments (collection periods: 26 January to 13 May 1988, 22 May to7 September 1988, 27 September 1988 to 12 January 1989). The sediment trapsampling interval was set at 18 days for all three periods. All programmed sampleswere recovered, except the last sample from the "rst deployment of the 200 and 600 mtraps. Current meters recorded pressure, temperature, and current speed and directionat 30-minute intervals.

The mooring lines were maintained taut by the distribution of #oats at the mooringhead and along the line. Examination of the current meter pressure sensors showedthat tilting and deepening of the canyon mooring line started when current speedsmeasured by the current meter at 80 m exceeded 12 cm s~1. Such speeds wererecorded 40% of the time. However, deepening of instruments during the course ofeach deployment never exceeded 16 m, except for the instruments at 80 m during the"rst deployment, which deepened to a maximum of 122 m. These important depthvariations (larger than 20 m) observed for the shallowest instruments occurred mostlyin April-May during strong current outbreaks and represented less than 7% of the"rst deployment period. Mooring dynamics were computed for each line using actualcurrent values. The results showed that tilting never exceeded a few degrees, that is,with no detectable e!ect on the collection rate of cylindrical traps (Gardner, 1985).For the sake of convenience, we will therefore refer in the following part of the text tothe nominal depths of the traps.

3.2. Analytical methods

3.2.1. Sediment trap sample processingA complete description of the PPS3 sediment trap and the sample processing was

given by Heussner et al. (1990). The Te#on receiving cups of the traps were "lledbefore deployment with a bu!ered 5% (v/v) formaldehyde solution in 0.45 lm-"lteredsea water. This poisoning solution was used to limit degradation of trapped particles.It also prevented the mechanical disruption of swimming organisms (swimmers) thatoccasionally entered the traps during sampling. After recovery, the cups were stored inthe dark at 2}43C until they could be processed in the laboratory, within a maximumdelay of a few months. Swimmers were removed in a two-step procedure. Afterdecantation of the supernatant, particles were wet-sieved through a 1 mm nylon meshin order to retain the largest organisms. Swimmers smaller than 1 mm were manuallyremoved under a dissecting microscope using "ne tweezers. Even though no checkother than visual examination was performed, it can be considered that the bulk ofswimmers was removed by this method. The original samples were then precisely


divided into subsamples for subsequent analyses using a rotary splitting method(Heussner et al., 1990). Sample dry weight was determined on four subsamples "lteredonto 0.45 lm Millipore "lters, rinsed with distilled water to remove salts, and dried at403C for 24 h.

3.2.2. Major constituentsReplicate subsamples were "ltered onto Whatman "lters (GF/F), rinsed with distilled

water, and dried in an oven at 403C. Total carbon was analyzed by combustion in a CS125 LECO carbon analyzer. Organic carbon was analyzed in the same way afterprogressive, controlled acidi"cation with 2 N HCl to remove carbonate. Organic mattercontent was estimated as twice the organic carbon content (Monaco et al., 1990b).Carbonate content was calculated assuming all inorganic carbon (total C!C

03') was

calcium carbonate and using the molecular mass ratio 100/12 (i.e. CaCO3/C). After

"ltration onto 0.45 lm Millipore "lters, biogenic silica (opal) content was measured asdescribed by Mortlock and Froelich (1989). The siliciclastic residue was de"ned as thedi!erence between the total mass and the sum of the biogenic components (i.e. organicmatter, carbonate and opal). It includes quartz, feldspars, heavy minerals andaluminosilicates, and is generally referred to as the lithogenic fraction.

3.2.3. Microscopic analysesSEM analyses were performed with a HITACHI 5000 scanning microscope equip-

ped with a TRACOR microprobe. Subsamples were taken from the original, undistur-bed samples prior to any further treatment. This helped to describe the nature of thecollected particles, the relative importance of "ne-grained and coarse material, aggreg-ates, and lithogenic material. These analyses were also intended to identify the maincomponents of #uxes, particularly the sources of chemical constituents during somepeculiar #ux events, such as diatoms and silico#agellates for opal, coccoliths, ptero-pods, and molluscan shells for carbonate.

4. Results

The detailed time series of the 18-day interval mass #uxes and the contents of majorconstituents are given in Table 1. Changes of particle composition with depth will bediscussed hereafter on the basis of average values. As the time variability of #ux mustbe taken into account in the computation of the means (Heussner et al., 1990; Monacoet al., 1990b; Biscaye and Anderson, 1994), a #ux-weighted mean concentration wastherefore computed for each constituent and each trap by weighting individual sampleconcentrations for their corresponding mass #uxes:

C&8"+C

iF

i/+F

i,

where C&8

is the #ux-weighted concentration for a given trap depth, and Ciand F

ithe

measured concentration and mass #ux for sample i. This way of computing a meanyields a `truea average value equivalent to what would have been obtained if settlingparticles had been collected as a single sample during the entire experiment.


Tab

le1

Tota

lm

ass#ux

(mg

m~2

d~

1),#uxe

s(m

gm

~2

d~1)

and

cont

ents

(wt%

)of

maj

orco

nst

itue

nts

ofth

ein

div

idua

lse

dim

ent

trap

sam

ple

sco

llect

edduring

the

3dep

loym

ents

(EC

OR

O)N

EIto

III)

ofth

eEC

OM

AR

GE-I

Iex

perim

ent.

Org

anic

mat

teris

calc

ulat

edas

org

anic

carb

on]

2;ca

rbonat

eis

calc

ulat

edas

sum

ing

all

inorg

anic

carb

on

isca

lciu

mca

rbon

ate

(inor

ganic

carb

on]

8.33

);th

elit

hoge

nic

frac

tion

was

de"

ned

asth

edi!er

ence

betw

een

the

tota

lm

ass

and

the

sum

oft

he

bioge

nic

com

pon

ents

(i.e.org

anic

mat

ter,

carb

onat

ean

dopal

)

EC

OR

HO)

NE

I1

23

45

61/

26-2

/13

2/13

-3/2

3/2-

3/20

3/20

-4/7

4/7-

4/25

4/25

-5/1

3

Dep

th(m

)Flu

xco

mpon

ents

%Flu

x%

Flu

x%

Flu

x%

Flu

x%

Flu

x%

Flu

x

Can

yon

80Tota

l#ux

100

136

100

193

100

163

100

172

100

1816

100

208

Org

anic

mat

ter

1926

1834

2134

2645

142

5122

45C

arbona

te13

1716

3023

3825

4210

189

2449

Opal

57

611

1017

814

1018

08

16Litho

geni

c63

8660

118

4674

4171

6611

9646

9820

0Tota

l#ux

100

445

100

243

100

300

100

472

100

796

Org

anic

mat

ter

625

615

1132

628

648

Car

bon

ate

2912

826

6326

7720

9622

172

No

sam

ple

Opal

941

818

928

1048

1291

Litho

geni

c56

251

6014

754

163

6430

060

483

600

Tota

l#ux

100

326

100

288

100

326

100

1418

100

3410

Org

anic

mat

ter

1240

1542

827

1115

69

300

Car

bona

te21

7013

3824

7922

306

2792

6N

osa

mple

Opal

824

720

825

811

29

313

Litho

geni

c59

192

6518

860

195

5984

455

1871

900

Tota

l#ux

100

889

100

360

100

8510

020

210

024

3210

045

9O

rgan

icm

atte

r5

437

246

56

138

199

521

Car

bon

ate

2320

820

7327

2325

5019

452

3415

6O

spal

874

726

76

713

718

111

53Litho

geni

c64

564

6623

760

5162

126

6616

0050

229

Ope

nsl

ope

900

Tota

l#ux

100

217

100

7110

080

100

159

100

1342

100

1305

Org

anic

mat

ter

1327

86

97

711

678

673


Car

bon

ate

No

valu

e21

1524

1925

4026

349

2330

9O

pal

1225

85

87

1015

1012

98

101

Litho

geni

cN

ova

lue

6345

5947

5893

5878

663

822

EC

OR

HO)

NE

II7

89

1011

125/

22-6

/96/

9-6/

276/

27-7

/15

7/15

-8/2

8/2-

8/20

8/20

-9/7

Dep

th(m

)Flu

xco

mpon

ents

%Flu

x%

Flu

x%

Flu

x%

Flu

x%

Flu

x%

Flu

x

Can

yon

80Tota

l#ux

100

8410

063

100

123

100

132

100

374

100

263

Org

anic

mat

ter

3731

4025

4657

5269

6323

671

187

Car

bona

te27

2323

1436

4528

3720

76N

ova

lue

Opal

87

85

67

No

valu

e3

123

8Litho

geni

c28

2329

1912

14N

ova

lue

1450

No

valu

e20

0Tota

l#ux

100

476

100

8510

071

100

7910

032

210

013

7O

rgan

icm

atte

r19

9234

2930

2123

1812

4014

19C

arbon

ate

1988

2723

3424

2923

3711

943

59O

pal

1153

108

129

2621

721

1115

Litho

genic

5124

329

2524

1722

1744

142

3244

600

Tota

l#ux

100

726

100

272

100

8710

025

910

063

710

034

8O

rgan

icm

atte

r10

7411

3017

1510

268

5210

34C

arbon

ate

2014

726

7211

1028

7326

164

2910

1O

pal

1176

1335

87

1127

956

1034

Litho

genic

5942

950

135

6455

5113

357

365

5117

990

0Tota

l#ux

100

1120

100

203

100

6610

016

110

022

810

042

1O

rgan

icm

atte

r7

769

1811

79

156

148

32C

arbon

ate

2528

321

4427

1727

4328

6331

130

Opal

1213

318

3610

710

169

2111

46Litho

geni

c56

628

5210

552

3554

8757

130

5021

3

Ope

nsl

ope

900

Tota

l#ux

100

778

100

120

100

4310

041

100

9910

015

3O

rgan

icm

atte

r10

7512

1419

817

712

1210

15

(con

tinu

edon

nextpa

ge)


Tab

le1

(con

tinu

ed).

EC

OR

HO)

NE

II7

89

1011

125/

22-6

/96/

9-6/

276/

27-7

/15

7/15

-8/2

8/2-

8/20

8/20

-9/7

Dep

th(m

)Flu

xco

mpon

ents

%Flu

x%

Flu

x%

Flu

x%

Flu

x%

Flu

x%

Flu

x

Car

bon

ate

1814

125

3034

1532

1327

2633

50O

pal

1410

712

145

210

412

128

12Litho

genic

5845

551

6242

1841

1749

5049

76

EC

OR

HO)

NE

III

1314

1516

1718

9/27

-10/

1410

/14-

11/1

11/1

-11/

1911

/19-

12/7

12/7

-12/

2512

/25-

1/12

Dep

thFlu

xco

mpon

ents

%Flu

x%

Flu

x%

Flu

x%

Flu

x%

Flu

x%

Flu

x

Can

yon

80Tota

l#ux

100

5510

086

100

3510

014

010

072

100

205

Org

anic

mat

ter

4022

2118

248

1217

2115

1225

Car

bon

ate

137

1916

3412

4259

3324

2653

Opal

74

87

62

46

43

49

Litho

geni

c40

2252

4536

1342

5842

3058

118

200

Tota

l#ux

100

9210

081

100

5410

042

910

043

210

055

1O

rgan

icm

atte

r26

2417

1435

1918

779

389

52C

arbona

te23

2128

23N

ova

lue

3012

933

141

3016

4O

pal

88

76

74

522

624

527

Litho

geni

c43

3948

38N

ova

lue

4720

152

229

5630

860

0Tota

l#ux

100

4510

028

100

2710

018

210

080

110

075

3O

rgan

icm

atte

r13

68

215

47

136

488

59C

arbon

ate

6027

5516

205

2851

4233

227

201

Opal

73

72

41

48

435

537

Litho

geni

c20

930

861

1761

110

4838

660

456

900

Tota

l#ux

100

120

100

3810

032

100

3910

044

810

072

9O

rgan

icm

atte

r9

1111

413

410

46

288

55C

arbon

ate

2530

2510

289

218

2611

423

170


Opal

79

62

72

62

627

642

Litho

geni

c59

7058

2252

1763

2562

279

6346

2

Ope

nsl

ope

900

Tota

l#ux

100

1210

013

100

1210

017

100

2010

030

Org

anic

mat

ter

233

192

172

122

82

83

Car

bon

ate

71

192

No

valu

e34

632

614

4O

pal

No

valu

eN

ova

lue

No

valu

eN

ova

lue

51

113

Litho

geni

cN

ova

lue

No

valu

eN

ova

lue

No

valu

e55

1167

20


Fig. 2. (A) Time-series plots of total mass #ux (mg m~2 d~1) of particulate matter collected by sedimenttraps during the course of the ECOMARGE-II experiment in the Grand-Rho( ne Canyon (GRC) and theadjacent open slope (OS). (B) Depth evolution of the annual mean total mass #ux (mg m~2 d~1).

4.1. Total mass yux

Temporal variability of mass #uxes and the variation with depth and mooring siteof the annual mean #uxes are illustrated in Fig. 2. Total mass #uxes in the Grand-Rho( ne region presented three periods of high values (Fig. 2A):

(1) In spring (April}May 1988), with the highest #uxes of the year at all depths. Themaximum #ux value reached 3410 mg m~2 d~1 at 600 m depth. A secondary peakappeared in late May for the 900 m canyon trap.


(2) In summer, #uxes peaked in early August 1988 for the traps between 80 and600 m depth, and in late August for the two 900 m traps, with values around300}400 mg m~2 d~1.

(3) In winter (December 1988}January 1989), with values ranging from 500 to1000 mg m~2 d~1, except at 80 m, where no increase appeared. A smaller peak wasobserved, particularly at 900 m in the canyon, in late January 1988.

During the rest of the year, mass #uxes were generally low (less than300 mg m~2 d~1). Minimum values, sometimes only a few tens of mg m~2 d~1, werefound for all traps in late February}early March, in June}July, and October}Novem-ber. Except for the maximum in winter 1988}89, which was not observed on the openslope, the seasonal variation of near-bottom #uxes at both sites was similar.

Mean total mass #uxes exhibited a linear increase from 80 m down to 600 m, thena decrease towards 900 m depth (Fig. 2B). The mean #ux computed for the 900 m trapon the open slope was about half of that determined at the same depth in the canyon.

4.2. Organic carbon

The highest C03'

concentrations were observed in the samples from the shallowesttraps (80 and 200 m) (Fig. 3A). For these two traps, C

03'contents showed an annual

trend with concentrations increasing during the spring}summer period (a noticeablemaximum value of 35.4% was observed in the 80 m trap in early September), and lowconcentrations around 4}10% in winter and autumn. In the deeper traps (600 and900 m), C

03'contents were lower and variations were weaker over the year, with values

generally varying between 3 and 6%. Slight increases were nevertheless observed,especially in June}July and November, the latter being observed throughout thewater column.

Temporal variations of C03'#uxes (Fig. 3B) essentially di!ered from those of total

mass #uxes (Fig. 2A) by a large summer peak recorded at 80 m depth, which wasassociated with the highest C

03'concentrations. SEM analyses were used as a criterion

for the recognition of organic matter contributors in the 80 m trap. During spring,these analyses revealed an abundant fresh siliceous phytoplanktonic component, butassociated with an important siliciclastic fraction (Plate 1A). During summer (Plate1B), the high organic matter values corresponded to large non-crystalline, often"lamentous particles, fecal pellets and phytoplankton cells (diatoms and coc-colithophorids).

Annual #ux-weighted mean C03'

contents constantly decreased with depth (Fig. 3C),sharply between the 80 and the 200 m traps (from 13 to 6%), and then quite linearlydown to &4% at 900 m depth. Both traps at this level presented similar meancontents. The vertical variation of mean #uxes of C

03'di!ered from those of the other

major constituents and of total mass #ux. Due to the large C03'

surface #uxes duringsummer, the annual mean #ux at 80 m (32 mg m~2 d~1) was higher than the one at200 m depth (17 mg m~2 d~1). The mean #ux at 600 m increased again to a valueclose to the surface #ux (27 mg m~2 d~1). The lowest mean #ux appeared at 900 mdepth on the open slope (10 mg m~2 d~1).


Fig. 3. Time-series plots of the particulate organic carbon content (A) and #ux (B) as measured by sedimenttraps during the course of the ECOMARGE-II experiment in the Grand-Rho( ne Canyon (GRC) and theadjacent open slope (OS). Depth evolution of the #ux-weighted annual mean organic carbon content (%)and annual mean #ux (mg m~2 d~1) (C).


4.3. Inorganic carbon

C*/03'

contents exhibited limited temporal variations (1}4%) at all depths during the"rst semester, but large #uctuations in summer and early autumn (C

*/03'contents as

high as 7% at 600 m depth or as low as 0.6% at 80 m depth) (Fig. 4A). Fluxes ofC

*/03'quite exactly matched those of total mass #uxes (Fig. 4B).

Annual #ux-weighted mean C*/03'

concentrations (Fig. 4C) increased from 2.1% at80 m to 3.2% at 200 m depth, and decreased to about 2.8% at 900 m depth. Assumingthat all C

*/03'was in the form of calcium carbonate, these results show that carbonate

was the second most important contributor to downward particulate #uxes throughthe slope waters, with annual mean values ranging from 18 to 27%. Contrary to theother biogenic constituents, C

*/03'mean content in the near-bottom trap was lower on

the open slope than in the canyon. The evolution of annual mean C*/03'#uxes was very

similar to that of mean total mass #uxes and increased from a minimum of5 mg m~2 d~1 at 80 m depth to a maximum of &19 mg m~2 d~1 at 600 m depth.

SEM observations showed that C*/03'

was associated with three kinds of particles:carbonate micritic grains (most frequently observed in the deeper layers), coc-colithophorids and foraminifera. Fresh coccolithophorids were constantly present inthe surface layers. Foraminifera were also abundant, especially during the au-tumn/winter #ux peaks.

4.4. Opal

All traps showed a smooth, dome-shaped trend over the year with opal concentra-tions being higher and also more variable during June}July, with opal concentrationsas high as 26% (Fig. 5A). It has to be noted that these large #uctuations precededthose of C

*/03'concentrations previously described (Fig. 4A). During the rest of the

year, the concentrations varied within a limited range of 5}12% in winter-spring andof 4}7% in autumn. Again, time series of opal #uxes closely matched those of totalmass (Fig. 5B).

As for C*/03'

, the annual mean opal concentration increased from 7.2% at 80 m toa maximum of 8.8% at 200 m (Fig. 5C). In the deeper traps, mean opal contentremained quite constant (8.2 and 8.7%) but increased signi"cantly near the bottom onthe open slope (9.7%). Annual mean opal #uxes ranged from &18 mg m~2 d~1 at80 m to 48 mg m~2 d~1 at 600 m. The two near-bottom traps recorded the samedecrease as for total mass #uxes.

4.5. Siliciclastic residue

All traps exhibited the same temporal trend. The siliciclastic residue (or lithogenicfraction) contents decreased constantly during the "rst semester, and increased pro-gressively during the second semester to contents close to the beginning of theexperiment (Fig. 6A). The seasonal signal was enhanced for the shallower traps(80 and 200 m depth), which presented a summer decrease by a factor of 2}3.Deeper traps showed only a limited decrease. Individual lithogenic contents rarely



&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&bPlate 1. Scanning electron micrographs of selected samples collected by the 80 m depth sediment trapduring the ECOMARGE-II experiment in the region of the Grand-Rho( ne canyon showing variouscontributors to the biogenic #ux: (A) Spring period (April 7}25, 1988); fresh siliceous and calcareousphytoplankton, diluted by an important lithogenic fraction. (B) Summer period (August 2}20, 1988); long,"lamentous material, including fresh phytoplankton (coccolithophorids and diatoms). (C) Autumn period(October 14 }November 1, 1988); foraminifera, fecal pellets of copepods and large crustaceans (e.g., shrimpsand euphausiids) during the minimum of #uxes.

exceeded 50% at 80 m, but this fraction constituted the main part of total mass #uxesin the deeper layers (600 and 900 m, individual contents rarely below 50%). The 200 mtrap showed a mixed character, with values below 50% in summer and over 50%during the rest of the year.

Due to the fact that the lithogenic fraction generally represented the "rst contribu-tor to the #ux of settling particles, time series of lithogenic mass #uxes obviouslyexhibited, at all depths, the closest match with those of total mass #uxes (Fig. 6B).

Variations in the lithogenic content of particles was mainly related to theC

03'fraction in the upper traps, whereas for the deeper traps most of the variations

were in#uenced by changes in C*/03'

and opal concentrations. A correlation coe$cientof !0.85 between the lithogenic and C

03'fractions was found. Accordingly, the

variation of the mean annual lithogenic content with depth (Fig. 6C) was the inverse ofthat observed for the C

03'content (Fig. 3C). It constantly increased with depth in the

canyon from 48% at 80 m to 61% at 900 m depth. Due to large opal concentrations,the siliciclastic residue on the open slope was signi"cantly lower (55%). Mean annualsiliciclastic residue #uxes linearly increased from &120 mg m~2 d~1 at 80 m depth to330 mg m~2 d~1 at 600 m depth, before decreasing again in the near-bottom canyontraps.

5. Discussion

5.1. Sources of particulate matter at the entrance of the Gulf of Lions

The particulate material caught by the shallow traps (80}200 m) in the slope watercolumn at the entrance of the Gulf of Lions can be distinguished from that caught bythe deeper traps (600}900 m) on the basis of the temporal variability but also the meanvalue of major biogeochemical constituents of particles. The temporal variabilitydi!erences between the traps can be summarised by the coe$cient of variation(CV"standard deviation as % of the mean) of contents of the major constituentscalculated for each trap (Table 2). The results indicate that variations around meanvalues of organic carbon and siliciclastic residue contents broadly decreased withincreasing depth. The trend was less obvious for inorganic carbon and particularly foropal, which kept almost the same degree of variability throughout the water column.At 900 m depth, the chemical composition of particulate matter within the canyonwas more homogeneous than on the open slope. In terms of mean contents (Fig.3C}Fig. 6C), it appears that the 80 m trap presented the highest organic carbon


Fig. 4. Time-series plots of the particulate inorganic carbon content (A) and #ux (B) as measured bysediment traps during the course of the ECOMARGE-II experiment in the Grand-Rho( ne Canyon (GRC)and the adjacent open slope (OS). Depth evolution of the #ux-weighted annual mean inorganic carboncontent (%) and annual mean #ux (mg m~2 d~1) (C).


Fig. 5. Time series plots of the particulate opal content (A) and #ux (B) as measured by sediment trapsduring the course of the ECOMARGE-II experiment in the Grand-Rho( ne Canyon (GRC) and the adjacentopen slope (OS). Depth evolution of the #ux-weighted annual mean opal content (%) and annual mean #ux(mg m~2 d~1) (C).


Fig. 6. Time-series plots of the particulate lithogenic content (A) and #ux (B) as measured by sediment trapsduring the course of the ECOMARGE-II experiment in the Grand-Rho( ne Canyon (GRC) and the adjacentopen slope (OS). Depth evolution of the #ux-weighted annual mean lithogenic content (%) and annualmean #ux (mg m~2 d~1) (C).


Table 2Temporal variability of total mass #ux and contents of major constituents of the particulate mattercollected by the di!erent sediment traps during the ECOMARGE-II experiment. The variability aroundmean values is expressed by the coe$cient of variation (standard deviation/mean]100)

Depth (m) Organiccarbon (%)

Inorganiccarbon (%)

Opal (%) Siliciclasticresidue (%)

Canyon 80 57 45 34 40200 60 22 50 29600 29 44 31 22900 30 15 33 9

Open slope 900 42 30 26 14

content, but the lowest value for the three other major constituents. At the otherextreme, the near-bottom canyon trap presented the lowest organic carbon content,the highest content in siliciclastic residue and intermediate mean contents for inor-ganic carbon and opal. The two other canyon traps and the near-bottom open slopetrap presented intermediate characteristics: samples from the 200 m level were closerto surface samples, whereas the 600 m samples and those from near-bottom on theopen slope were closer to the samples collected at 900 m inside the canyon. Theseglobal features suggest the existence of at least two distinct sources of material andperhaps di!erent transfer processes, which progressively shift the nature of settlingparticles from a sub-surface material characterised by an important, but highlyvariable, biogenic contribution to a deep material, more homogenous in compositionand with a much stronger lithogenic contribution.

The moorings were located at the frontier of three major hydrological provinces ofthe Gulf of Lions: the Gulf of Marseille area on the northeastern part of the shelf, thedilution zone of the Rho( ne river plume west of its mouth, and the southern open seaarea bordered by the Northern Current (Lefevre et al., 1997). Productivity within themain nutrient-poor water mass of the Northern current is low, but within the frontalzone along its continental border (i.e., close to the mooring sites) primary productionis generally higher. According to Conan (1996) and Lefevre et al. (1997) the typicalseasonal evolution in this frontal zone is characterized by a spring bloom aroundApril that lasts generally 6 weeks and is initiated by the increase of insolation andthermal strati"cation. Secondary blooms appear in summer and autumn and arelinked to upwelled or river nutrient inputs.

The shallower trap at 80 m depth, located immediatly below the seasonal thermoc-line, exhibited a well marked biological component throughout the year (Fig. 7). Thehighest biogenic #ux peak was observed in April and concerned all three biogenicconstituents (organic matter, opal and carbonate). The sample contained largeamounts of fresh siliceous and calcareous phytoplankton (Plate 1A). It is neverthe-less noteworthy that this strong biogenic #ux event represented only a small part(25%) of the total settling material collected at that period. In contrast, biogenic#uxes in summer contributed up to 60}90% of the total mass #ux. Diatom and


Fig. 7. Time series plot of the particulate organic carbon, inorganic carbon and opal #uxes measured at80 m depth during the course of the ECOMARGE-II experiment in the Grand-Rho( ne Canyon.

coccolithophorid cells were also abundant as well as di!erent species of pelagicforaminifera (Plate 1B). The period of low #uxes following this summer peak showedan increased contribution of fecal pellets produced by large copepods and pelagiccrustaceans (e.g., shrimps and euphausiids), indicating an enhanced grazing activity(Plate 1C). The late November biogenic secondary peak contributed 50}60% of thetotal mass #ux and was largely related to increased numerical #uxes of foraminiferaand thus of carbonate inputs (M. Mahfouf, unpublished data). These features clearlyshow that the time series of biogenic #uxes at 80 m depth qualitatively matched thetypical seasonal variability of the biological activity of the Gulf of Lions slope waters.The biological contribution was also con"rmed by speci"c 210Po/210Pb ratios(Radakovitch et al., 1999a). These ratios were largely above unity over the whole year} a feature characteristic of biogenic material (Heyraud and Cherry, 1983) } and moreparticularly during the April and July peak of mass #uxes and also in early Octoberduring the lowest mass #uxes.

The two mass #ux peaks observed in July and November were largely associatedwith the settling of aggregates of phytoplanktonic cells through the thermocline,which probably scavenged lithogenic matter present in the surface layer. However, thesituation seems more complicated for the April biogenic #ux peak. Though it wasobserved during the typical spring bloom period, this peak was accompanied bya very high lithogenic #ux in the upper trap. Despite its short duration ()1 samplinginterval), this #ux event delivered a large fraction of the yearly #ux: in the 80 m trap,the 18-day sampling period of April represented 42% of the material collected duringthe whole experiment. Durrieu de Madron et al., 1999 showed that this period wascharacterized by peculiar climatological events (Rho( ne river #oods, seaward wind)that probably allowed large amounts of river material to be transported } eitherdirectly or through successive short steps of deposition/resuspension events } towardsthe Grand-Rho( ne canyon located south of the Rho( ne river mouth. This period was


further characterized by an intense dynamical activity (meandering of the along-slopecurrent) and weak strati"cation, which enhanced horizontal (shelf/slope) and verticalexchanges. The riverine origin of the particles during the April event was attested bythe neodymium isotopic composition of the 80 m trap material that was close to thatof the Rho( ne particulate material, whereas it di!ered during the rest of the year(Henry et al., 1994). The small biogenic component of the April #ux peak, at 80 mdepth but also in deeper waters, suggests that the particulate #uxes were relatedlargely to advection of lithogenic SPM from the shelf and to a lesser extent to thesettling of phytoplanktonic remains from the surface layer. The simultaneity of thesephysical and biological driving mechanisms is probably not fortuitous. As suggestedby Ittekot et al. (1991) for the Bay of Bengal, the spreading of nutrient-rich riverplumes favors phytoplanktonic blooms. As they settle, the produced biogenic particlesincorporate suspended lithogenic particles (silts and clays), which are also supplied bythe river, leading to enhanced biogenic and lithogenic #uxes.

The particulate matter collected in the deepest layers (600 and 900 m) exhibiteda prominent lithogenic contribution, representing more than 50% of the constituents.Though the material trapped at 600 m depth still showed an input of biogenicparticles during the summer and autumn #ux peaks, the more stable composition ofsettling particles suggests an additional and more homogeneous source of material.This source provides the bulk of the material caught at depth on the slope (inside andoutside the Grand-Rho( ne canyon) and dilutes the biological signal exported fromsur"cial waters. Several collateral observations suggest that the general additionalparticle source at depth could be of sedimentary origin, i.e. resuspended material.First, the covariation of contents in major constituents with total mass #ux (Fig. 8)reveals that the degree of scatter is highest when total mass #ux is low (it broadlyvaries within a factor of 5}7), but quite rapidly decreases when total mass #uxincreases. When the latter exceeds 500}1000 mg m~2 d~1, the chemical compositionof settling particles rapidly tends towards a narrow range of values: 3}5% for organiccarbon, &3% for inorganic carbon, 10% for opal, and 60}70% for the lithogenicfraction. These asymptotic values } slightly higher than those observed in sur"cialsediments from the northwestern Mediterranean, particularly regarding organic car-bon and opal } indicate that, at least for high mass #uxes, resuspended particlesrepresent a signi"cant contribution to the settling material collected by traps. Theresuspension process is further attested by the 210Po and 210Pb analyses presented inRadakovitch et al., 1999a (this issue). In the 600 m trap, a 210Po/210Pb ratio nearunity was observed all year round; such a ratio is characteristic of sediments (Heyraudand Cherry, 1983). Finally, microscopic examination of the samples showed that someof the foraminifera from the deeper trap samples were occasionally "lled with sedi-ment (M. Mahfouf, unpublished data).

All these observations now raise the question of the location of the source ofresuspended particles, which can only be hypothesized so far. In considering resuspen-sion processes as a large if not a major contributor to the #uxes observed at depth onthis margin, we exclude mechanisms of deep bottom resuspension such as thosedescribed in open ocean locations (e.g., Gardner et al., 1983; Au!ret et al., 1994).Episodic resuspension and subsequent lateral transport by the along-slope circulation


Fig. 8. Variation of the content of major constituents (%) with total mass #ux (mg m~2 d~1) of particulatematter collected by sediment traps during the course of the ECOMARGE-II experiment in the region of theGrand-Rho( ne canyon. Data from all traps were pooled together.

was shown by Durrieu de Madron (1994) and Durrieu de Madron et al., 1999. Theseauthors observed episodes of high current speeds at 600 m depth, especially inautumn, strong enough to resuspend sediments. They also noticed the presence ofintermediate nepheloid layers that detached from the open slope east to the Grand-Rho( ne canyon, at depths around 400}800 m, and were transported downstream overthe canyon, providing a source of particles to the deeper canyon traps. In a recentwork, Buscail et al. (1996) reported the presence of a speci"c amino acid ({-amino-glutaric acid) in oxic sur"cial sediments of the NW Mediterranean upper continentalslope. Preliminary analytical runs at the time of our experiment showed that thisamino acid was found only in the 600 and 900 m trap samples, a result that also arguesfor a contribution of reworked sedimentary material of upper slope origin (R. Buscail,personal communication).

Shelf particles can also be considered as important contibutors to slope #uxes ina way similar to what was observed in the SW part of the Gulf of Lions (Monaco et al.,1990a) or during the Shelf Edge Exchanges Processes (SEEP) experiments on theMiddle-Atlantic Bight (Biscaye and Anderson, 1994). These experiments demon-strated the role of climatic and hydrodynamic processes in resuspending shelf materialand subsequently transferring it to the slope environment. Repeated nephelometricsurveys show the existence of a continuous bottom nepheloid layer on the shelfadjacent to our trap sites that detaches from the bottom at the shelf-break and spreadsinto slope waters along isopycnals (e.g. AlomK si et al., 1982; Durrieu de Madron and


Panouse, 1996; Durrieu de Madron et al.). Even if pulses of fresh material aredeposited on the shelf, resuspended particles delivered to the slope may be homogene-ous in their composition, since bioturbation is active on the shelf. Indeed, 210Pbpro"les have revealed that surface sediments from the shelf are mixed over depths of4}10 cm (Zuo et al., 1991; Radakovitch et al., 1999b). By inducing a rapid mixing ofthe `fresh deposited materiala with old sediment and also by favoring resuspension(Davis, 1993), bioturbation may thus provide particles of homogeneous compositionable to be resuspended. In the slope water column, this source would not only easilydilute the biogenic signal issued from the upper water layers but would also homogen-ize and reduce the overall variability of its chemical composition.

5.2. Comparison with yuxes at the exit of the Gulf of Lions (ECOMARGE-I experiment)

The "rst ECOMARGE experiment was carried out during 1985}1986 in theLacaze-Duthiers canyon, on the Pyrenean continental slope, at a distance of approx-imately 60 nautical miles to the southwest (Fig. 1). Four traps were moored, at 50, 100,300 and 600 m depth, on a single line at 650 m depth in the axis of the canyon(Monaco et al., 1990b). Despite the limitations inherent to the comparison of experi-ments that took place 3 years apart and involving slightly di!erent mooring designs,#uxes measured at the two sites located at the NE entrance and the SW exit of theGulf of Lions (with respect to the general circulation in the gulf) displayed signi"cantquantitative di!erences that revealed and emphasized the role of the hydrodynamicconditions.

The mean annual mass #uxes in the Lacaze-Duthiers canyon were much strongerthan in the Grand-Rho( ne canyon, except around the shelf break depth (100}200 m),where #uxes were mimimal and comparable (Fig. 9). From the entrance to the exit ofthe Gulf of Lions, the near-surface mass #ux increases by a factor of about three, anda ten-fold increase is observed at the depth of 600 m. Qualitatively, the materialtrapped at equivalent depths at both sites (80}100 and 600 m) presented similarchemical characteristics, underlining the predominance of lithogenic material over theentire Gulf of Lions. The fundamental characteristics that distinguish the two sites aretherefore merely expressed in quantitative terms rather than in terms of chemicalcomposition.

The quantitative aspect of this comparison was quite expected since some majorclimatic and hydrodynamic processes } that either increase or control the amount andthe transport of particulate material in shelf waters } di!er between the two sites. Thecirculation on the shelf is globally cyclonic and entrains the suspended mattersupplied by the Rho( ne river and smaller local rivers preferentially southwestward. Inthe same way, particles resuspended from shelf deposits will be preferentially en-trained towards the southwestern part of the gulf. A direct relationship betweenparticle #uxes and climatic forcings was actually inferred in the Lacaze-Duthierscanyon (Monaco et al., 1990b). Mass #ux events were correlated with rainy andstormy periods during autumn and winter that favored both river discharge andsediment resuspension. In the contrast, on the Rho( ne margin, correlations with thelocal climatology (Rho( ne river discharge, winds) were only occasional (Durrieu de


Fig. 9. Comparison of annual mean total mass (A) and organic carbon (B) #ux pro"les with depth ofparticulate material collected at the entrance of the Gulf of Lions (present study) and at the exit, in theLacaze-Duthiers canyon (data from the ECOMARGE-I experiment from July 1985 to April 1986; afterMonaco et al., 1990b).

Madron et al., 1999). Furthermore, Durrieu de Madron and Panouse (1996) traced thecascading of cold and turbid shelf water over the slope down to a few hundred metersdepth, and showed that this export was more intense on the western part of the Gulf ofLions than on the eastern part. This enhanced southwestern export likely resultedfrom an increased dense water formation and spreading on the western shelf, linked tohigher northwesterly wind frequencies and the presence of wind-induced downwell-ings along the North}South Pyrenean coast. All these processes converge in sucha way that the southwestern exit of the Gulf of Lions acts as an outlet for the entiregulf } which is of course probably not restricted to the Lacaze-Duthiers region but


encompasses a more or less larger portion of the southwestern slope }through which a signi"cant amount of the material introduced into shelf watersescapes to the slope. Durrieu de Madron (1991) e!ectively observed an increase to thesouthwest of the SPM standing stock along the slope. This author estimated, fromrepeated nephelometric surveys, that the relative SPM stock in slope waters regularlyincreased to the southwest by a factor of 1.9 to 7.5 (depending on the season) betweenthe Grand-Rho( ne canyon and the Lacaze-Duthiers canyon.

These results are very similar to those obtained during the SEEP experiments(Biscaye et al., 1994). The overall drift of shelf waters of the NE}SW orientedMiddle-Atlantic-Bight (MAB) is to the southwest. Along the northern MAB (o!NewEngland), the particulate matter exits the shelf mostly by di!usive exchange across theshelf-edge front (e.g. mixing of slope water with the surrounding shelf water duringstrong intrusion of o!shore water onto the shelf). The particulate matter not exportedby di!usive exchange is advected southwestward along the shore and runs o! the shelfalong the southern MAB to the slope waters or the Gulf Stream. The long-term average particulate and organic carbon #uxes on the southern MAB are largerthan in the northern MAB and present a strong increase with depth (Biscaye andAnderson, 1994) indicative of a lateral transport of particles to the slope waters. Themajor #ux episodes were also related to resuspension of shelf sediment mainly causedby storms.

The comparaison of the results from MAB and Gulf of Lions experiments clearlyshows the spatial disparity of the export of particulate matter due to the geometry ofthe shelf, the continental water discharge, the overall drift of the shelf waters andthe shelf-slope exchange processes. It has to be noted here that the increase with depthof the total mass and organic carbon #uxes is a recurrent feature of particle transferon continental margins observed world-wide, e.g., in the Panama Basin (Honjoet al., 1982,1992), the Santa Monica Basin o! Los Angeles (Nelson et al., 1987;Huh et al., 1990), the northeastern Paci"c (Lao et al., 1993), the northeastern Atlantic(Heussner, 1995; Etcheber et al., 1996), the Okinawa Trough (Narita et al., 1990) andthe MASFLEX site in the East China Sea (Iseki et al., 1995), the Black Sea (Buesseleret al., 1990), and several locations throughout the entire Mediterranean Basin (Heu-ssner et al., 1997). Flux increase with depth characterizes exporting margins andrepresents the contrasting feature with typical open ocean #uxes that generallydecrease with depth. Nevertheless, despite the abundant evidence accumulated theunderlying fundamental physical mechanisms still remain poorly understood.

6. Conclusions

Several conclusions may be drawn from this one year study of biogeochemicalparticulate #uxes in the region of the Grand-Rho( ne canyon, on the northeastern sideof the Gulf of Lions (northwestern Mediterranean):

(1) Mean total mass #uxes slightly increased with depth inside the canyon, a resultthat indicates lateral transport of particulate material that is added to the material


exported from the sur"cial water layers. This feature is quite typical of #uxes incontinental margin environments, though it is less marked in this part of the continen-tal slope than elsewhere. The near-bottom trap inside the canyon collected about2 times more material than its counterpart deployed at equivalent depth on the openslope, indicating a preferential transport of material within the canyon.

(2) The di!erences in particulate composition between the sub-surface (80 m) anddeeper (200}900 m) traps indicated the existence of distinct sources of material. Theshallower trap showed a clear biological signal; #ux peaks in summer and autumnwere largely related to periods of surface biological production. This biogenic contri-bution was con"rmed by SEM observations and results from natural radionuclidestudies (high 210Po/210Pb ratio; Radakovitch et al., 1999). The particulate mattertrapped at deeper levels inside the canyon but also on the open slope was character-ized by a more stable composition with a major lithogenic contribution. The biogenic#uxes immediatly below the thermocline (trap at 80 m depth) varied seasonally by oneorder of magnitude and co-varied qualitatively with the typical cycle of primaryproductivity in the surface slope waters of the Gulf of Lions.

(3) Several independent arguments (reduced compositional variability at highermass #uxes, presence of a sediment related speci"c amino acid ({-aminoglutaricacid) only in the 600 and 900 m samples, foraminifera "lled with sediment and nearunity of 210Po/210Pb ratios) support the hypothesis of an input of resuspendedmaterial that explains both the observed increase of total mass #ux with depth anddecrease of the variability in chemical composition. This source of resuspendedsediment probably fed the slope throughout the year. Occasionnally, strong lithogenic#uxes a!ected the entire water column } as for the April major #ux event } thatprobably originated in riverine material more or less directly transported over theshelf into slope waters.

(4) The #ux variations at depth were largely in#uenced by the cross-slope #uctu-ations of currents. This mechanism was particularly e!ective throughout the watercolumn, in late winter/early spring as the weak strati"cation favors vertical mixings(Durrieu de Madron et al., 1999).

The present experiment at the entrance of the Gulf of Lions, with respect tothe general circulation of water masses, completes a similar experience performedthree years earlier in the Lacaze-Duthiers canyon at the SW exit of the gulf (Monacoet al., 1990b). The comparison of the main characteristics of particle transferat both sites emphasizes the strong di!erences in the intensity of particle #uxes.Annual #uxes are 3}10 times higher at the exit of the Gulf of Lions than atthe entrance, particularly below the shelf break depth. Such marked di!erencesare related to changes in the local hydrodynamic conditions, that favours shelf-slope exchanges at the SW exit of the system. Di!erences in the chemical composition(major constituents) of the material transferred are, on the contrary, quite limited.This result suggests (i) that the particle reservoir feeding shelf-slope exchanges ishomogenous at the scale of the entire Gulf of Lions and (ii) that the importanceof carbon export to slope waters on this margin is controlled, above all, byhydrodynamic factors.


Acknowledgements

This research was supported by CNRS-INSU as a contribution to theECOMARGE programme (France-JGOFS). We are grateful to B. Deniaux forperforming the SEM analyses, to R. Buscail for giving us access to her unpublishedresults on amino acids in trap samples and to M. Mahfouf for the foraminifera data.We warmly thank the o$cers and crews of RVs Le NoronL t, Le SuronL t from IFREMERGenavir (France), and Aegaio from NCMR Athens (Greece) for assistance withmooring deployments.

References

Aloisi, J.C., Cambon, J.P., Carbonne, J., Cauwet, G., Millot, C., Monaco, A., Pauc, H., 1982. Origine et ro( ledu neH pheH loide profond dans le transfert des particules en milieu marin. Application au Golfe du Lion.Oceanologica Acta 5, 481}491.

Au!ret, G., Khripouno!, A., Vangriesheim, A., 1994. Rapid post-bloom resuspension in the northeasternAtlantic. Deep-Sea Research I 41, 925}939.

Beckers, J.M., Brasseur, P., Nihoul, J.C.J., 1997. Circulation of the western Mediterranean: from global toregional scales. Deep-Sea Research II 44, 531}550.

Biscaye, P.E., Anderson, R.F., 1994. Fluxes of particulate matter on the slope of the southern MiddleAtlantic Bight: SEEP-II. Deep-Sea Research II 41, 459}510.

Biscaye, P.E., Flagg, C.N., Falkowski, P.G., 1994. The Shelf Edge Exchange Processes experiment, SEEP-II:an introduction to hypotheses, results and conclusions. Deep-Sea Research II 41, 231}252.

Buesseler, K.O., Livingston, H.D., Honjo, S., Hay, B.J., Konuk, T., Kempe, S., 1990. Scavenging and particledeposition in the southwestern Black Sea } evidence from Chernobyl radiotracers. Deep-Sea Research37, 413}430.

Buscail, R., Gadel, F., Kerherve, P., Serve, L., Charriere, B., Faganeli, J., Heussner, S. Mazaudier, D., 1996.Source and transformation of organic matter: biomarkers in the settling particles and the surfacesediments in the North Balearic basin. In: Wassman, P., Tselepides, T. (Eds.), Second Workshop of theMediterranean Targeted Project, February 1}3, Iraklion, Crete. pp. 91}101.

Conan, P., 1996. VariabiliteH et bilan de la production primaire en zone co( tiere (MeH diterraneH e nordoccidentale; entreH e du Golfe du Lion) en relation avec les systemes biologique, chimique et hy-drodynamique (courant nord-meH diterraneH en). Ph. D. Thesis, UniversiteH de la MeH diterraneH e, pp. 205.

Davis, W.R., 1993. The role of bioturbation in sediment resuspension and its interaction with physicalshearing. Journal Experimental of Marine Biology and Ecology 171, 187}200.

Durrieu de Madron, X., 1991. Transport de la matiere particulaire en suspension sur les marges continen-tales du golfe du Lion (MeH diterraneH e occidentale), du bassin des Sporades (MeH diterraneH e orientale) et dugolfe de gascogne (Atlantique Nord-est). Ph. D. Thesis, University of Neucha( tel, 159 pp.

Durrieu de Madron, X., 1994. Hydrography and nepheloid structures in the Grand-Rho( ne canyon.Continental Shelf Research 14, 457}477.

Durrieu de Madron, X., Panouse, M., 1996. Transport de matiere particulaire en suspension sur le plateaucontinental du Golfe du Lion: Situation estivale et hivernale. Comptes Rendus de l'AcadeHmie desSciences, Paris 322, 1061}1070.

Durrieu de Madron, X., Ny!eler, F., Godet, C.H., 1990. Hydrographic structure and nepheloid spatialdistribution in the Gulf of Lion continental margin. Continental Shelf Research 10, 915}930.

Durrieu de Madron, X., Radakovitch, O., Heussner, S., Monaco, A., Loye-Pilot, M.D., 1999. Role of theclimatological and current variability on the shelf-slope exchanges of particulate matter. Evidence fromthe Rho( ne continental margin (NW Mediterranean). Deep-Sea Research I 46, 1513}1538.

Etcheber, H., Heussner, S., Weber, O., Dinet, A., Durrieu De Madron, X., Monaco, A., Buscail, R., Miquel,J.C., 1996. Organic carbon #uxes and sediment biogeochemistry on the French Mediterannean and


Atlantic margins. In: Ittekot, V., SchaK fer, P., Honjo, S., Deptris, P.J. (Eds.) Particle Flux in the Ocean.SCOPE Report 57. Wiley, Chichester, pp. 223}241.

Gardner, W.D., 1985. The e!ect of tilt on sediment trap e$ciency. Deep-Sea Research 32, 349}361.Gardner, W.D., Richardson, M.J., Hinga, K.R., Biscaye, P.E., 1983. Resuspension measured with sediment

traps in a high-energy environment. Earth and Planetary Science Letters 66, 262}278.Henry, F., Jeandel, C., DupreH , B., Minster, J.-F., 1994. Particulate and dissolved Nd in the western

Mediterranean Sea: sources, fate and budget. Marine Chemistry 45, 283}306.Heussner, S., 1995. Source and fate of settling particles on a northeastern Atlantic margin: recent issues from

the ECOMARGE program (France-JGOFS). In: Global Fluxes of Carbon and its Related Substancesin the Coastal Sea}Ocean}Atmosphere System. Proceedings of the 1994 Sapporo IGBP Symposium,pp. 183}188.

Heussner, S., Ratti, C., Carbonne, J., 1990. The PPS 3 time-series sediment traps and the trap sampleprocessing techniques used during the ECOMARGE experiment. Continental Shelf Research 10,943}958.

Heussner, S., Monaco, A., Durrieu De Madron, X., 1997. Particle #ux in the Mediterranean Sea: howimportant is the 'biological pump'? In: International Conference Progress in Oceanography of theMediterranean Sea, Rome, November 17}19, 1997, Sponsored and co-organized by the EC MarineScience and Technology Programme. Abstracts volume, pp. 37}38.

Heyraud, M., Cherry, R.D., 1983. Correlation of 210Po and 210Pb enrichments at the sea-surface microlayerwith neuston biomass. Continental Shelf Research 1, 283}293.

Honjo, S., Spencer, D.W., Farrington, J.W., 1982. Deep advective transport of lithogenic particles inPanama Basin. Science 216, 516}518.

Honjo, S., Spencer, D.W., Gardner, W.D., 1992. A sediment trap intercomparison experiment in thePanama Basin, 1979. Deep-Sea Research 39, 333}358.

Huh, C.A., Small, L.F., Niemnil, S., Finney, B.P., Hickey, B.M., Kachel, N.B., Gorsline, D.S., Williams,P.M., 1990. Sedimentation dynamics in the Santa Monica-San Pedro Basin o! Los Angeles:radiochemical, sediment trap and transmissometer studies. Continental Shelf Research 10, 137}164.

Iseki, K., Okamura, K., Tsuchiya, Y., 1995. Seasonal variability in particle distributions and #uxes in theEast China Sea. In: Global Fluxes of Carbon and Its Related Substances in the CoastalSea}Ocean}Atmosphere System. Proceedings of the 1994 Sapporo IGBP Symposium, pp. 189}197.

Ittekot, V., Nair, R.R., Honjo, S., Ramaswamy, V., Bartsch, M., Manganini, S.J., Desai, B.N., 1991.Enhanced particle #uxes in Bay of Bengal induced by injection of water. Nature 351, 385}387.

Lao, Y., Anderson, R.F., Broecker, W.S., Hofmann, H.J., Wol#i, W., 1993. Particulate #uxes of 230Th,231Pa, and 10Be in the northeastern Paci"c Ocean. Geochimica et Cosmochimica Acta 57, 205}217.

Lefevre, D., Minas, H.J., Minas, M., Robinson, C., Le, P.J., Williams, B., Woodward, E.M.S., 1997. Reviewof gross community production, primary production, net community production and dark communityrespiration in the Gulf of Lions. Deep-Sea Research II 44, 801}832.

LOICZ (Land-Ocean Interactions In The Coastal Zone), 1993. Science plan. In: Holligan, P.M., De Boois,H. (Eds.), Report 25, 27 pp.

Millot, C., 1990. The Gulf of Lion's hydrodynamics. Continental Shelf Research 10, 885}894.Monaco, A., Biscaye, P.E., Soyer, J., Pocklington, R., Heussner, S., 1990a. Particle #uxes and ecosystem

response on a continental margin: the 1985}1988 Mediterranean ECOMARGE experiment. Continen-tal Shelf Research 10, 809}839.

Monaco, A., Courp, T., Heussner, S., Carbonne, J., Fowler, S.W., Deniaux, B., 1990b. Seasonality andcomposition of particulate #uxes during ECOMARGE-I, western Gulf of Lion. Continental ShelfResearch 10, 959}988.

Mortlock, R.A., Froelich, P.N., 1989. A simple method for the rapid determination of biogenic opal inpelagic marine sediments. Deep-Sea Research 36, 1415}1426.

Narita, H., Harada, K., Tsunogai, S., 1990. Lateral transport of sediment particles in the Okinawa Troughdetermined by natural radionuclides. Geochemical Journal 24, 207}216.

Nelson, J.R., Beers, J.R., Eppley, R.W., Jackson, G.A., McCarthy, J.J., Soutar, A., 1987. A particle #ux studyin the Santa Monica-San Pedro Basin o! Los Angeles: particle #ux, primary production, and trans-missometer survey. Continental Shelf Research 7, 307}328.


Pont, D., 1997. Les deH bits solides du Rho( ne a proximiteH de son embouchure: donneH es reH centes (1994}1995).Revue de GeH ographie de Lyon 72, 23}33.

Radakovitch, O., Cherry, R.D., Heussner, S., 1999a. 210Pb and 210Po: tracers of particle transfer on theRho( ne continental margin (NW Mediterranean). Deep-Sea Research I 46, 1539}1563.

Radakovitch, O., Charmasson, S., Arnaud, M., Bouisset, P., 1999b. 210Pb and caesium accumulation in theRho( ne delta sediments. Estuarine, Coastal and Shelf Research 48, 77}92.

Zuo, Z., Eisma, D., Berger, G.W., 1991. Determination of sediment accumulation and mixing rates in theGulf of Lions. Mediterranean Sea. Oceanologica Acta 14, 253}262.


Origin and variability of downward biogeochemical fluxes on the Rhone continental margin (NW mediterranean)

Documents