Journalof Marine Research, 59, 633–656, 2001

Particle uxinthe Rainbow hydrothermal vent Ž eld (Mid-AtlanticRidge): Dynamics, mineral and biological composition

byAlexis Khripounoff 1,Annick Vangriesheim 1,PhilippeCrassous 1, MichelSegonzac 1,Ana Colac¸o 2,DanielDesbruye `res 1 andRoxane Barthelemy 3

ABSTRACT Inorder to provideinformation about the export and the distribution of hydrothermalparticulate materialto the surrounding deep ocean, four moorings were deployed in the vicinity of the hydrothermalRainbow vent Ž eld(Mid-Atlantic Ridge, 36° 14 9N,2250m depth).The Ž rstmooring wasa sedimenttrap with a currentmeter deployed at 2 mfroma chimneyof the Rainbow vent Ž eld and1.5 m abovethe bottom(a.b.) for 16 days. It representedthe reference for the initial composition ofparticles produced by the vent. The total mean mass particle  ux(6.9 g m 22 d21 )wasdistinctly higherthan the  uxmeasured at the shallower hydrothermal vents on the MAR segment.This particulate uxshoweda hightemporal variation at thescale of afewdaysand was characterized by ahighconcentration of sulphur(17.2%) and copper (3.5%) and a verylow concentrationof organic carbon(0.14%). Several hundred bivalve larvae belonging to thehydrothermalmytilid Bathymodio- lusazoricus werecollected in this trap at the beginning of the experiment. The density of larvae decreasedstrongly at theend, indicating a patchinessdistribution or adiscontinuousreproduction of thisspecies. The other three moorings, including sediment traps, current-meters and thermistor chains,were deployed for 304 days at differentdistances and altitudes from the Rainbow vent Ž eld. Themean speed of the current in the rift valley was low (6 cm s 21 )andwas oriented toward the north.The total mean particle mass  uxmeasuredwith the Ž vesedimenttraps varied little, from 10.6 to 25.0 mg m m22 d21 ,anddisplayed temporal variations which are typical of deep-sea environ- mentswith seasonal changes in the overlying production. However, in the trap at 500 m fromthe vents150 m a.b.,the presence of the hydrothermal plume can be observed: the sulphur, iron and copperconcentrations of particleswere signiŽ cantly higher compared to the particles sampled in the pelagicreference trap. The plume composition was about 50% hydrothermal particles and 50% pelagicparticles and its upper limit reached 300 m a.b.at this distance. In thetraps at 1000 m from thevents, the elemental composition of particleswas similar to thepelagic particles and we assume thatthese traps were not in the plume during the experiment. The zooplankton obtained in the long-termtrap samples revealed high density variations in relationto thedistance from the vent site. Thenutrient enrichment around the hydrothermal area and the abundance of free living bacteria explainthese variations in zooplankton density.

1.IFREMER Centre deBrest, De´partement DRO/EP,BP70,29280 Plouzane ´,France. email: [email protected] 2.IMAR, Laborato ´rioMaritimo da Guia,Faculdade da Ciencias deLisboa,Estrada doGuincho,2750 Cascais, Portugal. 3.Universite ´deProvence, Laboratoire de Biologie Animale, 3place VictorHugo, 13331 Marseille Cedex3, France. 633 634 Journalof MarineResearch [59, 4

1.Introduction Themost spectacular manifestations of hydrothermalactivity are the high-temperature ventŽ elds,which create particle-rich plumes rising hundreds of metersabove the sea  oor. Theproduction of this particulate material and its dispersal to the deep ocean mostly dependson theemission of vent uidsthat are precipitated as Žneparticlescarried by the neutrallybuoyant hydrothermal plume. The hydrothermal plumes, transported by thedeep currents,can be detectedup to100 km from theemitting vent (Dymond and Roth, 1988; Germanand Sparks, 1993). The study of plumesis nowan importantpart of hydrothermal researchbecause of the in uence of vent  uidson the ocean chemical balance (see ElderŽeld and Schultz, 1996). Hydrothermal activity also supports the production of biomasssuch as planktonic larval stages or juvenilesof vent species. These are carried by thenear-bottom currents and/ orthebuoyant hydrothermal plume and are then transported severalhundred meters above the bottom (Herring and Dixon, 1998; Kim et al., 1994; Kim andMullineaux, 1998; Mullineaux and France, 1995; Mullineaux et al., 1995).Through theseprocesses, hydrothermal plumes participate in the dispersal of ventspecies along the mid-oceanridges and constitute a possiblefood source for thesurrounding deep-sea fauna. Theaim of this work is to assess the in uence of the hydrothermal plume in the surroundingocean. The Rainbow hydrothermal Ž eld(36° 15 9N)was chosenbecause it producesthe strongest plume of itskind known on theMid-Atlantic Ridge (German et al., 1996;Fouquet et al., 1997),making it an ideal natural laboratory to study plume processes. Toprovideinformation about production and exportof hydrothermalmaterial by ventsand plumesto the deep ocean, four moorings with sediment traps, current meters and thermistorchains were deployedat differentdistances from theRainbow site for periods varyingfrom 16daysto 10 months.

2.Materials and methods a.Study area (Fig. 1a) TheRainbow site (36° 14 9N, 33°549W,2250m depth)was detectedon the Azores Mid-AtlanticRidge segment (AMAR) in1996 (German et al., 1996)and observed by submersiblein 1997 (Fouquet et al., 1997,2001; German et al., 2001).It is situated on the ankof araisedblock of ultramaŽc basementin the discontinuity between two segments of theridge and produces a strongand persistent hydrothermal plume. It is oneof themost activehydrothermal vent Ž eldsdiscovered in the North Atlantic with at least ten major groupsof extremelyactive black smokers without visible fauna. Some diffusive vents are present,located 25– 100 m from blacksmokers with small mussel beds and shrimp swarms (Desbruye`res et al., 2001). b. Sampling i.Mooring S (Short-termmooring with small sediment traps). OntheRainbow hydrother- malvent Ž eld(Fig. 1a), one triplicate sediment trap frame (trapS) was deployed 2001] Khripounoffet al.:Hydrothermal particulate material 635

Figure1. (a) Bathymetric map showing the location of theRainbow Ž eldwith the positions of the differentmoorings. (b) Scheme of thelong-termmoorings.

(36°13.8N, 33° 54.1W) during the Marvel cruise (August 1997) by thesubmersible Nautile for 16daysat 2 minthesouthwest of anactive vent Ž eld(2275 m depth).The description ofthe deployment method is detailed in Khripounoff and Alberic (1991). The top- collectingsurface of traps was situated1.5 m abovethe bottom (a.b.), while an Aanderaa current-meterwith a recordinginterval of 5 minuteswas positioned2 mabovethe trap. Thesettling particles were collectedin 3 epoxyŽ breglasstraps of cylindrical-conic shape with5 collectionbottles each and assembled (S trap)on a singleframe (Khripounoffand Alberic,1991). Only 4 collectorswere usedwith a samplingperiod of 4 dayseach. Each sedimenttrap had a samplingaperture of 0.07 m 2 andwas coveredwith a honeycomb 636 Journalof MarineResearch [59, 4 bafe of1 cmdiameter and 10 cmdeepcells at the top. Prior todeployment, the sampling bottlesof twotraps were Žlledwith Ž lteredseawater containing sodium borate buffered formalinto give a Žnalconcentration of 3% in order to prevent in situ microbial decomposition.The bottles of the third trap were Žlledwith a buffercomposed of 5M NaCl,10% DMSO and0.001% bromophenol blue to prevent in situ DNA degradation (Comtet et al., 2000).After recovery,the samples were storedin the dark at 4° C pending analyses. ii.Moorings L (Long-termmoorings with large traps). Threemoorings (Fig. 1b) with sedimenttraps, current meters and thermistor chains were deployedon theRainbow Ž eld duringthe Marvel cruise (August 1997) and recovered during the Flame 2 cruise(June 1998).The settling particles were collectedover 304 days using cone-shaped traps (PPS5 model,Technicapž ) with22 collection bottles. These traps had a samplingaperture of 1 m2 andwere coveredwith a honeycombbaf e of1cmdiameter and 10 cmdeepcells at thetop. As inthe S mooring,the sampling bottles were Žlledwith Ž lteredseawater and sodiumborate buffered formalin to givea Žnalconcentration of 3%.The sampling periods were 14dayseach except for thelast collector (10 days). Theaim of theŽ rst mooring(H500) was tosample the rising hydrothermal plume. Its positionwas selectedaccording to theCTD resultsobtained in this area by German et al. (1998).It was deployed ;500m tothe north of the Rainbow vents at 2300 m depth (M500)with two traps (150 m and300 m a.b.),two current-meters and a thermistorchain inbetween.Unfortunately, the trap at 150m a.b.did not work properly between February andApril (7 collectors). The location (34° 14.0N, 33° 54.0W) and depth are those obtained onthevessel at the mooring launching (Fig. 1). Due to a possibledrift during the mooring descent,its exact location on the bottom and its depth are given at 6100m. Thesecond mooring(M1000), with the same conŽ guration as the Ž rst,was usedto studythe neutrally buoyanthydrothermal plume. It was deployed ;1000m (36°14.2N, 33° 53.6W) to the north-northeastof thevents at 2250m depth.The last mooring (M Pelagic),with only one trapat 200m a.b.and one current-meter, was positioned(36° 13.3N, 33° 52.8W) at 1950 m depth,away from theaxis at ;2kmtothesoutheast of thevents. It was intendedto sample thepelagic  uxonlyas areference(Fig. 1). On each of thethree moorings, an Aanderaa current-meterwas installed10 maboveeach trap. The current-meters were equippedwith anarrowrange temperature sensor (resolution: 0.006° C) anda pressuresensor. The current andits direction as well as temperature and pressure were recordedevery hour. For the mooringsM500 and M1000, thermistor chains, consisting of ten thermistors (narrow range),10 mapart,were Žxedbetween the two traps, i.e. between 210 m and310 m above thebottom. All the current, temperature and pressure sensors were calibratedbefore and afterthe experiment. c.Current data Time-seriesanalysis of currentdata has been performed following methods as described inEmery and Thomson (1998). To examine the  uctuationsof the residual current and 2001] Khripounoffet al.:Hydrothermal particulate material 637

Table1. Current data obtained with current-meters Ž xedon theshort(Mooring S) andthelong-term moorings(M Pelagic,M500 and M1000). Residual Residual Duration Altitude current current Moorings (days) (a.b.) directionMean speed Max. speed speed Mooring S 16 1.5 m N330° 9 cm s2 1 19 cm s21 2.9 cm s21 MPelagic304 210 m N310° 5 cm s2 1 14 cm s21 1.1 cm s21 M500 304 310 m N005° 6 cm s2 1 17 cm s21 3.1 cm s21 M500 304 160 m N020° 6 cm s2 1 18 cm s21 4.5 cm s21 M1000 304 310 m N355° 6 cm s2 1 17 cm s21 1.5 cm s21 M1000 304 160 m N355° 6 cm s2 1 18 cm s21 3.4 cm s21

temperature,the instantaneous current data were low-passŽ ltered(Lanczos Ž lter,cut-off periodof 48 hours) to remove the tidal and inertial oscillations. The calculation of the eastwardand northward current components (current vector components on theeastward andnorthward directions taken as x and y axis)and statistics (current, temperature, pressure)were madefrom theŽ lteredand unŽ ltered data. d.Faunal sorting and particle analysis Inthe laboratory,each sample from thetraps was sievedand the fraction . 250 mm was examinedunder a dissectingmicroscope to sort and to count the organisms (particles smallerthan 250 mmcontainednegligible proportion of nauplii and juvenile copepods). Theshell hinge structure was examinedusing a PhilipsXL30 Scanning Electron Micro- scopeto identifybivalve larvae (Comtet et al., 2000).Then, the remaining particles were rinsedwith Milli-Q puriŽ ed water(pH ; 7),freeze-dried and weighed. Total sulphur and carbonwere determinedin duplicate with a LecoCS-125 auto-analyzer. Organic carbon concentrationwas measuredwith a LecoWR12 elemental analyzer after removing carbonateswith a 2NHCl solution (Weliky et al., 1983).Inorganic carbon content was calculatedas the difference between total and organic carbon. Analysis of the chemical compositionof particleswas undertakenby EDAX 1DX-4i X-ray spectrometry.Standards were preparedin the laboratory from purechemical compounds. Particles and standards were stronglycompressed (5 tons) to obtain a pelletof 3mm indiameter with a very at surface.The averaged accuracy of the analysis was 15%for theelements with a concentration . 1%.Carbon and nitrogen stable isotope analyses were performedusing a dualinlet Isotope Ratio Mass Spectrometer (Finnigan Mat Delta E) coupledwith a Carlo ErbaNA 1500auto-analyzer following the method described in Marguillier et al. (1997).

3. Results a.Currents and temperatures Closeto the vent, maximum current speed was 19cm s 21,andthe mean speed was 9 cm s21 duringthe 16 daysof experiment (Table 1). Currents had strong semi-diurnal oscillations(oriented to N040° – N220°) whichwere responsiblefor mostof thespeed and 638 Journalof MarineResearch [59, 4 directionvariability (Fig. 2a). A residualcurrent of 2.9cm s 21 was superimposedon the short-term uctuations.During the 16 daysof the experiment, its direction was almost steadilytoward N330, perpendicular to the semi-diurnal current (Fig. 2a). The temperature uctuatedbetween 3.6° C and10.1° C. Thehighest temperatures were observedwhen the currentdirection was betweenN070° and N295° (Fig. 3) corresponding to the periods whenthe trap was underthe venteffect. The temperatureanomalies $ 4°C recordedby the current-meterwere observed11% of thetime. Currentswere alsodominated by the semi-diurnaloscillations oriented to N045°– N225° atM500and M1000 (160 m and310 m a.b.)and N070° – N250°at MPelagic(210 m a.b.) (currentdata are summarized in Table1). At themooring situated 2 kmfrom thevent (M Pelagic),the current was low(mean speed was 5cms 2 1 andmaximum speed was 14 cm s21)andthe direction was morevariable which led to a muchlower residual current (1.1 cm s21)towardthe northwest (Fig. 4). This result indicates that this mooring, chosen asa pelagicreference, was outof the in uence of the hydrothermal vents during the experiment. At themoorings situated 500 m (M500)and 1000 m (M1000)from theRainbow vents, themean current speed was around6 cms 21 whilethe maximumspeed reached 18 cms 21 (Table1). The currents measured at both moorings were similarat thetwo levels (310 m a.b.and 160 m a.b.).The residual current (Fig. 4) was steadilytoward the north at the upper leveland veered slightly to the east at the deepest levels (parallel to thelocal bathymetry). Theresidual current was higherat thedeepest levels and at the mooring nearest to thevent (4.5and 3.4 cm s 21 at160 m a.b.and 3.1 and 1.5 cm s 21 at310 m a.b.,at M500 and M1000,respectively). The steady direction was favorablefor advectionof the vent emissionstoward the traps of thesetwo moorings. Duringthis experiment, no temperature anomaly exceeding the background of the naturaltemperature variations was recorded.No variation could be connectedto vent input. Allthe temperature records exhibited semi-diurnal oscillations. These oscillations were alwayshigher at 310 m a.b.than at 160 m a.b.(maximum amplitudes: 0.5° C and0.3° C, respectively).The temperature minimum values appeared when the current was oriented northward;i.e., when the current went from thevent to the moorings. This observation suggeststhat the temperature oscillations were onlydue to the background internal tides andnot to thevent in uence. b.Total mass particle uxes Near theRainbow vents (trap S), meanparticle  uxwas 6.9g m 22 d21 (range0.28 – 19.2 g m22 d21).Figure5a showsthe high temporal variations of themass uxesduring the16 daysof the experiment. During the Ž rst andthird sampling periods, the particle uxeswere similar.The  uxwas maximumin sample2 andwas 50timeshigher than the lastsampling (sample 4). Over the304 days, the mean particle  uxmeasuredwith the sediment trap far from the hydrothermalin uence (M Pelagic)was 11.2mg m 22 d21 (range0.45– 61.7 mg m 22 d21). 2001] Khripounoffet al.:Hydrothermal particulate material 639

Figure2. (a) Progressive vector diagram of thecurrent at 3.5 m abovethe bottomand at2mfromthe Rainbowblack smokers (Flores 5). (b) Scatter plots of thesame current for each period of particle sampling. 640 Journalof MarineResearch [59, 4

Figure3. Temperature versus current direction recorded every 5 minutesby thecurrent-meter 2 m abovethe S trapclose to Rainbowvent. The area limited by thetwo vertical lines is the direction intervalfor whichthe trap was under the direct vent effect.

At500m northwardof theRainbow vents (M500), it was similarto the pelagic  uxand variedfrom 11.3mg m 22 d21 (150m a.b.)to 10.6 mg m 22 d21 (300m a.b.).At 1000m (M1000),the particle  uxwas slightlyhigher than those obtained in the other traps, and equalto 17.5mg m 22 d21 (150m a.b.)and 25.0 mg m 22 d21 (300m a.b.)(Table 2). Some trendsare evident in allthe traps (Fig. 6). The total  uxesdecreased regularly in autumn, thenremained very weak during the winter (except for asmallincrease in February)and Žnishedby apeakin June.A fourthlarge peak appeared at M1000 in January 150 m a.b. and300 m a.b.explaining the increase of themean particle  ux. c.Particle composition Themain features of thevent particle composition in the S trapwere thedominance of sulphurand calcium (as CaSO 4),thehigh concentration of iron (7%) andcopper (3.5–5.5%) and the low concentration of particulate organic carbon (0.13%) (Fig. 5, Table3). Inorganic carbon (carbonate) was notdetected. The isotopic composition in d15N and d13C(Fig.7) oftheparticulate organic matter collected in the S trapare, respectively, 11.9‰ and 221‰.The major chemical elements did not show spatial or temporal uctuations(Fig. 5b) indicating that the particle composition of the emission was homogeneousduring the 16 daysof experiment. 2001] Khripounoffet al.:Hydrothermal particulate material 641

Figure4. Progressive vector diagram of the current measured with current-meters positioned on M500,M1000 and M Pelagicmoorings (Note the scale change for theM Pel.Ž gure.)

Theelemental composition of particlesfrom thesediment traps at M Pelagic,M500 and M1000is shown in Table 3. Theorganic carbon concentration varied from 4.9%to 7.3% andthe organic carbon  uxwas regularlylower (150 m a.b.)than 300 m a.b.in theM500 moorings.Inorganic carbon concentration was high( .8%)at M Pelagic,M500 (300 m) andM1000 (150 m and300 m) butit was clearlyweak (5.3%) at M500 (150 m). The inorganiccarbon  uxat M1000 and at M Pelagicdid not show particular temporal variationand was alwayshigher than at M500 (Fig. 8). The sulphur concentration (0.13–0.54%) was lowthroughout the experiment at M pelagicand M1000. In contrast, particlesfrom theM500 (150 m a.b.)were characterizedby high sulphur concentrations duringthe Ž rst fourmonths of theexperiment(Table 3, Fig.8). During the same period, the FeandCu concentrationswere particularlyhigh (respectively, 22.5 and 9.1%) compared 642 Journalof MarineResearch [59, 4

Figure5. Total particle  ux(a),particle composition (b) and larval bivalve  ux(c)in theS trap(Bars representthe SD of3replicates).

withthose measured in the other traps (Table 3). The isotopic composition in d13C was similarat M500 and M1000 ( ; 223.5)while the M500 (150 m) was enrichedin d15N (14.2)in comparison with M1000 ( 11.8)(Fig. 7). d. Fauna Numerousliving organisms were presentin theS trapmoored close to thevent. Bivalve larvaebelonging to the mytilid species Bathymodiolusazoricus (Comtet et al., 2000) were 2001] Khripounoffet al.:Hydrothermal particulate material 643

Table2. Mean particle  uxsampled in the largetraps (Long-term moorings). M Pelagic M 500 M 1000 200m a.b.150 m a.b.300 m a.b.150 m a.b.300 m a.b. Mass ux(mgm 22 d21 ) 11.2 11.3 10.6 17.5 25.0 OrganicC. (mgm 22 d21 ) 0.62 0.47 0.66 0.82 1.28 Inorg.C. (mgm 2 2 d2 1 ) 1.32 0.63 1.02 1.43 2.10 Sulphur(mg m 22 d21 ) 0.04 0.94 0.10 0.05 0.06 characterizedby theuniform size of theirshells (500 6 35 mmlong)suggesting that they were atthe same stage of development. No major difference was observedbetween the bivalvelarvae abundance during the three Ž rst periodsof sampling.However, the larvae uxstrongly decreased during the last period (Fig. 5c) although the number of other organismsdid not change (Table 4). A singlebivalve larva of theNuculidae family was collectedin the trap. Polychaetes (23) found in the trap were essentiallyat the larval or juvenilestage. Gastropods (13) belonging to two vent species ( Shinkailepas sp. and Phymorhynchus sp.)and two holoplanktonic species, halacarids and a singlepycnogonid were alsosampled in thetrap S. Numerouscopepods were alsofound in thistrap (Table 4). Benthicand planktonic organisms were collectedat different stages of developmentin thetraps M500 and M1000 (Fig. 9). Bivalves, polychaetes, gastropods, euphausiaceans andcopepods (harpacticoid and calanoid copepods, nauplii) were themost common taxa. Otherorganisms were presentin low abundance (nematodes, ostracods, amphipods, tanaids,epicarid isopods, larvae of cirripeds, ophiurids, asterids). Bivalvelarvae, belonging to the same species as the mytilid larvae found in the S trap( B. azoricus),occurredlargely at thebeginningof theexperimentin thetrap M500, 150 m a.b. Onlya few individualsof B. azoricus were presentat M500,300 m a.b.and M1000,300 m a.b.Several bivalve larvae belonging to another unknown species were alsofound at M1000,150 m a.b.No bivalve larva was presentin the pelagic trap. Polychaete larvae and

Table3. Mean chemical composition (%) ofparticlessampled in thesediment traps (Only the Ž rst4 month’s resultswere take into account in thetrapM500, 150 m a.b.). M500 S trap (Vent 150m a.b.(Plume M500 M1000 M1000 M Pelagic (%) composition) composition) 300 m a.b. 150 m a.b. 300 m a.b. 200 m a.b. Organic C. 0.13 4.9 7.3 5.5 5.9 7.2 InorganicC. ’0 5.3 8.0 8.4 8.4 9.0 Phosphorus0.45 0.8 0.3 0.21 0.5 0.2 Sulphur 17.2 13 0.90 0.31 0.35 0.31 Calcium 22.6 13.5 22.7 21.2 22.6 22.6 Iron 7.2 22.5 2 0.83 0.84 0.58 Copper 5.0 9.1 0.59 0.45 0.41 0.33 Silica 0.35 3.8 7.1 6.44 7.2 7.21 644 Journalof MarineResearch [59, 4

Figure6. Temporal variation of total uxmeasuredwith traps positioned on theMPelagic(a), M500 (b)and M1000(c) moorings. juvenilesbelonging to several families (Spionidae, Hesionidae, Glyceridae, Syllidae, Poecilochaetidae,Polynoidae, Aphroditidae, Archinomidae, Ophelidae, Typhloscoleci- dae)were alsoobserved in all traps. Only one polychaete Archinomidae, unequivocally belongingto the hydrothermal fauna, was sampledin thetrap M500, 150 m a.b.Gastropod 2001] Khripounoffet al.:Hydrothermal particulate material 645

Figure7. Isotopic composition of the sediment trap material and the organisms (Colac ¸o,2001)from Rainbowvent Ž eld.

Figure8. Temporal variation of theparticle composition in the traps positioned on theM500, M1000 andM Pelagicmoorings. 646 Journalof MarineResearch [59, 4

Table4. Number (min-max) of livinganimals collected close to the Rainbow vent in the Strap. Dates Bivalves PolychaetesGastropods Copepods (24/08/97–28/ 08/97)249– 355 0–3 1–5 30–53 28/08/97–01/ 09/97 59–302 2–5 0–3 13–52 01/09/97–05/ 09/97 80–237 0–5 0–2 30–38 05/09/97–09/ 09/97 4–32 0–2 0–2 12–27 larvalabundance (two benthic species) did not exhibit a temporaltrend over the collection period.Juvenile euphausids, belonging to the genus Thysanoessa, were observedin high abundancebetween August to December 1997 in the traps M500, 300 m a.b.,M1000 150m and3000 m a.b.,but they were rare inthe trap M500 (150 m a.b.)and in the pelagic trap(Fig. 9).

4.Discussion a.Pelagic  uxinthe Rainbow area ThePelagic trap, positioned 2 kmawayfrom theridge axis, was takenas areferencefor thebackground  uxnot in uenced by the hydrothermal vents. The mean particle  ux

Figure9. Temporal variations of commontaxa sampled with the M500, M1000 and M Pelagictraps. 2001] Khripounoffet al.:Hydrothermal particulate material 647

(11.2 mg m22 d21),measuredwith this trap, was comparableto the  uxobtained on the LuckyStrike area in theAzores region (7.7 mg m 22 d2 1,Khripounoff et al., 2000).These uxesare among the lowest recorded in the oligotrophic Atlantic Ocean (Honjo and Manganini,1993; Deuser, 1986; Auffret et al., 1994;Khripounoff et al., 1998).This result isin agreement with the very low primary production in the Azores region observed by Angel(1989). The relatively high d15Nofmaterialcollected in thePelagic trap (Fig. 6) is consistentwith the presence of plankton detritus from oligotrophic,nitrate-poor environ- ments(Holmes et al., 1996; Waser et al., 1998). Theparticulate  uxcoming from thesurface showed large  uctuationstypical of deep-seaenvironments with pronounced seasonal changes (Honjo and Manganini, 1993; Newton et al., 1994),with a maximumobserved in June due to the spring bloom of the surfaceprimary production and a minimumin winter. The very low concentration of typicalhydrothermal compounds (S, Fe,Cu) in the particles and the absence of hydrother- malfauna conŽ rm thatthe pelagic trap was notunder the direct effect of the vents. The highorganic carbon concentration indicates the freshness of thepelagic origin particles. b.Particles emitted by the Rainbow vents Theparticle  uxin the Rainbow vent area (average 6.9 g m 22 d21)was 500times higherthan that at the pelagic site. A similarhigh  uxwas observedclose to the East PaciŽc Risehydrothermal vent Ž eldsusing the same sampling technique (Khripounoff and Alberic,1991) and the Endeavour Ridge (2200 m) (Rothand Dymond, 1989). On the Mid-AtlanticRidge, the Rainbow vent emitted a comparable uxas on the TAG site (3680m) (9.0g m 22 d21:Desbruye`res et al., 2000)but was signiŽcantly higher than that measuredat MenezGwen (850m) (0.64g m 22 d21,Desbruye`res et al., 2001)and Lucky Strike(1600 m) (0.26g m 22 d21,Khripounoff et al., 2000).At Rainbow,in common with TAG andPaciŽ c vents,active black smokers emit dark, particle-rich  uid,while the shallowvents such as Lucky Strike and Menez-Gwen emit translucent  uidwith low particlecontent. These high differences in particle production result from differencesin uidchemistry which depends on the temperature and pressure conditions met during hydrothermalcirculation (Douville et al., 1999).Phase separation events due to the shallowdepth at Menez-Gwen and Lucky Strike (resulting in alowmetal concentration in the uid)and the extensive alteration of the basaltic substrate explain the low particle productionat these shallower stations (Klinkhammer et al., 1995;Wilson et al., 1996). Inaddition, the amountof particlescollected in thetrapalso depends on thetrapposition relativeto the chimney and on the current direction. The occurrences of the temperature anomalies . 4°C (Fig.3) indicate that the trap was directlyunder the chimney in uence duringapproximately 11% of the time without regular temporal trend (Fig. 3). This relativelylow percentage is easilyexplainable by the positionof thetrap (south) in relation tothe black smokers and the main current direction (northeast). The current varied very littlein direction and speed (Fig. 2) and the hydrodynamic conditions were verysimilar duringthe different periods of particle sampling. The strong  uctuationsof theparticulate 648 Journalof MarineResearch [59, 4

uxduring the 16 daysof experiment(Fig. 5) were likely,therefore, the result of irregular uidproduction rather than current variations. Temporal changes in ventproduction have beenreported at other hydrothermal sites over time scales of minutes (Converse et al., 1984),hours (Chevaldonne ´ et al., 1991)and weeks (Murton and Redbourn, 1997). For Chevaldonne´ et al. (1991),aperiodic variability of temperatureobserved in differentvents canbe explained by a changein the plumbing of the porous diffusive system in the chimney,which can explain also the variation of theparticle  uxemission. Theelemental composition of the particles sampled near the vent (Table 2) was characterizedby ahighsulphur concentration (S . 17%)of whichthe main origin was the hydrothermal uidprecipitation. The particulate sulphur content at Rainbowwas slightly higherthan that measured near the Lucky Strike vents (11%) (Khripounoff et al., 2000) andsimilar to the sulphur concentration at Menez-gwen (18.2%) (Desbruye `res et al., 2001).But it was lowerthan that observed close to the PaciŽ c vents(S $ 25%)(Dymond andRoth, 1988; Khripounoff and Alberic, 1991). One of themain features of theRainbow ventcomposition is the high concentration of Cu(3.6%),which is consistent with the  uid composition(Douville et al., 1997). The d13Ccompositionof the sediment trap materials was quitedifferent from thatof the mussel and shrimpadults collected on thebottom on the Rainbowsite (Fig. 7, Colac¸o et al., 2001). d13C (224‰)in thesediment traps was more depletedthan in the hydrothermal shrimps ( 210‰).Mussels ( B. azoricus)tendto have morenegative d13Cvaluesand are much more depleted in d15N.ThesesigniŽ cant differencessuggest that the hydrothermal mussels did not directly consume the organic particulatematerial emitted by the vents although these organisms demonstrate Ž ltering behavior(Le Pennec et al., 1990). c.Export of hydrothermalparticles to thesurrounding ocean Ventparticle export to thesurrounding area depends Ž rst onthe speed and the direction ofthe local current. To knowwhen the vent particle transport was favorableto the traps at 500m, the percentage of instantaneous current  owingin the right direction has been computedfor eachparticle-sampling interval. The favorable direction has been chosen as 21°, correspondingto the direction of Rainbow vents to the M500, plus an angleof 20°on eachside (to take into account the diffusion). The minimum speed required to cover the vent-trapdistance is equal to 500 m/ 3h(halftime of the tidal current reverse) 5 4.6 cm s21.Figure10 shows the result of this empirical estimation versus time. For the durationof thesampling, the current was favorablefor transportingvent particles to the M500during 36% of the time on averageat 150m a.b.and 22% at 300m a.b.The result of thisfavorable current orientation has been that sulphur, as a hydrothermalchemical signature,was foundin large quantity (max 5 15%)during the Ž rst fourmonths of the experimentin the trap M500 at 150 m a.b.and also at 300 m a.b.with a lower concentration(Fig. 8). This result indicates that the correct orientation of thecurrent from thevents to the traps is necessary in order to sample hydrothermal material in this equipment.However, Figure 10 showsthat the favorable direction of thecurrent is notthe 2001] Khripounoffet al.:Hydrothermal particulate material 649

Figure10. Temporal variations of sulphurconcentration in the particles and percentage of instanta- neouscurrent  owingin therightdirections from the vents toward the mooring M500. onlyparameter required to capturehydrothermal particles in the traps with certainty. The sulphurconcentration, derived from theplume, showed large temporal  uctuationsin the trap150 m a.b.without direct relationship with our estimationof theright current direction percentage(Fig. 10). This observation suggests variations either in the altitude of the plume(the plume  owedirregularly under the lower trap, 150 m a.b.)or in the particle 650 Journalof MarineResearch [59, 4 emissionby the vents on the scale of few days,or modiŽ cations of the plume vertical expansion.Multi-peaked plumes (German et al., 1996;Thurnherr and Richards, 2001) indicatethat the plume vertical structure might change with the time and that the plume horizontalextent might be compared to patches of particle rich  uid(Thurnherr and Richards,2001). We observethat the sulphur concentration was lowin the trap moored at theupper limit of theplume (300 m a.b.).It corresponds generally to sulphurbackground variabilityobserved in the pelagic particles with a notableexception in November when themaximum(3%) correspondsto themaximum found in thetrap150 m a.b.(Fig. 10). No evidentrelationship exists between the sulphur concentration at 300m a.b.and the current directionat this altitude. Finally, the complete absence of an expected variation in temperaturemeasured with the thermistor chain can be explained by therapid uiddilution andonlythe backgroundvariability of theseawatertemperature was measuredmasking the hydrothermalsignal (Wilson et al., 1995;Thurnherr and Richards, 2001). Althoughthe trap M500, 150 m a.b.was signiŽcantly enriched with particles from the plumeat the beginning of theexperiment, the total mass uxmeasuredwith this trap did notdiffer from thepelagic  uxmeasuredover the same period (Table 2). This discrepancy canbe explainedif the trap M500, 150 m a.b.did notreceivethe totality of thepelagic  ux. Assumingthat the ventparticles did not contain inorganic carbon (no inorganic carbon was measuredin the Strap),the estimation of therelativecontribution of pelagicparticles to the totalmass uxat150m a.b.for trapM500 during the plume sampling (Ž rst fourmonths) is:

Inorg.C.flux,M500 3 100 5 48%. Inorg.C.flux,MPelagic

Theparticle composition in the plume consisted of only 48% of pelagicparticles (52% of hydrothermalparticles). The same calculation indicates that this ratio was near100% in the trap300 m a.b.The main factor, which may account for thelow abundance of pelagic particlesin the plume, is that the plume had not reached its altitude of neutralbuoyancy (iso-density).The result was adiscontinuitywhich disrupted the settling of particlesinto theplume water (umbrella effect). The multi-peaked plumes observed by German et al. (1996)and Thurnherr and Richards (2001) at Rainbow imply different plume densities withdifferent heights of rise and possible different composition which sorts pelagic particles.In addition, there may have been some dissolution of thepelagic carbonates by theacidic plume water. This acidiŽ cation could be theresult of the very acid  uidorigin of theplume water orofthehigh bacteria activity on theplume particles (McCollom, 2000). Chemicalbehavior of major elements in hydrothermalplumes has generally been found toshow an enrichment of Fe oxidesand a decreaseof sulphideparticles. In comparison to theinitial conditions of emission(S trap),the increase of C,Fe,Cu andthe decrease of S andCa in the trap M500, 150 m a.b.conŽ rms thesechemical modiŽ cations during the plumetransport (Table 3). The high concentration of Fe isa characteristicof theneutrally buoyantplume at Rainbow(Edmonds and German, 2001) and apparently results from Fe 2001] Khripounoffet al.:Hydrothermal particulate material 651 hydroxideformation and lesser sedimentation of Fe sulŽdes as comparedwith other vent sites(Douville et al., 1997).The increase of Ca and C resultsfrom thepresence of the pelagicparticles in theplume (see above).The large decrease of Sindicatesrapid settling ofthiselement (as sulŽde) near the emission site and dilution of theplume particles with thepelagic particles. The d15Nenrichmentin theparticlesfrom M500,150 m a.b.suggests thatthe organic compounds of the plume are composed by mixing of particles from differentorigins with a dominanceof thepelagic organic particles. The d13Ccomposition, whichoverlaps with the one reported for phytoplanktonfrom temperateregions (Fontugne andDuplessy, 1981), conŽ rms thisobservation. Thesame calculation at M1000 as atM500(see above)of thepercentage of instanta- neouscurrent  owingin the right direction indicates that the current was inthe favorable directionduring 9% of the time (the parameters chosen for thisestimation are: current direction38° 6 20°, minimumspeed 1000 m/ 3h 5 9.2 cm s21).Thislow percentage couldbe enoughto explain the absence of hydrothermalsignature (sulphur anomaly) in the M1000traps. However, a smallincrease of Fe-Cu(Table 2) was observedcompared with ourreference (M pelagic).But the low accuracy of ourmeasurementsmade with the X ray spectrometryfor theseelements (due to weak concentrations) makes the conŽ rmation of thehydrothermal particle presence in these sediment traps difŽ cult. Other possibilities can alsoexplain this lack of signiŽ cant hydrothermal signal. The dilution of theplume at 1km from thevents was toogreat for itto be detected in the trap. But this explanation is insufŽcient because German et al. (1996)and German et al. (1998)found that particles in theplume, observed by nephelometry,could reach greater distances (50 km). The second hypothesisis that the neutrally buoyant plume passed over the upper trap (300 m) anddid notproduce settling particles at this altitude. This explanation seems unlikely because Thurnherrand Richards (2001) indicate a maximumplume elevation of 400 m ata distance of1kmfrom thevents and the plume thickness exceeds several hundred meters (German et al., 1998).The other possibility is that the traps were regularlyin the plumebut the large hydrothermalparticles, rich in sulphur, settled before reaching the traps or theirconcentra- tionwas tooweak to be detected in the samples at this distance. This explanation is consistentwith the strongdecrease of sulphurconcentration observed in thisstudy between theparticles in the S trapand the particles in M500 traps. The size distribution of suspendedparticles in neutrallybuoyant hydrothermal plumes rich in Fe ischaracterized bythedominance of particles with very small diameter (Walker and Baker, 1988; German andSparks, 1993). These observations suggest that, even if theplume particle density is high,the settling could be very slow and the plume particle  uxmeasured at 1000 m representsonly a smallfraction of the regional pelagic  ux.The d13Ccompositionat M1000,which was similarto the value obtained from thepelagic trap, conŽ rms the prevalenceof theorganic material from thepelagic  uxatthis distance. The depletion in term of d15NatM1000compared with the pelagic material can be the result of different ratesof bacterial degradation which produces a variationin the d15N(Altabetand McCarthy,1986) or ofthein uence of theparticle sizes (Altabet, 1988). 652 Journalof MarineResearch [59, 4 d. Fauna Thesediment traps, designed for thestudy of particle uxinthe ocean, proved to bea particularlywell-adapted tool to the collection and the observation of small living organisms.In this study, the sampled fauna varied greatly with the time and between traps locatedat differentdistances from thevents (Fig. 5 and9). The occurrence of highdensities ofhydrothermalvent mytilid populations (identiŽ ed as B.azoricus, Comtet et al., 2000) in theS trapat the Rainbow site suggests that they originated from themussel beds observed 120m tothesouthwest. The majority of thelarvae shells were ofsimilarsize, indicating a similarstage of development.This implies that the larvae may have stayed on thesite over alongtime (3– 6 weeksif we comparewith the larvaedevelopment duration of the shallow mussel Mytilusedulis ).Thevery high density of bivalvelarvae in the S trapwas dueto the existenceof retention factors. Vent larvae may remain aggregated because they have a positivechemosensory response to sulŽ des (Renninger et al., 1995).Alternatively, the hydrodynamicturbulence (vortex) produced by the active black smokers, could concen- tratelarvae around the trapand increase their residence time near the vent(Mullineaux and France,1995). The bivalve larvae  uxdecreased signiŽ cantly (Anova: p , 0.05) at the endof August(Fig. 5) probablydue to thestrong decrease of theparticle  uxatthisperiod andto thespatial heterogeneity of theorganism distribution around the vent. However, the hypothesisof a discontinuous(seasonal?) mytilid reproduction cannot be dismissed (Comtetand Desbruye `res,1998). Its startingsignal could be the peak of particle  ux duringthe spring bloom in surface waters. Although a lotof bivalve larvae were found aroundthe vent, the black smokers were nearlyazoic in August 1997 (Marvel cruise) and oneyear later, in June 1998 (Pico cruise). The development of musselsaround the black smokerswas hinderedby thevery high  uxofmineralparticles which can affect the water Žltrationand also juvenile growth (Bricelj et al., 1984).In contrast, the common holoplanktongroups, such as thecopepods, were clearlymore abundant in thetrap closer tothe ventsthan in thetrapsmoored far away.Their prevalence near the vents suggests that theyfound a localnutrient enrichment and it appears that they are highly tolerant to reducedcompounds and heavy metals present in the hydrothermal  uids.These groups maybe speciŽcally associated with vent habitats (Kim andMullineaux, 1998). Bivalvelarvae were alsocaught in the trap M500 (150 m a.b.).Temporal difference in abundanceof these organisms, belonging to the hydrothermal species B.azoricus, was foundwith a maximumof larvaein summer,suggesting again a discontinuoushydrother- malbivalve reproduction, in agreement with our results obtained with the trap S moored nearthe vent. The presence of few individualsof thesame species in theM1000 traps (but notin thepelagic trap) indicates that the hydrothermal plume is ableto transport few small organisms,at leasta kilometrefrom thevent site, and to dispersethe vent species along the mid-oceanridge (Kim et al., 1994;Mullineaux and France, 1995). Previousstudies have highlighted that zooplankton biomass often increases near the hydrothermalŽ elds(Berg and Van Dover, 1987; Burd and Thomson, 1994; Kaartvedt et al., 1994;Mullineaux et al., 1995;Herring and Dixon, 1998). The presence of high 2001] Khripounoffet al.:Hydrothermal particulate material 653 euphausiaceandensity in thetraps moored in therift valley in comparisonwith the pelagic trapconŽ rms thisrise. The water nutrient enrichment, due to anabundanceof free-living bacteriasupporting the hydrothermal trophic food web, is the common explanation for this biomassincrease (McCollom, 2000). The depletion of d15Ninthe traps moored in the rift (Fig.6) isan indication of local organic matter production in the valley which was also conŽrmed by theincrease of organiccarbon  uxintheM1000 traps (Table 3). This local modiŽcation of thefood web creates a newenvironment which facilitates the zooplank- toniccongregation (Vereshchaka and Vinogradov, 1999). Burd et al. (1992)and Burd and Thomson(1995) suggest that several zooplanktonic taxa (like the euphausiaceans in this study)may approach the plume closely, but the water toxicity prevents them from living withinthe plume. This hypothesis can explain the very low density of euphausiaceans foundin the trap at M500, 150 m a.b.On the contrary, other zooplanktonic groups, like copepods,seem insensitive to thistoxicity problem (see aboveresults from theS trap).

5.Conclusions Thehydrothermal particle production and transport at Rainbow have been shown to be controlledŽ rstlyby the variation of  uidemission and then by the current direction dependingon thetopography of therift valley. The results presented in this work provide importantinformation on the particleproduction close to the ventsand on thedispersion of thehydrothermal plume. Hydrodynamical study and biochemical analyses of theparticles form thebasis of a detaileddescription of the hydrothermal Rainbow area. For closer investigation,several studies would have to be improved. Long-term particle sampling closeto a ventshould provide deŽ nite proof of theirregular emission of vent uidsand the possiblediscontinuous reproduction of somehydrothermal organisms. Because our knowl- edgeof thewater circulation is nowwelldocumented for theRainbowarea, it willallow us todeŽ ne the best position for themooring of instruments in the future to measure the dispersionand the composition of theplume and the consequences for faunaldistribution. Inconclusion, the Rainbow site offers anideal natural laboratory to study hydrothermal processes.

Acknowledgments. Wewouldlike to thank the ofŽ cers and the crews of theFrench N/ O Atalante andthe German R/ V Poseidon duringthe Marvel and Flame 2 cruisesand the chief scientists, A.-M.Alayse (IFREMER) andC. German(Southampton Oceanography Centre). The authors thank T.Comtet,B. Casanova,P. Bouchetand A. Warenfor providing collaboration and zoological determinations.We also acknowledge the thoughtful comments of threeanonymous reviewers. This researchwas undertaken in part in the framework of the AMORES project.We acknowledge the supportfrom the European Commission’ s MarineScience and Technology Programme (MAST 3) undercontract MAST3-CT95-0040.

REFERENCES Altabet,M. A.1988.Variations in nitrogen isotopic compositions among particle classes: Implica- tionfor particle transformation and uxintheopen ocean. Deep-Sea Res., 35, 535–554. 654 Journalof MarineResearch [59, 4

Altabet,M. A.andJ. J.McCarthy.1986. Vertical patterns in 15N natural abundance in PON fromthe surfacewaters of several warm core rings in the SargassoSea. J. Mar.Res., 44, 185–201. Angel,M. V.1989.Vertical proŽ les of pelagic communities in thevicinity of theAzores Front and theirimplications to deep ocean ecology. Prog. Oceanogr., 22, 1–46. Auffret,G., A. Khripounoffand A. Vangriesheim.1994. Rapid post-bloom resuspension in the North-EasternAtlantic. Deep-Sea Res., 41, 925–939. Berg,C. J.,Jr.and C. L.VanDover. 1987. Benthopelagic macrozooplankton communities at and neardeep-sea hydrothermal vents in the eastern PaciŽ c Oceanand the Gulf of California. Deep-SeaRes., 34, 379–401. Bricelj,V. M.,R.E.Maloufand C. DeQuillfeldt. 1984. Growth of juvenile Mercenariamercenaria andthe effect of resuspendedbottom sediments. Mar. Biol., 84, 167–173. Burd,B. J.andR. E.Thomson.1994. Hydrothermal venting at Endeavour Ridge: effect on zooplanktonbiomass throughout the water column. Deep-Sea Res. I, 41, 1407–1423. 1995.Distribution of zooplanktonassociated with the Endeavour Ridge hydrothermal plume. J. PlanktonRes., 17, 965–997. Burd,B. J.,R.E.Thomsonand G. S.Jamieson.1992. Composition of a deepscattering layer overlyinga mid-oceanridge hydrothermal plume. Mar. Biol., 113, 517–526. Chevaldonne´,P.,D. Desbruye`resand M. LeHa ˆ‡tre.1991. Time-series of temperature from three deep-seahydrothermal vent sites. Deep-Sea Res., 38, 1417–1430. Colac¸o,A., F. Dehairs,A. Colac¸oandD. Desbruyeres.2001. Nutritional relations of deep-sea hydrothermalŽ eldsat the Mid-Atlantic Ridge: a stableisotope approach. Mar. Ecol. Prog. Ser. (submitted). Comtet,T. andD. Desbruye`res.1998. Population structure and recruitment in mytilid bivalves from theLucky Strike and Menez Gwen hydrothermal vent areas (37° 17 9N and 37°509N on the Mid-AtlanticRidge). Mar. Ecol. Prog. Ser., 163, 165–177. Comtet,T., D. Jollivet,A. Khripounoff,M. Segonzacand D. R.Dixon.2000. Molecular and morphologicalidentiŽ cation of Bathymodiolusazoricus (Bivalvia:Mytilidae) in situ-preservedin sedimenttraps at Rainbowhydrothermal vent Ž eld(Mid-Atlantic Ridge). Limnol. Oceanogr., 45, 1655–1661. Converse,D. R.,H.D.Hollandand J. M.Edmond.1984. Flow rates in the axial hot springs of the EastPaciŽ c Rise(21N): Implications for the heat budget and the formation of massive sulŽ de deposits.Earth Planet. Sci. Lett., 69, 159–175. Desbruye`res,D., A. Almeida,M. Biscoito,T. Comtet,A. Khripounoff,N. LeBris, P. M.Sarradin andM. Segonzac.2000. A reviewof the distribution of hydrothermalvent communities along the NorthernMid Atlantic Ridge: Dispersal vs. environmental controls. Hydrobiologia, 440, 201–216. Desbruye`res,D., M.Biscoito,J. C.Caprais,A. Colac¸o,T. Comtet,P. Crassous,Y. Fouquet, A.Khripounoff,N. LeBris, K. Olu,P. M.Sarradin,M. Segonzacand A. Vangriesheim.2001. Variationsin deep-sea hydrothermal vent communities on theMid-Atlantic Ridge when approach- ingthe Azores plateau. Deep-Sea Res., 1, 48, 1325–1346. Deuser,W. G.1986.Seasonal and interannual variations in deep-water particle  uxesin the Sargasso Seaand their relation to surfacehydrography. Deep-Sea Res., 33, 225–246. Douville,E., J.-L. Charlou, J.-P. Donval, D. Hureauand P. Appriou.1999. Le comportement de l’arsenic (As) et del’antimoine(Sb) dans les  uidesprovenant de diffe´rentssyste `meshydrother- mauxoce ´aniques.C. R.Acad.Sci. Paris, Se ´r.II, 328, 97–104. Douville,E., J. L.Charlou,J. P.Donval,J. Knoery,Y. Fouquet,P. Bienvenuand P. Appriou.1997. Traceelements in uidsfrom the new Rainbow hydrothermal Ž eld(36° 14 9N,MAR): acompari- sonwith other Mid-Atlantic  uids.Eos Trans. Am. Geophys.Union, 78, 832. Dymond,J. andS. Roth.1988. Plume dispersed hydrothermal particles: A time-seriesrecord of settling uxfrom the Endeavour Ridge using moored sensors. Geochim. Cosmochim. Acta, 52, 2525–2536. 2001] Khripounoffet al.:Hydrothermal particulate material 655

Edmonds,H. N.andC. R.German.2001. Particle geochemistry and cycling in the Rainbow hydrothermalplume. Geochim. Cosmochim. Acta, (submitted). ElderŽeld, H. andA. Schultz.1996. Mid-ocean ridge hydrothermal  uxesand the chemical compositionof theocean. Ann. Rev. Earth Planet. Sci., 24, 191–224. Emery,W. J.andR. E. Thomson.1998. Data Analysis Methods in Physical Oceanography, Pergamon,London, 634 pp. Feely,R. A.,J. H.Trefry,G. T. Lefonand C. R.German.1998. The relationship between P/ Fe and V/Fe ratiosin the hydrothermal precipitates and dissolved phosphate in seawater. Geophys. Res. Lett., 25, 2253–2256. Fontugne,M. R.andJ.C.Duplessy.1981. Organic carbon isotopic fractionation by marine plankton inthe temperature range 21to31° . Oceanol.Acta, 4, 85–90. Fouquet,Y., J.-L.Charlou, J.-P. Donval, J. Radford-Knoe¨ry,H. Ondre´as,P. Cambon,H. Bougault, J.Etoubleau,F. Barriga,I. Costa,N. Lourenc¸oandM. K.Tivey.2001. Hydrothermal processes on shallowvolcanic segments: Mid-Atlantic Ridge near the Azores Triple Junction. J. Geophys.Res., (submitted). Fouquet,Y., J.-L. Charlou, H. Ondre´as,J.Radford-Knoe¨ry,J.-P.Donval, E. Douville,R. Apprioual, P.Cambon,H. Pele´,J.Y.Landure´,A.Normand,E. Ponsevera,C. R.German,L. Parson, F.Barriga,I. Costa,J. Relvasand A. Ribeiro.1997. Discovery and Žrstsubmersible investigations ontheRainbow hydrothermal Ž eldon theMAR (36°14 N).Eos Trans. Am. Geophys.Union, 78, 832. German,C. R.,G. P.Klinkhammerand M. D.Rudnicki.1996. The Rainbow hydrothermal plume. 36°149N.Mid-AtlanticRidge. Geophys. Res. Lett., 23, 2979–2982. German,C. R.,K.Richards,M.-M. Lam, A. Thurnherr,P. J.Baptiste,A.-M. Le Clerc, A. Dapoigny, M.Cooper,M. Rudnicki,H. ElderŽeld, J.-L. Charlou, J. Knoery,J.-P. Donval, H. Edmond, D.Green,A. Khripounoff,P. Crassous,A. Vangriesheim,J. D.O’Brien,J. Patching,D. Dixonand P.Herring.2001. A segmentscale study of  uxesthrough the Rainbow hydrothermal plume, 36°14 N, Mid-AtlanticRidge. Deep-Sea Res., (submitted). German,C. R.,K. J.Richards,M. D.Rudnicki,M. M.Lam andJ. L.Charlou.1998. Topographic controlof adispersinghydrothermal plume. Earth Planet. Sci. Lett., 156, 267–273. German,C. R.andR. S.J.Sparks.1993. Particle recycling in theTAG hydrothermalplume. Earth Planet.Sci. Lett., 116, 129–134. Herring,P. J.andD. R.Dixon.1998. Extensive deep-sea dispersal of postlarval shrimp from a hydrothermalvent. Deep-Sea Res., 45, 2105–2118. Holmes,M. E.,P.J.Mu¨ller,R. R.Schneider,M. Segl,J. Pa¨tzoldand G. Wefer.1996. Stable nitrogen isotopesin Angola Basin surface sediments. Mar. Geol., 134, 1–12. Honjo,S. andS. J.Manganini.1993. Annual biogenic particle  uxesto the interior of the North AtlanticOcean studied at 34N–21W and48N– 21W. Deep-Sea Res. II, 40, 587–607. Kaartvedt,S., C. L.VanDover, L. S.Mullineaux,P. H.Wiebeand S. M.Bollens.1994. Amphipods onadeep-seahydrothermal treadmill. Deep-Sea Res., 41, 179–195. Khripounoff,A. andP. Alberic.1991. Settling of particlesin hydrothermalvent Ž eld(East PaciŽ c Rise13N) measured with sediment traps. Deep-Sea Res., 38, 729–744. Khripounoff,A., T. Comtet,A. Vangriesheimand P. Crassous.2000. Near-bottom biological and mineralparticle  uxin the Lucky Strike hydrothermal vent area (Mid-Atlantic Ridge). J. Mar. Syst., 25, 101–118. Khripounoff,A., A. Vangriesheimand P. Crassous.1998. Vertical and temporal variations of particle uxesin thedeep tropical Atlantic. Deep-Sea Res., 45, 193–216. Kim,S. L.andL. S.Mullineaux.1998. Distribution and near-bottom transport of larvaeand other planktonat hydrothermalvents. Deep-Sea Res. II, 45, 423–440. Kim,S. L.,L.S.Mullineauxand K. R.Helfrich.1994. Larval dispersal via entrainment into hydrothermalvent plumes. J. Geophys.Res., C 99, 12655–12665. 656 Journalof MarineResearch [59, 4

Klinkhammer,G. P.,C. S. Chin,C. Wilsonand C. R.German.1995. Venting from the Mid-Atlantic Ridgeat 37° 17 9N:TheLucky Strike hydrothermal site, in Hydrothermalvents and processes, L.M.Parson,C. L.Walkerand D. R.Dixon,eds., Geological Society, London, 87– 96. LePennec, M., A.Donvaland A. Herry.1990. Nutritional strategies of thehydrothermal ecosystem bivalves.Prog. Oceanogr., 24, 71–80. Marguillier,S., G. VanDer Velde, F. Dehairs,M. A.Hemmingaand R. Rajagopal.1997. Trophic relationshipsin an interlinked mangrove-seagrass ecosystem as traced by d1 3 C and d1 5 N. Mar. Ecol.Prog. Ser., 151, 115–121. McCollom,T. M.2000.Geochemical constraints on primary productivity in submarinehydrother- malvent plumes. Deep-Sea Res. I, 47, 85–101. Mullineaux,L. S. andS. C.France.1995. Dispersal mechanisms of deep-sea hydrothermal vent fauna, in Seaoor Hydrothermal Systems: Physical, Chemical, Biological and Geological Interac- tions,S. E.Humphris,R. A.Zierenberg,L. S.Mullineauxand R. E.Thomson,eds., American GeophysicalUnion, Washington, 408 –424. Mullineaux,L. S.,P. H.Wiebeand E. T.Baker.1995. Larvae of benthic invertebrates in hydrothermalvent plumes over Juan de FucaRidge. Mar. Biol., 122, 585–596. Murton,B. J.andL. J.Redbourn.1997. Segment-scale hydrothermal energy  uxesfrom the Mid-AtlanticRidge. EOS Trans.Am. Geophys.Union, 78, 842. Newton,P. P.,R.S.Lampitt,T. D.Jickells,P. Kingand C. Boutle.1994. Temporal and spatial variabilityof biogenicparticle  uxesduring the JGOFS northeastAtlantic process studies at 47N, 20W.Deep-Sea Res., 41, 1617–1642. Renninger,G. H.,L.Kass,R. A.Gleeson,C. L. VanDover, B.-A. Batelle, R. N.Jinks,E. D.Herzog andS. C.Chamberlain.1995. SulŽ de as a chemicalstimulus for deep-sea hydrothermal vent shrimp.Biol. Bull., 189, 69–76. Roth,S. E.andJ. Dymond.1989. Transport and settling of organic material in the deep-sea hydrothermalplume: evidence from particle  uxmeasurements.Deep-Sea Res., 36, 1237–1254. Thurnherr,A. M.andK. J.Richards.2001. Hydrographic setting of the Rainbow hydrothermal plume(36° 14 9N,Mid-AtlanticRidge). J. Geophys.Res., (in press). Vereshchaka,A. L.andG. M.Vinogradov.1999. Visual observations of thevertical distribution of planktonthroughout the water column above Broken Spur vent Ž eld,Mid-Atlantic Ridge. Deep-SeaRes. I, 46, 1615–1632. Walker,S. H.andE. T.Baker.1988. Particle-size distributions within hydrothermal plumes over the Juande Fucaridge. Mar. Geol., 78, 217–226. Waser,N. A.D.,P. J.Harrison,B. Nielsen,S. E.Calvertand D. H.Turpin.1998. Nitrogen isotope fractionationduring the uptake and assimilation of nitrate, nitrite, ammonium, and urea by a marinediatom. Limnol. Oceanogr., 43, 215–224. Weliky,K., E. Suess,C. Ungere,P. Mullerand K. Fischer.1983. Problems with accurate carbon measurementsin marine sediments and water column particulates: a newapproach. Limnol. Oceanogr., 28, 1252–1259. Wilson,C., J.-L.Charlou, E. Ludford,G. Klinkhammer,C. Chin,H. Bougault,C. German,K. Speer andM. Palmer,1996. Hydrothermal anomalies in theLucky Strike segment on theMid-Atlantic Ridge(37° 17 9N).EarthPlanet. Sci. Lett., 142, 467–477. Wilson,C., K. Speer,J.-L. Charlou and H. Bougault.1995. Hydrography above the Mid-Atlantic Ridge(33– 40N) and withinthe Lucky Strike segment. J. Geophys.Res., B, 100, 20555–20564.

Received: 24April,2000; revised: 26April,2001.