..............................................................

Long-lived vortices as a mode of deepventilation in the Greenland SeaJean-Claude Gascard*, Andrew J. Watson†, Marie-Jose Messias†,K. Anders Olsson‡, Truls Johannessen§ & Knud Simonsenk{

* Universite Pierre et Marie Curie, Laboratoire d’Oceanographie Dynamique et deClimatologie, 75252 Paris cedex 05, France† School of Environmental Sciences, University of East Anglia, Norwich NR4 7TJ,UK‡ Department of Analytical and Marine Chemistry, Goteborg University,S-412 96, Goteborg, Sweden§ Geophysical Institute and the Bjerknes Centre for Climate Research, Universityof Bergen, Allegaten 55, N-5007 Bergen, NorwaykNansen Environmental and Remote Sensing Center, Edvard Griegsvei 3A,N-5037 Solheimsviken, Bergen, Norway.............................................................................................................................................................................

The Greenland Sea is one of a few sites in the world ocean whereconvection to great depths occurs1 – 4—a process that forms someof the densest waters in the ocean. But the role of deep convectiveeddies, which result from surface cooling and mixing acrossdensity surfaces followed by geostrophic adjustment5, has notbeen fully taken into account in the description of the initiationand growth of convection6. Here we present tracer, float andhydrographic observations of long-lived (,1 year) and compact(,5 km core diameter) vortices that reach down to depths of2 km. The eddies form in winter, near the rim of the GreenlandSea central gyre, and rotate clockwise with periods of a few days.The cores of the observed eddies are constituted from a mixtureof modified Atlantic water that is warm and salty with polarwater that is cold and fresh. We infer that these submesoscalecoherent eddies contribute substantially to the input of Atlanticand polar waters to depths greater than 500 m in the centralGreenland Sea.

As part of the European Sub-Polar Oceans programme (ESOP-2),an extensive series of oceanographic observations were made duringthe period 1996–97 in the Greenland Sea. Figure 1a shows the studyarea. During summer 1996, a release of sulphur hexafluoride (SF6)tracer was made at ,300 m depth in the centre of the GreenlandSea7, and its fate followed by surveys in November 1996, and Marchand May 1997. Early in November 1996, four pairs of floats (one ofeach pair at 250 m and one at 500 m depth) were deployed along748 30 0 N at 28 W, 38 W, 48 W and 58 W respectively. Four more pairsof floats were deployed four months later throughout the gyre.Figure 1b shows the tracks of five of the 16 floats during the sixmonths March–August 1997. All these floats show evidence ofbecoming entrained for most of their trajectories in anticycloniceddies. The eddies tend to populate the rim of a region ,200 kmacross, the central gyre of the Greenland Sea, which circulatescyclonically and in which deep convection has been previouslyreported to occur1 – 4.

Figure 2b shows the north–south component of the velocity ofone of these floats. Float no. 02 was launched in November 1996.Late in December 1996 (day 50 of the experiment), it was entrainedin an anticyclonic eddy at a time when sea-to-air heat fluxesincreased drastically (Fig. 2a). On January 20 (day 80), this floatmoved in towards the eddy centre, spending 150 days circling theeddy centre at a constant temperature of 21 8C (Fig. 2d) within aradius of 2–3 km, before spiralling out again until its recovery inAugust 1997, indicating an eddy lifetime of at least nine months. Aplot of the orbital speeds of the float against distance from the centreshows that the vortex had a ‘core’ of about 5 km diameter in which itrotated like a solid body, and a ‘skirt’ surrounding this out to radii of,15 km in which angular velocity decreased. The core had relative

vorticity y equal to about 2 f/2, compared to y ¼ 2f =8 at 8 kmradius in the skirt, where f is the planetary vorticity.

In May 1997 while the floats were circulating the eddy, hydro-graphic and tracer measurements of the same feature were made.Figure 3a and b show sections of density, and SF6 tracer, passingthrough the eddy at 08 E along 758 N. The core of the eddy wascomposed of homogeneous water at about 21 8C (in agreementwith float no. 02 temperatures, Fig. 2d), with high levels ofchlorofluorocarbons (CFCs) and oxygen (O2), and low SF6 extend-ing from approximately 300 m to 2,200 m depth.

Calculation of geostrophic currents confirms a strong anti-cyclonic spin to the core at mid-depth. Comparison with the floatvelocities shows that the geostrophic currents contributed abouttwo-thirds of the total rotation, the discrepancy being accounted forby centrifugal force, the eddy being in ‘gradient wind’ balance. Theeddy core was characterized by a Rossby number R ¼ 0:5(R ¼ y=f ), Burger number B ¼ 0:25 (B ¼ ðN2=f 2ÞðH2=L2Þ, whereN is the Brunt-Vaisala frequency) and aspect ratio a ¼ 0:25(a ¼ H=L, where H and L represent vertical and horizontal eddydimensions respectively).

The observations of Fig. 2 indicate that the eddy resulted from anintrusion of 21 8C homogeneous water into a stratified rotating

Figure 1 The Nordic seas (NS) and an enlargement of the central Greenland Sea (GS).

a, The study area showing the 3,000 m depth contour, and b, six-month trajectories of

five floats drifting at depths from 240 m to 530 m from March until August 1997 in the

region marked by the box in a. Depth contours .2,500 m are also indicated.{ Present address: University of the Faroe Islands, PB 2109, FO-165 Argir, Faroe Islands.

letters to nature

NATURE | VOL 416 | 4 APRIL 2002 | www.nature.com 525© 2002 Macmillan Magazines Ltd

environment, and that this intrusion occurred in January 1997when air–sea heat fluxes were at their most intense. Models8 androtating tank experiments9 have shown that intrusions of fixedvolumes of homogeneous fluid into a stratified rotating mediumproduce anticyclonic lenses. These have interior cores that rotate assolid bodies, and approximately obey the relation R/

ffiffiffiBp

for arange of values of R from 0.1 to 0.4. This relationship fits theGreenland Sea eddy for which R varies from 0.5 near the centre,where maximum negative vorticity (very low potential vorticity) isobserved, to 0.125 in the skirt. Hedstrom and Armi9 observed thatlenses had a ‘fast spin-down’ phase in rotating tanks, and the ,70rotations that float no. 02 executed in the eddy core correspondroughly to the expected lifetime of this phase. This may suggest thatthe float was ejected from the core as it began to collapse.

To locate the source of the water in the eddy core, we searched forbinary combinations of surface and subsurface waters that wouldreproduce its salinity, temperature, and concentrations of CFCs, SF6

and O2 (Table 1). The high concentrations of oxygen and CFCs inthe core of the eddy make it apparent that the water there containeda substantial component that had recently contacted the atmos-phere. Scaling from these tracers, it is clear that the low SF6 contentin the eddy (,1 fmol l21) was also entirely of atmospheric origin,and not from the tracer release, despite the fact that the eddy wasembedded in water containing relatively high concentrations(,40 fmol l21) of SF6 from the release experiment. A combination

of ,36% surface Arctic waters (parent 2: P2 in Table 1) asencountered in March and May 1997 and ,64% ‘return Atlanticwater’—relatively salty water of Atlantic origin recirculatingsouthward along the East Greenland current (parent 1: P1)—which we encountered at ,500 m depth just west of the gyre inNovember 1996, fits the composition of the eddy core closely(Table 1). Regarding surface water, we have measurements fortracers in May 1997 but not for March 1997, which would havebeen preferable because it is closer to the time during whichconvection occurred (in January 1997). However, surface water inMay 1997 should be a good representative for winter surfaceconditions, except that it had warmed by ,0.3 8C. In agreementwith previous findings10, our observations indicate that surfacesalinity close to 34.80 p.s.u. at nearly freezing temperatures(21.8 8C) is necessary to trigger mixing with subsurface saltierand warmer modified Atlantic waters (,34.90 p.s.u.). This pro-duces a new vintage of Greenland Arctic Intermediate Water(GAIW), the densest water mass found within the Greenland Seagyre in this salinity range.

Our observations suggest a pattern for deep convection substant-ially different from that described in ref. 6, in which eddies are notconsidered important in initiating or sustaining convection, butonly in the dissipation of larger-scale ‘mixed patches’ formed byconvection. Following air–sea heat loss, it appears that GreenlandSea submesoscale eddies are directly generated by geostrophicadjustment and diapycnal mixing (mixing across density surfaces)between surface polar waters and subsurface modified Atlanticwaters. This results in intrusions of homogeneous GAIW charac-terized by very low potential vorticity in a stratified environment. Asimilar process has been proposed5 to be the most probable forgenerating anticyclonic submesoscale coherent vortices in theLabrador Sea. In our case, it would be expected to occur dominantly

Figure 2 Time series for the period November 1996–August 1997. a, Total sea-to-air

heat flux from NCEP (National Centers for Environmental Prediction) re-analysis. b, Float

no. 2 north-south horizontal velocity component; c, float no. 2 in situ pressure;

d, float no. 2 in situ temperature. The reanalyses data were provided through the NOAA

Climate Diagnostic Center at http://www.cdc.noaa.gov.

Figure 3 Vertical sections taken along 758 N in May 1997 in the central Greenland Sea.

a, Seawater potential density (ju), and b, SF6 concentration (fmol l21).

letters to nature

NATURE | VOL 416 | 4 APRIL 2002 | www.nature.com526 © 2002 Macmillan Magazines Ltd

at the rim of the gyre, where the parent originating from returnAtlantic water and that coming from inside-gyre surface waters,meet and mix. Because these coherent eddies are stable for ,1 year,they may also precondition water masses for convective activity inthe following winter season. They could then form foci to concen-trate further convection4,11,12 after erosion of the layer of less densewater that caps the core during the summer.

On the basis of float data only, we have direct evidence for abouteight such eddies during ESOP-2. This probably underestimates thetrue number, because almost every float we released that strayedinto the rim region of the Greenland gyre in 1997 became entrainedin an anticyclonic eddy. When these eddies eventually decay, theypresumably release their core water at mid-depths in the GreenlandSea, ventilating the intermediate water. Each eddy core has a volume,50 km3. Comparing the volume of water in eight eddy cores to atotal volume involved in convection of (2–4) £ 1012 m3 during1996–97, a figure we have previously calculated from analysis of thetracer release experiment7, suggests a contribution of 10–20% tothe total amount of convection from the eddies. However, only asmall proportion—less than 10% of the total amount of waterinvolved in convection—penetrated to around 1,000 m or deeper,most being confined to the upper ,500 m. Thus the eddies made asignificant contribution to the total volume of deep convection—and dominated the water injected to substantial depth—in theGreenland Sea in 1996–97. Greenland Arctic Intermediate Wateris thought to contribute to the overflows leading into the NorthAtlantic deep water, and the eddies thus provide a pathway forventilation of the deep North Atlantic. Long-lived submesoscaleanticyclonic vortices have also been observed in the LabradorSea13,14 (another site known for deep ocean convection), indicatingthat such vortices may be ubiquitous features of deep oceanconvection.

Received 19 October 2001; accepted 1 February 2002.

1. Rudels, B., Quadfasel, D., Friedrich, H. & Houssais, M. N. Greenland Sea convection in the winter of

1987–1988. J. Geophys. Res. 94, 3223–3227 (1989).

2. Schott, F., Visbeck, M. & Fischer, J. Observations of vertical currents and convection in the central

Greenland Sea during the winter of 1988–1989. J. Geophys. Res. 98, 14401–14421 (1993).

3. Morawitz, W. M. L., Sutton, P. J., Worcester, P. F. & Cornuelle, B. D. Three-dimensional observations

of a deep convective chimney in the Greenland Sea during winter 1988/1989. J. Phys. Oceanogr. 26,

2316–2343 (1996).

4. Lherminier, P., Gascard, J.-C. & Quadfasel, D. The Greenland Sea in winter 1993 and 1994:

Preconditioning for deep convection. Deep Sea Res. II 46, 1199–1235 (1999).

5. McWilliams, J. C. Vortex generation through balanced adjustment. J. Phys. Oceanogr. 18, 1178–1192

(1988).

6. Marshall, J. & Schott, F. Open-ocean convection: observations, theory and models. Rev. Geophys. 37,

1–64 (1999).

7. Watson, A. J. et al. Mixing and convection in the Greenland sea from a tracer-release experiment.

Nature 401, 902–904 (1999).

8. Gill, A. E. Homogeneous intrusions in a rotating stratified fluid. J. Fluid Mech. 103, 275–295

(1981).

9. Hedstrom, K. & Armi, L. An experimental study of homogeneous lenses in a stratified rotating fluid.

J. Fluid Mech. 191, 535–556 (1988).

10. Roach, A. T., Aagaard, K. & Carsey, F. Coupled ice-ocean variability in the Greenland Sea. Atmos.

Ocean 31, 319–337 (1993).

11. Legg, S., McWilliams, J. & Gao, J. Localisation of deep ocean convection by a mesoscale eddy. J. Phys.

Oceanogr. 28, 944–970 (1998).

12. Straneo, F. & Kawase, M. Comparisons of localized convection due to localized forcing and to

preconditioning. J. Phys. Oceanogr. 29, 55–68 (1999).

13. Gascard, J.-C. & Clarke, R. A. The formation of Labrador Sea Water. Part 2: mesoscale and smaller-

scale processes. J. Phys. Oceanogr. 13, 1779–1797 (1983).

14. Lilly, J. M. & Rhines, P. B. Coherent eddies in the Labrador Sea observed from a mooring. J. Phys.

Oceanogr. (special issue) (in the press).

AcknowledgementsWe thank the staff and crew of RV Hakon Mosby, RV Johan Hjort and RRS James Clark Rossfor their support. The main support for this work was from the programmes of the EU:European sub-polar ocean project (ESOP), Tracer and Circulation in the Nordic SeasRegion (TRACTOR) and Monitoring the Atlantic Inflow toward the Arctic (MAIA).Additional support from the following national agencies was also important: IFRTP(France), NERC (UK). ESOP required the involvement of a large number of individuals.We thank E. Jansen, coordinator of ESOP 2 and F. Rey for leading the RV Johan Hjortcruise; E. Fogelqvist, T. Tanhua, D. Wallace, C. Rouault and A. Lourenco for theircontributions to the tracers and floats programmes respectively. NCEP-NCAR re-analysisdata were provided through the NOAA Climate Diagnostics Center.

Competing interests statement

The authors declare that they have no competing financial interests.

Correspondence and requests for materials should be addressed to J.-C.G.

(e-mail: gascard@lodyc.jussieu.fr).

..............................................................

Fin development in a cartilaginousfish and the origin of vertebrate limbsMikiko Tanaka*, Andrea Munsterberg*†, W. Gary Anderson‡,Alan R. Prescott§, Neil Hazon‡ & Cheryll Tickle*

* Division of Cell and Developmental Biology, § Division of Cell Biology andImmunology, Wellcome Trust Biocentre, University of Dundee, Dow Street,Dundee DD1 5EH, UK‡ School of Biology, Division of Environmental and Evolutionary Biology,Gatty Marine Lab, University of St Andrews, St Andrews, Fife, KY16 8LB, UK.............................................................................................................................................................................

Recent fossil finds and experimental analysis of chick and mouseembryos highlighted the lateral fin fold theory, which suggeststhat two pairs of limbs in tetrapods evolved by subdivision of anelongated single fin1. Here we examine fin development inembryos of the primitive cartilaginous fish, Scyliorhinus canicula(dogfish) using scanning electron microscopy and investigateexpression of genes known to be involved in limb positioning,identity and patterning in higher vertebrates. Although we didnot detect lateral fin folds in dogfish embryos, Engrailed-1expression suggests that the body is compartmentalized dorso-ventrally. Furthermore, specification of limb identity occursthrough the Tbx4 and Tbx5 genes, as in higher vertebrates. Incontrast, unlike higher vertebrates, we did not detect Shh tran-scripts in dogfish fin-buds, although dHand (a gene involved inestablishing Shh) is expressed. In S. canicula, the main fin axisseems to lie parallel to the body axis. ‘Freeing’ fins from the bodyaxis and establishing a separate ‘limb’ axis has been proposed tobe a crucial step in evolution of tetrapod limbs2,3. We suggest thatShh plays a critical role in this process.

The continuous fin fold theory was once considered to be “morean established fact than a theory”3 but was subsequently questionedbecause of inconsistencies in the fossil record and in the embryologyof cartilaginous fish4. Recently discovered fossils of the earliest-

Table 1 Properties of the eddy core and its parent waters

ReturnAtlantic water*,

P1

Surfacewater†,

P2

Mixture,64% P1þ 36% P2

Eddycore‡

.............................................................................................................................................................................

Salinity (p.s.u.) 34.894 34.810 34.864 34.872Oxygen (mmol kg21) 320.5 358.7 334.2 329.7CFC-11 (pmol kg21) 4.63 7.00 5.48 5.74CFC-12 (pmol kg21) 2.22 3.34 2.62 2.69SF6 (fmol l21) 0.78 1.61 1.08 1.17Potential temperature (8C) 20.47 21.55† 2 1.8§ 20.86

2 0.9520.998

.............................................................................................................................................................................

Water mass characteristics observed at various locations and times.* 748 15 0 N, 68 W, 500 m depth, November 1996.† 758 N, 28 W, 10 m depth, May 1997.‡ 758 N, 08 E, average of 300–900 m depth, May 1997.§ Lowest temperature observed in March 1997 during winter cooling nearer the time of convection.

† Present address: School of Biological Sciences, University of East Anglia, Norwich, NR4 7TJ, UK.

letters to nature

NATURE | VOL 416 | 4 APRIL 2002 | www.nature.com 527© 2002 Macmillan Magazines Ltd