biogeochemistry of trace metals in the ocean

C S I R O P U B L I S H I N G

Volume 50, 1999© CSIRO Australia 1999

A journal for the publication of original contributions in physical oceanography, marine chemistry,marine and estuarine biology and limnology

w w w. p u b l i s h . c s i ro . a u / j o u r n a l s / m f r

All enquiries and manuscripts should be directed to Marine and Freshwater ResearchCSIRO PUBLISHINGPO Box 1139 (150 Oxford St)Collingwood Telephone: 61 3 9662 7618Vic. 3066 Facsimile: 61 3 9662 7611Australia Email: [email protected]

Published by CSIRO PUBLISHINGfor CSIRO Australia and

the Australian Academy of Science

&Marine

FreshwaterResearch

http://www.publish.csiro.au/journals/mfr

http://www.publish.csiro.au

Reliability of trace metal analysisThe practical problems involved in the chemical analysis

of trace metals in aquatic systems are now widely known(Turekian 1977; Hunter 1982; Bruland 1983; Bruland et al.1991). The concentrations of most trace metals in aquaticsystems are so low in comparison with those prevailing in thematerials and flotsam of the average chemistry laboratory(dust, skin and paint flakes) that, without special precautions,a sea-water sample is easily overwhelmed by external, spuri-ous sources of trace metals. Much of our progress in devel-oping sample collection and analysis methods appropriate tothe very low concentrations found in environmental samplescan be traced to the pioneering efforts of Clair Patterson atthe California Institute of Technology, whose painstakingwork on the measurement of Pb and stable Pb isotopes(Patterson and Settle 1976) has been the inspiration for thosewho have followed.

Reliable measurements of trace metals in natural watersrequires scrupulous attention to all steps in the analytical pro-tocol from sample collection to analysis: the preparation ofsample containers free from trace metals, rigorously con-trolled collection and handling of the sample; carefullydesigned analytical techniques, including the preparation ofspecially purified reagents; and the strict control throughoutof external sources of contamination by use of specializedclothing and clean-room working conditions (Patterson andSettle 1976; Bruland et al. 1979; Ahlers et al. 1990). Thegeneral experience is that it is the working methods andscrupulous attention to detail that makes for reliable tracemetal measurements, with clean rooms and their associatedequipment mainly providing the right kind of environmentfor those working methods.

Biologically active trace metals in the cceanThe biologically-active trace metals Fe, Mn, Co, Cu and

Zn, as well as Cd, have figured prominently in the develop-

ment and use of clean room methods of analysis (Bruland,1983; Bruland et al., 1991). Consequently, the geographicaldistributions of these elements have been systematicallystudied by several research groups in almost all of the majorregions of the global ocean and the results obtained are largelyconsistent with our understanding of oceanographic processes(Boyle et al. 1976, 1977, 1981; Bender et al. 1977; Bruland1980; Bruland et al. 1978a, 1978b, 1994; Danielsson 1980;Knauer and Martin 1981; Bruland and Franks 1983; Martinand Gordon 1988; Martin and Fitzwater 1988; Hunter and Ho1991; Frew and Hunter 1992).

The first important observation arising from these studiesis that trace metal concentrations are generally orders ofmagnitude lower than was believed in the mid-1970s(Turekian 1977; Bruland 1983; Bruland et al. 1991). Indeed,they are generally well below the practical detection limits ofmost �conventional� analytical techniques when the blankcontribution from spurious contamination is properlyaccounted for.

A second important observation is that trace metals showvertical concentration profiles that are related to those of wellunderstood, conventionally measured parameters such as tem-perature, salinity, dissolved oxygen, dissolved inorganiccarbon or the phytoplankton macronutrients phosphate, nitrateand silicate; this is a property that Boyle termed �oceano-graphic consistency� (Boyle et al. 1976). This property notonly provides a means for assessing the reliability of data, butalso provokes a number of interesting questions about the bio-logical role of trace metals (Morel and Hudson 1985).

The general behaviour of bioactive trace metals in theocean can be illustrated with data reported by Bruland (1980)for a station at 32°418N,144°598W in the North Pacific; thevertical profiles for Cd, Zn and Cu (Fig. 1) show the samesmooth trend of increasing concentration with depth in thewater column that is exhibited by the correspondingprofilesof silicate, phosphate and nitrate (Fig. 2). This similarity

© CSIRO 1999 Mar. Freshwater Res., 1999, 50, 739�53

10.1071/MF99070 1323-1650/99/080739

Biogeochemistry of trace metals in the ocean

Keith A. HunterA and Philip BoydB

ADepartment of Chemistry, University of Otago, Dunedin, New ZealandBNIWA Centre for Chemical & Physical Oceanography, University of Otago, Dunedin, New Zealand.

email: [email protected]

Abstract. A variety of metallic elements, particularly Fe and Zn, are considered essential for biologic-al processes in a variety of natural ecosystems, but are generally present only in trace amounts. Althoughtheir essential nature in terrestrial ecosystems has been known for many decades, it is only in recent yearsthat scientific methodology has improved to the point that the biogeochemical behaviour of trace metalscan be adequately studied in aquatic ecosystems such as the ocean. This paper reviews some of the recentscientific progress and demonstrates how conventional thinking about the biogeochemical role of tracemetals in the ocean has been advanced by the findings of this research. In particular, it is now clear thatthe availability of several trace elements is a factor that controls phytoplankton growth in the ocean.

Keith A. Hunter and Philip Boyd740

implies, but does not prove, that these metals are directlyinvolved, like the classical macronutrients, in the biologicalcycle of algal uptake in the surface water and heterotrophicremineralization at depth that is typical of oceanic ecosystems.

What is not revealed by the similarity in behaviour iswhether the metals are assimilated by surface-layer phyto-plankton actively or passively, i.e. whether the metals areessential for phytoplankton growth and therefore able tolimit the productivity. Resolving that question requires addi-tional evidence. We now consider three trace metal exam-ples, the first two of which exhibit some puzzling features.

Cadmium � is it an essential metal?As first noted independently by Martin et al. (1976) and

Boyle et al. (1976) and confirmed by later work, the verticaldistribution of cadmium in the ocean closely mimics that ofthe labile macronutrients phosphate and nitrate and is there-fore highly interesting because of the special role of the lattermacronutrients in regulating algal growth in the ocean.

Figure 3 compares the correlation of both cadmium andnitrate with phosphate for the results of Bruland (1980) given

in Figs 1 and 2. In this case, Cd is linearly related to phos-phate as least as well as is nitrate. Similar close correlationsbetween Cd (or nitrate) and phosphate have been reported bymany authors (Boyle et al. 1976; Bruland et al. 1978a;Danielsson 1980; Bruland and Franks 1983; Hunter and Ho1991; Frew and Hunter 1992, 1995).

However, later work has shown that the slope of theCd�phosphate relationship is not constant throughout theglobal ocean, exhibiting marked depletion in Subantarcticwaters (Nolting et al. 1991; Frew and Hunter 1992, 1995) andenrichment in some Antarctic waters (Frew 1995). Significantdepartures from the Cd�phosphate ratio typical of deepPacific waters have been noted in open-ocean surface watersand in shelf waters (Boyle et al. 1981; Hunter and Ho 1991).Figure 4 presents a summary of Cd�phosphate data for sub-surface waters of the global ocean taken from Boyle (1984)and Frew and Hunter (1992). It shows a distinctive �kink�that corresponds to the transition from Atlantic and mid-depth waters to the deep waters of the Pacific and IndianOceans. This kink is thought to be caused by injection of low-Cd Subantarctic waters into the intermediate waters of theglobal ocean (Frew and Hunter 1992) and by formation ofhigh-Cd Antarctic bottom waters near the Antarctic continent(Frew 1995).

Despite these variations, it is clear that Cd is closelyinvolved in the Redfield model of organic matter formationthrough photosynthesis and its subsequent remineralizationthrough oxidation deeper in the water column. Exactly whyit is involved so intimately is not understood at present.Cadmium has no known or specific biochemical functionand is often regarded as exhibiting only toxic properties(Kulaev 1979). Although Cd-containing metallothioneinshave been reported in many organisms, their formation isgenerally regarded to be a detoxification mechanism (Webb1979). However, Price and Morel (1990) and Lee et al.(1995) have shown that Cd2+ can substitute for Zn2+ in essen-

2.5 5.0 7.5Zinc

2 4Copper

5000

4000

3000

2000

1000

Dep

th (

m)

0.5 1.0Cadmium

Fig. 1. Vertical concentration profiles of copper, zinc and cadmium at astation in the North Pacific (all concentrations in nmol kg�1). Drawn fromdata gathered by Bruland (1980).

50 100 150Si(OH)4

1 2 3PO4

3-

5000

4000

3000

2000

1000

Dep

th (

m)

10 20 30 40NO3

-

Fig. 2. Vertical concentration profiles of nitrate, phosphate and silicate ata station in the North Pacific (all concentrations in mmol kg�1). Drawn fromdata gathered by Bruland (1980).

10

20

30

40

NO

3- (µm

ol/k

g)

1 2 3

PO43- (µmol/kg)

0.25

0.50

0.75

1.00

Cad

miu

m (

nmol

/kg)

1 2 3

PO43- (µmol/kg)

Fig. 3. Comparison of the relationships of (left panel) cadmium and (rightpanel) nitrate with phosphate for the North Pacific station used in Figs 1 and2. Drawn from data gathered by Bruland (1980).

741

tial growth mechanisms of the diatom Thalassiosira weiss-flogii. The substitution is highly effective, allowing Zn-defi-cient cells to grow at 90% of their maximum rate whensupplied with Cd. Whether Cd is sufficiently abundant inocean waters relative to Zn for this substitution mechanismto be globally important, however, remains in doubt (videinfra). Very recently, Morel�s group have succeeded in iso-lating a strain of the enzyme carbonic anhydrase that uses Cdinstead of Zn in the active site (Morel, unpublished).

Walsh and Hunter (1992) showed that the macroalgaMacrocystis pyrifera accumulates Cd 40�50 times more effi-ciently during formation of intracellular polyphosphatebodies (PPB) than in controls without PPB formation.Moreover, both Cd and phosphate concentrations in the algawere seasonally correlated and Cd concentrations weremaximal when PPB were present in the cells. PPB are ubiq-uitous phosphate storage products in most organisms andform rapidly during periods of excess phosphate supply(Kulaev 1979). Walsh and Hunter (1992) argued that if thismechanism also applies to marine microalgae, then it mayprovide an important link between the oceanic behaviours ofCd and phosphate. In particular, it may explain why Cdappears to be depleted relative to phosphate in high-latitudeand upwelling regions of the ocean (Nolting et al. 1991;Frew and Hunter 1992), because these phosphate-rich waterswould provide the best conditions for PPB formation.

Although the biochemical mechanism for Cd uptake bymarine microalgae is not yet known with certainty, the con-sistent relationship between Cd and phosphate concentra-tions in sub-surface waters of the global ocean (Fig. 4) hasbeen elegantly exploited for the estimation of historical andpaleo-chemical macronutrient levels in the ocean (Boyle andKeigwin 1982, 1985, 1987; Hester and Boyle 1982; Boyle1984, 1986; Shen and Boyle 1987). This technique is basedon the close similarity in ionic radius of the divalent ionsCd2+ and Ca2+, which means that Cd2+ substitutes into thecrystal lattice of CaCO3 minerals formed by foraminifera,coccoliths and aragonitic corals. Saager (1994) has reviewedthe relationships between measured concentrations of Cd andphosphate in the global ocean and their implications for theuse of Cd as a paleo-oceanographic tracer.

Zinc � another puzzleEarly reliable measurements of zinc made by Bruland�s

group (Bruland et al. 1978a; Bruland 1980; Bruland andFranks 1983) showed a close similarity between the verticalprofiles of Zn and the non-labile macronutrient silicate (Figs1 and 2). For example, the close correlation (Fig. 5) betweenZn and silicate for the North Pacific station discussed earlierimplies that Zn is taken up into the mineral parts of phyto-plankton such as the CaCO3 or SiO2 exoskeleta of coccol-ithophores and diatoms respectively, or the separate siliconpool in diatoms. These biogenic mineral components of phy-toplankton are generally more refractory than the organic


0.25

0.50

0.75

1.00

[Cd]

(nm

ol k

g-1 )

1 2 3

[PO4] (µmol kg-1)Fig. 4. The global cadmium�phosphate relationship for sub-surfacewaters. Drawn from data compiled by Boyle (1984). Large open circles rep-resent samples collected in low-Cd Subantarctic waters (Frew and Hunter1992). Solid lines are least-squares regression lines drawn through datahaving [PO4

3�] < 1.4 mmol kg�1 (mainly intermediate and Atlantic waters)and having [PO4

3�] > 1.4 mmol kg�1 (mainly Pacific and Indian deep waters).

2.5

5.0

7.5

Zin

c (n

mol

/kg)

50 100 150

Si(OH)4 (µmol/kg)Fig. 5. Relationship between zinc and silicate concentrations at a stationin the North Pacific. Drawn from data compiled by Bruland (1980, see alsoFigs 1�3).


tissue containing the labile macronutrients and thereforehave a deeper regeneration cycle in the ocean, as can be seenby comparing the vertical profile of silicate with those ofphosphate and nitrate (Fig. 1).

Examination of further datasets reveals that, in fact, theZn�Si relationship curves downwards slightly (Fig. 6); thisimplies that a fraction of the Zn is remineralized at shallowerdepths than silicate, perhaps indicating that some of the Znhas been taken up into the soft tissue component of the phy-toplankton.

The observation that Zn is closely correlated with silicatedoes not imply, by itself, that Zn is actually carried by bio-genic SiO2 phases rather than CaCO3 phases, because bothmaterials have rather similar dissolution profiles down thewater column. Indeed, measurements of Zn in diatom opalcollected in plankton tows (Martin and Knauer 1973; Collierand Edmond 1984) and in fossil diatoms sampled from deep-sea sediment cores (Ellwood and Hunter 1999) show thatonly a minor fraction of Zn becomes incorporated into theopal exoskeleton.

Unlike Cd, Zn is a recognized essential trace element forplankton growth. It is an essential cofactor in several impor-tant enzyme systems, including alkaline phosphatase and thecarbonic anhydrase needed by diatoms to dehydrate seawaterHCO3

� to provide CO2 for photosynthesis (Morel et al. 1994).

Speciation of trace metals in the oceanThe data for trace metal concentrations discussed above

have all been obtained by variations of the same analyticaltechnique in which metal ions have been extracted from sea

water by complexing with a strong chelating ligand such asammonium pyrrolidine dithiocarbamate, or equivalentmethods using ligand functional groups immobilized on asolid phase (Boyle and Edmond 1975; Bruland et al. 1979;Danielsson et al. 1978). Because the ligands used are verystrong complexing agents for the metals determined (i.e. havevery high stability constants), almost all complexes formedby metal ions with organic and inorganic ligands naturallypresent in sea water undergo ligand exchange with the ana-lytical complexing agent. In addition to the high stability ofthe analytical complexing agent, it is also common to carryout metal ion extraction at a solution pH much lower than theoriginal sea water, further assisting in the release of free metalion for solvent extraction. Thus, these techniques measure thetotal concentration of metal ion present in solution.

It has long been recognised that the total solution concen-tration of a metal ion is not particularly valuable in assessingits bioavailability (Anderson and Morel 1982), or indeed fora variety of chemical processes such as metal ion�particlereactivity (Davis and Leckie 1978; Morel 1983). It nowseems clear that the uptake of metal ions by cellular organ-isms is controlled by the concentration of the free metal ionin solution (strictly speaking, it should be the activity of thefree metal ion; however, concentration and activity arerelated to each other through the activity coefficient of theion, which in sea-water medium will have a constant value).This leads to important considerations about how metal ionsinteract with organisms. In the presence of complexingligands, the free ion concentration may be orders of magni-tude lower than the total concentration in solution. Moreover,the free ion concentrations of different waters cannot neces-sarily be compared with each other by referring to total metalion concentrations because the concentration and nature ofligands may be different. Therefore, it is important to be ableto measure the speciation of metal ions, i.e. the concentra-tions (or activities) of all the complexes formed with natu-rally occurring organic and inorganic ligands.

The inorganic speciation of metal ions is readily calculatedfrom the known concentrations of the main complex-forminganions in seawater (OH�, Cl�, SO 4

2�, CO32� Br� and F�) and

available thermodynamic data for their complexes formedwith different metal cations (Turner et al. 1981). By compar-ison, isolation and direct study of metal�organic complexesin sea water is much more difficult, with the exception oftetrapyrrole complexes such as Vitamin B12 and the chloro-phylls. This is because the concentrations of both metal ionand ligands are extremely low, making attempts to measurespeciation highly susceptible to contamination artifacts. Inrecent years, however, the techniques of clean chemical anal-ysis discussed above have been applied to the study of tracemetal speciation in sea water, with very satisfactory results.

Several techniques have been used for measuring the con-centration of metal�organic complexes in sea water. Theseinclude competitive techniques in which the sea water is

2.5

5.0

7.5

50 100 150

Si (mmol/kg)

Zn

(nm

ol/k

g)

Fig. 6. Relationship between concentrations of Zn and reactive Si atselected stations in the global ocean. n, Northeast Pacific (Bruland et al.1978b); +, North Atlantic (Bruland and Franks 1983); ´, Indian Ocean(Saager et al. 1992); h, Drake Passage, Antarctica (Martin et al. 1990c); s,N. Equatorial Pacific (Bruland 1980).

743

equilibrated with an synthetic complexing ligand of knownconcentration and stability constant that closely matches thenatural ligands in the sea water (Van den Berg 1982, 1984;Moffett and Zika 1987; Moffett et al. 1990) and kinetic tech-niques in which electrochemical methods are used to distin-guish between electrochemically labile forms of the metaland inert organic complexes (Tuschall and Brezonik 1981;Huizenga and Kester 1983; Waite and Morel 1983; Coale andBruland 1988, 1990; Bruland 1989, 1992).

These methods yield data for the (operationally defined)concentrations of metal�organic complexes in a sea-watersample. However, if the uncomplexed free ligands in thesample are titrated with known additions of the metal ion ofinterest, more useful information on the concentrations andapparent stability constants of the ligand(s) may be obtained(Coale and Bruland 1988, 1990; Bruland 1989, 1992).

Speciation of zinc(II) Speciation has been measured by an electrochemical

kinetic technique for water depths down to 600 m in thenorth-eastern Pacific Ocean(Bruland 1989). From the resultsit is evident that the ligand concentration exceeds that of totalZn throughout the upper 360 m, with the result that only aminor fraction of Zn is present in an inorganic form over thisdepth range; below 500 m, the ligand concentration becomessmall relative to that of total Zn(II) and it exercises littlecontrol over Zn speciation (Fig. 7left). The free ion concen-tration of Zn2+ (calculated from the measured ligand concen-trations and complex stability constants at different depths)is maintained well below 1 pmol L�1 throughout the upper200 m, three orders of magnitude lower than the total Zn con-centration (measured by the conventional analytical methodsdescribed earlier) (Fig. 7right; note that a logarithmic scale isused in the figure). The dominance of such a strong ligand incontrolling the speciation of Zn over this biologically impor-tant depth range is compelling indirect evidence that theligand has a biological origin and may be produced by one ormore organisms for a specific biological purpose.

Speciation of copper(II)Like Zn(II), Cu(II) speciation in the ocean is dominated

by a highly selective strong ligand in the biologically activeupper ocean. Coale and Bruland (1988) found two Cu-binding ligands: a strong ligand L1 having a conditional for-mation constant logK1 = 11.5, found only in surface waters(0�200 m depth, maximum concentration 2 nmol L�1); and aweaker ligand L2 having a conditional formation constantlogK2 = 8.5, found at higher concentrations (8�10 nmol L�1)throughout the water column. This weaker ligand is probablya component of the ubiquitous dissolved organic matter inseawater generated by degradation of plant and animaltissues, and is similar in nature to the humic acids present infreshwater and soil systems.

The strong ligand L1 exists only in a narrow depth rangenear the surface (Fig. 8left), strongly suggesting that it has adirect biological origin. It exceeds total Cu(II) in concentrationin the euphotic zone (upper 200 m), with the result that the freeCu2+ concentration (Fig. 8right) is decreased by two orders ofmagnitude relative to other depths in the water column.

At depths greater than 200 m, the concentration of L1decreases to zero and that of uncomplexed, inorganic Cu(II)accordingly increases. However, inorganic Cu(II) remains lessthan 50% of total dissolved Cu(II) because of the effects of theless-strongly binding ligand L2. The free Cu2+ concentration inthe euphotic zone is kept well below 0.1 pmol L�1 (Fig. 8right)as a consequence of complexing with the strong ligand.


10-12 10-11 10-10 10-9

Free Zn2+ (mol/L)

600

500

400

300

200

100

Dep

th (

m)

1 2 3 4Concentration (nmol/L)

Ligand

Zninorganic

Zntotal

Fig. 7. (left panel) Vertical profiles of (h) total dissolved Zn, (j) inor-ganic Zn and (n) ligand concentrations at a station in the Northeast Pacific.(right panel) Vertical profiles of (s) free Zn2+ concentration and (h) totalZn(II) concentration. Drawn from data reported by Bruland (1989).

10-13 10-12 10-11

Free Cu2+ (mol/L)

1400

1200

1000

800

600

400

200

Dep

th (

m)

0.5 1.0 1.5 2.0Concentration (nmol/L)

Ligand L1

Cuinorganic

Cutotal

Fig. 8. (left panel) Vertical profiles of (h) total dissolved Cu, (j) inor-ganic Cu and (n) strong ligand L1 concentrations at a station in theNortheast Pacific. (right panel) Vertical profile of free Cu2+ concentration.Drawn from data reported by Coale and Bruland (1988).


These findings, which provide compelling evidence of abiological source for strong Cu-complexing ligands, havebeen confirmed by several more recent studies which confirmthat strong ligand complexing of Cu(II) is greatest at the depthof the biological productivity maximum and decreases by 1�3orders of magnitude down the water column. This has beenfound elsewhere in the north-eastern Pacific (Coale andBruland 1990) and in the Sargasso Sea (Moffett et al. 1990).The latter authors provided the first direct evidence that astrong Cu-binding ligand is produced by a marine phyto-plankter: in laboratory cultures involving four marine phyto-plankton species (three eucaryotes and one procaryote), thecyanobacterium Synechococcus sp. produced a Cu-bindingligand whose conditional formation constant was identical tothe strong Cu-binding ligand, termed L1, found in sea waterby the same authors and by Coale and Bruland (1988, 1990).This organism is widespread in the ocean and may thereforehave a strong influence on the global speciation of Cu(II) insurface waters. These results are also consistent with earlywork by McKnight and Morel (1979) who reported that onlycyanobacteria seem to exude strong Cu-binding ligands.

The production of a strong Cu-binding ligand bycyanobacteria in the ocean may represent the result of selec-tive evolutionary pressure to detoxify the environment bylowering the Cu2+ concentration. This is supported by mea-surements of the reproduction rates of 38 marine phyto-plankton species in which the Cu2+ concentration wascontrolled through the use of a metal ion buffer system(Brand et al. 1986); cyanobacteria were the most sensitive tothe toxic effects of Cu and diatoms were the least sensitive.The growth rates of the cyanobacteria were diminished atconcentrations of free Cu2+ greater than 0.4 pmol L�1,whereas most of the eucaryotic algae were able to maintainoptimal growth rates at free ion concentrations as high as 4nmol L�1. If these results are compared with the Cu2+ con-centrations observed in the right panel of Fig. 8, it is clearthat in the euphotic zone, the free Cu2+ concentrations areuniformly low enough not to limit the growth rates of eventhe most copper-sensitive cyanobacteria.

Speciation of cadmium(II)The general findings of speciation studies on Cd(II) in the

north-eastern Pacific (Bruland 1992) are very similar (Fig. 9)to those already discussed for Zn and Cu. A strong Cd-com-plexing ligand is found in the upper 150 m of the watercolumn, having a maximum concentration of 0.1 pmol L�1

and a conditional formation constant of logK = 12. In thiscase, the ligand exceeds total Cd(II) in concentration down toalmost 200 m depth, with the result that the free Cd2+ con-centration in this depth range is as much as 3 orders of mag-nitude lower than in the deep ocean. This must also beviewed in the light of the very strong inorganic complexingundergone by this ion which means that 97% of the inorganicCd(II) exists as complexes formed with chloride ion.

Price and Morel (1990) showed that the growth rates ofphytoplankton are reduced at low Zn2+ concentration (<0.2pmol L�1) as a result of insufficient Zn(II) availability. Thisconcentration is comparable to that observed in the north-eastern Pacific (Fig. 7). They also showed that such Zn-defi-cient cultures could maintain up to 90% of the maximumgrowth rate when supplied with Cd(II). They suggested thatthis biochemical substitution of Cd for Zn could be theprocess that accounts for the pronounced similarity betweenCd and the labile macronutrients. However, Fig. 10, whichcombines the results in Figs 7�9, shows that the free Cd2+

10-13 10-12 10-11

Free Cd2+ (mol/L)

600

500

400

300

200

100

Dep

th (

m)

0.2 0.4 0.6 0.8Concentration (nmol/L)

Ligand

Cdtotal

Fig. 9. (left panel) Vertical profiles of (h) total dissolved Cd and (n)ligand concentrations at a station in the Northeast Pacific. (right panel)Vertical profile of free Cd2+ concentration. Drawn from data reported byBruland (1992).

300

200

100

Dep

th (

m)

10-13 10-12

Free ion concentration (mol/L)

Zn2+

Cu2+

Cd2+

Fig. 10. Comparison of the vertical depth profiles of free ion concentra-tion of Zn2+, Cu2+ and Cd2+ taken from Figs 7�9. Drawn from data reportedby Bruland (1989, 1992) and Coale and Bruland (1988).

745

concentration is well below that of Zn2+, at least in the north-eastern Pacific, ruling out the likelihood of significant sub-stitution of Cd2+ for Zn2+. Thus, the unusual behaviour of Cdremains a mystery to be uncovered by further research.

How does speciation affect plankton growth?The knowledge gained from speciation studies of this type

has opened a door to studying how trace metals interact withphytoplankton, because it provides information on the freemetal ion concentrations that must be reproduced in labora-tory cultures in order to provide culture conditions that real-istically model the ocean. In particular, the realization thatfree metal ion concentrations are controlled at levels of 1pmol L�1 or less by the presence of strong ligands explainsthe puzzling, but widely known, observation that marinephytoplankton cannot be successfully cultured in the labora-tory unless a chelating agent such as EDTA is added to theculture medium. We can now see that the role of the chelat-ing agent is to reduce the free ion concentrations of the metalions in the medium to the sub-pmol L�1 range.

The concentrations of free metal ion in a culture mediumcan be set to a desired value by adjusting the proportions ofcomplexing agent (EDTA) and total metal ion added, takingaccount of the known thermodynamic properties of themedium constituents (Morel et al. 1979). This has nowbecome a standard approach for the study of metal ion�phy-toplankton interactions (e.g. Rueter and Morel 1981; Morelet al. 1991, 1994; Sunda and Huntsman 1992, 1995). As anexample, Fig. 11 shows how the growth rates of four phyto-plankton species are affected by free Zn2+ concentration overranges typical of the ocean as reported by Sunda andHuntsman (1992). The growth rates of the two oceanicspecies (Thalassiosira oceanica and Emiliania huxleyii)were unaffected by Zn2+ concentration in the range 1�1000

pmol L�1, whereas those of the two neritic species(Thalassiosira pseudonana and Thalassiosira weissflogii)were significantly reduced at concentrations below 10 nmolL�1. In addition, T. weissflogii growth rates were diminishedat Zn2+ concentrations greater than 10 nmol L�1, presumablythrough toxic effects.

From the same study, the specific uptake rate of Zn(II) bythe phytoplankton, again expressed as a function of the freeZn2+ concentration, exhibits an �S�-shaped curve characteris-tic of regulated uptake (Fig. 12). Indeed, in the large range ofconcentrations (5�5000 pmol L�1), the specific Zn(II) uptakerate changes by less than a factor of 10. This dramaticallydemonstrates the ability of these phytoplankters to regulatetheir Zn(II) uptake under conditions of widely-varying Zn2+

availability. Moreover, the close similarity in the uptake ratesof the four species suggests a common chemical mechanismfor Zn(II) uptake. This may be, for example, a Zn(II)-basedenzyme system such as alkaline phosphatase on the exteriorwall of the cell. Sunda and Huntsman (1992) showed that theZn:C ratios observed in their cultures in the regulated regionwere similar to those deduced from concentration profiles inthe oceanic nutricline.

Distribution of iron in the oceanIron is perhaps the most bioactive of the trace metals in the

ocean, but its chemistry is very complex and poorly under-stood (Bruland et al. 1991; Wells et al. 1995). The main oxi-dation state is Fe(III), an ion that is strongly hydrolysed at thepH of sea water, resulting in a high ratio of hydrolysed prod-ucts to free Fe3+ ions. Turner et al. (1981) and Byrne et al.(1988) separately estimated from thermodynamic data thatthe ratio of hydrolysed to free Fe3+ is about 1012 at pH 8.2,25°C. Their calculations use thermodynamic data that indi-


0.5

1.0

1.5

Spe

cific

Gro

wth

Rat

e (d

ay-1

)

10-11 10-10 10-9 10-8

Free Zn2+ (mol/L)

E. huxleyii BT6

T. pseudonana

T. weissflogii

T. oceanica

Fig. 11. Effect of free Zn2+ concentration on the growth rates of four phyto-plankton species. Drawn from data reported by Sunda and Huntsman (1992).

101

102

µmol

Zn

mol

C-1

day

-1

10-11 10-10 10-9 10-8

Free Zn2+ (mol/L)

T. oceanicaE. huxleyii BT6T. pseudonanaT. weissflogii

Fig. 12. Effect of free Zn2+ concentration on the specific uptake rate of Znrelative to carbon. Drawn from data reported by Sunda and Huntsman (1992).


cate that Fe(OH)30 is the predominant hydrolysis product. By

contrast, Hudson and Morel (1990) discounted the impor-tance of Fe(OH)3

0 and calculated a ratio of 1010, withFe(OH)2

+ being the major hydrolysis product.Chemical analysis of �dissolved� Fe(III) in sea water uses

filtration of the water sample through a membrane filterhaving a pore size in the range 0.2�0.5 mm as an operationaldefinition of the dissolved state. For most trace metals theproportion of metal found on the filter is very small com-pared with that passing through, which makes it unlikely thata significant fraction of these elements is present in a col-loidal state having a particle size close to that of the filterpores. However, for Fe, a major fraction is clearly particu-late; hence, it is highly likely that most, if not all, of the Fepassing through a membrane filter is actually in the form ofcolloidal particles.

Landing and Bruland (1981) and Gordon et al. (1982)made the first reliable studies of dissolved and particulate Fe,in the north-eastern Pacific Ocean. They found that verticalprofiles of dissolved iron had the following features incommon: iron was severely depleted (0.15�0.30 nmol kg�1)in surface waters and varied little in mid-depth waters(0.5�1.0 nmol kg�1). They also reported that total iron pro-files followed a similar pattern. More recently, numerousstudies have increased the global ocean coverage for irondata and increased the iron fractions under investigation(Danielsson et al. 1985; Symes and Kester 1985; Landingand Bruland 1987; Martin and Gordon 1988; Sherrell andBoyle 1988; Martin et al. 1989, 1990c; Saager et al. 1989,1993; Lewis and Landing 1991; Nolting et al. 1991;Westerlund and Ohman 1991).

In some regions, surface waters show evidence of atmo-spheric Fe input in waters above the thermocline (Bruland etal. 1991). Duce (1986) estimated that >95% of the Fe supplyto oceanic surface waters comes from atmosphere transportand deposition. However, more recently it has become clearthat upwelling of subsurface waters and ice melting are alsoimportant components of the iron supply budget (de Baar etal. 1995). The main source of iron in equatorial Pacificwaters is upwelling (Coale et al. 1996a).

Johnson et al. (1997a, 1997b) presented a thoughtful butprovocative analysis of information on the concentration dis-tribution of dissolved Fe in the world ocean as the leadingpart of a commentary series in the journal Marine Chemistry.They argued that this distribution is remarkably uniform,with very low values (<0.2 nM) in surface waters and aver-aging 0.76 nM in deep waters. Depth profiles were verysimilar to those of the labile nutrients NO3

� and PO43�, indi-

cating surface water uptake followed by deep water regener-ation as biogenic organic matter is oxidized. However, theypoint out that, unlike the labile nutrients, there is no inter-hemispheric fractionation. They ascribe this to control ofdeep-water Fe concentrations through complexing withorganic ligand(s) having concentration(s) up to 0.6 nM. They

put forward a simple 1-dimensional model in which Fe atconcentrations exceeding that of the ligand is rapidly scav-enged, giving an Fe supply to deep waters that is limited bythe rate of carbon export, rather than by local aeolian input.Their model produced calculated Fe concentration profilesthat were in good agreement with the measured data.

Boyle (1997) argued that this 1-dimensional model wastoo simplistic because it failed to account for simple proper-ties such as temperature. Moreover, he pointed out thatorganic complexation was not a necessary component of Febehaviour, since the radionuclide Th-230 also shows auniform distribution in deep waters and is not considered tobe complexed in the same way. Finally, he argued that thereare substantial gaps in the geographical distribution of mea-sured Fe concentrations, particularly in the South Pacificwhere the available data suggest much lower concentrationsthan argued by Johnson et al. (1997a). Indeed, this has beenconfirmed by measurements of de Baar et al. (1999).

Sunda (1997) criticized the Johnson et al. (1997a) modelon the grounds that it implies a constant Fe:C ratio in theorganic matter exported into deep waters. He indicated thatculture experiments show that Fe uptake by diatoms dependson the concentration of available Fe in the medium, whichwould imply a variable efficiency of Fe export given the largevariations in available Fe concentrations in surface waters.Furthermore, he points out that the correlations between dis-solved Fe concentration and apparent oxygen utilization pre-sented by Johnson et al. (1997a) support this notion.

Finally, Luther and Wu (1997) presented further evidencefor the organic complexation of Fe in sea water that supportsthe contention of Johnson et al. (1997a).

Iron limitation of plankton growthThe hypothesis that iron is a limiting factor for phyto-

plankton growth in regions of the open ocean (usually termedthe Iron Hypothesis) is not new. For example, Gran (1931,p.41), on the basis of growth experiments in culture, con-cluded that the low concentration of iron in sea water proba-bly limited plant growth in areas where it was not replenishedby land drainage. Later Hart (1934) suggested �Among the ...chemical constituents of sea water ... possibly limiting phy-toplankton production, iron may be mentioned ... it may helpto explain the observed richness of the neritic plankton ... theland being regarded as a source of iron�.

It is important to note that reliable techniques for measur-ing trace metals did not exist at that stage. Subsequently, theiron content of sea water was erroneously believed to rangefrom 50 to 400 nmol kg�1 (Harvey 1937a, 1937b).

In more recent years, the possibility that iron limitation maycontrol phytoplankton productivity in large regions of theocean has become widely accepted. This is particularly true forAntarctic waters, where phytoplankton stocks and productionremain relatively low all year round (even during the growthseason) despite persistently high surface concentrations of the

747

macronutrients. A high supply of macronutrients from depth isassured by continuous upwelling at the circumpolar continen-tal margins. Two other large regions of the world�s oceansexhibit these so-called high-nitrate, low-chlorophyll (HNLC)conditions in their surface waters: the Subarctic and Equatorialregions of the Pacific (Cullen 1991, 1995). In all, HNLCregions compose about 30% of the world ocean.

Martin and Gordon (1988) compared dissolved iron con-centrations at three deep-water stations along a 1600 kminshore�offshore VERTEX transect in the north-easternPacific with four shallow California continental-margin sta-tions. They deduced that advective processes within the oceancould supply only a few percent of the iron required by open-ocean phytoplankton. The majority of iron in surface watersin this region must be supplied by atmospheric deposition.Martin et al. (1989) extended this work to a second series ofVERTEX stations in the Gulf of Alaska. They postulated thatatmospheric deposition rates to the HNLC Subarctic regionwould not be sufficient to meet phytoplankton demand.

Analyses of the seasonal and spatial distribution of con-centrations of phytoplankton pigment in the Southern Ocean,as measured by the Coastal Zone Color Scanner, found thatphytoplankton blooms were not uniformly distributed(Comiso et al. 1993; Sullivan et al. 1993). Blooms appearedto be located in regions of ice retreat such as the Scotia andRoss Seas, in relatively shallow areas (e.g. the Patagonian andNew Zealand shelves) and in some regions of Ekmanupwelling (e.g. in the Tasman Sea). Comiso et al. (1993) sug-gested that the �high negative correlation (especially in bloomregions) with bathymetry is consistent with the abundance ofiron in shallow waters and its paucity in the open ocean�.However, in some open ocean areas where aerosol radianceswere high, the chlorophyll pigment concentrations were alsohigh, suggesting a possible influence of aeolian iron input.

In addition to atmospheric deposition and iron-rich sedi-ments in shallow waters, iron may also be supplied to off-shore waters by melting sea ice (de Baar et al. 1995). Thebioavailability of iron in sea ice is unknown. Blooms of phy-toplankton often occur at the edge of the ice in spring becauseof the formation from melting ice of a shallow verticallystable upper layer. Subsequent seeding by actively growingalgae from the sea ice may promote bloom formation.

The eastern and central equatorial Pacific ocean exhibitsHNLC conditions. Surface concentrations of nitrate rangefrom 4 to 12 mmol kg�1 whereas concentrations of chloro-phyll a never exceed 0.5 mg kg�1 (Chavez and Barber 1987).Natural phytoplankton communities in this region are domi-nated by picoplankton and nanoplankton, with a combinationof high primary production rates and high grazing rateskeeping standing stocks of phytoplankton low (Chavez et al1991). Atmospheric dust levels in the equatorial Pacific areamong the lowest in the world.

Iron limitation has also been observed in the coastalwaters of the California Current, which receive periodic

inputs of macronutrients from deep-water upwelling, withthe primary source of iron for phytoplankton growth beinginput from shelf sediments (Johnson et al. 1999). A similarbenthic iron source operates in coastal waters to the east ofsouthern New Zealand (Croot and Hunter 1998).

Phytoplankton growth in vitro with ironIn their second VERTEX study in the Gulf of Alaska,

Martin et al. (1989) included in vitro phytoplankton growthexperiments in which the effect of adding nanomolaramounts of iron was investigated. They found that ironenhanced the growth of phytoplankton from Subarctic watersand induced a change in species abundances. After additionof iron, the larger diatom cells became more abundant thanthe smaller nanoflagellates which normally dominate undernatural, low-iron conditions. They concluded that the inputof atmospheric iron to the Gulf of Alaska was sufficient tosupport moderate productivity, but not large enough to fullydeplete available macronutrient levels.

Martin et al. (1993) performed further iron enrichmentexperiments as part of the JGOFS North Atlantic BloomExperiment. They found no evidence of iron deficiency.They concluded that the iron requirements of the phyto-plankton were probably being met by lateral transport fromthe continental margins. In agreement with this, the NorthAtlantic does not typically exhibit HNLC conditions.

Martin et al. (1990a, 1990b) investigated the possibility ofiron limitation on Antarctic phytoplankton. They found thatdissolved iron levels were high (7.4 nmol kg�1) in the high-productivity neritic waters of the Gerlache strait, but verylow (0.16 nmol kg�1) in the offshore Drake Passage waters.Iron-enrichment experiments were also performed on watersamples from the Ross Sea. Nitrate uptake rates were 2 to 10times higher after addition of unchelated iron than in thecontrol samples. Total decreases in nitrate were balanced byincreases in particulate nitrogen. By contrast, addition ofmanganese to water samples produced no measurableincrease in productivity.

These iron-enrichment experiments and their results weresoon subject to much scrutiny as controversy raged overMartin�s iron-limitation hypothesis (Banse 1990, 1991a,1991b; Martin et al. 1990a, 1990b, 1990c, 1991a, 1991b). Atissue was the interpretation of growth rates and whether theexclusion of grazing zooplankton from the sample bottleshas a significant effect. For example, the absence of grazingpressure may allow phytoplankton to grow unhindered (Frost1991). In addition, the observation that control samplesexhibited at least some growth was considered to be signifi-cant evidence that other, unknown processes might be occur-ring in the experiments. Attention also became focused onthe rates of nitrate uptake as a means of determining theamount of new production versus regenerated production(Banse 1991a). Finally, concern was also raised over thedegree to which iron was adsorbed to the walls of the poly-



carbonate culture bottles and how this might be wronglyascribed to iron uptake by the plankton (Banse 1991a). Boydet al. (1996) showed that the likely reason for elevatedchlorophyll levels in the controls was inadvertent iron con-tamination, demonstrating a need for careful preparation ofsamples and controls. They also reported that the problem ofunder-representation of grazers could be overcome bycareful sampling techniques and the use of large-volumecarboys for incubations.

Dugdale and Wilkerson (1990) reanalysed the data fromthe iron-enrichment experiments in terms of specific uptakerates and found no evidence for enhanced phytoplanktongrowth. They suggested that adding iron might have a toxiceffect on grazers or favour the growth of larger phytoplank-ton cells indigestible to grazers. The unknown effects of het-erotrophic bacterial populations were also thought toinfluence growth rates.

Biochemical indicators of iron limitationOne strategy by which phytoplankters can reduce their

iron quotas and adapt to low iron conditions is by substitu-tion of iron-containing proteins such as ferredoxin by func-tionally similar proteins that do not contain iron, e.g. theredox protein flavodoxin (Entsch et al. 1983). La Roche et al.(1996) showed that flavodoxin can be used in situ as amarker of iron stress in phytoplankton growing in HNLCwaters. Using polyclonal antisera raised against flavodoxinthat had been isolated from the diatom Phaeodactylum tri-cornutum, they measured flavodoxin abundance relative tototal protein in extracts from phytoplankton samples alongan east�west transect from the British Columbia coast to theformer site of ocean station Papa (P26) on two occasions. Ingeneral, flavodoxin expression in microorganisms is regu-lated by iron availability, although there is taxonomic vari-ability ranging from constitutive expression to a completeabsence of flavodoxin. For example, when La Roche et al.(1996) added iron to iron-depleted water at station P26,flavodoxin levels showed transient reductions. Cellularflavodoxin levels in the low-iron samples (stations P26 andP20, farthest from the coast) were greatly enhanced (Fig. 13).Data for Subantarctic waters in the Otago, New Zealand,shelf system (Boyd et al. 1999) exhibit the same trends asthose for the Papa transects (Fig. 13). These results implythat in these HNLC waters, diatoms are able to reduce theiriron requirements and hence reduce their iron stress byexpressing flavodoxin in substitution for the iron-containingenzyme ferredoxin.

Iron and climate changeThe discovery that glacial atmospheric CO2 concentra-

tions were lower than those in interglacial time has led tomany hypotheses. Several explanations for this change dealwith variations in phytoplankton productivity in the Southern

Ocean and the related use of macronutrients. It is suggestedthat if a major fraction of the abundant macronutrients in theSouthern Ocean has been consumed in additional phyto-plankton growth, the resulting bloom may have resulted in asignificant draw-down of atmospheric CO2.

Martin (1990, 1992) extended his iron-limitation hypoth-esis to suggest that iron may have been more available toSouthern Ocean phytoplankton during glacial times. Martinused dust data (De Angelis et al. 1987) obtained from theVostok ice core to compare iron input to the glacial and inter-glacial Southern Ocean. Figure 14 compares the glacial CO2record with estimates of the iron concentration in glacial icearising from the presence of atmospheric dust. High dustlevels, such as those seen in glacial times, imply greatererosion of continental soils, e.g. through stronger windsand/or larger arid areas. The results suggest that duringglacial times, the atmospheric input of iron to remote areas,including the ocean, may have been about 50 times higherthan it is during present-day, interglacial times. Thus, it wassuggested that �iron-rich dust blew into the Antarctic, thephytoplankton bloomed, the biological pump turned on, andCO2 was withdrawn from the atmosphere� (Martin 1990).

Evidence for increased productivity in the Southern Oceanduring glacial periods, however, is still controversial.Sedimentary records of silicate preservation in the SouthernOcean have, owing to the lower abundance of opal, beeninterpreted as showing decreased glacial productivity (Bergerand Wefer 1991). However, the picture is complicated by thenon-preservation of CaCO3 in Southern Ocean waters and bythe absence of opal from many major species of plankton in

20

40

60

80

% M

ax F

lavo

doxi

n

1 2

[Fe] (nmol L-1)

Sep 95

May 95

Subantarctic

Fig. 13. Comparison of flavodoxin abundance (as a percentage of themaximum observed in each dataset) with dissolved iron concentrations (nmolL�1) along an east�west transect off the British Columbia coast to the site ofthe former ocean station Papa P26 (drawn from data of La Roche et al. 1996);and in Subantarctic waters to the east of southern New Zealand (drawn fromdata of Boyd et al. 1999)

749

the Southern Ocean (Martin 1992). Non sea-salt sulfur asfound in the Vostok ice core (Legrand et al 1991), when usedas a proxy for dimethyl sulfide, indicates an increase in phy-toplankton productivity during glacial times.

Glacial-age productivity in the equatorial Pacific and ineastern-boundary upwelling systems was enhanced, proba-bly because of stronger glacial winds promoting increasedupwelling of macronutrients (Berger and Wefer 1991). Prahl(1992) presented organic geochemical evidence in support ofthe iron hypothesis for the eastern tropical Pacific: the depo-sition of total organic carbon (primarily derived from marineproductivity) was highly correlated with the input of aeoliandust,and biomarkers for prymnesiophyte productivity pre-ceded the total organic carbon maximum by 4�6 thousandyears. This, Prahl suggested, is evidence for �an ecologicalchange in phytoplankton community during the last glacialtransition as a consequence of systematic variation in thesupply of eolian dust and associated bioavailable iron�.

Iron enrichment experiments in situ on the meso scaleOwing to the problems exposed in using small sample

bottles to perform iron enrichment studies, plans were devel-oped to enlarge the system under investigation. Feasibilitystudies were made for a meso-scale iron enrichment experi-ment to be carried out in situ in the Equatorial Pacific (Anon.1992; Watson et al. 1991). Martin (1990) proposed that CO2could be removed from the atmosphere by fertilizing theocean with iron, which might amplify the biological uptakeof carbon from surface waters. By such a mechanism, theenhanced �greenhouse� effect from increased anthropogeniccarbon emissions may be reduced. This idea (dubbed theGeritol solution to Greenhouse warming) ignited a storm ofcontroversy over the ethics of deliberately perturbing healthyecosystems (Baum 1990; Roberts 1991).

The first meso-scale experiment, IRONEX I, took place inthe equatorial Pacific Ocean in October 1993 (Martin et al.1994). Surface sea water at the study site was inoculated with15 600 L of Fe(III) solution pumped in to the propeller washof a research ship during a 24-h period, adding a total of 450kg Fe to the surface water. At the same time, the readilydetected and chemically inert tracer sulfur hexafluoride (SF6)was also added as a tracer for the movement of the Fe-enriched patch of water. Measurements of Fe and SF6 con-centrations, as well as those of macronutrients andchlorophyll pigments, were performed in a grid patternacross the study site for several days after the iron enrich-ment. Dissolved Fe concentrations as high as 6.2 nmol L�1

were measured in the core of the patch within 4 h of fertil-ization. These decreased rapidly on the first day as a result ofnight-time convective mixing, with the highest values on thesubsequent day being 3.6 nmol L�1. Thereafter, concentra-tions in the core of the patch decreased by about 15% per day.Contour plots of both dissolved Fe and the SF6 tracer were ingood agreement 3 days after the fertilization.

Photosynthetic competency of the phytoplankton (activefluorescence) was the first biological response observed afterthe fertilization (Kolber et al. 1994). Active fluorescence dra-matically increased during the first transect across the patch,indicating a large physiological response during the first 24 h,the time required to add the Fe(III). After fertilization, 14C-pro-ductivity increased 3�4 times and chlorophyll concentrationsincreased about three-fold (Martin et al. 1994). Macronutrientmeasurements showed little change in measured nitrate, phos-phate or silicate concentrations in the patch, and any changesin macronutrients expected on the basis of extra chlorophyllproduction were calculated to be within the limits of analyticalprecision. Much more use of these macronutrients would havebeen expected on the basis of single-bottle incubation studies.Significant changes in the distribution of ammonium ion wereobserved, and measurable increases in both particulatedimethylsulfaniopropionate and methyl iodide were recorded,inside the patch. Measurements of CO2 fugacity and total CO2were significantly lower inside the patch than outside (Watsonet al. 1994). However, as with the macronutrient results, thedecreases were much less than those expected on the basis ofsingle-bottle incubation studies using comparable inoculationsof Fe(III).

Subsequently, it was decided that the primary cause of thereduced magnitude of the biological and chemical responsesto iron enrichment was subduction of the iron-enriched patchinto a low-light environment only 4 days after the enrich-ment. Therefore, for the second experiment, IRONEX II,iron addition was made as an initial 225 kg aliquot, followedby two 112 kg aliquots on Days 3 and 7, rather than a singleinitial dose of 450 kg. This experiment, which took placeover a 17-day period starting on 29 May 1995 (Coale et al.1996b), was considerably more successful. The enrichmentgenerated a massive phytoplankton bloom with parallel


0.5

1.0

1.5

Fe (µm

ol/kg ice)

180

200

220

240

260

280

300

[CO

2] (

ppm

v)

20 40 60 80 100 120 140 160

Age (1000 yr)

Iron

CO2

Fig. 14. Comparison of CO2 concentrations measured in air bubblestrapped in layers of ice taken from the Vostok core, East Antarctica (Barnolaet al. 1987) with iron concentrations derived from measurements of dustconcentration in the same core (De Angelis et al. 1987). The horizontal agescale is based on ice stratigraphy.


decreases in macronutrient concentrations and CO2 fugacityin the surface waters (Coale et al. 1996b; Cooper et al.,1996). In addition, surface concentrations of the biologicallygenerated gas dimethyl sulfide increased by 3.5 times duringthe enrichment (Turner et al. 1996).

Finally, during February 1999, a third meso-scale ironenrichment experiment SOIREE (Southern Ocean IronEnrichment Experiment) was conducted in polar waters ofthe Southern Ocean by a team of scientists from NewZealand, Canada and Europe. Detailed results from thiscruise are not yet released, but the findings were very similarto those reported for IRONEX II, i.e. a substantial biologicalresponse to iron enrichment was observed.

Although very exciting, the results of these meso-scaleexperiments may not offer much hope for stimulating addi-tional CO2 absorption by the oceans, because of limitations onthe rate of vertical water mixing. Peng and Broecker (1991)applied a box model to examine the dynamical aspects of apossible iron fertilization in Antarctic waters. From theirmodel they concluded that even after 100 years of totally suc-cessful fertilization, the CO2 content of the atmosphere wouldbe reduced to only 10 ± 5% below what it was projected to bewithout fertilization. They identified that the rate of verticalmixing in Antarctic waters was too slow to create a significantdecrease in the CO2 content of the atmosphere. Other studiesusing different box models also concluded that iron fertiliza-tion of the Southern Ocean would reduce atmospheric CO2levels only slightly over a 100-year fertilization programme.This in all cases was due to dynamical limitations (Joos et al.1991a, 1991b; Kurz and Maier-Reimer 1993).

EpilogueIt is clear that the past 20 years have seen enormous

progress in our understanding of the biogeochemical behaviourof trace metals in the ocean. But has the field come of age?We think so, for the important reason that 20 years ago, theterm �trace metal biogeochemistry� itself was scarcely rec-ognized, and trace metals were studied largely because theirerratic behaviour made them interesting and challenging tounderstand (Turekian 1977). Although that �erratic behaviour�is now known to be the result of artifacts, the interest andchallenge remains.

AcknowledgmentsOur research on trace metal biogeochemistry is supported

by grants from the New Zealand Public Good Science fund,NIWA and the University of Otago. We are grateful to KenBruland for his comments on this paper.

ReferencesAhlers, W. W., Reid, M. R., Kim, J. P., and Hunter, K. A. (1990).

Contamination-free sample collection and handling protocols for traceelements in natural freshwaters. Australian Journal of Marine andFreshwater Research 41, 713�20.

Anderson, M. A., and Morel, F. M. M. (1982). The influence of aqueousiron chemistry on the uptake of iron by the coastal diatom Thalassiosiraweissflogii. Limnology and Oceanography 27, 789�813.

Anon. (1992). US Joint Global Ocean Flux Study: Southern Ocean processstudy. US JGOFS Planning Report No. 16, May 1992. (US JGOFSPlanning and Coordination Office: Woods Hole OceanographicInstitution, MA.) 114 pp.

Banse, K. (1990). Does iron really limit phytoplankton production in theoffshore arctic Pacific? Limnology and Oceanography 35, 772�5.

Banse, K. (1991a). Iron, nitrate uptake by phytoplankton and mermaids.Journal of Geophysical Research 96, 20701.

Banse, K. (1991b). Iron availability, nitrate uptake and exportable new pro-duction in the Subarctic Pacific. Journal of Geophysical Research 96,741�8.

Barnola, J. M., Raynaud, D., Korotkevich, Y. S., and Lorius, C. (1987).Vostok ice core provides 160,000-year record of atmospheric CO2.Nature 329, 408�14.

Baum, R. (1990). Adding iron to ocean makes waves as way to cut green-house CO2. Chemical Engineering News July 2, 21�4.

Bender, M. L., Klinkhammer, G. P., and Spencer, D. W. (1977). Manganesein seawater and the marine manganese balance. Deep-Sea Research 24,799�812.

Berger, W. H., and Wefer, G. (1991). Productivity of the glacial ocean: dis-cussion of the iron hypothesis. Limnology and Oceanography 36,1899�918.

Boyd, P. W., Muggli, D. L., Varela, D. E., Goldblatt, R. H., Chretien, R.,Orians, K. J., and Harrison, P. J. (1996). In vitro iron enrichmentexperiments in the NE subarctic Pacific. Marine Ecology ProgressSeries 136, 179�93.

Boyd, P. W., LaRoche, J., Gall, M., Frew, R., and McKay, R. (1999).Phytoplankton dynamics in oceanic waters south of New Zealand.Journal of Geophysical Research 104, 13391�404.

Boyle, E. A. (1984). Cadmium in foraminifera and abyssal hydrography:Evidence for a 41 kyr obliquity cycle. In �Climate Processes and ClimateSensitivity�. (Eds J. E. Hansen and T. Takahashi.) pp. 360�8. (AmericanGeophysical Union: Washington D.C.)

Boyle, E. A. (1986). Paired carbon isotope and cadmium data from benthicforaminifera: implications for changes in oceanic phosphorus, oceaniccirculation and oceanic carbon dioxide. Geochimica et CosmochimicaActa 50, 265�76.

Boyle, E. (1997). What controls dissolved iron concentrations in the worldocean?�a comment. Marine Chemistry 57, 163�7.

Boyle, E. A., and Edmond, J. M. (1975). Copper in surface waters south ofNew Zealand. Nature 253, 107�9.

Boyle, E. A., and Keigwin, L. D. (1982). Deep circulation of the NorthAtlantic over the last 200,000 years: geochemical evidence. Science 218,784�7.

Boyle, E. A., and Keigwin, L. D. (1985). Comparison of Atlantic andPacific paleochemical records for the last 250,000 years: changes in deepocean circulation and chemical inventories. Earth and Planetary ScienceLetters 76, 135�50.

Boyle, E. A., and Keigwin, L. D. (1987). North Atlantic thermohaline cir-culation during the last 20,000 years linked to high latitude surface tem-perature. Nature 330, 35�40.

Boyle, E. A., Sclater, F. R., and Edmond, J. M. (1976). On the marine geo-chemistry of cadmium. Nature 263, 42�4.

Boyle, E. A., Sclater, F., and Edmond, J. M. (1977). The distribution ofdissolved copper in the Pacific. Earth and Planetary Science Letters 37,38�54.

Boyle, E. A., Huested, S. S., and Jones, S. P. (1981). On the distribution ofcopper, nickel and cadmium in the surface waters of the North Atlanticand North Pacific Oceans. Journal of Geophysical Research 86, 8048�66.

Brand, L. E., Sunda, W. G., and Guillard, R. R. L. (1986). Reduction ofmarine phytoplankton reproduction rates by copper and cadmium.Journal of Experimental Marine Biology and Ecology 96, 225�50.

751

Bruland, K. W. (1980). Oceanographic distributions of cadmium, zinc,nickel and copper in the North Pacific. Earth and Planetary ScienceLetters 47, 176�98.

Bruland, K. W. (1983). Trace elements in seawater. In �ChemicalOceanography�. (Eds J. P. Riley and R. Chester.) Vol. 8, pp. 157�215.(Academic Press: London.)

Bruland, K. W. (1989). Complexation of zinc by natural organic ligands inthe central north Pacific. Limnology and Oceanography 32, 269�85.

Bruland, K. W. (1992). Complexation of cadmium by natural organicligands in the central north Pacific. Limnology and Oceanography 37,1008�17.

Bruland, K. W., and Franks, R. P. (1983). Mn, Ni, Cu, Zn and Cd in thewestern North Atlantic. In �Trace Metals in Seawater�. (Eds C. S. Wong,E. A. Boyle, K. W. Bruland and J. D. Burton.) pp. 395�414. (PlenumPress: New York.)

Bruland, K. W., Knauer, G. A., and Martin, J. H. (1978a). Cadmium innortheast Pacific waters. Limnology and Oceanography 23, 618�25.

Bruland, K. W., Knauer, G. A., and Martin, J. H. (1978b). Zinc in north-east Pacific waters. Nature 271, 741�2.

Bruland, K. W., Franks, R. P., Knauer, G. A., and Martin, J. H. (1979).Sampling and analytical methods for the determination of copper,cadmium, zinc and nickel at the nanogram per liter level in seawater.Analytica Chimica Acta 105, 233�45.

Bruland, K. W., Donat, J. R., and Hutchins, D. A. (1991). Interactiveinfluences of bioactive trace metals on biological production in oceanicwaters. Limnology and Oceanography 36, 1555�77.

Bruland, K. W., Orions, K. J., and Cowens, J. P. (1994). Reactive tracemetals in the stratified central North Pacific. Geochimica etCosmochimica Acta 58, 3171�82.

Byrne, R. H., Kump, L. R., and Cantrell, K. J. (1988). The influence oftemperature and pH on trace metal speciation in seawater. MarineChemistry 25, 163�81.

Chavez, F. P., and Barber, R. T. (1987). An estimate of new production inthe equatorial Pacific. Deep-Sea Research 34, 1229�43.

Chavez, F. P., Buck, K. R., Coale, K. H., Martin, J. H., DiTullio, G. R.,Welschmeyer, N. A., Jacobson, A. A., and Barber, R. T. (1991).Growth rates, grazing, sinking and iron limitation of equatorial Pacificphytoplankton. Limnology and Oceanography 26, 1816�33.

Coale, K. H., and Bruland, K. W. (1988). Copper complexation in thenortheast Pacific. Limnology and Oceanography 33, 1084�101.

Coale, K. H., and Bruland, K. W. (1990). Spatial and temporal variabilityin copper complexation in the North Pacific. Deep-Sea Research 47,317�36.

Coale, K. H., Fitzwater, S. E., Gordon, R. M., Johnson, K. S., andBarber, R. T. (1996a). Control of community growth and export pro-duction by upwelled iron in the equatorial Pacific Ocean. Nature 379,621�4.

Coale, K. H., Johnson, K. S., and 17 others (1996b). A massive phyto-plankton bloom induced by an ecosystem-scale iron fertilization experi-ment in the equatorial Pacific Ocean. Nature 383, 495�501.

Collier, R., and Edmond, J. M. (1984). The trace element geochemistry ofmarine biogenic particulate matter. Progress in Oceanography 13,113�99.

Comiso, J. C., McLain, C. R., Sullivan, C. W., Ryan, J. P., and Leonard,C. L. (1993). Coastal Zone Color Scanner pigment concentrations in theSouthern Ocean and relationships to geophysical surface features.Journal of Geophysical Research 98, 2419�51.

Cooper, D. J., Watson, A. J., and Nightingale, P. D. (1996). Largedecrease in ocean-surface CO2 fugacity in response to in situ iron fertil-ization. Nature 383, 511�13.

Croot, P. L., and Hunter, K. A. (1998). Trace metal distributions across thecontinental shelf near Otago Peninsula, New Zealand. Marine Chemistry62, 185�201.

Cullen, J. J. (1991). Hypotheses to explain high-nutrient conditions in theopen sea. Limnology and Oceanography 36, 1578�99.

Cullen, J. J. (1995). Status of the iron hypothesis after the Open-OceanEnrichment Experiment. Limnology and Oceanography 40, 1336�43.

Danielsson, L. G. (1980). Cadmium, cobalt, copper, iron, lead, nickel andzinc in Indian Ocean water. Marine Chemistry 8, 199�215.

Danielsson, L. G., Magnusson, B., and Westerlund, S. (1978). Animproved metal extraction procedure for the determination of tracemetals in sea water by atomic absorption spectrometry with electrother-mal atomization. Analytica Chimica Acta 89, 47�58.

Danielsson, L. G., Magnusson, B., and Westerlund, S. (1985). Cd, Cu, Fe,Ni and Zn in the northeast Atlantic Ocean. Marine Chemistry 17, 23�41.

Davis, J. A., and Leckie, J. O. (1978). Effect of adsorbed complexingligands on trace metal uptake by hydrous oxides. Environmental Scienceand Technology 12, 1309�15.

De Angelis, M., Barkov, N. I., and Petrov, V. N. (1987). Aerosol concen-trations over the last climatic cycle (160 kyr) from an Antarctic ice core.Nature 325, 318�321.

de Baar, H. J. W., de Joeng, J. T. M., Bakker, D. C. E., Loscher, B. M.,Veth, C., Bathmann, U., and Smetacek, V. (1995). Importance of ironfor plankton blooms and carbon dioxide drawdown in the SouthernOcean. Nature 373, 412�15.

de Baar, H., de Jong, J., Nolting, R., Timmermans, K., van Leeuwe, M.,Bathmann, U., Rutgers van der Loeff, M., and Sildam, J. (1999).Low dissolved Fe and the absence of diatom blooms in remote Pacificwaters of the Southern ocean. Marine Chemistry 66, 1�34.

Duce, R. A. (1986). The impact of atmospheric nitrogen, phosphorus andiron species on marine biological productivity. In �The Role of Air�SeaExchange in Geochemical Cycling�. (Ed. P. Buat-Menard.) pp. 497�529.(Reidel: Dordrecht, Netherlands.)

Dugdale, R. C., and Wilkerson, F. P. (1990). Iron addition experiments inthe Antarctic: a reanalysis. Global Biogeochemical Cycles 4, 13�19.

Ellwood, M., and Hunter, K. (1999). Determination of the Zn/Si ratio indiatom opal: a method for the separation, cleaning and dissolution ofdiatoms. Marine Chemistry 66, 149�60.

Entsch, B., Sim, R. G., and Hatcher, B. G. (1983). Indication from photo-synthetic components that iron is a limiting nutrient in primary produc-ers on coral reefs. Marine Biology 73, 17�30.

Frew, R. D. (1995). Antarctic Bottom Water formation and the globalcadmium to phosphorus relationship. Geophysical Research Letters 22,2349�52.

Frew, R. D., and Hunter, K. A. (1992). Influence of Southern Ocean waterson the Cadmium-phosphate properties of the global ocean. Nature 360,144�6.

Frew, R. D., and Hunter, K. A. (1995). Cadmium-phosphate properties inthe Subtropical Convergence south of New Zealand. Marine Chemistry51, 223�37.

Frost, B. W. (1991). The role of grazing in nutrient-rich areas of the opensea. Limnology and Oceanography 36, 1239�44.

Gordon, R. M., Martin, J. H., and Knauer, G. A. (1982). Iron in northeastPacific waters. Nature 299, 611�12.

Gran, H. H. (1931). On the conditions for production of plankton in thesea. Rappels Procès-Verbaux Réunions, Conseil Internationale pourl�Exploration de la Mer 75, 37�46.

Hart, T. J. (1934). On the phytoplankton of the Southwest Atlantic and theBellinghausen Sea 1929�1931. Discovery Reports 8, 1�268.

Harvey, H. W. (1937a). The supply of iron to diatoms. Journal of theMarine Biological Association of the United Kingdom 22, 221�5.

Harvey, H. W. (1937b). Note on colloidal ferric hydroxide in sea water. Journalof the Marine Biological Association of the United Kingdom 22, 205�19.

Hester, K., and Boyle, E. A. (1982). Water chemistry control of thecadmium content of recent benthic foraminifera. Nature 298, 260�2.

Hudson, R. J. M., and Morel, F. M. M. (1990). Iron transport in marinephytoplankton: kinetics of cellular and medium coordination reactions.Limnology and Oceanography 35, 1002�20.

Huizenga, D. L., and Kester, D. R. (1983). The distribution of total andelectrochemically available copper in the northwestern Atlantic Ocean.Marine Chemistry 13, 281�91.



Hunter, K. A. (1982). The marine geochemistry of trace elements: a per-spective. Chemistry in New Zealand 46, 28�32.

Hunter, K. A., and Ho, F. W. T. (1991). Phosphorus-cadmium cycling inthe Northeast Tasman Sea 35�40oS. Marine Chemistry 33, 279�98.

Johnson, K., Gordon, R., and Coale, K. (1997a). What controls dissolvediron concentrations in the world ocean? Marine Chemistry 57, 137�61.

Johnson, K., Gordon, R., and Coale, K. (1997b). What controls dissolvediron concentrations in the world ocean? Authors� closing comments.Marine Chemistry 57, 181�6.

Johnson, K., Chavez, F., and Friederich, G. (1999). Continental-shelf sed-iment as a primary source of iron for coastal phytoplankton. Nature 398,697�9.

Joos, F., Sarmiento, J. L., and Siegenthaler, U. (1991a). Estimates of theeffect of Southern Ocean iron fertilization on atmospheric CO2 concen-trations. Nature 349, 772�5.

Joos, F., Siegenthaler, U., and Sarmiento, J. L. (1991b). Possible effectsof iron fertilization in the Southern Ocean on atmospheric CO2 concen-tration. Global Biogeochemical Cycles 5, 135�50.

Knauer, G. A., and Martin, J. H. (1981). Phosphorus-cadmium cycling innortheast Pacific waters. Journal of Marine Research 39, 65�76.

Kolber, Z. S., Barber, R. T., Coale, K. H., Fitzwater, S. E., Greene, R. M.,Johnson, K. S., Lindley, S., and Falkowski, P. G. (1994). Iron limita-tion of phytoplankton photosynthesis in the equatorial Pacific Ocean.Nature 371, 145�9.

Kulaev, I. S. (1979). �The Biochemistry of Inorganic Polyphosphates.�(Wiley: New York.)

Kurz, K. D., and Maier-Reimer, E. (1993). Iron fertilization of the australocean�the Hamburg Model assessment. Global Biogeochemical Cycles7, 229�44.

Landing, W. M., and Bruland, K. W. (1981). The vertical distribution ofiron in the Northwest Pacific. EOS: Transactions of the AmericanGeophysical Union 62, 906.

Landing, W. M., and Bruland, K. W. (1987). The contrasting biogeo-chemistry of iron and manganese in the Pacific Ocean. Geochimica etCosmochimica Acta 51, 29�43.

La Roche, J., Boyd, P. W., McKay, R. M. L., and Geider, R. J. (1996).Flavodoxin as an in situ marker for iron stress in phytoplankton. Nature382, 802�5.

Lee, J., Roberts, S., and Morel, F. (1995). Cadmium: a nutrient for themarine diatom Thalassiosira weissflogii. Limnology and Oceanography40, 1056�63.

Legrand, M., Feniet-Saigne, C., Satlzmann, E. S., Germain, C., Barkov,N. I., and Petrov, V. N. (1991). Ice-core record of oceanic emissions ofdimethylsulphide during the last climate cycle. Nature 350, 144�6.

Lewis, B. L., and Landing, W. M. (1991). The biogeochemistry of man-ganese and iron in the Black Sea. Deep-Sea Research 38, S773�S803.

Luther, G. W. III, and Wu, J. (1997). What controls dissolved iron con-centrations in the world ocean?�a comment. Marine Chemistry 57,173�9.

Martin, J. H. (1990). Glacial-interglacial CO2 change: the iron hypothesis.Paleoceanography 5, 1�13.

Martin, J. H. (1992). Iron as a limiting factor in oceanic productivity. In�Primary Productivity and Biogeochemical Cycles in the Sea�. (Eds P.G.Falkowski and A.D. Woodhead.) pp. 123�37. (Plenum Press: New York.)

Martin, J. H., and Fitzwater, S. E. (1988). Iron deficiency limits phyto-plankton growth in the north-east Pacific subarctic. Nature 331, 341�3.

Martin, J. H., and Gordon, R. M. (1988). Northeast Pacific iron distribu-tions in relation to phytoplankton productivity. Deep-Sea Research 35,177�96.

Martin, J. H., and Knauer, G. A. (1973). The elemental composition ofplankton. Geochimica et Cosmochimica Acta 37, 1639�53.

Martin, J. H., Bruland, K. W., and Broenkow, W. W. (1976). Cadmiumtransport in the California Current. In �Marine Pollutant Transport�. (EdsH. L. Windom and R. A. Duce.) pp. 159�84. (Heath: Lexington, MA.)

Martin, J. H., Gordon, R. M., Fitzwater, S., and Broenkow, W. W.(1989). VERTEX: Phytoplankton/iron studies in the gulf of Alaska.Deep-Sea Research 36, 649�80.

Martin, J. H., Broenkow, W. W., Fitzwater, S. E., and Gordon, R. M.(1990a). Yes it does: a reply to the comment by Banse. Limnology andOceanography 35, 775�7.

Martin, J. H., Fitzwater, S. E., and Gordon, R. M. (1990b). Iron defi-ciency limits phytoplankton growth in Antarctic waters. GlobalBiogeochemical Cycles 4, 5�12.

Martin, J. H., Gordon, R. M., and Fitzwater, S. E. (1990c). Iron inAntarctic waters. Nature 345, 156�8.

Martin, J. H., Gordon, R. M., and Fitzwater, S. E. (1991a). The case foriron. Limnology and Oceanography 36, 1793�802.

Martin, J. H., Fitzwater, S. E., and Gordon, R. M. (1991b). We still sayiron deficiency limits phytoplankton growth in the subarctic Pacific.Journal of Geophysical Research 96, 20699�700.

Martin, J. H., Fitzwater, S. E., Gordon, R. M., Hunter, C. N., andTanner, S. J. (1993). Iron, primary production and carbon-nitrogen fluxstudies during the JGOFS North Atlantic Bloom Experiment. Deep-SeaResearch 40, 115�34.

Martin, J. H., and 43 others (1994). Testing the iron hypothesis in theecosystems of the equatorial Pacific Ocean. Nature 371, 123�9.

McKnight, D. M., and Morel, F. M. M. (1979). Release of weak and strongcopper-complexing agents by algae. Limnology and Oceanography 24,823�37.

Moffett, J. W., and Zika, R. G. (1987). Solvent extraction of copper acety-lacetone in studies of copper(II) speciation in seawater. MarineChemistry 21, 301�13.

Moffett, J. W., Zika, R. G., and Brand, L. E. (1990). Distribution andpotential sources and sinks of copper chelators in the Sargasso Sea.Deep-Sea Research 37, 27�36.

Morel, F. M. M. (1983). �Principles of Aquatic Chemistry.� (Wiley-Interscience: New York.)

Morel, F. M. M., and Hudson, R. J. M. (1985). The geobiological cycle oftrace elements in aquatic systems: Redfield revisited. In �ChemicalProcesses in Lakes�. (Ed. W. Stumm.) pp. 251�81. (Wiley: New York.)

Morel, F. M. M., Rueter, J. G., Anderson, D. M., and Guillard, R. R. L.(1979). Aquil: a chemically-defined phytoplankton culture medium fortrace metal studies. Journal of Phycology 15, 135�41.

Morel, F. M. M., Hudson, R. J. M., and Price, N. M. (1991). Limitationof productivity by trace metals in the sea. Limnology and Oceanography36, 1742�55.

Morel, F. M. M., Reinfelder, J. R., Roberts, S. B., Chamberlain, C. P.,Lee, J. G., and Yee, D. (1994). Zinc and carbon co-limitation of marinephytoplankton. Nature 369, 740�2.

Nolting, R. N., De Baar, H. J. W., Masson, A., and Van Bennekom, A. J.(1991). Cadmium, copper and iron in the Weddell Sea and Weddell/Scotia Confluence (Antarctica). Marine Chemistry 35, 219�43.

Patterson, C. C., and Settle, D. M. (1976). The reduction of orders of mag-nitude errors in lead analysis of biological materials and natural watersby controlling external sources of industrial Pb contamination intro-duced during sample collection, handling and analysis. In �Accuracy inTrace Analysis�. (Ed. D. M. La Fleur.) pp. 321�3. (National Bureau ofStandards: Washington, D.C.)

Peng, T-H., and Broecker, W. S. (1991). Factors limiting the reduction ofatmospheric CO2 by iron fertilisation. Limnology and Oceanography 36,1919�27.

Prahl, G. (1992). Prospective use of molecular paleontology to test for ironlimitation on marine primary productivity. Marine Chemistry 39,167�85.

Price, N. M., and Morel, F. M. M. (1990). Cadmium and cobalt substitu-tion for zinc in a marine diatom. Nature 344, 658�60.

Roberts, L. (1991). Reports nixes �Geritol� fix for global warming. Science253, 1490�1.

753

Rueter, J. G., and Morel, F. M. M. (1981). The interaction between zincdeficiency and copper toxicity as it affects the silicic acid uptake mech-anisms in Thalassiosira pseudonana. Limnology and Oceanography 26,67�73.

Saager, P. M. (1994). On the relationships between dissolved trace metalsand nutrients in seawater: implications for the use of cadmium as anpaleoceanographic tracer. Ph.D. Thesis, University of Amsterdam.

Saager, P. M., De Baar, H. J. W., and Burkill, P. H. (1989). Manganeseand iron in Indian Ocean waters. Geochimica et Cosmochimica Acta 53,2259�67.

Saager, P. M., De Baar, H. J. W., and Howland, R. J. (1992). Cd, Zn, Niand Cu in the Indian Ocean. Deep-Sea Research 39, 9�35.

Saager, P. M., Schijf, J., and De Baar, H. J. W. (1993). Trace metal distri-butions in seawater and anoxic brines in the eastern Mediterranean Sea.Geochimica et Cosmochimica Acta 57, 1419�32.

Shen, G. T., and Boyle, E.,A. (1987). Lead in corals: reconstruction of his-torical industrial fluxes to the surface ocean. Earth and PlanetaryScience Letters 82, 289�304.

Sherrell, R. M., and Boyle, E. A. (1988). Zinc, chromium, vanadium andiron in the Mediterranean Sea. Deep-Sea Research 35, 1319�34.

Sullivan, C. W., Arrigo, K. R., McLain, C. R., Comiso, J. C., andFirestone, J. (1993). Distributions of phytoplankton blooms in theSouthern Ocean. Science 262, 1832�7.

Sunda, W. (1997). Control of dissolved iron concentrations in the worldocean:�a comment. Marine Chemistry 57, 169�72.

Sunda, W. G., and Huntsman, S. A. (1992). Feedback interactions betweenzinc and phytoplankton in seawater. Limnology and Oceanography 37,25�40.

Sunda, W. G., and Huntsman, S. A. (1995). Regulation of copper concen-tration in the oceanic nutricline by phytoplankton uptake and regenera-tion cycles. Limnology and Oceanography 40, 132�7.

Symes, J. L., and Kester, D. R. (1985). The distribution of iron in the north-west Atlantic. Marine Chemistry 17, 57�74.

Turekian, K. K. (1977). The fate of metals in the oceans. Geochimica etCosmochimica Acta 41, 1139�44.

Turner, D. R., Whitfield, M., and Dickson, A. G. (1981). The equilibriumspeciation of dissolved components in freshwater and seawater at 25Cand 1 atm pressure. Geochimica et Cosmochimica Acta 45, 855�81.

Turner, S. M., Nightingale, P. D., Spokes, L. J., Lidicoat, M. I., and Liss,P. S. (1996). Increased dimethyl sulfide concentrations in sea water fromin situ iron enrichment. Nature 383, 513�17.

Tuschall, J. R., and Brezonik, P. L. (1981). Evaluation of the copperanodic stripping voltammetry complexometric titration for complexingcapacities and conditional stability constants. Analytical Chemistry 53,1986�9.

Van den Berg, C. M. G. (1982). Determination of copper complexationwith natural organic ligands in seawater by equilibration with MnO2.Part 1: Theory; Part 2: Experimental procedures and application tosurface seawater. Marine Chemistry 11, 307�42.

Van den Berg, C. M. G. (1984). Organic and inorganic speciation of copperin the Irish Sea. Marine Chemistry 14, 201�12.

Waite, T. D., and Morel, F. M. M. (1983). Characterization of complexingagents in natural waters by copper(II)/copper(i) amperometry. AnalyticalChemistry 55, 1268�74.

Walsh, R. S., and Hunter, K. A. (1992). Influence of phosphorus storageon the uptake of cadmium by marine algae and implications for cadmiumbiogeochemistry. Limnology and Oceanography 37, 1361�9.

Watson, A. J., Liss, P. S., and Duce, R. A. (1991). Design of a small-scalein situ iron fertilization experiment. Limnology and Oceanography 36,1960�5.

Watson, A. J., Laws, C. S., Van Scoy, K. A., Millero, F. J., Yao, W.,Friedrich, G. E., Liddicoat, M. I., Wanninkhof, R. H., Barber, R. T.,and Coale, K. H. (1994). Minimal effect of iron fertilization on sea-surface carbon dioxide concentrations. Nature 371, 143�5.

Webb, M. (1979). �The Metallothioneins.� 195 pp. (Elsevier: NorthHolland, Amsterdam.)

Wells, M. L., Price, N. M., and Bruland, K. W. (1995). Iron chemistry inseawater and its relationship to phytoplankton: a workshop report.Marine Chemistry 48, 157�82.

Westerlund, S., and Ohman, P. (1991). Cadmium, copper, cobalt, nickel,lead and zinc in the water column of the Weddell sea, Antarctic.Geochimica et Cosmochimica Acta 55, 2127�46.

Manuscript received 7 June 1999; accepted 13 September 1999.


http://www.publish.csiro.au/journals/mfr

biogeochemistry of trace metals in the ocean

Documents