GLOBAL
IRON
IN
DEFICIENCY
LIMITS
ANTARCTIC
R.
CYCLES,
PHYTOPLANKTON
VOL.
4, NO.
1, PAGES
5-12, MARCH
1990
GROWTH
WATERS
John H. Martin,
and
BIOGEOCHEMICAL
Steve
Michael
E.
Fitzwater
Gordon
Moss Landing Marine Laboratories
Moss Landing, California
Abstract.
Enrichment
experiments
were
performed in the Ross Sea to test the
hypothesis that iron deficiency
is
responsible
for the phytoplanktonfs
failure
to use up the luxuriant
major nutrient
supplies found in these and all other
INTRODUCTION
uptake rates in the controls without added
trace elements ranged from 0.58 to 1.22 •mol
We have been testing
the hypothesis
that
global ocean waters that have excess major
nutrients
(NO3, P04) , such as those in the
Gulf of Alaska
[Martin
et al.,
1989],
the
central
equatorial
Pacific
[Chavez and
Barber,
1987],
the northeast
Atlantic
[Strass and Woods, 1988], and the southern
kg-1 d-l; the addition of 1 to 5 nmolof
ocean[Bainbridge,1981], all suffer from
unchelated Fe per liter
resulted in rates
that were 2 to 10 times higher (2.54 to 6.00
the same condition:
commonFe deficiency
[Martin et al.,
1989].
Lack of this
•molNO
3 kg-1 d-l).
essentialelement[Weinberg,
1989]prevents
offshore
Antarctic
ocean
waters.
Nitrate
Ratesin bottles with 2
nmol Mn added were identical
to those in the
the phytoplankton from blooming, and
controls(0.57 to 1.04 •molNO
3 kg-1 d-l).
thereforemajornutrientsremainunused.
Total decreases in NO3 were balanced by
increases in particulate
organic N. These
results suggest that Fe deficiency is the
primary reason that the present-day southern
ocean biological pump is shut off.
In
contrast, iron was 50 times more abundant
during the last glacial maximum;greater Fe
availability
may have stimulated the
biological pumpand contributed to the ice
age drawdown of atmospheric CO2. These
Studies of Fe deficiency in the southern
ocean are of special importance, since
several models [Sarmiento and Toggweiler,
1984; Knox and McElroy, 1984] indicate that
the use or nonuse of major nutrients and
resulting increases and decreases in
photosynthesis could be responsible for the
results
also
imply that
ocean Fe fertilization
in
terms
required'
yr_l.
of
'
the
i.e.,
total
low atmospheric CO2 concentrations observed
during glacial maxima and high CO2 levels
found in the interglacials.
Here we report
large-scale
southern
the results
is feasible,
at least
experiments performed in the Ross Sea during
amounts
of
Fe
100 000 to 500 000 tons
,
,
of four
January to February
of these
findings
Fe enrichment
1990.
are
The implications
discussed.
METHODS
Copyright 1990
by the American Geophysical
Paper number 90GB02146,
0886-6236/90/90GB-02146510.00
Union.
Raw seawater with its resident
phytoplankton population was collected
from the
mixed layer at four Ross Sea stations
(Figure 1) using 30-L Go-Flo bottles
Martin
170øE
\
"•
et al.'
Iron and Antarctic
180 ø
"
'
t..........
I'
170øW
'
•
.
74øS
•/,ctoria;.•""
J2 L\
. Lan
d'.':::{.i...•
("
ß•
•
1
/
•,;:,,..•.
,•4 •.
-
Growth
160øW
'
Cape Sta.
3
72øS
Phytoplankton
'
'
"
' Sta.4
'
.
•
•
/Sta.
P,
/
74øS
76øS
'•':':•
76øS
..........
,,,.•20
"• •
Ross
Ice
Sh;i•
...................
/ .........
'"'Edward '....
VII
McMurdo Sound
•
72os
t
, %
,,
Peninsula
•
170øE
I
I
!
180 ø
I
170øW
Fig. 1. Ross Sea station
locations.
Contours are for opal content of bottom
sediments,
in percent (original
figure and opal data from Dunbar et al.
[1985]).
suspended on Kevlar line.
Aliquots
of the
water were placed in 2-L acid-washed
polycarbonate
bottles
to which 1 to 5 nmol
near the Victoria
Land coast (Figure 1)
where present and past high productivity
conditions
are evidenced by relatively
low
added.
Identical
replicates
with nothing
added served as controls.
Experiment and
control bottles
were kept in deck-top
well as the richness of opal (• 40% [Dunbar
et al.,
1985]) in the underlying
sediments.
In view of the proximity
to the coast and
incubators at ambient light
the abundance of ice, adequate amounts of
of unchelated Fe L -1 or 2 nmol Mn L-1
levels
were
(• 50 mol
surfacenutrients (• 5 •mol NO
3 kg-1) as
photons
m-2 d-l); temperature
wasmaintained iron wereexpected
[MartinandGordon,
1988;
at 0ø ñ 1ø with running seawater. Chlorophyll and nitrate concentrations were
monitored to estimate comparative growth.
To prevent spurious Fe contamination, all
Martin et al., 1989, 1990]; hence it was not
surprising that the phytoplankton in the
controls grew nearly as well as those with 5
nmol Fe added (Table 1; Figure 2).
procedures were carried out using ultraclean
techniques [Martin et al.,
1976]. Additional
The next experiment was performed at
Station 2, located 440 km farther offshore
aliquots
of water
and about
filtered
through precombusted GFF fiberglass
(100
to 125 mL) were
240 km north
(Figure 1).
of the Ross Ice
Shelf
On day 4, the treatments with
filters; particulates were analyzedfor
addedFe (1, 2.5, and 5.0 nmolL-1) beganto
their organic C and N content using a
Control Equipment C,H,N analyzer.
separate from those without Fe; by day 10,
chlorophyll
concentrations
were about twice
RESULTS
was used up by the tenth
as high with Fe as without.
in
Four sets
of experiments
were performed
using water and phytoplankton from a variety
of environments that presumably range from
Fe-rich to Fe-poor.
Station 1 is located
the
controls
and bottles
Available
day with
with
NO3
Fe, while
added Mn (2
nmolL-l), abouthalf of the NO
3 originally
available remained (Figure 2).
Essentially,
results with the three Fe
treatments were the same. Only slight
Martin
et al.'
Iron
and Antarctic
TABLE 1.
Linear
Phytoplankton
Regressions
and
Station
1
2
Added
Fe
or
2) With
Mn
Type
n
a
b
6-9
0 Fe
4
0.998
26.20
-2.54
5 Fe
4
0.999
28.75
-3.11
0 Fe
4
0. 998
24.58
-1.22
2 Mn
4
1.000
23.47
-1.04
1 Fe
4
0.999
27.50
-2.64
Fe
5 Fe
4
4
0.998
0.996
28.95
29.16
-2.85
-2.93
2.5
6-13
6-11
9-11
r
0 Fe
5
0.998
29.51
-0.91
2 Mn
5
0. 998
30.27
-1.00
1 Fe
5
0.999
37.61
-2.54
Fe
5
0.998
38.97
-2.84
5 Fe
5
1.000
40.45
-3.04
0 Fe
3
0.954
29.18
-0.58
2 Mn
3
0.986
29.02
-0.57
5 Fe
2
2.5
4
for NO3 Uptake (see Figure
Without
Day
4-10
3
Growth
-6.00
NO
3 (•molkg-1) -- a + b(day)o
,•.-.'ø.
2
.• ,,
w/'-
30 ß ,,. ........
20
.•
''
i
.
<
'...........
(
(• ,
i
ß
i
ß
i
ß
,
i'
ß
i
.
i'
ß
i
ß
ß
,'
i
ß
0 2 4 6 S 0 2 4 6 S 10 12 0 2 4 6
Day
Day
S 10 12 0
•
•'
•'
Day
•'
1'0 12
Day
Fig. 2.
Chlorophyll
and nitrate
concentrations
versus experiment day at
Stations 1 to 4 (see Figure 1) with added Fe or Mn versus control.
Station 2
and 3 chlorophyll
values for the three samples with added Fe were averaged;
means and standard deviations
are shown. Solid lines in nitrate
figures are
from Table 1 linear
regressions.
increases
in NO3 uptake were observed with
the
control
than half
increasing
FeNO
addition,
kg-1 d-l).
and
2.93 •mol
3 kg-1 d•ie.,
for2.64,
the 12,.85,
2.5,
and 5 nmol of
added Fe (Table
1).
Rates
in
and
Mn-added
as high
bottles
(1.22
With the depletion
were
less
and 1.04 •mol NO3
of NO3 on day 10, the
Martin
et al.:
bottles with added Fe were processed.
However, the phytoplankton in the control
and
Mn-added
bottles
were
allowed
Iron
and Antarctic
Phytoplankton
Growth
sufficient
light for photosynthesis during
the austral
spring and summer and that
Antarctic
phytoplankton grow well at low
to
continue growing for an additional
3 days.
Although no further increase in chlorophyll
temperatures.
Nevertheless,
their metabolic
activities
are slowed, which may result
in
was observed, nitrate
about the same uptake
inefficient
uptake continued
rates as those
observed on days 4 to 10 (Figure
Nearly
identical
from Station
results
3, which
is
at
in
an area
about 140 km SE of Cape Adare (Figure 1).
Again, with Fe, the phytoplankton took up
about 3 times as much NO3 per day as without
Fe (2.54, 2.84, and 3.04 versus 0.91 and
1.00 •mol NO
3 kg-1 d-l).
of
nutrients
and
radiant energy [Tilzer
et al.,
1986]; this
may explain the linear
decreases in
nutrients
and perhaps the rather low (in
2).
were obtained
located
utilization
view of the NO3 taken up) chlorophyll
levels
found at the ends of the experiments.
In addition to light and temperature,
water column stability,
grazing pressure,
and trace element availability
are almost
Amounts
of chloro- alwaysmentioned
as factors affecting
phyll produced with added Fe were also
markedly higher (Figure 2).
The final
experiment was performed about
500 km east of Cape Adare and 650 km north
of the Ross Ice Shelf,
i.e.,
far from
phytoplankton growth.
As early as 1934,
Hart [1934] presented strong circumstantial
evidence for Fe deficiency,
while E1-Sayed
[1988] notes it only in passing.
Our
experiments were performed taking all of the
shallow bottom and ice-iron sources (Figure
1).
The initial
chlorophyll level was only
aforementioned
0.4 •g L-1 and it appearedthat little
any of the available
if
NO3 (26.2 •mol NO3
kg-1) hadbeentakenup. Onday6 the
effects
of
added
Fe
became
noticeable
(Figure 2), andby day 11 the phytoplankton
had produced 11.8 •g chlorophyll
L -•,
in
comparison
to the 1.1 •g chlorophyllL-1
environmental
factors
into
account.
Water column stability
was
provided,
of course, by the containment of
the water in bottles;
light levels were
realistic,
since a deck-top incubator
was
used.
Running seawater through the
incubator maintained the phytoplankton at
temperatures
close to those in situ (• 0øC).
Grazing pressure was assumed to be the same
in all
containers.
Under
these
conditions,
made by the phytoplankton with added Mn and
the 1.4 •g produced in the control
containers.
The phytoplankton with Fe were
were only 0.57
also consuming about 10 times as muchNO3
2.54-6.00 •mol NO
3 kg-1 d-1 with Fe.
supplemental Fe during the last days of the
experiment.
(We had to stop observations 1
day early,
since we ran out of ship time.)
Concentrations
of particulate
organic
(0.08 nmolkg-1) we measured
in offshore
(• 6 •mol NO
3 kg-1 d-1) as thosewithout
carbon and nitrogen
(POC and PON) were also
measured at the ends of the experiments
(Table 2).
Excellent agreement was obtained
between the total amounts of NO3 removed and
PON quantities
found (Figure 3).
The ratio
between POC and NO3 taken up (6.79) was near
the Redfield C:N ratio of 6.6 (Figure 3).
Comparisons
of
and without
Fe are shown in Figure 4.
final
POC concentrations
with
In general,
these
Ross Sea results
uptake rates
for NO3 (Table 1)
to 1.22 without
Because of the very
Fe versus
low Mn concentrations
Drake Passage surface waters [Martin et al.,
1990], we performed Mn enrichment experiments
at
the
three
offshore
stations
to
see
if a deficiency
of this element might be
limiting
phytoplankton
growth.
As shown in
Figure 2 and Table 1, nitrate
uptake and
chlorophyll
production rates with added Mn
were indistinguishable
from those obtained
in the controls; thus no evidence for Mn
deficiency
was found.
With the NO3 uptake rates (Table 1), we
made simple mass balance calculations
to
estimate how long it would take to deplete
DISCUSSION
similar
the linear
the available NO3. We assumed that light
levels were adequate for phytoplankton
growth only in the upper 5 m of the water
column. We also assumed that the beginning
are
to those we obtained previously
[Martin et al.,
1989]; however, with the
exception
of Station
4, linear
decreases
NO
3 concentration
was25 mmolm-3' thus
in
NO3 were observed instead of the logarithmic
there was a total of 125 mmol NO3 in the
surface layer, 5-m-thick "box." An
nutrient
Alaska.
additional 6 mmol of NO3 was provided by
water with a NO3 concentration of 25 mmolm-
decreases found in the Gulf of
Perhaps this results from low
temperatureandlight levels; however,ElSayed[1988] points out that there is
3 upwellinginto the "box"at a rate of 0.25
m d-1 [Gordon,1971]. Whenwe usedthe
Martin
et al.'
Iron
TABLE 2.
and Antarctic
Total
Phytoplankton
Growth
Nitrate
Removed Together With Particulate
Carbon and Nitrogen
(POC, PON) Found in
Experiments
2, 3 and 4
Identification
NO3 Removed
a
0
Control
Organic
PON Found
Experiment
Initial
9
POC Found
2
10.5
76
16.1 (12.4) b
20.2 (16.5)
2 Mn
13.4
18.1
1 Fe
22.1
24.9
174
2.5
Fe
(10.3)
133 (108)
(15.0)
122
23.0
25.8
176
5 Fe
23.2
25.8
188
Initial
Control
0
10.6
Experiment
3
2.4
12.9
16
90
2 Mn
11.2
13.8
87
1 Fe
2.5
Fe
24.1
26.9
23.2
28.4
154
180
5 Fe
27.7
30.3
211
Initial
0
1.6
Control
3.7 (4.3) c
4.9
Experiment
2 Mn
5 Fe
3.3
18.9
(3.9)
(24.9)
(101)
4
9.7
(5.5)
5.2
19.4
(5.8)
(25.4)
28.0
(31.9)
32.2
106.5
(36.1)
(147)
Values are in units of #moles•er kilogram. Regressionsare #mol PON
kg-1 = 2.28 + 0.993 (#molNO
3 kg-•), n = 15; r = 0.993; #molPOCkg-1 =
12.36 + 6.79 (•molNO
3 kg-1), n = 15; r = 0.975.
aNot used in regressions.
bExperiment
2 control andMn-added
bottles wereallowedto run 3 days
longer
than Fe-added
added bottles
bottles.
To enable
on day 10, estimates
rates from Table
parentheses).
1 were made for
direct
comparisons
based on regressions
with
Fe-
and NO3 uptake
day 10 Mn-added and control
bottles
(in
CExperiment 4 values in parentheses are estimates if experiment
had been allowed
control
to run 1 more day.
and added Mn NO3 uptake rates
(Table
1) from experiments
2 to 4, it appears that
no depletion
would occur, since the removal
rates (from -0.57 to -1.22 mmolNO
3 m-3 d-1)
when multiplied
by the 5-m thickness
are
by upwelling.
On the other
when Fe is made available,
hand,
the 125 mmol NO3
in the 5-m box would be used up in about 2
weeks at the -3.0 mmol NO3 uptake rate
(5 X
-3.0 = -15 + 6 upwelled = -9; 125/9 = 14
days).
With the Station
1 control
bottle
rate of -2.54 mmolNO
3 m-3 d-1 it would
take about 3 weeks to deplete
scenario
is
the surface
assumedly
consistent
with
NO3 values
in this
Fe-rich
environment
the
the NO3.
fact
This
that
nearshore,
[Martin
and
et al.,
80% depleted,
1989, 1990]
that
is,
on the
order of 4 to 5 •mol NO
3 kg-1.
At the end of the Station
the phytoplankton
equal to or less than the 6 mmol NO3
provided
Gordon, 1988; Martin
were about
4 experiment,
were consuming NO3 at a
rate of 6 mmolm-3 eachday. At this very
high rate (6 X 5 = 30 mmolNO
3 d-l), it
would take only 5 days to completely deplete
the 125 mmol of NO3 in the 5-m-thick surface
box.
Assuming the Redfield
C:N ratio
of
6.6:1,
this NO3 uptake is equivalent
to a
newproductivity rate of 40 mmolm-3 d-1.
Again,
limits
assuming that light availability
this growth to the upper 5 m, the new
C fixation
rate would be 200 mmol C m-2 d-1
or 2.4 g m-2 d-1
4 controls
ß
without
In contrast the Station
,
Fe were removing only
60,
,,
,
-200
Initial
50-
40-
0
0
Control
•,
Mn
e
Fe
80-
o
-100 D.
o
o
E
•0-
10-
i
I
i
I
I
I
I
I
i
I
I
I
I
0
5
10
15
20
25
0
5
10
15
20
25
30
pmol
NO
3kg'1consumed
Fig.
3. Amounts of particulate
organic
carbon and nitrogen
(POC, PON) versus
nitrate consumed
with and without addedFe or Mn (data from Table 2•' •mol PON
kg-1 = 2.28 + 0.993 (•molNO
3 kg-1), r = 0.99, n = 15' •mol POCkg-z = 12.36 +
6.79 (•molNO
3 kg-1), r = 0.98, n = 15.
ß
..:.::..:.:.:.::•
200
....
:.•:•
,--I
Initial
•
Control
150
o 100
E
2
Fig.
4.
Stations
Initial
2,
3,
3
Station
and final
POC concentrations
and 4 (estimated
and measured
4
from experiments
values
from Table
performed
2).
at
Martin
et al.'
Iron
and Antarctic
Phytoplankton
Growth
0.58 mmolNO
3 m'3 d'1. Thenewproductivity
over 5 mwouldbe only 19 mmolC m'2 d'1 or
0.23 g m'2 d'1.
implications,
Fe deficiency
is also of
importance in relation
to the current
rapid
These open-sea Station
demonstrate the potential
buildup of greenhouse gas CO2 [Tans et al.,
1990].
Glacial to interglacial
CO2
fertilization.
With
the
4 findings
power of Fe
addition
of
5 nmol
In
addition
increases
to
these
occurred
at
historical
rates
of
the
order
Fe, an estimated100 •mol of POCkg'1 were
0.03 billion tons C y-1 [Martin, 1990;
produced (Table 2; Figure 4); this
represents
a molar C:Fe ratio
of 20,000:1.
If we assume that the phytoplankton
at
Station
4 were like the phytoplankton
at
Stations 2 and 3 and could have grown almost
Barnola
current
et al.,
1987].
activities
are
atmospheric CO2 at rates
magnitude higher
ways to actively
CO2 are being sought.
C to 1 Fe.
On a weight
method
20,000
1.0
basis
this
is about
of
C for
contemporary
a
small
as well
at
2 orders of
Iron fertilization
ocean appears
least
in
terms
of
to be a feasible
of
the
amounts
required. Theupwelling (• 0.25 m d'1
Fe.
In previous papers [Martin et al.,
1989;
Martin,
1990],
we have commented on the
potential
importance of this very large
return
the southern
man's
[Tans et al.,
1990].
Hence
stimulate
the removal of
as well with 1 nmol of Fe as with 5, the
molar C:Fe ratio
would be more like
100,000
C to
In contrast,
increasing
of
investment
in
Fe
in
as historic
[Gordon, 1971]) of major nutrients
during
the austral
spring and summer in the area
(2
X 1013 m2) from 56øSto the Antarctic
continent
could support
production of the order
phytoplankton
of 1.8 billion
new
tons
environments. The findings we present here,
C y-1.
together with the Fe distribution
data
published previously [Martin et al., 1990],
strongly suggest that Fe deficiency is
required for man to fertilize
this region
during the austral spring and summerrange
from 100,000 to 500,000 tons,
depending on
common in the offshore
southern ocean.
waters
of the
This has to be the primary
limiting factor responsiblefor the
phytoplankton's
failure
to use the luxuriant
major nutrient
sources.
Even when the
phytoplankton
are well adapted to cold
temperatures,
the water column is stable,
the sun is bright,
major nutrients
are
abundant, and grazers are absent, the
population
will
not be able to take
advantage of these optimal conditions
because they will not have the Fe required
for the synthesis of chlorophyll
pigments
and important enzymes such as nitrate
reductase
[Weinberg, 1989].
The direct
evidence we provide here, that
southern ocean phytoplankton
growth is
limited
by Fe deficiency,
provides support
for the hypothesis
[Martin,
1990] that
interglacial
atmospheric CO2 concentrations
may be high because very small amounts of
Fe-rich
atmospheric dust fall
out over the
southern ocean during these warm periods [De
Angelis et al.,
1987].
Because of this
limited
Fe input, the phytoplankton
cannot
use the abundant major nutrients,
and
essentially
the biological
pump is turned
off.
In contrast,
Fe input to the
Antarctic
was 50 times higher during recent
glacial
maxima [De Angelis et al.,
1987].
This indicates
that Fe deficiency
was not a
problem in the Antarctic
during the ice
ages; however, proof that this resulted
in
massive phytoplankton blooms that could have
contributed
to the glacial
drawdown of ice
age CO2 is yet to be obtained.
Estimatesof the amountof Fe
the C:Fe ratio
used.
Nevertheless,
spreading very small amounts of Fe (• 100 •g
m'2 d'l) over very large areas is not a
trivial
problem.
Even if
could be accomplished
bloom resulted,
there
fertilization
and even if a massive
is no guarantee that
atmospheric CO2 levels would be
significantly
affected
in view of the
dynamics of CO2 removal [Peng and Broecker,
1990].
Acknowledgments.
We thank the Walker
Smith group for the chlorophyll
analyses,
Joe Jennings and Paul Treguer for the
nitrate
analyses,
and Craig Hunter for the
POC and PON analyses.
This research was
supported by grants from the Biology and
Medicine Section of NSF Polar Programs (DPP87716460),
the ONR Ocean Chemistry Program
(N 000 14-84-C-0619),
Chemistry
and the
NSF Marine
Program (OCE-8813565).
REFERENCES
Bainbridge A. E., GEOSECSAtlantic
Expedition:
Hydrographic Data,
Vol.
1,
National
Science
1972-73,
Foundation,
Washington, D.C.,
1981.
Barnola, J. M., D. Raynaud, Y. S.
Korotkevich,
and D.
Lorius,
Vostok
ice
core provides 160,000-year
record of
atmospheric CO2, Nature, 329, 408-414,
1987.
Chavez,
F.
P.,
and R.
of new production
T.
Barber,
An estimate
in the equatorial
12
Martin
Pacific,
1987.
De Angelis,
Deep Sea Res.,
M.,
Petroy,
N. I.
Aerosol
34,
Barkov,
1229-1243,
and V.N.
concentrations
last climatic
cycle
Antarctic
ice core,
et al.'
over
the
(160 kyr) from an
Nature, 325, 318-321,
1987.
Dunbar,
R.
B.,
J.
B. Anderson,
Domack, Biogenic
Sound, Antarctica,
Antarctic
and E. W.
sedimentation
in McMurdo
in Oceanology of the
Continental
Shelf,
Antarct.
Res.
Scr., vol. 43, edited by S.S. Jacobs, pp.
291-312, AGU, Washington, D.C.,
1985.
E1-Sayed, S. Z, Productivity
of the southern
ocean:
A closer look, Comp. Blochem.
Physiol.,
90B, 489-498,
1988.
Gordon, A.L.,
Oceanography of Antarctic
waters,
in Antarctic
Oceanology I,
Antarct.
Res. Ser., vol. 15, edited by
J.L. Reid, pp. 169-203,
AGU, Washington,
D.C.,
Hart,
1971.
T. J.,
On the phytoplankton
south-west Atlantic
sea,
1929-31,
Rep.,
VIII,
1-268,
1934.
Knox, F., and M. B. McElroy, Changes in
atmospheric CO2'
Influence of the marine
biota at high latitude,
J. Geophys. Res.,
89,
4629-4637,
Martin,
J. H.,
1984.
Glacial-interglacial
change'
The iron
Paleoceanography.,
Martin,
Pacific
J.
H.,
iron
phytoplankton
CO2
Northeast
distributions
in
productivity,
relation
Growth
Phytoplankton/iron
studies in the Gulf of
Alaska,
Deep Sea Res., 36, 649-680,
1989.
Martin,
J. H., R. M. Gordon, and S. E.
Fitzwater,
Iron in Antarctic
waters,
Nature, 345, 156-158, 1990.
Peng, T.-H.,
and W. S. Broecker, The
Antarctic
iron fertilization
strategy:
Dynamic considerations,
Nature,
(in
press),
1990.
Sarmiento,
new
model
J. L.,
for
and L. R. Toggweiler,
the
role
of
the
oceans
A
in
determining
atmospheric carbon dioxide
levels,
Nature,
308, 621-624,
1984.
Strass, V., and J.D. Woods, Horizontal
and
seasonal variation
of density and
chlorophyll
profiles
between the Azores
and Greenland,
in Towards a Theory of
Biological-physical
Interactions
in the
World Ocean, edited by B. Rothschild,
pp.
113-136, D. Reidel,
Hingham, Mass., 1988.
Y. Fung, and T. Takahashi,
constraints
on the global
atmospheric CO2 budget, Science, 247,
1431-1438,
1990.
Tilzer,
M. M., M. Elbrachter,
W. W. Gieskes,
and B. Beese, Light-temperature
interactions
photosynthesis
phytoplankton,
in
the
control
in Antarctic
Polar Biol.,
of
•,
105-112,
Weinberg, E. D., Cellular
regulation
of iron
assimilation,
O. Rev. Biol.,
64, 261-290,
1989.
to
Deep Sea Res.,
3__5, 177-196,
1988.
Martin,
J. H., K. W. Bruland,
and W. W.
Broenkow, Cadmium transport
in the
California
Current,
in Marine Pollutant
Transfer,
Duce, pp.
Phytoplankton
1986.
hypothesis,
•, 1-13, 1990.
R. M. Gordon,
and Antarctic
Tans, P. P., I.
Observational
of the
and the Bellingshausen
Discovery
Iron
S.
E.
Fitzwater,
R. M. Gordon,
Martin,
Moss Landing Marine
P.O. Box 450, Moss Landing,
edited
by H. Windom and R.
159-184,
D.C. Heath, Lexington,
Mass.,
1976.
Martin,
J. H., R. M. Gordon,
S.
and W. W. Broenkow,
VERTEX'
Fitzwater,
(Received August 28, 1990;
revised September 28, 1990;
accepted October 1, 1990.)
and J.
Laboratories,
CA 95039
H.