Articles
https://doi.org/10.1038/s41561-018-0200-y
Cenozoic record of
δ
34
S in foraminiferal calcite
implies an early Eocene shift to deep-ocean
sulfide burial
Victoria C. F. Rennie
1
*, Guillaume Paris
2,3
, Alex L. Sessions
2
, Sigal Abramovich
4
,
Alexandra V. Turchyn
1
and Jess F. Adkins
2
1
Department of Earth Sciences, University of Cambridge, Cambridge, UK.
2
Division of Geological and Planetary Sciences, California Institute of Technology,
Pasadena, CA, USA.
3
Centre de Recherches Pétrographiques et Géochimiques, CRPG, UMR 7358, CNRS, Université de Lorraine, Vandoeuvre-lès-Nancy,
France.
4
Department of Geological and Environmental Sciences, Ben Gurion University of the Negev, Beer Sheva, Israel. *e-mail: vcr22@cam.ac.uk
SUPPLEMENTARY INFORMATION
In the format provided by the authors and unedited.
NATuRE GE
oSCiENCE
|
www.nature.com/naturegeoscience
1
A novel
Cenozoic record of seawater
sulfur isotopes
from foraminiferal calcite
S
u
pp
l
ementary Material
1
. Foraminiferal cleaning
Foraminiferal calcite is associated with several other phases that could contain
sufficient sulfur to contaminate the measured CAS.
Ca
lcitic foraminiferal tests
contain abundant associated organic
matter
, which foraminifera use to lay down a
template
for calcification
1
.
This organic
matter
is likely to contain sulfur,
typically at
~1% abundance,
although its contribution to the overall sulfur concentration in
foraminifera has bee
n suggested to be small
2
.
Sediments have other constituents that may contain sulfur, and which may be
adhered onto the for
aminiferal tests.
These include
exogenous
organic
molecules
, clay
minerals, recrystallized calcite, pyrite (from bacterial sulfate reduction), barite (from
the water column, or precipitated within pore fluid), and ferromanganese crusts. If
barite is adhere
d to the carbonate, its high sulfate
content
makes
it likely to
contaminate the overall CAS
δ
34
S
value
.
Diagenetic barite forms as barium reacts
with sulfate in pore fluid.
Sources of barium into pore fluid include its in
-
situ release
from remineralized or
ganic matter and dissolving carbonates, or diffusion from below
(pelagic, or seawater, barite dissolves when the pore fluid sulfate concentrations have
dropped to zero).
The sulfur content and associated
δ
34
S of ferromanganese crusts are
less well known.
F
erromanganese crusts form as iron and manganese are reduced at
depth (
coupled
to the microbial oxidation of organic
matter
in the sediment column).
Dissolved iron and manganese can diffuse upwards through the sediment, and
precipitate as oxides in oxygenat
ed parts of the sediment column.
Ferro
-
manganese
2
oxides will coat sedimentary grains (including foraminiferal tests), and may
incorporate trace concentrations of other elements, such as neodymium, and lead.
It is
not clear whether these coatings contain su
lfur, but if this is the case then they could
provide another source of contamination. Because of these possible contaminants, it
becomes important to test which phases associated with foraminifera
l tests
in the
sediment column contain sufficient sulfur to
affect
measured
CAS
δ
34
S
values
.
The long history of geochemical research into foraminifera has resulted in the
optimisation of methods to remove clays,
(hydr)oxides,
organics, ferromanganese
crusts, and barite from samples of handpicked foraminifera.
The methods used for t
his
study are modified from
3,4
to account for larger sample sizes.
Methods used to re
move
barite
5
were not used here
because the methods employed strongly chelating cations,
resulting in the dissolution of a significant proportion of calcite, as well as any barite
(although barium concentrations were measured to look for possible contamination
–
see below).
Implementing a barite removal step would triple the amount of calcite
needed.
Additionally, standard cleaning methods do not attempt to chemically remove
pyrite
6
,
although
in some
cases
microcrystalline pyrite can be
oxidized
during CAS
e
xtraction
7
.
This is addressed by
visually ensuring as much as possible that the
samples do not contain pyrite,
replacing all the
oxidi
z
ing
acids used (such as nitric
acid) with non
-
oxidi
z
ing acids (such as hydrochloric acid), and dissolving samples in
as weak an acid as possible.
2.1 Cleaning Tests
and core
-
top results
3
The typical sequence for cleaning foraminiferal calcite for trace element work
involves the sequential removal of clays, organic material (using an oxidizing
solution
)
and oxidative coa
tings (using a reducing solution).
Samples were crushed between two glass plates. Samples for clay removal were then
mixed with Aristar methanol and agitated in an ultrisonicator for 30 s, before
removing the overlying fluid.
This step was repeated twice
with methanol and
multiple times with 18.2 Milli
-
Q water until the agitated solu
tion was clear.
The
samples were
t
hen inspected under the microscope for the manual removal of any
visual
contaminants. Samples undergoing oxidative cleaning were treated with
100
μ
l
of ultraclean hydrogen peroxide in 10 ml of 0.1 M sodium hydroxide
,
in a water bath
with ultrasonication for 10 minutes. Samples that were
additionally
reductively
cleaned were
initially
treated
with
100
μ
l hydrazine solution (250
μ
l hydrous
hydrazine i
n 2
ml ammonium hydroxide and 2 ml 0.25 M citric acid/16 M ammonia
solution) for 45 mins in a water bath, with intermittent ultrasonification.
The final
cleaning step test
(steps 4 and 6)
was a leaching step with
500
μ
l of 0.001 M HCl
solution for 1 minute
-
half of the reductively cleaned samples and half of the
reductively
-
then oxidatively cleaned samples
(post steps 3&5)
underwent leaching
.
Four species (
G. siphonifera, O. universa, G. menardii, and G. sacculifer
) in the
>250
μ
m size fraction were picke
d from the coarse fraction of the top 0.5
cm from
WIND 10B in the Western Indian Ocean. Species were divided up into ~4
mg
subsamples prior to crushing.
Splitting the sample before crushing has one advantage
over crushing the sample prior to splitting, bec
ause the final
δ
34
S will express the true
variability expected from any given foraminiferal sample
–
both the intra
-
specific
variability and any that is the result of different cleaning approaches.
The
4
disadvantage of this method is that it tests for two s
ources of variability at once, and
therefore cann
ot distinguish between them
8
.
For the purposes of quantifying the total
“external variability” in a sample, ho
wever, this method provides a more useful test.
Each 4
mg sample was treated
with
a variety of different cleaning steps, as outlined in
Figure S
.1
.
Where sample volume allowed, duplicate or triplicate samples were
processed (for
O. universa
, and
G. mena
rdii
), however, for other species with low
sample volumes, some steps were omitted (e.g. steps 1, 3 and 5).
Figure S
.
1. Flow diagram of the cleaning steps for the cleaning test.
Samples were dissolved (in 0.2
M
ultraclean
HCl) prior to analysis, and a
n
aliquot was run on the ICP
-
AES to determine calcium concentrations.
Additionally,
samples were run for the standard suite of trace elements: Ba/Ca, Fe/Ca, Mn/Ca by
FORAMINIFERAL CAS
δ
34
S
CHAPTER 4
112
are much higher than this natural variability could help indicate the presence of contaminants
within the samples.
Figure
4
.
8
-
Flow diagram of the different cleaning steps for the second cleaning test.
The most likely sedimentary contaminants of
foraminiferal
CAS
δ
34
S
are
pyrite and
barite, because they contain high concentrations of sulfur
that may have di
stinct
δ
34
S
compared to primary foraminiferal CAS
δ
34
S
, and there is no specific cleaning step employed
to remove them.
However, the natural range of barium and iron concentrations within clean
foraminiferal calcite is well known, and so anomalously high
concentrations of either barium
or iron will indicate possible barite or pyrite contamination within the sample. This approach
is also useful for
the identification of ferro
-
man
ganese crusts (using Fe/Mn ratios as well as
Fe/Ca and Mn/Ca ratios), clay cont
amination (using Al/Ca ratios) and glove/lab
contamination (using Zn/Ca ratios
-
Zn is in very low concentrations in natural carbonate, but
in high concentrations on surfaces and gloves in the laboratory).
5
ICP
-
AES.
The natural abundance of these elements in Quaternary planktonic
foraminifera is
well known, due to the long
-
standing body of research into these
microfossils.
Any trace element concentration that is
significantly
higher than this
natural variability indicates the
likely
presence of contaminants within the samples.
The use of trace e
lement analysis by ICP
-
AES
is also useful for the
identification of ferro
-
manganese crusts (using Fe/Mn ratios as well as Fe/Ca and
Mn/Ca ratios), clay contamination (using Al/Ca ratios) and glove/lab contamination
(using Zn/Ca ratios
-
Zn is in very low co
ncentrations in natural carbonate, but in high
concentrations on surfaces and gloves in the laboratory).
The remaining solution was dried down in a laminar flow hood and
redissolved in 0.2 M ultraclean HCl and then processed through columns after
2,9
.
Each sample batch was run with a blank, and each column batch had at least one
column with seawater (and for the long
-
term record, an additional consistency
standard). Samples were analysed on a ThermoScientific Neptun
e Plus MC
-
ICP
-
MS
using a desolvating membrane (Aridus, Cetac) at the California Institute of
Technology, with sample
-
standard bracketing to correct for mass fractionation within
the instrument
10
and reported relative to V
-
CDT. The V
-
CDT calibration was made
using an in
-
house Na
2
SO
4
standard using the absolute ratio measured previously
10
.
With each batch of columns, we ran a seawater and a carbonate consistency standard
consistin
g of a dissolved coral, modified by the addition (1%) of a pure calcium
solution
. Column seawater results were 21.01 ± 0.12‰, in agreement with previously
published values
11
and carbonate consistency standards were 22.1 ± 0.3‰.
6
2.2 Results
The r
esults of the cleaning test
for the four species of core
-
top foraminifera
(
>250
μ
m)
from WIND 10B in the Western Indian Ocean
are displayed in Figure S
.
2.
In three species (
G. menardii, O. universa
and
G. sacculifer
)
,
the average
δ
34
S
decreases from raw to clay removal to oxidatively cleaned samples (Steps 1
-
3), while
reductively cl
eaned samples generally have the highes
t average
δ
34
S.
This is in
contrast to samples of
G. siphonifera
, which
has measured CAS
δ
34
S
values for clay
removed, oxidatively cleaned (leached) and reductively cleaned (leached) which are
only 0.14‰ apart, simila
r to the analytical uncer
tainty for the Neptune MC
-
ICPMS
Figure S
.
2.
δ
34
S values
for different species, subdivided by cleaning step.
Where more than one sample
was
analyzed
the average isotope composition is shown, along with the standard deviation (1
σ
)
d
isplayed as an error bar, and the number of samples
analyzed
.
The blue horizontal bar represents
seawater
δ
34
S
(
21.1‰
)
.
24
23
22
21
20
19
18
δ
34
S (‰)
G. siphonifera
G. menardii
O. universa
G. sacculifer
Raw
Clay removal
Oxidative cleaning (unleached)
Oxidative cleaning (leached)
Reductive cleaning (unleached)
Reductive cleaning (leached)
n=4
n=4
n=4
n=4
n=2
n=2
n=2
n=2
n=2
n=2
n=2
n=2
n=2
7
Where more than one sample was
analyzed
, the standard deviation (1
σ
) of the
results is reported as an error bar.
The variability in
δ
34
S (as measured by the standard
deviation) is higher across the first three steps (raw to oxidatively cleaned samples),
than for the final three cleaning steps (oxidatively cleaned (leached) samples, and
reductively cleaned samples
-
Steps 4
-
6).
The reductiv
ely cleaned foraminifera had
δ
34
S
,
which were closer to those expected for seawater, and there was lower
variability between replicates.
This suggests that reductively cleaning foraminifera
removes more contaminants than the previous steps.
This observation is unexpected
because there
are no known sulfur bearing phases in ferromanganese coatings.
In
general, while the oxidation of pyrite by HCl has been demonstrated
7
, it appears that,
for very small carbonate samples, this is not the case
6
.
However, Fe
3+
is an efficient
oxidi
zing
agent for pyrite, so perhaps the removal of this phase limits pyrite oxidation
during the
subsequent samples dissolution
12
.
Three species (
G. siphonifera, O. universa,
and
G. sacculifer
) have a foraminiferal
CAS
δ
34
S that is at or below the
δ
34
S of seawater.
Reductively cleaned samples for
these species
have
δ
34
S values
within 0.2
–
0.6‰ of seawater (21.1‰).
This sulfur
isotope composition is slightly higher than that expected for
O. universa
from
culturing experi
ments, in which
O. universa
(which were only oxidatively cleaned)
was shown to incorporate sulfate with a
δ
34
S
value
that was 1‰ lower than
that
of
sulfate in the culturing media
2
.
The sulfate concentrations in cultured
O. universa
are
also twice as high as any sulf
ate concentrations in
O. universa
in this core
-
top.
The
difference between cultured and core
-
top
O. universa
are not easy to reconcile
-
although oxidatively cleaned samples in this study were also ~1‰ lower than
reductively cleaned samples.
If different c
l
eaning methods
have an impact on the
8
measured foraminiferal CAS
δ
34
S
(as is suggested by this data), then to investigate
real differences between cultured and core
-
top foraminifera, they must have been
cleaned identically.
In contrast to the other foramin
ifera,
G. menardii
is 1‰ higher than the other samples,
across all cleaning steps except Step 3 (oxidatively cl
eaned but unleached) with an
average
foraminiferal CAS
δ
34
S of 21.7‰ rather than of 19.9‰.
The average
CAS
δ
34
S
value
for reductively cleaned
G.
menardii
is 22‰, 1‰ higher than the other
species and modern seawater.
This highlights the necessity of using single species
foraminifera, and splicing different records of single species foraminifera in order to
remove any offsets related to
inter
-
species
differen
ces.
The offsets related t
o inter
-
species differences in the
δ
34
S of calcite are possibly related to different
biomineralization conditions within different foraminifera (
e.g.,
rates and mechanisms
of biomineralization, cellular conditions, presence/absence of symbionts
1
), however
distinguishing the mechanism is beyond the scope of these data.
3. Sulfur isotope record
The sulfur isotope record was collected from six different
,
globally distr
ibuted core
sites (see Figur
e S
.
3.
).
9
Figure S
.
3
-
Map of the core
-
sites used to make the second Cenozoic record of foraminiferal CAS
δ
34
S
,
labeled
with the age int
ervals used for each site. Colo
red symbol
s on the map correspond to colo
rs of
the symbols used in Figure 1
. Age model
s for DSDP 588, ODP 757, 758, 926, 1262 and 1265 were
13
-
19
.
We addressed the likely sulfur isotope offset between species by comparing
overlapping records of single species foraminifera,
pinning the record to mod
ern
seawater values, using the offsets found in modern core
-
top samples
(see Table S.1
for offset
-
corrected species
)
.
Table S.1 Species splicing offsets
APPENDIX
171
Appendix
Table A.1
-
Species offsets in
δ
34
S used to boot
-
strap the second foraminiferal CAS
δ
34
S
record together. Only Species which were adjusted are listed.
Species
Species spliced to
δ
34
S of offset
G. bulloides,
praebulloides
D. venezuelana
,
1
2Ma, 15Ma
+1‰
O. universa
D. venezuelana
at
12Ma, modern
cultures
+1‰
G. index
D. galavisi,
36 Ma,
Catapsydrax. U
37 Ma
+0.7‰
A. bullbrooki
G. index,
45 Ma
M. aragonensis,
47 Ma
+0.8‰
G. menardii
Core
-
top
-
1‰
Table A.2
-
Species offsets in sul
fate concentrations used to boot
-
strap the foraminiferal
CAS concentration record together. Only Species which were adjusted are listed.
Species
Species spliced to
ppm
offset
G. index
D. galavisi
,
35 Ma
-
200 ppm
Catapsydrax u.
D.
galavisi
at 32
Ma
,
+
150 ppm
A. bullbrooki
Catapsydrax. U
44.5
Ma
-
220 ppm
M. aragonensis
A. bullbrooki,
48 Ma
-
450 ppm
M. caucascia
M. aragonensis,
48 Ma
-
450 ppm
M. lensiformis
M. caucascia
50
M
a
-
450 ppm
M. aequa
M. lensiformis
52 Ma
-
450 ppm
10
Most overlapping foraminiferal samples had consistent
δ
34
S values.
Only six
species
(see Table S.1
-
G. b
ulloides
&
praebulloides
,
O. universa
,
G. index
,
A. bullbrooki
,
and
G. menardii
)
of foraminifera had measured CAS
δ
34
S
values that were
isotopically offset from
overlapping sample of
foraminiferal CAS
δ
34
S
.
Isotopic
offsets were determined by comparing the foraminiferal CAS
δ
34
S
values
from two
different
species picked out of the sample sediment horizon (
and therefore from the
same geological time interval, assuming that each sediment horizons represents a
d
iscrete time interval, and that bioturbation and sediment reworking are negligible
relative to the residence time and sampling intervals associated with sulfate in the
ocean.
). Where two species co
-
exist in more than one time interval, the measured
foramin
iferal CAS
δ
34
S
offset between them remained constant, engendering
confidence in the values used for splicing single species records.
Certain features of the Cenozoic foraminiferal CAS
δ
34
S record appear to be
confirmed independently by different foraminiferal speci
es, for example, from 10 to 6
Ma there is a 1‰ decrease in foraminiferal CAS
δ
34
S both in
G. menardii
and in
D.
venezuelana
across two different sites (758 and 926), and a similar offset between the
two species at both time intervals. An additional example
is from
~
53
–
48
Ma, where
three different foraminiferal species (
M. lensiformis, M. aragonensis,
and
M.
caucascia
)
all
show steep increases in CAS
δ
34
S, and where overlap exists between
species there is a good match between each species’ measured CAS
δ
34
S.
4. Screening for possible contamination
Using the ICP
-
AES data, it is possible to rule out blank contamination by comparison
11
of the foraminiferal CAS
δ
34
S
against total sulfate in each sample (see Figure
S
.
4
)
.
T
here is no correlation between foraminif
eral CAS
δ
34
S
and total sulfate in the
samples
, as would be expected if a blank were present in varying amounts.
Figure S
.
4.
-
Foraminiferal CAS
δ
34
S
(un
-
spliced) against total sulfate for the samples.
It
is also possible to rule out barite contaminati
on by comparing foraminiferal CAS
δ
34
S
and sulfate concentrations against barium concentrations, and total sulfate in the
sample against total bariu
m in the sample (see Figure S
.
5
). Neither foraminiferal CAS
δ
34
S
,
SO
4
/
Ca
ratios, or total sulfate in the sam
ples show any relationship to either
Ba/Ca or total barium in the samples, indicating that barite contamination
, if any, is
insignificant
.
The detection limit of Ba over the experimental runs was a maximum of
0.0001 ppm (results recorded below this concent
ration were assumed to be too poorly
calibrated), which corresponded to a sub
-
sample volume of 10
μ
l out of a total sample
volume of 470
μ
l, or 1ng Ba. or 0.007 nmol. Any “unseen barium” would therefore
FORAMINIFER
AL CAS
δ
34
S
CHAPTER 4
130
Figure
4
.
20
-
Foraminiferal CAS
δ
34
S (
un
-
spliced
) against total sulfate for the samples. Symbol shapes
denote different species, and the different colours represent the different core sites shown in Figure 4.18.
Figure
4
.
21
-
foraminiferal C
AS
δ
34
S
(un
-
spliced
)
against Ba/Ca (a) and Total barium (b), and sulfate/calcium
ratios
(un
-
spliced
)
against Ba/Ca (c), and total sulfate against total barium (d). Symbol shapes denote different
species, and the different colours represent the different c
ore sites shown in Figure 4.18.
24
23
22
21
20
19
18
δ
34
S (‰)
15x10
-3
10
5
0
Total Sulfate (
μ
mol)
G. menardii
G. praebulloides
M. aragonensis
Red- ODP 1265
D. venezuelana
Catapsydrax u
M. caucascia
Green- ODP 758
O. universa
D. galavisi
M. lensiformis
Black- ODP 926
G. premenardii
G. index
M. aequa
Purple -ODP 588
G. bulloides
A. bullbrooki
Grey - ODP 1262
Blue- ODP 757
24
23
22
21
20
19
18
δ
34
S (‰)
200x10
-9
150
100
50
0
Total Ba (mmol)
(b)
1200
800
400
0
SO
4
/Ca (ppm)
14x10
-3
12
10
8
6
4
2
Ba/Ca (mmol/mol)
(c)
G. menardii
G. premenardii
Catapsydrax u.
A. bullbrooki
M. lensiformis
Purple - ODP 588
Black - ODP 926
D. venezuelana
G. bulloides
D. galavisi
M. aragonensis
M. aequa
Blue - ODP 757
Red - ODP 1265
O. universa
G. praebulloides
G. index
M. caucascia
Green - ODP 758
Grey - ODP 1262
24
23
22
21
20
19
18
δ
34
S (‰)
14x10
-3
12
10
8
6
4
2
Ba/Ca (mmol/mol)
(a)
16x10
-3
12
8
Total Sulfate
(
μ
mol)
200x10
-9
100
0
Total Ba (mmol)
(d)
12
contribute a maximum sulfur contribution of 0.2 ng to a 5 ng sample, or ~5%.
Figure S
.
5
-
foraminiferal CAS
δ
34
S
(un
-
spliced) against Ba/Ca (a) and
Total barium (b), and
sulfate/calcium ratios (un
-
spliced) against Ba/Ca (c), and total sulfate against total barium (d).
I
ron con
centration data (Figure S
.6
)
also
shows no
correlation with
δ
34
S
,
indicating
that there is
negligible
pyrite contamination in
the Cenozoic foraminiferal CAS
samples.
However, there does appear to be a correlation between total iron and
foraminiferal CAS
δ
34
S
for the
G. menardii
samples from Site 758, which spans the
most recent 10
Ma.
No such correlation is apparent between sulf
ate concentrations
and iron concentrations, or total sulfate against total iron (See Figure
S
.
6
).
This
suggests that there
could be
an iron phase contributing to the measured foraminiferal
CAS
δ
34
S
of these samples, but not supplying sufficient sulfate to
alter the sulfate
concentration of the sample.
FORAMINIFER
AL CAS
δ
34
S
CHAPTER 4
130
Figure
4
.
20
-
Foraminiferal CAS
δ
34
S (
un
-
spliced
) against total sulfate for the samples. Symbol shapes
denote different species, and the different colours represent the different core sites shown in Figure 4.18.
Figure
4
.
21
-
foraminiferal C
AS
δ
34
S
(un
-
spliced
)
against Ba/Ca (a) and Total barium (b), and sulfate/calcium
ratios
(un
-
spliced
)
against Ba/Ca (c), and total sulfate against total barium (d). Symbol shapes denote different
species, and the different colours represent the different c
ore sites shown in Figure 4.18.
24
23
22
21
20
19
18
δ
34
S (‰)
15x10
-3
10
5
0
Total Sulfate (
μ
mol)
G. menardii
G. praebulloides
M. aragonensis
Red- ODP 1265
D. venezuelana
Catapsydrax u
M. caucascia
Green- ODP 758
O. universa
D. galavisi
M. lensiformis
Black- ODP 926
G. premenardii
G. index
M. aequa
Purple -ODP 588
G. bulloides
A. bullbrooki
Grey - ODP 1262
Blue- ODP 757
24
23
22
21
20
19
18
δ
34
S (‰)
200x10
-9
150
100
50
0
Total Ba (mmol)
(b)
1200
800
400
0
SO
4
/Ca (ppm)
14x10
-3
12
10
8
6
4
2
Ba/Ca (mmol/mol)
(c)
G. menardii
G. premenardii
Catapsydrax u.
A. bullbrooki
M. lensiformis
Purple - ODP 588
Black - ODP 926
D. venezuelana
G. bulloides
D. galavisi
M. aragonensis
M. aequa
Blue - ODP 757
Red - ODP 1265
O. universa
G. praebulloides
G. index
M. caucascia
Green - ODP 758
Grey - ODP 1262
24
23
22
21
20
19
18
δ
34
S (‰)
14x10
-3
12
10
8
6
4
2
Ba/Ca (mmol/mol)
(a)
16x10
-3
12
8
Total Sulfate
(
μ
mol)
200x10
-9
100
0
Total Ba (mmol)
(d)
13
Figure S
.
6
-
foraminiferal CAS
δ
34
S
(un
-
spliced) against Fe/Ca (a) and Total iron (b)
, with
a zoomed in
view of Site 758 samples (see insets), and sulfate/calcium ratios (un
-
spliced) against Fe/Ca (c), and
total sulfate against total iron (d).
A
similar correlation between foraminiferal CAS
δ
34
S
and total manganese in the
G.
menardii
samples from Site
758 can be seen (see Figure S
.
7
). There is also a
correlation between sulfate concentrations and manganes
e concentrations/total
manganese in these samples. Furthermore,
t
here is a distinct
correlation
between iron
and manganese concentrations for the Site 758
G menardii
samples
(see Figure
S.8
)
.
The other foraminiferal samples analy
z
ed for this Cenozoic recor
d either have high
manganese and high iron concentrations (likely ferro
-
manganese oxides), or high
manganese and low iron concentrations (likely manganese carbonates), however these
G. menardii
samples have high manganese concentrations, with intermediate
iron
concentrations. This is therefore likely to be some form of manganese precipitate that
is rich in both iron and sulfur, and has formed on foraminifera in contact with the pore
fluid of this site. There are only two additional samples from Site 758 (bo
th
D.
venezuelana
), too few to confirm that the excess manganese is from a diagenetic
FORAMINIFER
AL CAS
δ
34
S
CHAPTER 4
131
The iron concentration data (see Figure 4.22) shows no relationship between samples
with high iron concentrations/total iron, and CAS
δ
34
S, suggesting that there is no pyrite
contamination in the Cenozoic foraminiferal CAS
samples. However, there does appear to be
a correlation between total iron and foraminiferal CAS
δ
34
S for the
G. menardii
samples from
Site 758, which spans the most recent 10
Ma. No such correlation is apparent between sulfate
concentrations and iron co
ncentrations, or total sulfate against total iron (See Figure 4.22).
This suggests that there is an iron phase that is contributing to the measured foraminiferal
CAS
δ
34
S of these samples, but not supplying sufficient sulfate to alter the sulfate
concentra
tion of the sample.
Figure
4
.
22
-
foraminiferal CAS
δ
34
S
(un
-
spliced
) against Fe
/Ca (a) and T
otal iron
(b),
with a zoomed in view of
Site 758 samples (see insets),
and
sulfate
/calcium ratios (un
-
spliced
) against Fe
/Ca (c), and tot
al sulfate against
total iron
(d). Symbol shapes denote different species, and the different colours represent the different core sites
shown in Figure 4.18.
Examining the manganese data, a similar correlation between foramini
feral CAS
δ
34
S
and total manganese in the
G. menardii
samples from Site 758 can be seen (see Figure 4.23).
There is also a correlation between sulfate concentrations and manganese concentrations/total
manganese in these samples. Furthermore, when compari
ng the iron and manganese data (see
Figure 4.24), it can be seen that there is a distinct relationship between iron and manganese
24
23
22
21
20
19
18
δ
34
S (‰)
5x10
-6
4
3
2
1
0
Total Fe (mmol)
24
23
22
δ
34
S (‰)
2
1
0
x10
-6
(b)
1400
1200
1000
800
600
400
SO
4
/Ca (ppm)
0.8
0.6
0.4
0.2
0.0
Fe/Ca (mmol/mol)
(c)
24
23
22
21
20
19
18
δ
34
S (‰)
0.8
0.6
0.4
0.2
0.0
Fe/Ca (mmol/mol)
24
22
δ
34
S (‰)
0.2
0.1
0.0
(a)
16x10
-3
12
8
Total Sulfate
(
μ
mol)
5x10
-6
4
3
2
1
0
Total Fe (mmol)
(d)
24
23
22
21
20
19
18
δ
34
S (‰)
200x10
-9
150
100
50
0
Total Ba (mmol)
(b)
1200
800
400
0
SO
4
/Ca (ppm)
14x10
-3
12
10
8
6
4
2
Ba/Ca (mmol/mol)
(c)
G. menardii
G. premenardii
Catapsydrax u.
A. bullbrooki
M. lensiformis
Purple - ODP 588
Black - ODP 926
D. venezuelana
G. bulloides
D. galavisi
M. aragonensis
M. aequa
Blue - ODP 757
Red - ODP 1265
O. universa
G. praebulloides
G. index
M. caucascia
Green - ODP 758
Grey - ODP 1262
24
23
22
21
20
19
18
δ
34
S (‰)
14x10
-3
12
10
8
6
4
2
Ba/Ca (mmol/mol)
(a)
16x10
-3
12
8
Total Sulfate
(
μ
mol)
200x10
-9
100
0
Total Ba (mmol)
(d)
14
coating that impacts all foraminifera at this site. However these two
D. venezuelana
samples show a similar relationship between
SO
4
/
Ca
vs. Mn/Ca as the
G. menar
d
ii
sampl
es at this site, which indicates that the
D. venezuelana
samples may also be
contaminated by
an
Mn
-
rich phase in the sediments.
Figure S
.
7
-
foraminiferal CAS
δ
34
S
(un
-
spliced) against Mn/Ca (
a) and Total manganese (b), and
sulfate/calcium ratios (un
-
sp
liced) against Mn/Ca (c), and total sulfate against total manganese (d).
R
2
values for best
-
fits for the
G. menardii
samples for Site 758 are displayed on the relevant graphs.
FORAMINIFER
AL CAS
δ
34
S
CHAPTER 4
132
concentrations for the Site 758
G. menardii
samples. The other foraminiferal samples
analysed for this Cenozoic record either
have high manganese and high iron concentrations
(likely ferro
-
manganese oxides), or high manganese and low iron concentrations (likely
manganese carbonates), however these
G. menardii
samples have high manganese
concentrations, with intermediate iron con
centrations. This is therefore likely to be some
form of manganese precipitate that is rich in both iron and sulfur, and has formed on
foraminifera in contact with the pore fluid of this
s
ite. There are only two additional samples
from Site 758 (both
D.
venezuelana
), too few to confirm that the excess manganese is from a
diagenetic coating that impacts all foraminifera at this
s
ite. However these two
D.
venezuelana
samples show a similar relationship between sulfate/calcium vs. Mn/Ca as the
G.
menarii
sa
mples at this
s
ite, which indicates that the
D. venezuelana
samples may also be
contaminated by a Mn
-
rich phase in the sediments.
Figure
4
.
23
-
foraminiferal CAS
δ
34
S
(un
-
spliced)
against
Mn
/Ca (a) and
T
otal manganese
(b), and
s
ulfate
/calcium ratios (un
-
spliced
) against Mn
/Ca (c), and to
tal sulfate against total manganese
(d).
R
2
values for
best
-
fits for the
G. menardii
samples for Site 758 are displayed on the relevant graphs.
Symbol shapes denote
di
fferent species, and the different colours represent the different core sites shown in Figure 4.18.
24
23
22
21
20
19
18
δ
34
S (‰)
12x10
-6
10
8
6
4
2
0
Total Mn (mmol)
(b)
R
2
=0.89
1400
1200
1000
800
600
SO
4
/Ca (ppm)
0.5
0.4
0.3
0.2
0.1
0.0
Mn/Ca (mmol/mol)
(c)
24
23
22
21
20
19
18
δ
34
S (‰)
0.8
0.6
0.4
0.2
0.0
Mn/Ca (mmol/mol)
(a)
R
2
=0.68
16x10
-3
12
8
Total Sulfate
(
μ
mol)
12x10
-6
10
8
6
4
2
0
Total Mn (mmol)
(d)
24
23
22
21
20
19
18
δ
34
S (‰)
200x10
-9
150
100
50
0
Total Ba (mmol)
(b)
1200
800
400
0
SO
4
/Ca (ppm)
14x10
-3
12
10
8
6
4
2
Ba/Ca (mmol/mol)
(c)
G. menardii
G. premenardii
Catapsydrax u.
A. bullbrooki
M. lensiformis
Purple - ODP 588
Black - ODP 926
D. venezuelana
G. bulloides
D. galavisi
M. aragonensis
M. aequa
Blue - ODP 757
Red - ODP 1265
O. universa
G. praebulloides
G. index
M. caucascia
Green - ODP 758
Grey - ODP 1262
24
23
22
21
20
19
18
δ
34
S (‰)
14x10
-3
12
10
8
6
4
2
Ba/Ca (mmol/mol)
(a)
16x10
-3
12
8
Total Sulfate
(
μ
mol)
200x10
-9
100
0
Total Ba (mmol)
(d)