230
Mackensen and Bickert
paleoceanographic reconstructions based on different proxies contradict each other in some locations.
A most important dicrepancy, for instance, exists
between the interpretation of Cd and 8 13 C values
as nutrient proxies in the glacial Southern Ocean
(Boyle 1994; Boyle and Rosenthal 1996). Also
there is a controversy in estimating northern source
deep water production and thermohaline circulation
strength during glacial times in the Atlantic Ocean
inferred from either 8 13 C values or radiochemical
data (Yu et al. 1996).
In this paper, we first give a brief state-of-theart compilation of 8 13 C research regarding both
benthic foraminiferal as well as dissolved inorganic
carbon isotopic composition in the Atlantic and Antarctic oceans. We then compile new results obtained during the last five years or so from the calibration of southern Atlantic live benthic
foraminiferal 8 13 C values with 813CLC02 values
from overlying bottom water masses. Most of these
data are published previously in various sources, but
are compiled here for the first time (Duplessy et
al. 1984; Bickert 1992; Mackensen et al. 1993;
Mackensen et al. 1996; Mackensen 1997). In addition, new data from the inner Weddell Sea are
presented that corroborate some of our earlier findings. In a third part of this paper we present two
depth transects of sediment cores from the equatorial Brazil and Guinea basins and one core located
some degrees north of the present-day Antarctic
Polar Front to discuss South Atlantic and Southern
Ocean paleoceanography as inferred from benthic
foraminiferal8 13 C values (Mackensen et al. 1994;
Bickert and Wefer 1996).
Calibration of Bottom Water 813CEC02
versus Phosphate
The correlation between carbon isotope ratios in the
ocean water masses and nutrients is not perfect.
This is because carbon circulates through the atmosphere and the sea, whereas nutrients just
circulate within the oceans. If there were no airsea fractionation, the distribution of 8 13 C values
within the sea would be tightly coupled with nutrient concentrations. During gas exchange between
the surface water and the atmosphere, a strongly
temperature dependent isotopic fractionation occurs, such that fractionation increases by approximately 0.12 %0 with temperature decreasing by 1 DC
(Mook et al. 1974; Zhang et al. 1995). Atlowtemperatures, the isotopic fractionation between atmospheric CO 2 and surface ocean LC0 2 decouples
13C/ 12 C ratios from nutrients. Consequently, highest positive deviations of813CIC02 values from the
expected correlation between phosphate and 8 1l C
occur in Antarctic surface and intermediate waters,
coinciding with the Polar Frontal Zone where Antarctic Intermediate Water (AAIW) is formed and
distributed (Charles and Fairbanks 1990; Broecker
and Maier-Reimer 1992). In addition, Mackensen
et al. (1993; 1996) and Charles et al. (1993) showed
that thermodynamic 8 13 C effects are not limited to
surface and intermediate water masses, but must
be considered at all depths, especially in areas of
deep and bottom water formation.
When Kroopnick (1985) observed the constant
relation between 813CLC02 and phosphate for the
bulk of the world's oceans (i.e., a 0.71 %0 increase
in 8 1l C for every Ilmollkg PO. decrease), he only
used samples with a potential density in excess of
27.7. In general, this density occurs at depths between 100 and 1800 m and it was chosen to avoid
thermodynamic isotopic fractionation during air-sea
gas exchange between atmospheric CO 2 and dissolved inorganic carbon near the ocean surface and
thus to keep the correlation restricted to the biologic
alteration processes. Turning around this idea,
Oppo and Fairbanks (\989) and Charles and
Fairbanks (1990) used a "Redfield" slope of 0.93,
i.e. a 0.93 %0 increase in 8 1l C for every Ilmollkg
decrease in phosphate, to normalize the 8IlCLC02
signal of Southern Ocean surface waters to
zero phosphate. Phosphate normalized 8 1l C
(!l.8IlCLCO) is the 8 1l C a water mass would have
if phosphate were entirely removed. Thus no biologically driven fractionation would mask the sense
and magnitude ofthe isotopic fractionation during
gas exchange at low temperatures. The same
approach based on high quality deep Indian and
Pacific data, with a slightly steeper "Redfield" tie
between 8 13 C and PO. (1.1 %o/lllmol kgl), was
used by Broecker and Maier-Reimer (\992) to
quantify the biologic versus the thermodynamic
effect on the 813CLC02 distribution in the surface
ocean and the deep ocean.
Mackensen and Bickert
paleoceanographic reconstructions based on different proxies contradict each other in some locations.
A most important dicrepancy, for instance, exists
between the interpretation of Cd and 8 13 C values
as nutrient proxies in the glacial Southern Ocean
(Boyle 1994; Boyle and Rosenthal 1996). Also
there is a controversy in estimating northern source
deep water production and thermohaline circulation
strength during glacial times in the Atlantic Ocean
inferred from either 8 13 C values or radiochemical
data (Yu et al. 1996).
In this paper, we first give a brief state-of-theart compilation of 8 13 C research regarding both
benthic foraminiferal as well as dissolved inorganic
carbon isotopic composition in the Atlantic and Antarctic oceans. We then compile new results obtained during the last five years or so from the calibration of southern Atlantic live benthic
foraminiferal 8 13 C values with 813CLC02 values
from overlying bottom water masses. Most of these
data are published previously in various sources, but
are compiled here for the first time (Duplessy et
al. 1984; Bickert 1992; Mackensen et al. 1993;
Mackensen et al. 1996; Mackensen 1997). In addition, new data from the inner Weddell Sea are
presented that corroborate some of our earlier findings. In a third part of this paper we present two
depth transects of sediment cores from the equatorial Brazil and Guinea basins and one core located
some degrees north of the present-day Antarctic
Polar Front to discuss South Atlantic and Southern
Ocean paleoceanography as inferred from benthic
foraminiferal8 13 C values (Mackensen et al. 1994;
Bickert and Wefer 1996).
Calibration of Bottom Water 813CEC02
versus Phosphate
The correlation between carbon isotope ratios in the
ocean water masses and nutrients is not perfect.
This is because carbon circulates through the atmosphere and the sea, whereas nutrients just
circulate within the oceans. If there were no airsea fractionation, the distribution of 8 13 C values
within the sea would be tightly coupled with nutrient concentrations. During gas exchange between
the surface water and the atmosphere, a strongly
temperature dependent isotopic fractionation occurs, such that fractionation increases by approximately 0.12 %0 with temperature decreasing by 1 DC
(Mook et al. 1974; Zhang et al. 1995). Atlowtemperatures, the isotopic fractionation between atmospheric CO 2 and surface ocean LC0 2 decouples
13C/ 12 C ratios from nutrients. Consequently, highest positive deviations of813CIC02 values from the
expected correlation between phosphate and 8 1l C
occur in Antarctic surface and intermediate waters,
coinciding with the Polar Frontal Zone where Antarctic Intermediate Water (AAIW) is formed and
distributed (Charles and Fairbanks 1990; Broecker
and Maier-Reimer 1992). In addition, Mackensen
et al. (1993; 1996) and Charles et al. (1993) showed
that thermodynamic 8 13 C effects are not limited to
surface and intermediate water masses, but must
be considered at all depths, especially in areas of
deep and bottom water formation.
When Kroopnick (1985) observed the constant
relation between 813CLC02 and phosphate for the
bulk of the world's oceans (i.e., a 0.71 %0 increase
in 8 1l C for every Ilmollkg PO. decrease), he only
used samples with a potential density in excess of
27.7. In general, this density occurs at depths between 100 and 1800 m and it was chosen to avoid
thermodynamic isotopic fractionation during air-sea
gas exchange between atmospheric CO 2 and dissolved inorganic carbon near the ocean surface and
thus to keep the correlation restricted to the biologic
alteration processes. Turning around this idea,
Oppo and Fairbanks (\989) and Charles and
Fairbanks (1990) used a "Redfield" slope of 0.93,
i.e. a 0.93 %0 increase in 8 1l C for every Ilmollkg
decrease in phosphate, to normalize the 8IlCLC02
signal of Southern Ocean surface waters to
zero phosphate. Phosphate normalized 8 1l C
(!l.8IlCLCO) is the 8 1l C a water mass would have
if phosphate were entirely removed. Thus no biologically driven fractionation would mask the sense
and magnitude ofthe isotopic fractionation during
gas exchange at low temperatures. The same
approach based on high quality deep Indian and
Pacific data, with a slightly steeper "Redfield" tie
between 8 13 C and PO. (1.1 %o/lllmol kgl), was
used by Broecker and Maier-Reimer (\992) to
quantify the biologic versus the thermodynamic
effect on the 813CLC02 distribution in the surface
ocean and the deep ocean.