244
Mackensen and Bickert
According to the discussion summarized in the
first chapters ofthis paper, three factors determine
late Quaternary variations in epibenthic carbon isotope ratios. (1) Global changes in the813C ofLC0 2
of seawater (Curry et a!. 1988; Duplessy et a!.
1988), (2) productivity/seasonality induced effects
(Macken sen et a!. 1993), and (3) changes in the
depth distribution or preformed properties of deep
water masses at a given position. The latter factor
includes a possible increased thermodynamic imprint due to air-sea fractionation in the cold source
areas of deep and bottom water masses (Broecker
and Maier-Reimer 1992; Charles et a!. 1993;
Mackensen et a!. 1996).
To estimate the global shift in813C values, we
compared the carbon isotope record of Core GeoB
1112, located on the Mid Atlantic Ridge south of
the equatorial upwelling zone (Figs. 8, 11), to the
record ofODP Hole 806B (OON, 159°E, 2529 m)
from the western equatorial Pacific, which is assumed to monitor exclusively the global variability
in 8 13 C (Bickert et a!. 1993). The Pacific site always has 8 13 C values lower than the Atlantic site.
The interglacial8 13 C difference between Atlantic
and Pacific water is about 0.8 %0, consistent with
the modern gradient between the two oceans.
During glacials, carbon isotope values of these sites
converge to a difference as small as 0.4 %0. Therefore, the variability ofthe global shift accounts for
about 50% of the glacial/interglacial amplitudes in
the mid-depth equatorial Atlantic (Fig. 10). This
global variability in 013C is assumed to be due to
global changes in the oceanic carbon reservoir.
To estimate the productivity/seasonality induced
effect two examples are presented in Fig. II. The
first one is a comparison of the records of Core
GeoB 1105, which is located in the central
upwelling area south of the equator, and Core GeoB
1112, located south of the highly productive area.
Although both cores are located at about the same
water depth, the glacial 8 13 C values of the
upwelling core are up to 0.5 %0 lower than the
values of Core GeoB 1112. The pattern of the difference of carbon isotopes between the two sites
is similar to the difference in accumulation rates of
total organic carbon (TOC), and is characterized
by a strong power in the 23-ky period. There is a
clear relationship between the deviation of the carbon isotopes to the TOC accumulation which here
is regarded as a proxy for productivity. This effect
on 8 13 C is explained to be caused by the decay of
organic matter, reducing the 13C/'2C ratio in the
"fluffY" layer, which influences the carbon isotopic
composition of the F. wuellerstorfi tests, as described in greater detail above.
Both effects on carbon isotopes - the global
variability and the productivity-related phytodetritus
effect - together are not sufficient to explain the
high glacial to interglacial amplitudes of up to
1.6 %0, which are observed especially in the carbon isotope records of the equatorial cores. An
additional effect is called for, namely a change in
water mass distribution and/or chemistry.
Starting with the records of the Brasil Basin
(Fig. 9), where the shallower core positions (GeoB
IllS, 1117) today are bathed in NADW and the
deepest location (GeoB 1118) is completely covered by AABW, it is evident that during glacial
periods the mid-depth carbon isotope record (GeoB
1117) is close to the values of the deepest record
(GeoB 1118) and shows therefore the highest glacial to interglacial amplitudes of 1.2 %0 (stages 211
transition) and 1.6 %0 (stages 10/9 and 8/7 transitions). Estimating the sum of global shift and
phytodetritus effect for the last termination to be
about 0.8 %0, an additional amplitude of 0.4 %0 remains, which equals about the today's difference
in 8 13 C between NADW (0.9 %0) and AABW (0.5
%0) in the western South Atlantic (Kroopnick
1985). It is therefore commonly explained by a glacial reduction ofNADW, substituted by a southern component water mass, as reported by many
authors (Boyle and Keigwin 1987; Labeyrie et a!.
1987; Curry eta!. 1988; Duplessy eta!. 1988; Oppo
and Fairbanks 1990; Raymo eta!. 1990; Sarnthein
et a!. 1994; Curry 1996; Raymo et al. 1997). This
change in deep water circulation is also obvious in
older glacials such as stages 4, 6, 8, and 10, except
for a few warmer periods in glacial times, such as
substages 3.3, 8.3, and 8.5. In these warmer periods, on the one hand, the 8 13 C values are intermediate between full-glacial and full-interglacial values, suggesting at least an admixture of a northern
component water mass to the glacial southern component water mass. On the other hand, values almost as low as during glacials in the cold intergla-
Mackensen and Bickert
According to the discussion summarized in the
first chapters ofthis paper, three factors determine
late Quaternary variations in epibenthic carbon isotope ratios. (1) Global changes in the813C ofLC0 2
of seawater (Curry et a!. 1988; Duplessy et a!.
1988), (2) productivity/seasonality induced effects
(Macken sen et a!. 1993), and (3) changes in the
depth distribution or preformed properties of deep
water masses at a given position. The latter factor
includes a possible increased thermodynamic imprint due to air-sea fractionation in the cold source
areas of deep and bottom water masses (Broecker
and Maier-Reimer 1992; Charles et a!. 1993;
Mackensen et a!. 1996).
To estimate the global shift in813C values, we
compared the carbon isotope record of Core GeoB
1112, located on the Mid Atlantic Ridge south of
the equatorial upwelling zone (Figs. 8, 11), to the
record ofODP Hole 806B (OON, 159°E, 2529 m)
from the western equatorial Pacific, which is assumed to monitor exclusively the global variability
in 8 13 C (Bickert et a!. 1993). The Pacific site always has 8 13 C values lower than the Atlantic site.
The interglacial8 13 C difference between Atlantic
and Pacific water is about 0.8 %0, consistent with
the modern gradient between the two oceans.
During glacials, carbon isotope values of these sites
converge to a difference as small as 0.4 %0. Therefore, the variability ofthe global shift accounts for
about 50% of the glacial/interglacial amplitudes in
the mid-depth equatorial Atlantic (Fig. 10). This
global variability in 013C is assumed to be due to
global changes in the oceanic carbon reservoir.
To estimate the productivity/seasonality induced
effect two examples are presented in Fig. II. The
first one is a comparison of the records of Core
GeoB 1105, which is located in the central
upwelling area south of the equator, and Core GeoB
1112, located south of the highly productive area.
Although both cores are located at about the same
water depth, the glacial 8 13 C values of the
upwelling core are up to 0.5 %0 lower than the
values of Core GeoB 1112. The pattern of the difference of carbon isotopes between the two sites
is similar to the difference in accumulation rates of
total organic carbon (TOC), and is characterized
by a strong power in the 23-ky period. There is a
clear relationship between the deviation of the carbon isotopes to the TOC accumulation which here
is regarded as a proxy for productivity. This effect
on 8 13 C is explained to be caused by the decay of
organic matter, reducing the 13C/'2C ratio in the
"fluffY" layer, which influences the carbon isotopic
composition of the F. wuellerstorfi tests, as described in greater detail above.
Both effects on carbon isotopes - the global
variability and the productivity-related phytodetritus
effect - together are not sufficient to explain the
high glacial to interglacial amplitudes of up to
1.6 %0, which are observed especially in the carbon isotope records of the equatorial cores. An
additional effect is called for, namely a change in
water mass distribution and/or chemistry.
Starting with the records of the Brasil Basin
(Fig. 9), where the shallower core positions (GeoB
IllS, 1117) today are bathed in NADW and the
deepest location (GeoB 1118) is completely covered by AABW, it is evident that during glacial
periods the mid-depth carbon isotope record (GeoB
1117) is close to the values of the deepest record
(GeoB 1118) and shows therefore the highest glacial to interglacial amplitudes of 1.2 %0 (stages 211
transition) and 1.6 %0 (stages 10/9 and 8/7 transitions). Estimating the sum of global shift and
phytodetritus effect for the last termination to be
about 0.8 %0, an additional amplitude of 0.4 %0 remains, which equals about the today's difference
in 8 13 C between NADW (0.9 %0) and AABW (0.5
%0) in the western South Atlantic (Kroopnick
1985). It is therefore commonly explained by a glacial reduction ofNADW, substituted by a southern component water mass, as reported by many
authors (Boyle and Keigwin 1987; Labeyrie et a!.
1987; Curry eta!. 1988; Duplessy eta!. 1988; Oppo
and Fairbanks 1990; Raymo eta!. 1990; Sarnthein
et a!. 1994; Curry 1996; Raymo et al. 1997). This
change in deep water circulation is also obvious in
older glacials such as stages 4, 6, 8, and 10, except
for a few warmer periods in glacial times, such as
substages 3.3, 8.3, and 8.5. In these warmer periods, on the one hand, the 8 13 C values are intermediate between full-glacial and full-interglacial values, suggesting at least an admixture of a northern
component water mass to the glacial southern component water mass. On the other hand, values almost as low as during glacials in the cold intergla-