sills (with a depth of less than one kilometer) mixes with the
much warmer waters of the North Atlantic, so that the newly
formed NADW is characterized by a temperature of +3 °C
and a salinity of 34.95 psu. Its density remains high, close to
but slightly less than that of AABW (temperature of −1 to
0 °C, salinity of 34.6 psu). It is therefore AABW that lines
the great ocean depths, and, in the Atlantic Ocean, is topped
by NADW. During the last glaciation, the sinking of very
cold water directly into the North Atlantic (and not into the
Norwegian Sea anymore) explains why the deep waters of
the world are all found to be at temperatures close to the
freezing point.
Estimating the Temperature Independently
of the Paleotemperature Formula
The simplest way to determine the temperature of the water
close to the sediment is to use the concentration of trace
metals (Mg/Ca) contained in the carbonate shells of benthic
foraminifera. This independent estimate of bottom water
temperature allows the calculation of the d w of deep water
using the paleotemperature formula.
However, the temperature dependence of Mg incorporation in benthic foraminiferal tests is species-specific, and
may depend on different hydrological factors such as salinity
or carbonate ion saturation (Elderfield et al. 2006). In
addition, the expected bottom water temperature variations
during glacial-interglacial cycles are small compared to
surface temperature changes, implying relatively small
Mg/Ca variations. The recently discovered D47 method is
only starting to be applied to this problem (Peral et al. 2018),
although more precise measurements in benthic foraminifera
are needed to confirm the utility of this technique.
Searching for the Geochemical Signature
of Ancient Waters in Pore Waters
Adkins et al. (2002) found that in long cores extracted by
drilling ships, pore water trapped within the sediments
shows measurable differences in salinity and in d w . These
differences increase initially with core depth, then reach a
maximum after which they decrease slowly. They interpreted this maximum as the signature of highly saline water
from the LGM that had diffused into the sedimentary column. Using a simple diffusion model, the authors estimated
the values for the salinity and the d w of the bottom waters
20 ka ago. Using this estimate of the d w of bottom waters,
and combining it with d
18 O measurements on the calcite of
benthic foraminifera, the paleotemperature formula confirmed that the deep waters of the glacial ocean were actually
at a temperature near freezing point and hypersaline. The
sediment cores that have been measured for d w are too few
to give a complete picture of the ocean during the last
glaciation. However, they show a significant disparity in the
salinities from one basin to another, with the Southern Ocean
being the saltiest, in contrast with the modern situation.
Reconstructing Changes in Water Mass
Distribution
Lynch-Stieglitz et al. (1999) and (2014) showed that an
approximate direct relationship could be established between
the d
18 O of benthic foraminifera and the density of seawater,
within the temperature range where the temperature-salinitydensity relationship is roughly linear (T greater than 2 °C).
In this way, the authors were able to study the geostrophic
deformations of the deep thermocline in the Straits of Florida
and propose estimates of the changes in the meridian flow
linked to the Atlantic thermohaline circulation between the
LGM and the present.
Reconstructions using benthic foraminiferal d
18 O have
also shown marked changes in the distribution of deep and
intermediate water masses during the LGM compared to the
present day. In the Atlantic Ocean, the temperature gradient
currently observed at the base of NADW at around 3000 m
was to be found at around 2000 m, and was much more
pronounced than the one that currently separates NADW and
AABW (Labeyrie et al. 1992). In the Indian Ocean, a strong
gradient separated two water masses with distinctly different
characteristics at a depth of around 2000 m (Kallel et al.
1988). More recent research even suggests the presence of a
third deep water mass in the deepest North Atlantic at the
LGM, formed by brine rejection and not by heat loss to the
atmosphere (Keigwin and Swift 2017).
Reconstructing the Circulation of Deep Waters
Searching for Lines of Current from the d
13 C
of Benthic Foraminifera
An original approach, independent of temperature and
salinity, tries to characterize the main features of deep water
circulation without trying to a priori understand the underlying physical mechanisms governing it. It is based on the
carbon cycle and its tracer, the
13 C/
12 C ratio usually denoted
as d
13 C. At the ocean surface, waters easily exchange their
gas content with the atmosphere; they contain carbon dioxide and are rich in dissolved oxygen. During photosynthesis,
phytoplankton preferentially absorbs
12 CO 2 over
13 CO 2 . The
organic material thus produced has a d
13
C close to −20‰,
while the d
13 C of dissolved CO 2 in surface waters varies
between +1 and +2‰. This surface organic matter forms the
base of the ocean’s food chain, and eventually falls to the
depths carried in fecal pellets of zooplankton and higher
animals. In the water column, settling organic matter
undergoes a slow remineralization, which consumes any
dissolved oxygen that may remain and produces CO 2
238
T. Caley et al.
much warmer waters of the North Atlantic, so that the newly
formed NADW is characterized by a temperature of +3 °C
and a salinity of 34.95 psu. Its density remains high, close to
but slightly less than that of AABW (temperature of −1 to
0 °C, salinity of 34.6 psu). It is therefore AABW that lines
the great ocean depths, and, in the Atlantic Ocean, is topped
by NADW. During the last glaciation, the sinking of very
cold water directly into the North Atlantic (and not into the
Norwegian Sea anymore) explains why the deep waters of
the world are all found to be at temperatures close to the
freezing point.
Estimating the Temperature Independently
of the Paleotemperature Formula
The simplest way to determine the temperature of the water
close to the sediment is to use the concentration of trace
metals (Mg/Ca) contained in the carbonate shells of benthic
foraminifera. This independent estimate of bottom water
temperature allows the calculation of the d w of deep water
using the paleotemperature formula.
However, the temperature dependence of Mg incorporation in benthic foraminiferal tests is species-specific, and
may depend on different hydrological factors such as salinity
or carbonate ion saturation (Elderfield et al. 2006). In
addition, the expected bottom water temperature variations
during glacial-interglacial cycles are small compared to
surface temperature changes, implying relatively small
Mg/Ca variations. The recently discovered D47 method is
only starting to be applied to this problem (Peral et al. 2018),
although more precise measurements in benthic foraminifera
are needed to confirm the utility of this technique.
Searching for the Geochemical Signature
of Ancient Waters in Pore Waters
Adkins et al. (2002) found that in long cores extracted by
drilling ships, pore water trapped within the sediments
shows measurable differences in salinity and in d w . These
differences increase initially with core depth, then reach a
maximum after which they decrease slowly. They interpreted this maximum as the signature of highly saline water
from the LGM that had diffused into the sedimentary column. Using a simple diffusion model, the authors estimated
the values for the salinity and the d w of the bottom waters
20 ka ago. Using this estimate of the d w of bottom waters,
and combining it with d
18 O measurements on the calcite of
benthic foraminifera, the paleotemperature formula confirmed that the deep waters of the glacial ocean were actually
at a temperature near freezing point and hypersaline. The
sediment cores that have been measured for d w are too few
to give a complete picture of the ocean during the last
glaciation. However, they show a significant disparity in the
salinities from one basin to another, with the Southern Ocean
being the saltiest, in contrast with the modern situation.
Reconstructing Changes in Water Mass
Distribution
Lynch-Stieglitz et al. (1999) and (2014) showed that an
approximate direct relationship could be established between
the d
18 O of benthic foraminifera and the density of seawater,
within the temperature range where the temperature-salinitydensity relationship is roughly linear (T greater than 2 °C).
In this way, the authors were able to study the geostrophic
deformations of the deep thermocline in the Straits of Florida
and propose estimates of the changes in the meridian flow
linked to the Atlantic thermohaline circulation between the
LGM and the present.
Reconstructions using benthic foraminiferal d
18 O have
also shown marked changes in the distribution of deep and
intermediate water masses during the LGM compared to the
present day. In the Atlantic Ocean, the temperature gradient
currently observed at the base of NADW at around 3000 m
was to be found at around 2000 m, and was much more
pronounced than the one that currently separates NADW and
AABW (Labeyrie et al. 1992). In the Indian Ocean, a strong
gradient separated two water masses with distinctly different
characteristics at a depth of around 2000 m (Kallel et al.
1988). More recent research even suggests the presence of a
third deep water mass in the deepest North Atlantic at the
LGM, formed by brine rejection and not by heat loss to the
atmosphere (Keigwin and Swift 2017).
Reconstructing the Circulation of Deep Waters
Searching for Lines of Current from the d
13 C
of Benthic Foraminifera
An original approach, independent of temperature and
salinity, tries to characterize the main features of deep water
circulation without trying to a priori understand the underlying physical mechanisms governing it. It is based on the
carbon cycle and its tracer, the
13 C/
12 C ratio usually denoted
as d
13 C. At the ocean surface, waters easily exchange their
gas content with the atmosphere; they contain carbon dioxide and are rich in dissolved oxygen. During photosynthesis,
phytoplankton preferentially absorbs
12 CO 2 over
13 CO 2 . The
organic material thus produced has a d
13
C close to −20‰,
while the d
13 C of dissolved CO 2 in surface waters varies
between +1 and +2‰. This surface organic matter forms the
base of the ocean’s food chain, and eventually falls to the
depths carried in fecal pellets of zooplankton and higher
animals. In the water column, settling organic matter
undergoes a slow remineralization, which consumes any
dissolved oxygen that may remain and produces CO 2
238
T. Caley et al.