glacial-interglacial cycles. Atmospheric CO 2 changes are
driven by changes in carbon storage in other reservoirs,
particularly on land and in the ocean.
On land, the lower sea level, by 120 m, during glacial
time increased the surface area where plants could develop,
although the larger ice sheets covering part of North
America and Eurasia reduced the available area. Overall, due
to the colder and drier climate, as well as reduced atmospheric CO 2 concentration, photosynthesis by land vegetation is reduced and the terrestrial biosphere tends to
represent a smaller carbon reservoir during glacial period,
which would increase atmospheric CO 2 , not lower it. This is
indicated by ocean d
13 C which decreased by around 0.03–
0.04‰ during the Last Glacial Maximum compared to the
pre-industrial level. This is explained by the transfer of
continental carbon with low d
13 C values (due to fractionation during photosynthesis) to the ocean (Shackleton 1977;
Bird et al. 1994), causing a reduction in continental carbon
of around 270–720 GtC. An understanding of the changes in
the terrestrial biosphere can also be obtained from
paleo-biomes, using pollen from sediment cores (Adams
et al. 1990; Crowley 1995; Maslin et al. 1995) which show a
carbon loss of around 750–1350 GtC. Pollen data have also
been used to reconstruct maps of vegetation types during the
LGM (Fig. 23.7).
However, frozen soils, i.e. permafrost, may have
increased in glacial times, potentially storing more carbon,
which could partly explain the lower CO 2 (Ciais et al. 2012)
and help resolve the d
13 CO 2 signal recorded in ice cores,
which strongly depends on land carbon changes (Crichton
et al. 2016).
Although permafrost probably played a role, most of the
change is likely to have come from the ocean, which is a
much bigger carbon reservoir. In addition, d
13 C measurements from sediment cores indicate large changes in the
ocean (Curry and Oppo 2005; Marchal and Curry 2008;
Hesse et al. 2011). In the ocean, the carbon cycle changes
could originate from modifications of biological activity and
physical or chemical changes. Known processes include
temperature change as colder temperatures lead to more
carbon being stored in the ocean. The sea level drop
of *120 m during glacial maxima results in higher concentrations of salinity, which causes a reduction in the
storage capacity of carbon in the ocean, and an increase in
nutrients, which increases biological activity and thus
increases ocean carbon storage. However, these processes
are not sufficient to fully explain the decrease in CO 2
decrease (see review by Sigman and Boyle 2000; Archer
et al. 2000), and additional mechanisms are needed.
Four main hypotheses have been proposed to explain the
CO 2 lowering by increased ocean carbon storage: increased
biological pump, isolation of the ocean from the atmosphere
due to sea ice coverage, changes in ocean dynamics, and
carbonate compensation.
The biological pump causes more carbon to be stored
when it is stimulated, for example when more nutrients are
delivered to the ocean (Broecker and Peng 1982). In regions
of high nutrients low chlorophyll (HNLC), biological
activity is limited due to the lack of iron. An influx of iron to
these zones during glacial periods would increase biological
activity. Alternatively, the biological pump could also store
more carbon if it becomes more efficient, for example with a
greater carbon to nutrient ratio (Broecker and Peng 1982) or
a switch of plankton species with higher productivity
(Archer and Maier-Reimer 1994). However, both data and
model simulations have shown that changes in biological
activity are not sufficient to sufficiently account for the
decrease in atmospheric CO 2 (Kohfeld et al. 2005; Bopp
et al. 2003a; Tagliabue et al. 2009; Lambert et al. 2015).
Increased sea ice coverage has also been proposed, as this
could isolate the ocean, preventing carbon from getting to
the atmosphere, hence lowering atmospheric CO 2 (Stephens
and Keeling 2000). But such an impact has only been simulated in very simple models, more complex models do not
show such an effect on CO 2 (Archer et al. 2003).
Most current theories involve changes in ocean dynamics,
and point to the Southern Ocean (Fischer et al. 2010).
A larger ocean volume occupied by AABW, or slower
overturning, could result in more carbon stored in the deep
ocean, reducing atmospheric CO 2 . Ocean circulation changes are supported by data generally indicating a reduced
NADW and a more stratified Southern Ocean (Adkins
2013). In particular, d
13 C measurements show lower d
13 C
values in the deep glacial ocean, especially around Antarctica, and higher values near the surface (Curry and Oppo
2005; Marchal and Curry 2008; Hesse et al. 2011). In
addition, very salty water has been measured in the deep
Southern Ocean (Adkins et al. 2002). Complementary data,
such as from neodymium isotopes (Basak et al. 2018), B/Ca
ratio (Yu et al. 2016) and D
14 C (Skinner et al. 2010), also
point towards changes in the circulation in the Southern
Ocean.
Comparison of model simulations over the last decade
have shown that models simulate a large range of ocean
circulation changes, which are generally opposite to those
deduced from data. In PMIP3, most models simulate a
strengthening and deepening of the NADW with LGM
boundary conditions (Muglia and Schmittner 2015). Yet
simulations have shown that better agreement with d
13 C and
CO 2 data requires lower NADW intensity and/or shoaling of
NADW (Tagliabue et al. 2009; Tschumi et al. 2011; Menviel
et al. 2017). This is also seen in terms of water mass volume
with a smaller volume of NADW and a larger volume of
AABW filling the ocean. The latter has a larger DIC content,
23 Biogeochemical Cycles and Aerosols Over the Last Million Years
279
driven by changes in carbon storage in other reservoirs,
particularly on land and in the ocean.
On land, the lower sea level, by 120 m, during glacial
time increased the surface area where plants could develop,
although the larger ice sheets covering part of North
America and Eurasia reduced the available area. Overall, due
to the colder and drier climate, as well as reduced atmospheric CO 2 concentration, photosynthesis by land vegetation is reduced and the terrestrial biosphere tends to
represent a smaller carbon reservoir during glacial period,
which would increase atmospheric CO 2 , not lower it. This is
indicated by ocean d
13 C which decreased by around 0.03–
0.04‰ during the Last Glacial Maximum compared to the
pre-industrial level. This is explained by the transfer of
continental carbon with low d
13 C values (due to fractionation during photosynthesis) to the ocean (Shackleton 1977;
Bird et al. 1994), causing a reduction in continental carbon
of around 270–720 GtC. An understanding of the changes in
the terrestrial biosphere can also be obtained from
paleo-biomes, using pollen from sediment cores (Adams
et al. 1990; Crowley 1995; Maslin et al. 1995) which show a
carbon loss of around 750–1350 GtC. Pollen data have also
been used to reconstruct maps of vegetation types during the
LGM (Fig. 23.7).
However, frozen soils, i.e. permafrost, may have
increased in glacial times, potentially storing more carbon,
which could partly explain the lower CO 2 (Ciais et al. 2012)
and help resolve the d
13 CO 2 signal recorded in ice cores,
which strongly depends on land carbon changes (Crichton
et al. 2016).
Although permafrost probably played a role, most of the
change is likely to have come from the ocean, which is a
much bigger carbon reservoir. In addition, d
13 C measurements from sediment cores indicate large changes in the
ocean (Curry and Oppo 2005; Marchal and Curry 2008;
Hesse et al. 2011). In the ocean, the carbon cycle changes
could originate from modifications of biological activity and
physical or chemical changes. Known processes include
temperature change as colder temperatures lead to more
carbon being stored in the ocean. The sea level drop
of *120 m during glacial maxima results in higher concentrations of salinity, which causes a reduction in the
storage capacity of carbon in the ocean, and an increase in
nutrients, which increases biological activity and thus
increases ocean carbon storage. However, these processes
are not sufficient to fully explain the decrease in CO 2
decrease (see review by Sigman and Boyle 2000; Archer
et al. 2000), and additional mechanisms are needed.
Four main hypotheses have been proposed to explain the
CO 2 lowering by increased ocean carbon storage: increased
biological pump, isolation of the ocean from the atmosphere
due to sea ice coverage, changes in ocean dynamics, and
carbonate compensation.
The biological pump causes more carbon to be stored
when it is stimulated, for example when more nutrients are
delivered to the ocean (Broecker and Peng 1982). In regions
of high nutrients low chlorophyll (HNLC), biological
activity is limited due to the lack of iron. An influx of iron to
these zones during glacial periods would increase biological
activity. Alternatively, the biological pump could also store
more carbon if it becomes more efficient, for example with a
greater carbon to nutrient ratio (Broecker and Peng 1982) or
a switch of plankton species with higher productivity
(Archer and Maier-Reimer 1994). However, both data and
model simulations have shown that changes in biological
activity are not sufficient to sufficiently account for the
decrease in atmospheric CO 2 (Kohfeld et al. 2005; Bopp
et al. 2003a; Tagliabue et al. 2009; Lambert et al. 2015).
Increased sea ice coverage has also been proposed, as this
could isolate the ocean, preventing carbon from getting to
the atmosphere, hence lowering atmospheric CO 2 (Stephens
and Keeling 2000). But such an impact has only been simulated in very simple models, more complex models do not
show such an effect on CO 2 (Archer et al. 2003).
Most current theories involve changes in ocean dynamics,
and point to the Southern Ocean (Fischer et al. 2010).
A larger ocean volume occupied by AABW, or slower
overturning, could result in more carbon stored in the deep
ocean, reducing atmospheric CO 2 . Ocean circulation changes are supported by data generally indicating a reduced
NADW and a more stratified Southern Ocean (Adkins
2013). In particular, d
13 C measurements show lower d
13 C
values in the deep glacial ocean, especially around Antarctica, and higher values near the surface (Curry and Oppo
2005; Marchal and Curry 2008; Hesse et al. 2011). In
addition, very salty water has been measured in the deep
Southern Ocean (Adkins et al. 2002). Complementary data,
such as from neodymium isotopes (Basak et al. 2018), B/Ca
ratio (Yu et al. 2016) and D
14 C (Skinner et al. 2010), also
point towards changes in the circulation in the Southern
Ocean.
Comparison of model simulations over the last decade
have shown that models simulate a large range of ocean
circulation changes, which are generally opposite to those
deduced from data. In PMIP3, most models simulate a
strengthening and deepening of the NADW with LGM
boundary conditions (Muglia and Schmittner 2015). Yet
simulations have shown that better agreement with d
13 C and
CO 2 data requires lower NADW intensity and/or shoaling of
NADW (Tagliabue et al. 2009; Tschumi et al. 2011; Menviel
et al. 2017). This is also seen in terms of water mass volume
with a smaller volume of NADW and a larger volume of
AABW filling the ocean. The latter has a larger DIC content,
23 Biogeochemical Cycles and Aerosols Over the Last Million Years
279