rescaling as iron supply rates (gray lines and right-hand scales) between the EPICA
Dome C ice core and a sediment core from the Atlantic sector of the subantarctic
(ODP-1090) for the three most recent cycles of glaciation. The bars indicate periods
during re-expansion of terrestrial ice sheets, which were periods of low dust
deposition on both Antarctica and the Southern Ocean.
(After Martinez-Garcia et al. 2009.)
There are other and earlier data suggesting only a small role for dust fertilization in
CO 2 cycling. Elderfield and Rickaby (2000) suggested, based on sediment profiles of
cadmium/calcium ratios in planktonic foraminifera from the subantarctic, that there
was little difference between modern and ice-age dissolved-phosphate levels in the
subantarctic sector of the Southern Ocean. The Cd/Ca reflects phosphate levels
because cadmium substitutes for zinc in a tiny fraction of enzymes in which zinc is a
cofactor, and dissolved zinc through its role as a nutrient is closely proportional to
phosphate. Greater iron availability should have allowed phytoplankton to consume
more major nutrients, reducing phosphate (and cadmium) concentration in the upper
water column. Like many more recent writers, Elderfield and Rickaby, having
downplayed iron effects, speculated that CO 2 was lower in the glacial because area-
related and seasonal increase in ice cover prevented some of the ventilation occurring
now in the Southern Ocean. They proposed that there was no change in biological
productivity involved, at least not in the Southern Ocean. There could have been a
switch to diatom production in some latitudes (again, Anderson et al. 2009) without a
general increase in productivity due to iron.
(^) In addition to the general glacial-interglacial cycling, ice-core and other records
show shorter-term variations: the Dansgaard–Oeschger (D–O) variations, mentioned
above, during glacial eras and Bond events during the Holocene. The D–O events,
most strongly evident in Greenland ice cores (Dansgaard et al. 1993), are rapid (∼ 40
years) warmings with much longer cooling phases and total cycle times around 1470