What Ends an Interglacial? Feedbacks Between Tropical Rainfall, Atlantic Climate and Ice Sheets During the Last Interglacial - Integrated Analysis of Interglacial Climate Dynamics (INTERDYNAMIC) - page 24

Although the decrease in boreal summer insolation is the primary driver ending the
LIG, vegetation and ocean feedbacks are necessary to initiate ice sheet growth.
However, the role of the prolonged North Atlantic warmth observed during the LIG is
unclear: did it favor or delay the end of the LIG? Warm North Atlantic waters were
likely sustained by an active thermohaline circulation (THC) across the LIG. Climate
models, however, show no consistent temporal evolution of the THC throughout the
LIG (Bakker et al. 2013 ), making involved climatic mechanisms uncertain. Past
millennial-scale climate variability highlights strong interactions between tropical
hydrology and the THC (Krebs and Timmermann 2007 ). However, the lack of
detailed tropical proxy records and adequate simulations prevented investigating the
role of feedbacks between tropical rainfall, Atlantic climate and ice sheets in the
sequence of events ending the LIG.
By combining new high-resolution paleoclimate records and model experiments,
we determine the in

uence of South American precipitation changes on tropical
Atlantic salinity, North Atlantic mixed layer depth and the Atlantic THC. The
effects of these changes on ice-sheet inception and growth at the end of the LIG
presently remain under investigation.

2 Materials and Methods

We consider a transect of eight sediment cores from the South American margin
(12 S). To reconstruct past changes in South American rainfall and fresh-

water delivery to the ocean, we applied well-established geochemical methods on
the biogenic, terrigenous and organic fractions of the sediment. With the world

° N - 32 °

s
largest discharge, the Amazon River is the most likely river to impact tropical
Atlantic salinity and the THC. We hence focus here on the northernmost sites
(Fig. 1 ) located along the pathway of Amazon freshwater. We measured the sed-
imentary elemental composition which allows tracing the provenance of terrigenous
material and reconstructing past climate variations over the source regions. We
de

'

ned regional terrigenous endmembers based on six major elements and applied
an endmember unmixing model to deduce the relative proportions of Amazonian
Andean versus lowland material at 5 ° N and 9 ° N, and of Amazon versus Orinoco

material at 12 N. See Govin et al. ( 2014 ) for detailed methods.

We use the Community Climate System Model version 3 (CCSM3) to co-verify
the South American hydrologic reconstructions and investigate impacts on the
climate system. The model was con

°

gured with T31 atmospheric resolution cou-
pled to a nominally 3-degree ocean (Yeager et al. 2006 ). The fully coupled model
further includes sea ice and a land surface model. We used CCSM3 for two
equilibrated experiments, with orbital parameters and greenhouse gas concentra-
tions at 125 and 115 ka BP as boundary conditions. The 125 ka BP experiment was
initialized from a quasi-equilibrated 130 ka BP experiment. In turn, the 115 ka BP
experiment was branched from the 125 ka BP experiment following an initial
equilibration period. Both 115 and 125 ka BP simulations were allowed to