University of Cambridge, United Kingdom

The isotopic composition of carbon, delta (super 13) C, in seawater is used in reconstructions of ocean circulation, marine productivity, air-sea gas exchange, and biosphere carbon storage. Here, a synthesis of delta (super 13) C measurements taken from foraminifera in marine sediment cores over the last 150 000 years is presented. The dataset comprises previously published and unpublished data from benthic and planktonic records throughout the global ocean. Data are placed on a common delta (super 18) O age scale suitable for examining orbital timescale variability but not millennial events, which are removed by a 10 ka filter. Error estimates account for the resolution and scatter of the original data, and uncertainty in the relationship between delta (super 13) C of calcite and of dissolved inorganic carbon (DIC) in seawater. This will assist comparison with delta (super 13) C of DIC output from models, which can be further improved using model outputs such as temperature, DIC concentration, and alkalinity to improve estimates of fractionation during calcite formation. High global deep ocean delta (super 13) C, indicating isotopically heavy carbon, is obtained during Marine Isotope Stages (MIS) 1, 3, 5a, c and e, and low delta (super 13) C during MIS 2, 4 and 6, which are temperature minima, with larger amplitude variability in the Atlantic Ocean than the Pacific Ocean. This is likely to result from changes in biosphere carbon storage, modulated by changes in ocean circulation, productivity, and air-sea gas exchange. The North Atlantic vertical delta (super 13) C gradient is greater during temperature minima than temperature maxima, attributed to changes in the spatial extent of Atlantic source waters. There are insufficient data from shallower than 2500 m to obtain a coherent pattern in other ocean basins. The data synthesis indicates that basin-scale delta (super 13) C during the last interglacial (MIS 5e) is not clearly distinguishable from the Holocene (MIS 1) or from MIS 5a and 5c, despite significant differences in ice volume and atmospheric CO (sub 2) concentration during these intervals. Similarly, MIS 6 is only distinguishable from MIS 2 or 4 due to globally lower delta (super 13) C values both in benthic and planktonic data. This result is obtained despite individual records showing differences between these intervals, indicating that care must be used in interpreting large scale signals from a small number of records.