1. Program in Atmospheric and Oceanic Sciences, Princeton University, PO Box CN710, Princeton, New Jersey 08544, USA
2. Geophysical Fluid Dynamics Laboratory/NOAA Princeton University, PO Box 308, Princeton, New Jersey 08542, USA
3. Present address: Earth Frontier Research System, 7th Floor, Seavans Building-N, 1-2-1 Shibaura, Minato-ku, Tokyo 105, Japan.

Correspondence to: Jorge L. Sarmiento¹ Correspondence and requests for materials should be addressed to J.L.S. (e-mail: Email: jls@splash.princeton.edu).

A 1995 report¹ of the Intergovernmental Panel on Climate Change provides a set of illustrative anthropogenic CO₂ emission models leading to stabilization of atmospheric CO₂ concentrations ranging from 350 to 1,000 p.p.m. (refs 1–4). Ocean carbon-cycle models used in calculating these scenarios assume that oceanic circulation and biology remain unchanged through time. Here we examine the importance of this assumption by using a coupled atmosphere–ocean model of global warming⁵ for the period 1765 to 2065. We find a large potential modification to the ocean carbon sink in a vast region of the Southern Ocean where increased rainfall leads to surface freshening and increased stratification⁶. The increased stratification reduces the downward flux of carbon and the loss of heat to the atmosphere, both of which decrease the oceanic uptake of anthropogenic CO₂ relative to a constant-climate control scenario. Changes in the formation, transport and cycling of biological material may counteract the reduced uptake, but the response of the biological community to the climate change is difficult to predict on present understanding. Our simulation suggests that such physical and biological changes might already be occurring, and that they could substantially affect the ocean carbon sink over the next few decades.