Carbon isotopic fractionation in alkenones, biomarkers produced by haptophyte algae, relative to that of dissolved inorganic carbon (Ɛp) is a commonly applied proxy for aqueous CO concentration, [CO ], in the ocean. However, the apparent inability of Ɛp-derived CO records to capture the full magnitude of glacial-interglacial CO change as measured in ice cores using the traditional approach has led to scepticism regarding the application of this proxy to pre-Quaternary CO reconstructions. Here, we present a new detailed late Pleistocene reconstruction of alkenone Ɛp, together with records of coccolithophore productivity, upper water column stratification, and coccolith size, mass, and stable isotopes measured in the same samples, from sediments deposited at IODP Site U1443 in the equatorial Indian Ocean (5°N), spanning Marine Isotope Stages 1 to 9a (~275 kyr). As well as considering the classical diffusive model of CO uptake to interpret Ɛp, we also explore an alternative interpretive framework, in which the relationship between Ɛp and [CO ] and other variables in all existing coccolithophore culture data is considered. In this framework, active carbon concentrating mechanisms are inherent and growth rate is highly sensitive to temperature, resulting in a significantly lower sensitivity of Ɛp to [CO ] at the low [CO ] typical of the late Pleistocene tropical oceans. At Site U1443, we find that precessional-timescale variability in coccolithophore productivity has a significant effect on Ɛp that is amplified with cell size increases at times of higher productivity. In light of these cellular considerations, we explore which model provides the best fit for our data. Over the last glacial cycle, we also determine δ O and δ C in size-separated coccoliths, and calculate Ɛ (carbon isotopic fractionation into coccolith calcite) to assess the sensitivity of this parameter to small [CO ] changes. Ɛ is highest at low [CO ], consistent with the "CO effect" as seen in the few existing culture studies and in long-term late Neogene records, suggesting that this indicator could be diagnostic of CO supply relative to demand of the cell. 2 2 2 2 2 2 2 2 2 18 13 coccolith 2 coccolith 2 2 2