AbstractThe method of reconstructing paleoatmospheric CO2 levels using carbon isotope signatures of organic matter buried in sediments has been questioned due to the dubious foundation that carbon isotope fractionation during phytoplankton photosynthesis (εp) is controlled primarily by aquatic CO2 concentration ([CO2(aq)]). Consequently, what carbon isotope data from bulk sedimentary organic matter reflects is a puzzle. In this study, we determined the carbon isotope compositions of dissolved inorganic carbon and particulate organic carbon in a lake located in a carbonate area. Partial correlation analysis was employed to distinguish between direct and indirect factors in controlling εp. The results show that εp is more closely, and more steadily related with pH than with [CO2(aq)], which is in accordance with recent advances in our understanding of the physiology of carbon utilization by phytoplankton for CO2 and . Therefore, we propose that carbon isotope fractionation in phytoplankton is more suitable as a proxy of pH than of [CO2(aq)]. One advantage of this amendment is that information on , the main species of carbon uptake by phytoplankton, is likewise included. In the future, culture experiments aiming at revealing the relationship between pH and cellular carbon isotope signatures is necessary to construct a new isotope fractionation formula to couple the different effects of CO2 and , which is of critical importance to improve the understanding of carbon isotope fractionation, and to more precisely model pH and CO2.