LAUSR

Abstract Freshwater bivalve shell oxygen and carbon stable isotope ratios (δ18O, δ13C) may act as recorders of hydroclimate (e.g., precipitation-evaporation balance, discharge) and aquatic biogeochemistry. We investigate the potential of these hydroclimate proxies measured along the growth axis of shells collected from the Oubangui River (Bangui, Central African Republic) and the Niger River (Niamey, Niger). Biweekly water samples and in situ measurements collected over several years, along with daily discharge data from both sites allowed a direct comparison with proxies recorded in the shells. Data from a total of 14 unionid shells, including three species (Chambardia wissmanni, Aspatharia dahomeyensis, and Aspatharia chaiziana), confirmed that shells precipitate carbonate in oxygen isotope equilibrium with ambient water. Because water temperature variations were small, shell δ18O values (δ18Oshell) also accurately record the seasonality and the range observed in water δ18O (δ18Ow) values when calculated using an average temperature. Calculated δ18Ow values were in good agreement over the entire record of measured δ18Ow values, thus δ18Oshell records can be reliably used to reconstruct past δ18Ow values. Discharge and δ18Ow values from both rivers fit a logarithmic relationship, which was used to attempt reconstruction of past hydrological conditions, after calculating δ18Ow values from δ18Oshell values. A comparison with measured discharge data suggests that for the two rivers considered, δ18Oshell data are good proxies for recording discharge conditions during low(er) discharge levels, but that high discharge values cannot be accurately reconstructed due to the large scatter in the discharge-δ18Ow relationship. Moreover, periods of bivalve shell growth cessation due to high turbidity or air exposure should be taken into account. While δ13C values of dissolved inorganic carbon in both rivers showed clear seasonality and correlated well with discharge, most of the shells analyzed did not record these variations adequately, likely due to the complication of vital effects including the variable contribution of metabolic CO2. Thus, tropical African unionid δ18Oshell values can be used to reconstruct δ18Ow values with high confidence to provide insight on past hydroclimate such as precipitation-evaporation balance and periods of low discharge.