LAUSR

Abstract We test the applicability of bivalve shell oxygen isotope composition to reconstruct hydrological dynamics in four riverine sites in the Congo River basin. Twenty-three specimens from the Unionoida order were collected from locations where long-term discharge data are available, and in situ measurements and water samples were collected over several years. Due to the highly variable (species-specific) shell morphology, various sampling techniques were used to analyze the shell sections; however, every specimen recorded the seasonality of the host water oxygen stable isotope composition (δ18Ow) in its δ18Oshell record. Discharge data showed an inverse relationship with δ18Ow values, which was well described with a logarithmic fit. An exception was the Kasai River, where the δ18Ow record shows an additional peak occurring during the high discharge period, which renders the discharge-δ18Ow relationship more complex than in the other systems investigated. Low ratios of maximum to minimum discharge (Qmax/Qmin) were found to result in a low δ18Ow amplitude, which was reflected as low δ18Oshell variability. The Congo and Kasai rivers had Qmax/Qmin ratios ∼2 to 2.5, while the Oubangui showed a much higher Qmax/Qmin (∼19). Shells correspondingly showed a large δ18Oshell range (amplitude between 2.4 and 5.0‰) for individual Oubangui shells, and lower amplitude for other sites (1.0 to 2.2‰). Thus, shells have a high resolving power to be used to record hydrological variability, since long-term changes in precipitation pattern, discharge, land-use change, or other hydrological changes have an influence on δ18Ow values. Shells with wide range of δ18O values reflect high seasonal variability in rivers, while shells with lower δ18O amplitude correspond to sites with more steady river conditions over the year. Our study illustrates that fossil shell δ18O values could indicate Qmax/Qmin values in ancient African river systems.