Christine Y Chen , David McGee, Arielle Woods, Liseth Pérez, Robert George Hatfield, R. Lawrence Edwards, Hai Cheng, Blas Lorenzo Valero-Garcés, Sophie Lehmann, Joseph S Stoner

*----- NOTE: This is a peer reviewed preprint of a paper accepted in Quaternary Science Reviews as of June 8, 2020 -----* Deep sediment cores from long-lived lake basins are fundamental records of paleoenvironmental history, but the power of these reconstructions has often been limited by poor age control. Uranium-thorium (U-Th) dating has the potential to fill a gap in current geochronological tools available for such sediment archives. We present our systematic approach to U-Th date carbonate-rich sediments from the ~100 m drill core from Lake Junín, Peru. The results form the foundation of an age-depth model spanning ~700 kyrs. High uranium concentrations (0.3--4 ppm) of these sediments allow us to date smaller quantities of material, giving us the opportunity to improve sample selection by avoiding detrital contamination, the greatest factor limiting the success of previous U-Th dating efforts in other lake basins. The dates from 174 analyses on 55 bulk carbonate samples revealed significant scatter that could not be resolved with traditional isochrons, suggesting that at least some of the sediments have not remained closed systems. To understand the source of noise in the geochronological data, we first apply threshold criteria that screen samples by their U/Th ratio, reproducibility, and d234U initial value. We then compare these results with facies types, trace element concentrations, carbonate and total organic carbon content, color reflectance, mineralogy, and ostracode shell color to investigate the causes of open system behavior. We find that the greatest impediment to U-Th dating of these sediments is not detrital contamination, but rather post-depositional remobilization of uranium. After examining U-Th data in these contexts, we identify samples that have likely experienced the least amount of alteration, and use dates from those samples as constraints for the age-depth model. Our work has several lessons for future attempts to U-Th date lake sediments, namely that the samples geologic context is equally important as the accuracy and precision of analytical measurements when determining the age of sample materials. In addition, we caution that significant geologic scatter may remain undetected if not for labor intensive tests of reproducibility achieved through replication. As a result of this work, the deep sediment core from Lake Junín is the only continuous record in the tropical Andes spanning multiple glacial cycles that is constrained entirely by independent radiometric dates.