Estimating the age of poorly dated fossil specimens and deposits using a total-evidence approach and the fossilized birth-death process

Abstract

Bayesian total-evidence approaches under the fossilized birth-death model enable biologists to combine fossil and extant data while accounting for uncertainty in the ages of fossil specimens, in an integrative phylogenetic analysis. Fossil age uncertainty is a key feature of the fossil record as many empirical datasets may contain a mix of precisely dated and poorly dated fossil specimens or deposits. In this study, we explore whether reliable age estimates for fossil specimens can be obtained from Bayesian total-evidence phylogenetic analyses under the fossilized birth-death model. Through simulations based on the example of the Baltic amber deposit, we show that estimates of fossil ages obtained through such an analysis are accurate, particularly when the proportion of poorly dated specimens remains low and the majority of fossil specimens have precise dates. We confirm our results using an empirical dataset of living and fossil penguins by artificially increasing the age uncertainty around some fossil specimens and showing that the resulting age estimates overlap with the recorded age ranges. Our results are applicable to many empirical datasets where classical methods of establishing fossil ages have failed, such as the Baltic amber and the Gobi Desert deposits.