INACCURATE FOSSIL PLACEMENT DOES NOT COMPROMISE TIP-DATED DIVERGENCE TIMES

Time trees underlie the exploration of evolutionary processes across deep timescales, as well as attempts to link these to the history of the planet. Tip-dated inference under the fossilized birth-death prior has become the standard approach for generating time-calibrated phylogenies. By inferring (rather than enforcing) the placement of fossils, and using their associated ages as temporal constraints, this approach results in a statistically coherent prior on divergence times. At the same time, it also directly links topological and temporal accuracies, as incorrect fossil placement should misinform divergence times in a manner roughly proportional to the error in their inferred phylogenetic position. Given the overall difficulties of morphological phylogenetics, this could pose serious issues for accurate time scaling, yet the interaction between topological and temporal error has not been thoroughly tested before. We explore this using state-of-the-art approaches to simulate phylogenies and morphological datasets that incorporate natural selection, and are benchmarked to recapitulate numerous properties of empirical datasets. We characterize patterns of error and bias in the positioning of fossils and the timing of divergence events, including a systematic overestimation of the inferred ages of deeper nodes. While divergence times improve with the incorporation of a larger number of fossil terminals, we find no link between the accuracy in their placement and the accuracy of inferred divergence times. This result is encouraging, suggesting even fossils with uncertain affinities can improve time calibration analyses, while also emphasizing that no amount of paleontological information can overturn mismatches between model priors and the true diversification history.