Back

Combining fossil taxa with and without morphological data improves dated phylogenetic analyses

Nikolic, M. C.; Warnock, R. C. M.; Hopkins, M. J.

2025-04-09 evolutionary biology
10.1101/2025.04.07.646048 bioRxiv
Show abstract

The fossilized birth-death (FBD) model has become an increasingly popular method for inferring dated phylogenies. This is because it is especially useful for incorporating fossil data into such analyses, integrating fossils along with their age information directly into the tree as tips or sampled ancestors. Two approaches are common for placing fossil taxa in trees: inference based on morphological character data or using taxonomic constraints to control their topological placement. These approaches have historically been treated as alternatives, though combining the two is theoretically possible. For phylogenetic inference of entirely extinct organisms, additional related fossil taxa other than those for which morphology is available are generally overlooked. Here, we implement a combined approach on an empirical dataset for a group of trilobites. We use a morphological matrix and ages for 56 taxa and age information for another 196 taxa from the Paleobiology Database. To evaluate the effects of a combined approach, we conducted FBD dated phylogenetic analyses using a combined approach with morphology and taxonomic constraints and compared them to analyses of taxa with morphology alone. We find that a combined approach yields topologies that are more stratigraphically congruent, substantially more precise parameter estimates (e.g., divergence times), and more informative tree distributions. These findings are a consequence of the substantial increase in stratigraphic age information and a more representative sample of their temporal distributions.

Published in Biology Letters · training set

Matching journals

The top 1 journal accounts for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.