Format: ORAL

Daniele Silvestro, Rosane Collevatti, Carina Hoorn, Huasheng Huang, Viktoria Keller, Kelly Matsunaga, Robert Morley, Luis Palazzesi, Shalani Parmar, Vandana Prasad, David Sunderlin, Yaowu Xing, and Christine D. Bacon

) University of Fribourg, Switzerland

Throughout the long evolutionary history of life, species of all kingdoms have undergone staggering diversification and faced countless environmental changes and extinction events. Since the great majority of species that lived on Earth have since gone extinct, it is difficult to infer how the evolution of biodiversity unfolded over billions of years. The fossil record provides the most direct evidence of past biodiversity. Yet, using fossils to understand the time of origination of major lineages and how their diversity changed over time is challenged by the inevitable incompleteness of the paleontological record, plagued by taxonomic, temporal, and spatial biases. Here we present a suite of supervised and unsupervised models based on Bayesian inference and deep learning to infer clade age and species richness dynamics in deep time. We compile a comprehensive global database of palm micro- and macro- fossils to evaluate the age of the clade independently of assumptions made under phylogenetic molecular clock models. We also infer the diversity dynamics of the clade after accounting for the biases in the fossil record. Our findings provide new insight into palm evolution across more that 100 million years and the basis for data-driven priors to improve the use of fossil information in phylogenetic inference