Invasion Percolation Theory as a Tool for Computational Phylogenetics: A Case Study of the Evolution of Cenozoic Mammals Involving Geographical Distributio

Computational tools have given scientists new possibilities to process many different types of data, which has allowed for more decisive results, in-depth analysis, as well as new interpretations of fundamental questions. In paleobiology, computational tools are used to analyze data describing changes in organismal lineages and evaluate the evolutionary processes that produced them, as well as to investigate patterns of fossil distributions in space and time. These patterns provide insight into evolutionary dynamics and are summarized using evolutionary trees. In this paper, a percolation theory is explored as a model for lineage network growth within evolutionary trees, as well as probabilistic modeling of temporal and spatial patterns of dispersal of species from a biogeographic center of origin. In order to evaluate the number of parameters needed to accurately reconstruct paleobiogeography, the model was tested using fossil occurrence data from two groups of Cenozoic mammals, Camelidae and Proboscidea. I tested the model by simulating static and dynamic cases with varying parameters of geographic ranges and longevity frequencies. The biogeography models were then evaluated for best fit to known information about the locations of dated fossil occurrences, obtained from the Paleobiology Database (PBDB; http://paleobiodb.org). Constructed cladograms and videos were also compared with biogeographic dispersal events and evolutionary diversification events inferred from the fossil data. The results suggest that the model supports a hypothesis of the biogeographic controls on temporal patterns of diversification and extinction within clades. However, the extent of control differed between the two groups, demonstrating that clades vary in their biogeographic and evolutionary patterns, making it difficult to describe all with a single model. Overall, the results suggest that it is possible to simulate both phylogenetic and biogeographic patterns using the percolation model developed here, and utilizing such as both predictive and analytic tools in paleobiological analysis.