PALEOBIOGEOGRAPHIC ANALYSES OF MIDDLE TO LATE ORDOVICIAN TAXA WITH IMPLICATIONS FOR CONTINENTAL AND REGIONAL INVASION EVENTS

Parsimony methods have long been the primary means of reconstructing relationships among biogeographic areas occupied by organisms in deep time. However, recent advances in probabilistic models provide new quantitative tools for characterizing vicariance and dispersal events using maximum likelihood and Bayesian approaches. Whether probabilistic models or parsimony-based analyses provide more accurate reconstructions of biogeographic signals and events in deep time has yet to be tested.

In this study, ten published species-level phylogenetic hypotheses of trilobite and brachiopod taxa spanning the Middle through Late Ordovician were used to 1) compare the inference of biogeographic histories using Lieberman-modified Brooks Parsimony Analysis (LBPA) and the Bayesian statistical models implemented in the R package BioGeoBEARS (BGB), and 2) determine the pathways and migration routes of invasive taxa during the Late Ordovician Hiscobeccus expansion and Richmondian Invasion.

Results of BGB analyses indicate that speciation via long-distance dispersal models was best supported by the data, signifying that speciation between paleocontinents and across basins happened via “island hopping”, a form of speciation typical in modern island clades. BGB results further demonstrate that dispersal was the dominant mode of speciation across major tectonic and paleoclimatic shifts. Results of the LBPA analysis indicate that both vicariance and dispersal processes influenced speciation across major tectonic and paleoclimatic shifts, with dispersal the dominant speciation mode before the Taconic orogeny. Both methods recovered several dispersal events between basins within Laurentia, and several dispersal events between Baltica, Avalonia, and Laurentia as early as the Darriwilian Stage. For this dataset, Bayesian analyses may be more applicable than parsimony analyses because the commonly observed long-distance dispersal events cannot be detected from parsimony analyses.