The ghosts of Gondwana and Laurasia in modern liverwort distributions

Recent advances in phylogenetics and, in particular, molecular dating, indicate that transoceanic dispersal has played an important role in shaping plant and animal distributions, obscuring any effect of tectonic history. Taxonomic sampling in biogeographic studies is, however, systematically biased towards vertebrates and higher plants and the possibility remains that a much stronger signature of ancient vicariance might be evident among other organisms, particularly among basal land plants. Here, an explicit Bayesian model-based approach was used to investigate global-scale biogeographic patterns among liverwort genera and to determine whether the patterns identified are consistent with the expectations of vicariance or dispersal scenarios. The distribution of each genus was mapped onto the phylograms describing the floristic affinities among areas in order to define the synapomorphic transitions supporting the observed groupings. The probabilities of change in a branch were calculated by implementing the Markov model of BayesTraits. The consistent ambiguity in ancestral state reconstructions returned by the unconstrained, two-rate model indicated that the overall signal in the data was weak, leading us to test the performance of competing, explicit models. The analyses resolved clades of geographic areas that are mostly consistent with the kingdoms traditionally identified for plants and animals, but with strikingly lower rates of endemism. The major split observed in the phylograms is into almost entirely Laurasian and Gondwanan clades. Other patterns recovered by the analyses, including Wallace's line and the South Atlantic Disjunction, have also traditionally been interpreted in terms of vicariance. These observations contrast with the idea that, in spore-dispersed organisms like bryophytes and pteridophytes, dispersal obscures evidence of vicariance. However, some discrepancies between the liverwort trees and expectations from a continental drift scenario were observed, such as the sister-group relationship of the Australian and New Zealand floras, which is supported by the co-occurrence of many genera, often endemic to these two areas. Together with an interpretation of the results within a phylogenetic context, our analyses suggest that patterns, which are at first sight consistent with an ancient vicariance hypothesis, may, in fact, conceal a complex mixture of relictual distributions and more recent, asymmetrical dispersal events. Our results provide a framework for testing specific evolutionary hypotheses concerning the extremely low levels of endemism in bryophytes and in particular, the significance of dispersal and cryptic diversification.