Comparing Morphology and Molecules to Evaluate Phylogenetic Inference

Abstract

Phylogenetics, the reconstruction of evolutionary relationships between species, underpins evolutionary biology. These relationships are most often inferred from either or both morphology and molecules, but molecular data are often preferred because of the objective atomization and large number of characters, the relative neutrality of some genomic regions and the sophisticated statistical techniques that these benefits allow. Morphology, by contrast, remains relatively neglected. However, morphological data remain essential, particularly for the phylogenetic placement of fossil species for which no molecular data are available. Morphology nevertheless suffers from issues such convergence and non-independence of traits, the extent and distribution of which remain largely unknown. In this thesis I compare osteological, dental and other partitions of morphological data in tetrapods, using molecular trees as a benchmark. I assess differences in levels of homoplasy, ages of character transition, tree-based correlations and internal consistency. To do this, I compute the retention index of characters and partitions on molecular trees, perform ancestral state reconstructions to estimate character transition ages, apply correlated and uncorrelated models of character pair evolution, perform cluster analyses, and build trees using subsets of data. I find heterogeneity between these partitions, both with and without respect to molecular trees. Specifically, I find that osteological characters are more homoplasious and transition earlier than either dental or soft characters. Further, characters are more correlated within partitions than between partitions. These results highlight the importance of partitions, implying differences in convergence, evolutionary rates and integration between different morphological subsets. As well as shedding light on the nature of morphological evolution in tetrapods, these results have important implications for phylogenetic inference, and suggest the need for careful consideration of the properties of morphological data when reconstructing evolutionary history. Specifically, composite coding and partitioning may be necessary in model-based approaches.

Date of Award	1 Aug 2020
Original language	English
Awarding Institution	The University of Manchester
Supervisor	Chris Klingenberg (Supervisor), David Robertson (Supervisor) & Robert Sansom (Supervisor)