| Phylogenetics -- the discipline of inferring the evolutionary history of organisms -- has become central to the study of biology and its related fields. In light of the myriad applications of phylogenetics, ranging from understanding the evolution of complex structures and functions such as powered flight to illuminating disease transmission dynamics, it becomes crucial to ensure the robustness and accuracy of phylogenetic inference. However, except in very limited experimental circumstances, we are unable to directly observe the evolutionary history of organisms of interest. As a result, congruence and consilience across independent sources of data becomes an important barometer of phylogenetic accuracy (built upon the axiom that there is only one history of life, which arises naturally from the theory of evolution).;Discordance among datasets, however, is pervasive throughout the tree of life, often aligning across disciplinary lines: specifically, incongruence is rampant between phylogenetic trees inferred from morphological and paleontological data versus molecular sequence data. Although molecular phylogeneticists often herald the supremacy of trees inferred from genetic data over those inferred from phenotypic data, there exists a substantial body of evidence that phylogenetic analyses using molecular sequence data are susceptible to issues of systematic error as a result of biological processes (e.g., saturation, codon usage bias, gene convergence, etc.) and methodological practices (e.g., incomplete taxon sampling). In addition, the unique and indispensable ability of fossils to allow us to steal a glimpse of evolutionary history -- and the profound boon this affords phylogenetic inference -- cannot be overstated.;This dissertation investigates sources of phylogenetic incongruence, focusing on: 1) the causes of systematic bias in molecular sequence data; and 2) the importance of fossils in reconstructing evolutionary history. Two test cases are used to tackle these issues: snakes (Squamata: Serpentes), and the terrestrial egg-laying vertebrates (Amniota), in particular with regard to the position of turtles (Testudines) relative to the other major amniote clades.;Chapter 2 presents the first comprehensive analytical reconstruction of the ancestor of crown snakes, with consideration of the differences in topology between trees inferred from molecular sequence versus morphological data, and the consequences of including fossils on ancestral state reconstructions of behavioral, ecological, and biogeographic characters.;Chapter 3 presents evidence -- from both empirical and simulation studies -- that casts doubt on the validity of the turtle-archosaur hypothesis (which is supported by virtually all molecular studies targeting the turtle problem to date). Investigation of the effects of different taxon sampling schemes suggests that taxon sampling in most molecular phylogenetic analyses is inadequate. I further raise the contention that the phylogenetic affinity of turtles is likely a `hard phylogenetic problem' -- that is, the basal divergences in the reptile [avian inclusive] tree were so rapid and ancient that they appear to have occurred instantaneously, from our extremely temporally-removed vantage point in the present. This carries the implication that the turtle-archosaur reconstruction may thus be artifactual. Simulation studies investigating the effects of anomalous gene trees on phylogenetic inference also suggest that the turtle-archosaur relationship is inferred preferentially over the alternative turtle-diapsid and turtlelepidosaur hypotheses, suggesting the existence of systematic bias.;The final chapter builds on the conclusion of Chapter 3 and analyzes two phylogenomic datasets targeting the turtle problem in detail. Both the robustness of phylogenetic methods and the quality/appropriateness of data are investigated, through parametric bootstrapping, statistical tests, calculating differences in site-specific rates of evolution, profiling phylogenetic informativeness, and in-depth analysis of gene trees versus species trees. I conclude that the turtle-archosaur signal may arise misleadingly as a result of branch-length attraction and/or interference from the presence of anomalous gene trees. |
