Phylogenetic Perspectives on Viviparity, Gene-Tree Discordance, and Introgression in Lizards (Squamata)

Abstract

This dissertation consists of three works of phylogenetic biology. In each case, my collaborators and I contribute to the field by providing novel empirical results germane to each topic and study system, but also in a broader sense by facilitating the use and understanding of modern phylogenetic methods. In appendix A, I investigate the evolution of viviparity in phylogenetic test of the cold-climate hypothesis. This long-standing hypothesis states that the evolution of viviparity is an adaptation that prevents egg and/or juvenile mortality in cold climates. I apply a suite of cutting-edge phylogenetic comparative methods to phylogenetic, climatic, and life history data from the lizard family Phrynosomatidae. The results strongly support the cold-climate hypothesis, and help to explain two counter-intuitive patterns in Phrynosomatidae. In appendix B, I examine discordance in phylogenetic reconstruction in order to answer an unresolved question with significant practical implications for phylogeneticists: can disagreement between concatenated and multi-species coalescent estimates be predicted? Using an empirical example in higher-level scincid lizard relationships, I find that discordance between these methods is related to short, weakly supported branches, and the presence of conflicting gene trees. These results suggest that concatenation may provide a reasonable approximation of theoretically preferable species-tree methods under most circumstances. Moreover, standard methods for incorporating phylogenetic uncertainty may be sufficient to account for disagreements by species-tree method estimates. In appendix C, I demonstrate a novel workflow for the detection of nuclear introgression in an empirical case study of closely related spiny lizard (Sceloporus) species. I first show that mitochondrial data independently suggests a repeated history of introgression between the focal taxa. I then show that an exhaustive application of the DFOIL method to double-digest RADseq data from many individuals is effective at identifying introgression in the nuclear genome. This novel approach also reveals intra-specific geographic variation in in patterns of introgression, without the need to determine population structure. Finally, I find that that batch effects in ddRADseq data may mislead D-statistics (and DFOIL) under certain circumstances.