Phylogenomics of Rapid Avian Radiations

Abstract

I use data from sequence capture of ultraconserved elements to resolve three rapid radiations in the avian tree of life and in the process gain insights on applying analytical strategies with gene tree-based coalescent methods (GCM). In Chapter 1, I explore analytical strategies that can be employed with GCMs to increase phylogenetic resolution and minimize highly supported conflicting results, including subsampling taxa to increase the number of gene trees analyzed, trimming sequences to eliminate sequence length heterogeneity, and filtering loci based on information content. These strategies are used to reconstruct a highly resolved and consistent phylogenetic hypothesis for the relatively young avian family, Zosteropidae. I show how conflicting results from different GCMs can arise from biases introduced by sequence length heterogeneity and uninformative loci that can lead to strongly supported incorrect estimates of phylogeny. In Chapter 2, I examine higher-level relationships in the enigmatic core Corvoidea group of Oscine passerines. A highly resolved phylogeny of core Corvoidea is recovered, with a majority of nodes receiving high support from both ML and coalescent analyses. I show that short sequence lengths do not bias species tree estimates of GCMs if informative sites are present in these sequences. In contrast, some samples that have longer sequence lengths compared to most taxa but shorter sequence lengths compared to taxa in its clade can also bias species tree estimates of GCMs. In Chapter 3, I develop a hypothesis on the origins of the trogons (Trogonidae) based on a robust dated phylogeny estimated from thousands of genome-wide loci. I recover the first well-supported hypothesis of relationships among trogon genera. This topology, combined with the trogon fossil record, geologic, and climatic data, suggests an Old World origin in the Late Oligocene/Early Miocene for the crown group. In this iii chapter, I show that in some datasets in which loci have high information content, exclusion of less informative loci in analysis can lead to lower bootstrap support of species tree estimates of GCMs.