Content area
This dissertation focuses on identifying empirical systems that present challenging ‘edge cases’, to borrow from the engineering lexicon, where current methods for the reconstruction of evolutionary history from genomic data break down. The main methodological focus centers on the ability of the bifurcating phylogenetic tree, or phylogeny, to accurately and effectively explain the evolutionary history of a given group. Beyond this conceptual focus on testing the limits of phylogenomics, the dissertation centers on groups of closely related and rapidly diversifying birds from islands located throughout the Pacific Ocean. This choice of study system adds to the conceptual throughline of challenging phylogenetic histories by giving this dissertation a distinct sense of place. The first two chapters of this dissertation make use of reduced representation genomic sequencing (restriction enzyme-associated DNA sequencing, or RADseq in particular) to generate a few thousand loci, which constitutes a reasonable sample size to reliably characterize genome-wide patterns of relatedness among lineages. The third chapter meanwhile, uses a whole genome sequencing approach, resulting in a 100% complete (i.e., no missing data) DNA sequence alignment spanning more than 800 million base-pairs. In this way, the dissertation retains a sense of time, as its construction has spanned a transitional period in the genomic sequencing revolution (2018-2024) during which the promise of whole genome sequencing for non-model vertebrate organisms has become a tangible reality. While this dissertation focuses extensively on the details of these specific avian groups, their systematics, and the technical challenges of accurate phylogenetic reconstruction, its ultimate goal is to reveal novel insights into the evolutionary process—to help us understand how the biodiversity that surrounds us here on earth, came to be.
To that end, the first chapter focuses on reconstructing the evolutionary history of eight lineages within the widespread, highly diverse genus of kingfishers known as Todiramphus. I focused on this clade, which had been previously identified using mitochondrial DNA sequence data, because it contains three lineages T. sanctus, T. tristrami, and T. saurophagus, all of which diverged from a single common ancestor within the past one million years and occur in sympatry in the Solomon Islands. The evolution of sympatric distributions between closely related lineages is rare in birds, so I hypothesized that the patterns of genetic relatedness among these species and their closely related allies might reveal unique insights into the process of rapid evolution of reproductive isolating barriers and secondary sympatry. To investigate this, I leveraged the extensive KU ornithology tissue collection from the Solomon Islands region and across the Pacific to generate a genomic dataset (using RADseq) for this clade with nearly comprehensive geographic sampling across the distribution of each species. I began by describing the genome-wide patterns of relatedness among the 83 samples that I sequenced, which revealed that all samples could be confidently assigned to distinct clades corresponding to their species identity (i.e., I identified no individuals with signatures of recent hybrid ancestry). I then assigned samples to species according to these recovered clades and generated a coalescent-based summary species tree describing the overall history of the group. This genomic species tree recovered a well-supported topology that conflicted with previous well-supported reconstructions based on mitochondrial DNA. To investigate the source of this discordance, I tested for gene flow on the species tree topology using ABBA/BABA tests, which uncovered no reliable evidence for gene flow among the focal lineages. Additionally, I calculated quartet support for internal branches in our species tree, which revealed evidence for extensive incomplete lineage sorting (ILS), with the supported topology present in the genome at < 50% frequency for all branches. These results led us to conclude that the mitonuclear discordance I observe is caused by ILS, rather than gene flow, which is the commonly presumed explanation for this phenomenon. This investigation provides one of the few rigorously documented cases of ILS driving mitonuclear discordance in nature. I think that the ubiquitous, yet challenging to quantify nature of ILS has led to the phenomenon of ILS driven mitonuclear discordance being seriously underappreciated across the tree of life, and hope that this presentation of a clear empirical case can provide a roadmap for investigators in other systems across the globe.
Continuing the theme of attempting to reconstruct challenging and discordant phylogenetic scenarios, the second chapter of this dissertation dives into the evolutionary history of three geographically structured lineages within the Philippines endemic species known colloquially as the Buzzing Flowerpecker (Dicaeum hypoleucum). I again used a RADseq approach to sequence thousands of homologous loci from 60 samples distributed across a geographically comprehensive suite of localities from throughout the Philippines. This time, our phylogeographic approach revealed evidence of hybridization, specifically on Mount Busa, in western Mindanao, where five samples came out consistently admixed in our genomic analyses. Despite this evidence for incomplete reproductive isolation, I still identified three genomic clades which correspond tightly to the three a priori identified groups within the species based on differences in phenotype and mtDNA sequence. I then attempted to reconstruct the evolutionary history of these three lineages by assigning samples into these three lineages and constructing a single species tree describing their relationships. In our recovered species tree, I found that the internal branch was so short as to be indistinguishable from zero, indicating that the best explanation for the evolution of this group was the simultaneous branching of all three lineages. To further investigate this pattern, I used a demographic modeling approach where I simulated a set of three lineages evolving under four discrete evolutionary histories, and trained a neural network to classify sets of population genetic summary statistics describing the resulting lineages based on the scenario under which they were simulated. When I used this neural network to classify the summary statistics calculated from our observed dataset, I found that the model consistently classified the observed data as generated from a polytomy model with gene flow between lineages. Overall, this comprehensive investigation revealed an apparent empirical case of a hard polytomy, where three lineages diverged instantaneously, or at least so rapidly that current approaches cannot statistically distinguish the branching order. I document a clear example of the ability of rapid branching and gene flow to essentially erase phylogenetic signal in the genome, resulting in a set of relationships that are unresolvable or even misleading. I hope that this case study demonstrates the need for caution in phylogenetic reconstructions, as the ability to reconstruct the evolutionary history of a group only diminishes as time passes. Ultimately, certain complex evolutionary histories cannot be described satisfactorily by a single bifurcating species tree, and more research is needed to understand the frequency of this phenomenon across the tree of life.
Finally, in the third chapter, I pushed deeper into the realm of misleading evolutionary histories by attempting to understand the phylogeny and speciation dynamics of the leaf warblers (Genus Phylloscopus) of northern Melanesia. This time, I used whole genome resequencing to generate 8-10x coverage of the entire genome for 18 individuals spread across the region, including two historical samples collected in 1927 and 1930. I then used these samples to construct a phylogeny, with the hopes of understanding how two separate Phylloscopus lineages came to exist in sympatry on a single volcanic island known as Kolombangara, which has only an estimated 21 km2 of montane forest to support their sympatric coexistence. Our phylogenetic reconstruction revealed a sister relationship between the two lineages on Kolombangara. This sister relationship, combined with their current sympatric distributions and highly divergent ecologies suggest that the two lineages likely arose via ecological sympatric speciation. Unexpectedly, I also identified a strong statistical signature of gene flow between the canopy gleaning lineage found on Kolombangara and other canopy gleaning lineages found throughout the Solomon Islands. I thoroughly investigated whether this statistical signature of excess allele sharing was indeed due to gene flow, or instead might be explained by an incorrectly reconstructed species tree topology. Ultimately, I discovered that the Z chromosome, traditionally thought to resist introgression and be enriched for regions supporting the accurate species tree topology, was actually enriched for this “gene flow” topology between the canopy gleaning lineages across the region. Based on this evidence, I suggested that the true species tree topology for this group supports the monophyly of the canopy gleaning lineages and the extended isolation of the mossy trunk specialist lineage on Kolombangara. Based on this accurately reconstructed phylogeny, I find support for a speciation scenario where multiple lineages colonized Kolombangara island independently, having achieved partial reproductive isolation. Because this isolation was not complete, extensive gene flow between these lineages altered the patterns of relatedness throughout genome and misled phylogenetic reconstructions. Nonetheless, character displacement and the eventual evolution of reproductive isolating barriers has allowed for the stable coexistence of these lineages within the small montane forest found at the top of Kolombangara Island. This complex and unexpected evolutionary history sounds a major alarm bell for phylogenetic studies, demonstrating the ability of gene flow between non-sister lineages to positively mislead phylogenetic reconstruction. This phenomenon has important implications for our understanding of the diversification of specific groups of organisms, and also for our ability to accurately connect biogeographic dynamics to the speciation process.
Ultimately, these three chapters present a cohesive suite of in depth investigations into the challenges and promises of attempting to reconstruct evolutionary history using the genetic code of extant organisms. I identified multiple cases where the complexity of evolution in natural systems tests the limit of our theoretical and computational abilities to accurately reconstruct the history of evolutionary relatedness among organisms. In each of these cases, a consistent theme emerges: conflicting patterns of relatedness throughout the genome. These conflicting patterns are thoroughly dissected in each chapter using a myriad of population genetic and phylogenetic approaches to build a single coherent argument tying the observed patterns to the underlying evolutionary mechanisms that generated them. By deeply investigating the sources of genomic conflict rather than sweeping them under the rug or ignoring them, we can generate more rigorous and reliable reconstructions of evolutionary history and avoid being misled by edge cases to the greatest extent possible. Hopefully, the comprehensive investigations presented here can serve as a useful framework for future investigators seeking to better understand the relationship between evolutionary pattern and process in challenging systems with similarly complex histories across the tree of life.