Polyploidy, base composition bias, and incomplete lineage sorting in fish phylogenetics

Abstract

Understanding the evolutionary relationships between organisms is of fundamental importance in biology. Originally based on overall similarity in morphological traits, depiction of evolutionary relationships is now often pursued by constructing trees based on molecular data- molecular phylogenetics. Molecular phylogenetic inference uses variation in molecular data in a variety of frameworks to produce hypothetical relationships between organisms. As with many practices making use of biological data, the inherent noise and complexity challenges phylogeneticists. In this dissertation, I examine three empirical datasets while addressing three possible issues in phylogenetic inference: polyploidy, base composition bias and incomplete lineage sorting. Polyploidy leads to incorrect genes (paralogs) being analyzed, since it is often impossible to distinguish between gene copies generated as a result of polyploidization. My analysis indicates that incorrect assumptions of orthology have led to incorrect conclusions being drawn from phylogenetic studies including the polyploid salmons (Salmoniformes). Results indicate that pikes (Esociformes) and the polyploid salmons are not only sister taxa, but that the graylings (Thymallinae) and whitefishes (Coregoninae) are most closely related to each other. Base composition bias misleads inference through the overall similarity between sequences being a result of changes in base composition, not shared evolutionary history. Incomplete lineage sorting refers to the fact that the reconstructed relationships of different genes do not agree. Genetic variants may persist through speciation events and are not completely "sorted" between lineages, and require a methodology to reconcile the different genealogies. In two chapters I focused on base composition bias and incomplete lineage sorting in a detailed study of flatfish (Pleuronectiformes) origins. A major issue in fish phylogenetics is the question of whether flatfish are monophyletic with poor support from both morphological and molecular data. Often it appears that cranial asymmetry is the only characteristic uniting the group. I found very little evidence for a single evolutionary origin of the extant flatfishes. Base composition bias appears not to be a major contributor to flatfish non-monophyly; however incomplete lineage sorting likely results in the inability to generate robust statistical support for inferred relationships of flatfishes and relatives. Results of my work indicate that more care should be exercised in phylogenetics in determining orthology of genes. I also find that not acknowledging the presence of paralogs does indeed mislead analyses. With increased data availability and computational capabilities, non-neutral models of nucleotide evolution should be developed and included in further studies. Presenting the heterogeneity of datasets and actively accounting for incomplete lineage sorting will definitively improve the field of phylogenetics as well.