The importance of phylogenetic model assessment for macroevolutionary inference

Abstract

Several fields in biology rely on inference of evolutionary timescales using phylogenetics. As more data become available, estimates of phylogeny and evolutionary timescales can be used to answer long-standing questions in biology. Some examples include the resolution of deep taxonomic relationships or the causes of geographic gradients in species richness. Importantly, answering these questions largely depends on phylogenetic and timescale estimation methods that can reliably recover the molecular evolutionary process. If the methods used in phylogenetics suffer from systematic bias, the inferences that are now routinely made in several fields of biology might be misleading. For this reason, it is critical to identify the processes that can bias phylogenetic inference, and to propose solutions that can be used in practice. In this thesis, I apply empirical tests and simulation analyses to explore the way that molecular processes at the level of DNA sequences link with the inference of phylogeny and evolutionary timescales, and with broad macroevolutionary and macroecological patterns of biodiversity. In chapter 2 I use a published estimate of the phylogeny of the birds to address one of the major questions in biogeography: the causes of the latitudinal diversity gradient. I find support for the hypothesis that dispersal across latitudes is limited and higher latitudes are likely to contain younger clades, such that they have had less time available to accumulate diversity compared to tropical clades. Chapter 2 provides an example of the use of phylogenetic estimates to test hypotheses in macroevolution. However, macroevolutionary processes themselves might have an effect on the inference of phylogenetic timescales. In chapter 3, I use a data set for the plant family Proteaceae to demonstrate a link between the rate of diversification and the rate of molecular evolution. I find in chapter 4 that this link between the rate of diversification and the rate of molecular evolution could cause systematically biased estimates of evolutionary timescale. I also find in chapter 5 that phylogenetic imbalance, a phylogenetic pattern that arises from variation in macroevolutionary processes across lineages, can also be a source of systematic bias in estimates of evolutionary timescales. Finally, in chapter 6 I propose a method to assess the absolute performance of phylogenetic methods to estimate evolutionary timescale, as opposed to assessing the relative performance among methods. In this thesis, I show that despite the prominence and progress in methods to estimate phylogeny and evolutionary timescales, there is work to be done towards accounting for the effect of possible sources of bias. Methods to assess absolute model performance might provide a fruitful way forward to improve phylogenetic and evolutionary timescale estimates. A promising approach for the near future is to use the regions in the genome that are reasonably described by the existing models for phylogenetics.