Inferring Phylogeny

Chapter Summary

1. The task of creating a phylogenetic tree is a problem in statistical inference. That is, we wish to make inferences about the historical evolutionary relationships among populations based on some data set.
2. At the most basic level, to build a phylogenetic tree, we collect information about the characters (also called traits) of some species, and we look at which species have which characters in common. We begin by assuming that species with many characters in common are more likely to be closely related to one another than are species with fewer characters in common. This logic assumes that common characters are homologies—characters that are due to shared common ancestry.
3. Evolutionary biologists have developed a number of different phylogenetic methods to test whether characters that are shared across species are analogous rather than homologous.
4. Parsimony methods search for trees that have the minimum number of evolutionary changes. The best phylogeny is assumed to be the one that both explains the observed character data and posits the fewest evolutionary changes.
5. Phylogenetic distance methods are a second approach to inferring trees. The idea behind distance methods is that, if we can measure the pairwise "distances" between species, then we can use these distances to reconstruct a tree. First, researchers have to measure these distances, and then they have to use statistical methods to find the best tree given these distance data. The goal is to find a tree with branches arrayed so that the distance along the branches between any two species is as close as possible to the distance that we measured between those two species.
6. Maximum likelihood methods and Bayesian inference methods use explicit models of how characters change through the evolutionary process. By applying techniques of statistical inference, they attempt to find the phylogenetic tree that best explains the data.
7. For any comparison involving more than a few species, there are too many possible phylogenetic trees to search exhaustively, even with the fastest computers, and so researchers have devised clever ways to search within the "space" of possible trees.
8. Evolutionary biologists have developed numerous statistical measures of support to test between different phylogenetic hypotheses. Once they have used character data to infer a tree, they can test how certain they are that a tree—or some component of a tree—is correct. Bootstrap resampling is one technique for doing this; the odds ratio test is a second technique used to address such questions.
9. A common method for assigning absolute time to our molecular genetic phylogeny is to "anchor" our molecular genetic data to data obtained from the fossil record.
10. Phylogeography links the phylogenetic history to the geographic distribution of organisms in an effort to reconstruct migrations and patterns of speciation over time and space.
11. When using the comparative method for studying how natural selection operates, we must account for any shared evolutionary history among the species we are studying. The method of independent contrasts allows evolutionary biologists to do this.