Parsimony methods were among the first methods for inferring phylogenies. The central idea is that the preferred evolutionary tree requires the smallest number of evolutionary changes to explain the differences observed among the taxa under study. As an example, we will look at the 4 taxa case, in which there are 3 unrooted trees (Figure 1).
Figure 1. In the 4 taxa case, there are 3 possible trees
We examine the sequences to look
for informative sites, i.e., those sites that will favor one
tree over the other.
Site 1 2 3 4 1 A G G A 2 A G G G 3 A A C A 4 A A C G Site 1 is uninformative, as there is no change in any of the sequences. At site 2, the most parsimonious explanation is that there is a change from G to A which favors the first tree in Figure 1. At site 3, once again the most parsimonious explanation is one change from G to C and again the most parsimonious tree is the first tree. For both sites 2 and 3, the other 2 trees would require 2 changes. At site 4, the 2nd tree is the most parsimonious, only requiring once change from A to G. Based on these sites, the first tree is the most parsimonious with 2 informative sites favoring it. As the number of taxa being investigated increases, so will the number of trees that must be considered. Regardless of the complexity, the main idea is still to infer the minimum number of substitutions/changes for a given tree.
Although parsimony makes no explicit assumptions, there is the critical assumption of the parsimony criterion that a tree that requires fewer substitutions/changes is better than a tree that requires more (Li 1997). This can be contrasted with likelihood methods that make explicit assumptions about the rate of evolution and patterns of nucleotide substitution.