PHylogENy: THE UNITy ANd dIvERSITy of lIfE 411
scenarios that requires 4 changes is the true history shown in tree A. The first and
second bases each changed once, but the third base changed 2 times; it indepen-
dently evolved from A to G in both species 2 and 3. Under the hypothesis that tree
B is correct, at least 5 changes must have happened, while tree C would require at
least 6. Parsimony therefore identifies tree A as the best estimate of the phylogeny.
While parsimony often gives the correct result, it does a poor job of estimating
the phylogeny in some situations, in particular when there is a lot of homoplasy on
the phylogeny and when evolutionary rates vary among branches of the tree. Those
problems motivate methods that use statistical approaches based on likelihood, which
is described in the Appendix. These methods start with a set of assumptions about
how the characters evolved. With DNA sequences, for example, we might assume
that there was a constant substitution rate (a molecular clock; see Chapter 7). Prob-
ability theory can then be used to calculate the chance that the species would have the
observed DNA sequences for a given order of branching events, set of branch lengths,
and substitution rate. Finally, we search for the phylogeny and substitution rate that
maximize the probability of the observed data. This is called the maximum likelihood
estimate for the phylogeny. More details are given in BOX 16A. Bayesian inference is
a related approach that allows other information (for example, dates from the fossil
record) to be incorporated into the estimate (see the Appendix). It also can also esti-
mate phylogenies that are difficult for likelihood because of their size.Futuyma Kirkpatrick Evolution, 4e
Sinauer Associates
Troutt Visual Services
Evolution4e_16.12.ai Date 01-02-2017Species
1Species
2Species
3Outgroup
1Outgroup
2
TTA TTG AAG AAAAAAATATTATTGAAGAAATree ASpecies
1Species
2Species
3Outgroup
1Outgroup
2
TTA TTG AAG AAAAAAAAATree BSpecies
1Species
3Species
2Outgroup
1Outgroup
2
TTA AAG TTG AAAAAAAAATree CATATTAATATTAAAGATGTTG
AAGATG
AAGTTGFIGURE 16.12 Parsimony is one of the methods used to estimate
phylogenies. The example shown here uses data from three sites in the
genomes of five species: the three species (1, 2, and 3) whose relation-
ships are to be determined, and two outgroup species. The outgroups
suggest that the ancestral DNA bases were AAA. Changes in the DNA
bases that occurred during the evolution of these species are shown by
the horizontal bars, with the bases that changed highlighted in blue. Tree
A requires at least four changes to account for the data, tree B requires at
least five, and tree C requires at least six. Parsimony identifies tree A as the
best estimate of the phylogeny because it requires the fewest evolution-
ary changes.16_EVOL4E_CH16.indd 411 3/22/17 1:33 PM