Evolution__3rd_Edition

..
Percentage difference in DNA of
two species
100
0
c.75%
More than one substitution may
underlie one base difference
between two species
I II III
Multiple hits can be corrected for
Time since common ancestor
CHAPTER 15 / The Reconstruction of Phylogeny 443
Figure 15.13
As two species evolve apart over time, their DNA becomes
increasingly different. Initially, each evolutionary change
increases the difference between the two species and the line goes
up. After a while, a second change may occur at a site where a
change has already taken place; the second change then does not
increase the difference between the two species. The line starts to
level off. Eventually the two species are “saturated” with change
and evolution has no average effect on the difference between
them. The line is now flat. The line may flatten off at a 75%
difference because there are four bases, but the exact figure may
not be 75% for various reasons. See Figure 15.16 for examples.
The regions I, II, and III correspond to regions where the
molecular inference of phylogeny is (I) relatively easy, (II) possible
but requires correction for multiple hits, and (III) impossible.
distance between the species increases over time. The molecular distance increases
because each successive change is likely to be at a different site in the 100-nucleotide
stretch. After a while, a second change may occur at a site where a change has already
occurred. Maybe the species with C evolves to have G. This evolutionary change will
not increase the molecular distance between the species. When the first change happened,
and one species had T and the other species had C, that produced a 1% difference.
If the T or C now changes, the difference is still 1%. Thus, beyond a certain level,
the molecular distance between the two species flattens off even though they continue
to evolve apart. The later changes do not add the distance a there are multiple hits at
the same site.
The molecular distance between species is likely to level off at something like 75%
(Figure 15.13) because DNA has four bases. Suppose the nucleotide is C at a site in
one species. We then look at the equivalent site in a very different species a a species
that is evolutionarily so distant that the site has changed many times and is effectively
randomized. If we look at two random sequences of DNA, the chance of identity
between the sequences at one site is approximately 25%. If the nucleotide at the site is
C in one sequence, it could be C, G, T, or A in the other, chosen at random. Thus distance
levels off at about 75% (and identity at 25%) for very different species. (These
figures assume the base frequencies are equal. If C or G are more frequent than A or T in
the species, molecular distance will level off at a figure below 75%.)
Molecular distances can be corrected for multiple hits. We use a model of sequence
evolution (Box 15.1). The simplest model assumes that the chance that any nucleotide
will change (p) is the same. We can estimate the value of p from the sequence data for
the species. We then use an appropriate statistical model (such as the Poisson distribution)
to calculate how many changes underlie the observed sequence data. The calculation
might, for instance, show that in a 100-nucleotide stretch of DNA in two species,
30 sites have not changed, 30 have changed once, 20 have changed twice, 10 have
changed three times, six have changed four times, and four have changed five times.
We then add up the total number of changes: (30 × 1) + (20 × 2) + (10 × 3) + (6 × 4) +
(4 × 5) = 144. This is the corrected number of evolutionary events. Compare it with
the 70 sites that differ between the two sequences: we have corrected a raw number of