Biology Reference
In-Depth Information
data sets for which the measured mutation rate can be considered more
or less stable and uniform. For such molecular data, one can reasonably
make a “molecular clock” hypothesis and estimate durations from muta-
tional distances. Even if, in contrast to the radioactive-decay methods
used to date fossils, no simple formula exists to convert evolutionary
distances into an exact timescale, this approach has been very productive
for the evolutionary analysis of gene families (paralogs) and even of
taxons (orthologs).
As explained above, rates of evolution can be heterogeneous. To make
matters more complicated, this applies not only to lineages, but also
to characters within a single lineage. Some parts of a protein are more
conserved than others, i.e. an amino acid change within a conserved part
is often lethal to the organism and is rarely transmitted to future genera-
tions. This is another source of caution for phylogeneticists, yet they can
use this to their advantage: they use quickly evolving genes or parts of
genes to assess phyletic relationships between organisms having diverged
a short time ago, and slowly evolving genes or parts of genes for those
having diverged a long time ago.
The rationale for this differentiation is that the high frequency of micro-
mutations, combined with the small number (four) of possible nucleotide
states, can rapidly lead to signal saturation at large evolutionary distances.
After a certain number of mutations, it becomes impossible to estimate how
many mutations took place and therefore how large the phylogenetic dis-
tance between two taxa or two genes really is. To simplify, it is impossible to
estimate how many mutations occurred since two genes diverged if their dis-
similarity gets too close to the statistical upper bound of (4-1)/4
0.75.
l
Conversely, dissimilarities that are too small are not appropriate for a quan-
titative approach: if all pairwise dissimilarities have values ranging from, say,
0 to 2, for nucleotide sequence lengths in the hundreds, any phylogenetic
signal would be buried under random noise.
In summary, while there is a molecular clock that runs fairly precisely,
the observation of the clock is hampered by biological constraints such
=
l
Actually, the “twilight zone” for nucleotide sequences, i.e. the upper threshold
beyond which dissimilarity values start to be too large for multiple alignments to
make phylogenetic sense, is often estimated to be as low as 0.5.
