Molecular clock for beginners
I believe that there is some misunderstanding in the net about the
concept of molecular clock. Hence this post.
There are three unique features of a molecular clock: the locus
effect, the lineage effect, and the locus by lineage interaction
(residual) effect. Of course someone would say that the
stochasticity is the most prominent feature of the molecular clock, but
we will wait until such protest arises.
First, the locus effect.
A mechanical clock almost always has three needles ticking off at
almost constant rates: the "second" needle ticking once every
second, the "minute" hand once every minute and the "hour" hand
once every hour. the lock is specifically designed and assembled to
A molecular is not specifically designed and assembled to keep
time. It is assembled in a bizarre way, by sampling needles from a
large pool of needles. Some times you see a molecular clock consisting of
almost entirely of "second" needles, and the clock consequently
ticks very fast (a high substitution rate). Some other times you
see a molecular clock consisting of almost entirely of "hour"
needles, and the clock consequently ticks rather slowly. This is
what has been referred to as the locus effect.
In most cases, we have molecular clocks consisting of a mixture of
fast and slow needles. If we did not know the rate differences
among the needles, we read time wrong, just as we would read time
wrong if we could not distinguish the three needles on a mechanical
The locus effect refers specifically to the intrinsic differences
between two or more molecular clocks. For example, a protein coded
by a nuclear gene consisting entirely of methionine (coded by AUG
only) is expected to have lower synonymous substitution rate than
a protein consisting entirely of leucine (coded by 6 synonymous
codons) - the former simply cannot have synonymous substitutions.
Similarly, a protein consisting entirely of cysteine is expected to
have lower nonsynonymous substitution rate than a protein
consisting entirely of proline because there are more alternative
amino acids that are similar in physical and chemical properties to
proline than to cysteine. (By alternative amino acids I mean those
amino acids that are just one nucleotide substitution away from the
original amino acid)
The detection of a locus effect does not hurt the concept of
molecular clock. We will have more of the locus effect latter.
Second, the lineage effect.
The lineage effect refers to the phenomenon that the same clock,
e.g., the same (the "sameness" is hard to define) protein-coding
gene, ticks at different evolutionary lineage.
The detection of a lineage effect is quite tricky. If you found a
gene having a higher substitution rate in Lineage A than the
orthologous gene in Lineage B during one particular geological
period, it does not mean that the lineage effect has popped up its
ugly head. This is because of the stochastic nature of the
molecular clock. If you follow a stochastic process over time and
record, say, the occurrence of successful events, then you'll see
that sometimes a number of successful events occur in a short
stretch of time, and sometimes nothing happens over a long stretch
of time. So, for any particular period of limited length, the
number of substitutions in the two lineages are not expected to be
the same. So what you need to do is to assume that nucleotide
substitution occurs according to such and such stochastic
processes, and quantify the process by estimating its parameters.
If the parameters related to substitution rate differ much between
two lineages, then you claim to have documented the lineage effect
under the assumed substitution model.
Lineage effect can be caused by many factors, such as generation
time, mutation rate, population size, or any factor that changes
the mutation-selection balance.
Does the detection of a lineage effect hurt the concept of
molecular clock? It could, if the effect is rather dramatic, and if
it is found in almost all genes.
That genes may evolve at different rates in different lineages
suggests the possibility that genes may also evolve at different
rates within the same lineage at different geological periods. This
then implies the existence of the most troublesome feature of
molecular evolution, the locus by lineage interaction (residual)
Third, the locus by lineage interaction (residual) effect.
The locus by lineage interaction (residual) effect refers to the
phenomenon that Gene A evolves faster than Gene B in Lineage 1, but
slower than Gene B in Lineage 2. One of the consequences of this
residual effect is that you will frequently get different trees
from different genes for the same group of taxa.
John Gillespie has claimed that the residual effect is common. This
is a serious attack on the foundation of molecular systematics (You
see that the relationship between molecular systematics and the
study of molecular evolution is like that of the boat and the
water. The latter could support the boat, but could also topple the
boat). In addition to Gillespie, there are also many others who
attacked the foundation of molecular systematics, sometimes with
violent words, but few attacks have been based on real
For the time being, genes involved in generating the residual
effects seem to be rather few. I believe that Gillespie must have
done a rather exhaustive search, and the small number of examples
he could assemble in his 1991 book is somehow comforting. I
sincerely hope that the small number will remain small with time.
Of course, to get Gillespie's criticism out of the way, one really
needs to demonstrate that the residual effect is really rare, i.e.,
it is not visible in almost all genes in almost all lineages.
In short, there has been no devastating attack or even serious
threat to the concept of molecular clock. So let us bear with the
concept of molecular clock a bit longer.
Xuhua Xia |
Museum of Natural Science | Phone: (504) 388-2841
119 Foster Hall | Fax : (504) 388-3075
Louisiana State University | Email: xuhua at unix1.sncc.lsu.edu
Baton Rouge, LA 70803 |