Molecular clock? [switch to Fixed Pitch Font to see the beautiful trees]

Ludvig Mortberg lugmog96 at student.umu.se
Sat Dec 6 15:01:29 EST 1997


[switch to Fixed Pitch Font to see the beautiful trees]


Phylogenetic research today is based to a large extent on molecular
data interpreted by computer programs that automatically generates
trees. The assumptions behind these computer programs are seldomly
discussed, if ever. Let's do just that; discuss the clock.

Is there a clock?

No, it can be easily proven. The proof goes like this: if there is a
molecular clock the relative rate test should always give a positive
result, it doesn't so the clock doesn't exist.

QED

Now, what's the relative rate test? (from now on RRT) It's simply
checking for constant evolution in a phylogeny.

Consider the following tree:

A    B   C   D
 \  /   /   /
  \/   /   /
   \  /   /
    \/   /
     \  /
      \/

Every distance A-D, C-D, B-D should be the same otherwise the clock
has been going in different speeds. Let's enlarge the tree and see
what happens:

A    B   C   D  E    F
 \  /   /   /    \  /
  \/   /   /      \/
   \  /   /       /
    \/   /       /
     \  /       /
      \/       /
       \      /
        \    /
         \  /
          \/

Distance A-E, B-F, D-E, C-F and so on, should all be the same!
Translate it into real species for example a human, a bear, a badger
and a colibri:

be  ba  hu  co
 \  /   /   /
  \/   /   /
   \  /   /
    \/   /
     \  /
      \/

Badger to colibri should be the same as human to colibri! Can you
really use the clock at very large distances? Yes, you can, infact the
match should get better over large distances, small deviations from
clock like behavior should cancel out.

So what does the data say? Sometimes the RRT gives a positive result,
but just as often it doesn't. For the clock to exist it should always
give positive result. The clock doesn't exist.

QED

And this goes for all supposed clocks. Immunological distances,
DNA-DNA melting temperature distances should all be possible to
enlarge, and always should the RRT give positive results, even if you
measure very "large" distances. And so also for distance matrix
measurments on sequenced DNA etc etc.




More information about the Mol-evol mailing list