Molecular Clock Calibration

rhoneycu at NSF.GOV rhoneycu at NSF.GOV
Fri Feb 12 16:08:00 EST 1993


          mwooten at ducvax.auburn.edu had several questions concerning
          using fossils to calibrate the molecular clock.  This is a
          response to those questions and comments.

               I am not for sure what Mike Wooten is arguing.  The
          divergence time estimate for Mus and Rattus is based on the
          fossil record, and the age of fossils has nothing to do with
          the phenotype and the environment.  In fact, I don't even
          know what this has to do with calibration of the molecular
          clock.  The best estimate for divergence between Mus and
          Rattus is based on the earliest fossils and is between
          approximately 11 and 12 million years.  The controversy as
          to this divergence comes from arguments presented by Vincent
          Sarich based on an albumin clock.  Here, Sarich estimated a
          35 million year divergence.  This estimate has nothing to do
          with the fossil record.  To the contrary, it is based on an
          assumption that the albumin clock for rodents should be
          similar to other mammals.  Given this assumption, the amount
          of divergence seen in terms of AID units argues for a much
          older divergence time.  The problem with many molecular
          clocks for rodents is that generation time effects are known
          to exist, especially for synonymous substitutions.  Thus, it
          is difficult to calibrate a clock based on divergence rates
          of genes sequenced from other mammalian lineages.  In fact,
          the whole idea of calibrating clocks using some time of
          assumption that the same gene has a constant rate of change
          among all mammalian lineages is wrong minded, and in fact,
          there  is empirical data to suggest otherwise.

               Two observations are essential to deriving a molecular
          clock.  First, one must demonstrate linearity with time and
          evidence that rate heterogeneity is not apparent in the
          group under study.  Second, if an absolute rate is desired,
          one must have a calibration point, based either on the
          fossil record or biogeography.  It would be nice to have
          multiple calibration points, so one could get an idea about
          error (a point suggested by David Hillis).

               Molecules provide no insight into divergence time
          without a calibration point, and I feel that in too many
          cases the use of a molecular clock is abused.  If used
          correctly, calibration points based on fossils can be very
          informative, especially if one is interested in divergence
          times in other parts of the rodent phylogeny.  Circularity
          arises when one does not use the method correctly and
          assumes to much.

               I disagree with Wooten's statement that morphology and
          fossils have nothing to offer.  To the contrary, without
          this information it would be difficult to examine processes
          of molecular evolution, at least so far as rates are
          concerned.  Also, Wooten might also remember that
          convergence cannot be ruled out at the molecular level.
          There is beginning to more and more evidence of convergence
          and even positive selection, at least in some cases.



More information about the Mol-evol mailing list