[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Subject Index][Author Index]

Re: K-T mammals

> > fossils.  A new study in _Nature_ about a month ago found that if
> > you include the likely number of missing species (i.e., we know 
> > that only a small proportion of species actually fossilize) then 
> > the (statistical) fossil record of mammals is pushed back to agree
> > with molecular data

A few thoughts...

I'm assuming you refer to the April 18th paper by Tavare et al on
dating primate ancestry. (The diversification model is logistic
growth, and the authors parameterize it using the time at which 90% of
present-day diversity was reached, for those curious).  This is used
to derive a model for fossil finds, and this extrapolates to the last
common ancestor.  They estimate the LCA at about 81.5 mya (molecular
gives 90mya)

I thought that the methods were relatively strong (given it's supposed
to be a relatively broad estimate), but the diversification model may
be a bit simplistic (logistic growth is the most realistic simple
model, however).  Their technique is nice because it is fossil-based
and thus could, in theory, lead to a similar method for finding LCAs
for groups without molecular data. (Such as dinosaurian taxa, at least
in theory).  The model might be stronger if they tried to calibrate
the growth curve to multiple points, and compared the resulting
models. Another somewhat informal test of the model would be its use
on more groups, noting which were fossil-rich vs fossil-poor; and thus
check if the model gives estimates closer to the established LCA times
for fossil-rich groups, as would be expected.

Keep in mind for the molec dates: Different groups are likely to
diversify at different rates.  Hence, calibrating an entire tree
against one or two nodes can lead to rather ugly correlated errors
across the tree (ie. even if your molec clock is right on target for
one lineage, it may be completely wrong for everything else).

--Mike Habib