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

Re: [dinosaur] Fossil theropods, dated phylogenies: topology, divergence dates, and macroevolutionary inferences



Apologies for the late reply.

Thomas R. Holtz, Jr. <tholtz@geology.umd.edu> wrote:

> Some modern phylogeneticists can pretend that they can ID ancestors
> with varying degree of certainty, but in actual reality they can't. An 
> algorithm
> is all well and good, but they have no independent empirical evidence to
> back up their conclusions.

Why is that? What is the "independent empirical evidence" that is
available to phylogeneticists who estimate sister-group relationships
but not to those who estimate ancestor-descendant relationships?

> Furthermore, except for a theropod specialist (and indeed, to MANY of us),
> there is no difference between Archaeopteryx & Wellnhoferia! (Or at least,
> no confirmation that it lies outside the set of specimens otherwise called
> "Archaeopteryx").

That's not the point of the paper, though, and neither is it the point
of the above-quoted comment. What matters is not the validity of the
probabilities that Bapst et al. (2016) obtained for any particular
taxon, but the very fact that for the first time in the history of the
field, paleontologists are starting to acquire methodological tools
that allow them to actually _estimate_ ancestor-descendant
relationships (i.e., to put a number on the hypothesis that one fossil
was an ancestor of another) rather than just speculate about them or
_a priori_ reject them. This is a huge improvement over the common
paleontological practice in the first half of the 20th century, when
it was not unusual for a fossil to be regarded as a direct ancestor of
a modern group simply because, as Heilmann (1926) so nicely put it,
"nothing in [its] structure militate[d] against the view" that it
might have been one; but also over the cladist paradigm that came to
replace it, according to which direct ancestors were essentially
unknowable.

> This is a good paper for what it is about, namely showing the difference
> between different software packages and algorithms within them in trying to
> assess the same question.

That is difficult to reconcile with what you wrote above, since many
of the algorithms in question (cal3 in paleotree, SA-BDSS as
implemented in BEAST and MrBayes) attempt to estimate
ancestor-descendant relationships. Moreover, the authors repeatedly
point out that allowing one taxon to be a direct ancestor of one or
more other taxa is rather important for time-scaling phylogenies with
extinct tips. It might not be a problem for methods that combine a
parsimony-based cladogram with occurrence data and more or less
arbitrarily pad it out with a few million years here and there, but
for approaches that try to explicitly model sampling and evolutionary
processes, failure to account for the possibility of sampled ancestors
(i.e., what Bapst. et al. refer to as the noSA-BDSS model) amounts to
conflating fossilization with extinction.

If, however, it is impossible to detect sampled ancestors as you
suggest, then the paper is not just wrong but largely pointless --
even when evaluated on its own terms. For the record, I don't think
that's the case (far from it, in fact), but it is a logical
consequence of your earlier statement on the issue.

> But please don't go to this for details on theropod systematics!

I don't think anyone actually suggested doing that.


*Ref:*

Heilmann G 1926 _On the origin of birds._ London: Witherby. 209 p


-- 
David Černý