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

Re: Morpho v molecular (was Re: Tinamous: living dinosaurs)



 You want a case where molecular results keep finding phylogenies that
 are not well agreed upon, simply look at the _Gavialis_ - _Tomistoma_
 back and forths. Molecular phylogenies routinely found a sister group
 relationship of _Tomistoma_ and _Gavialis_ in a sister group to
 Crocodylidae, while morphological results split them off with
 _Tomistoma_ being a crocodylid and _Gavialis_ being a basal
 crocodylian.

Oh yeah, I should have remembered that case. There still seems to be nothing but puzzlement on both sides.

 Hillis (1994) gives a nice rundown of all the many ways in which
 convergences happen in sequence data. It is one of those things that,
 in hindsight, appears pretty obvious. With only 4 base pairs to
 choose f
 pure random shuffling is actually quite good.

The snipper demon was at work here.

 Also, I'm curious as to what papers you and David Marjanovic have
 been reading in regards to squamate phylogeny. The (or one of the)
 most recent phylogenies to come out using a "total evidence" approach
 found monophyly among amphisbaenians (Wiens et al. 2010). This
 supports the morphological data, but goes against what the molecular
 data has said (Townsend 2002, Townsend et al. 2004).

 According to the authors:

 "...although our main focus is the impact of molecular data on fossil
 taxa, we also show that addition of morphological data can change the
 placement of living taxa relative to analyses of molecular data
 alone. In our study, the molecular data alone (15,794 characters)
 show strong support for nonmonophyly of amphisbaenians (i.e.,
 Rhineura outside amphisbaenians) in both parsimony and Bayesian
 analyses (bs = 98%; Pp = 1.00), possibly due to long-branch
 attraction. Yet, monophyly of amphisbaenians is strongly supported
 when the 363 morphological characters are added in both analyses (bs
 = 83%; Pp = 1.00)."

 The authors found here, and previously (Wiens 2005) that despite the
 small number of morphological characters relative to molecular data,
 these phenotypic data carry much greater "weight' in analyses than
 the molecules.

Impressive.

 Despite David's complaints about small character samplings, when it
 comes to "total evidence" even a little bit of morph data can
 drastically change a phylogeny.

I complain about this within morphology all the time. :-) I didn't know there were cases where the signal in molecular data was so weak that morphology was able to override it.

I haven't read Wiens et al. (2010) because I'm no longer in the same room as a growing collection of Syst. Biol. issues. But I should have online access, actually...

I'm not sure if I've read Wiens (2005); probably I just took the answer "yes" to the title question for granted -- several papers saying "don't be afraid of missing data" had already been published by then -- and didn't look.

 Another good example
 ates back to the _Gavialis_ - _Tomistoma_
 debacle. Gatesy et al. (2003) took a total evidence approach and came
 up with a novel result that agreed with BOTH the molecular and
 morphological data; making these two taxa sister groups, but sinking
 them both deeply within Crocodylidae.

(Snippage again.)

Is this the best or the worst of both worlds? Where did the fossil gharials (all the way to the Cretaceous *Thoracosuchus*) go?