[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)

Tim Williams wrote-

> In the case of amphisbaenians, yes the fossil squamatan
> _Cryptolacerta_ appears to back up the molecular-based placement of
> amphisbaenians as the sister-group to Lacertidae. However, correct me
> if I'm wrong, but the paper that described _Cryptolacerta_ only
> included a total evidence analysis (morphological characters + nuclear
> gene sequences) to test its affinities. AFAIK, there was no
> morphology-only analysis. I'm not disputing the
> lacertid-amphisbaenian link; I'm merely pointing out that the
> amphisbaenians are not the best example of your claim that morphology
> is playing catch-up with molecules.

The analysis was total evidence, yes, but the authors also state Cryptolacerta 
shows "a mosaic of lacertid and amphisbaenian anatomical characters."

> > You never see more genes and sequenced taxa leading to molecular results 
> > that resemble the traditional
> > morphological results after all.  The changes always seem to go one way.
> Hmmm... I've noticed this too (at least for mammals), and I've
> wondered if this is an artifact. With molecular-based phylogenies,
> there is a limit to the number of taxa that can be sampled for
> phylogenetic analysis. So once we hit this limit, the topology of a
> molecular tree is pretty much "fixed". This gives the illusion of
> stability (especially if certain nodes are well-supported,
> statistically). By contrast, new fossils are being discovered all the
> time, and one 'key' fossil can radically change the topology of a
> morphology-based tree. So (again, at least for mammals)
> morphology-based trees can and do change on a fairly regular basis.
> So if a morphology-based tree happens to be congruent to a
> molecular-based tree, is that corroboration, or coincidence?
> Take the Cetartiodactyla, for example. It contains two highly
> speciose clades, Ruminantia (~180 species) and Cetacea (~90 species),
> but the other clades have much fewer species (Hippopotamidae,
> Camelidae, Suina). To obtain a molecular dataset, we would compile
> lots of genetic sequences for all hippopotamids, camelids, suines, and
> a representative cross-section of cetacean and ruminantian species (or
> you could sample *all* the cetacean and ruminantian species, but that
> probably would not help resolve the relative position of either clade
> better than a representative cross-section). The resulting
> molecular-based tree has effectively exhausted the possibilities for
> taxon sampling. We can try adding more sequences, but that probably
> won't change the topology very much. We could try a new algorithm,
> but that calls into question the efficacy of the original analysis.
> With fossils, discoveries of stem-cetaceans, and (to a lesser extent)
> the discoveries and re-interpretations of certain fossil artiodactyl
> taxa, have guided our views on how whales evolved. Future discoveries
> will undoubtedly provide further insights. Remember that despite all
> the confidence molecular biologists had in a whale-artiodactyl clade,
> it wasn't until _Artiocetus_ turned up - the proto-whale with the
> artiodactyl-style ankle - that we had hard evidence (quite literally)
> to support this link.

It's true that the sample of disparate taxa we can include grows faster for 
morphological than molecular analyses.  But the sample size is pitiful in most 
molecular analyses.  Sure we have a lot of cetartiodactyl sequences, but Iwabe 
et al.'s amniote analysis that found archosauromorph turtles used just the 
axolotl, human, mouse, iguana, caiman, chicken and slider.  What about 
platypus, echidna, opossum, armadillo, dibamid, gekko, pleurodire, trionychoid, 
gavial, ostrich, neoavian, etc?  Use basal forms, and multiple examples.

More importantly, we have the huge untapped resource of genes.  How many genes 
are used in most molecular analyses?  One?  Three?  Fifteen if they're really 
recent.  The turtle study I noted above used two.  I could understand if 
analyzing a few genes gave us bad results due to misalignment, LBA, etc., but 
as we sequence five, ten, a hundred, a thousand genes, not to mention SINEs, 
LINEs and other non-coding regions, things really should start to resemble the 
morpho trees if the latter are right.  Yet instead as molecular analyses 
acquire more data, the trees gain a strongly supported consensus which 
sometimes differs from morphology.  I think this is a strong indication the 
molecular analyses are finding something real, since none of the artifacts 
should bias the result in the same way across so many genes.  

Mickey Mortimer