RE: Shaking up the bird family tree

[Tim Williams <twilliams_alpha@hotmail.com>]

Mike Keesey wrote:


>> Kind of like how flying bats evolved from flightless bears, right?
>
> (Not quite the best analogy, but I'm having trouble thinking of a good one.)
>
>> Right? Doesn't that just make so much sense?!
>
> Yeah, given the many independent occurrences of flightlessness among
> avialans, it seems more likely to me that, if their topology is
> correct, flight was lost three times among palaeognathes: once in
> stem-ostriches, once in stem-rheas, and once in the Apteryx-casuariid
> stem group. Maybe more, depending on where the extinct flightless
> clades go. (Who knows, though?)


I think the operative clause here is "if their topology is correct." It's worth 
remembering that molecular phylogenies are not inherently superior to 
morphological phylogenies. They are both susceptible to the same weaknesses, 
despite the datasets being superficially very different (bases/amino acids 
versus morphological traits). In any dataset there may be many signals that are 
vying for attention, and the algorithm has no magical ability to pick out the 
'correct' one (= the historical/phylogenetic signal). Simply piling on the taxa 
and/or sequences does not necessarily help either. Very few people are actually 
interested in *why* a molecular phylogeny pulls certain taxa together; they 
just assume a given clade reflects a unique shared descent, especially if the 
bootstrap/posterior probability value is high enough. For example, we know 
almost nothing about which changes in beta-fibrinogen positions support a given 
node, and how the sequence is influenced by the secondary or tertiary
 structure of the protein. By contrast, a morphological phylogeny lets you know 
*explicitly* which characters are supporting a given clade, and it can be 
evaluated in light of what we know (or what we think we know) about the 
ecomorphology of these taxa.


When it comes to Aves, the advantage of a molecular over a morphological 
phylogeny is there is no shortage of extant taxa for expanded sampling. Also, 
Neornithes have a cruddy fossil record (at least when it comes to the early 
history of most extant 'orders'), and morphological analyses tend to lead to 
many poorly-supported clades. So molecular analyses are intuitively attractive. 
But frankly, I'm highly skeptical of some of the conclusions from the 
aforementioned phylogeny (e.g., nesting tinamous inside flightless ratites; 
diphyletic Falconiformes). The only thing that will convince me is independent 
corroboration from morphological/fossil data - not another (and even bigger) 
molecular analysis. I'm not saying the entire tree is wrong. I'm only saying 
that just because some expected, time-honored clades turn up (and with good 
support), doesn't mean that the entire tree is 'correct'.


In short: molecular phylogenies are not a panacea. The fact that different 
molecular phylogenies produce conflicting topologies is not simply due to the 
size of the respective datasets, or the identity of the sequences used. It can 
be a consequence of the limits of the phylogenetic analysis. There is no 
guarantee that ever-expanding datasets will improve the odds of the algorithm 
recovering the 'correct' signal for every lineage.


I'll get down off my soapbox now.


Cheers


Tim

_________________________________________________________________
The other season of giving begins 6/24/08. Check out the i’m Talkathon.
http://www.imtalkathon.com?source=TXT_EML_WLH_SeasonOfGiving