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

RE: Livezey and Zusi's paper plagued with convergence?

> of course they are, stupid".)  The thing is, if they
> are it doesn't necessarily rule these characters out
> as phylogenetically 
> significant.  For example, when Mayr (2004)'s
> analysis recovered plotopterids (extinct
> wing-propelled divers) as the sister group 
> to penguins, he bit the bullet and suggested that it
> wasn't just convergence (homoplasy) pulling them
> together, but that this 
> plotopterid-penguin relationship might be real.

Mirandornithes (flamingos+grebes) are possibly an even
better example. Who would have thought of than 10
years ago? But when it *was* started to think of it as
a possibility, the Mirandornithes are about as
robustly supported as Passeriformes monophyly. A
total-evidence showcase. Even the parasitological data
fits like a glove.

That being said, I eagerly anticipate the new
Pelagornithidae results. Because it's about time to
sort out the
assemblage, and this will likely help a lot.

> Secondly, does inclusion of these oddball taxa
> (Dromornithiformes, Diatrymiformes, _Sylviornis_)
> lower the bootstrap value for 
> a monophyletic Gruiformes?  In other words, is
> Gruiformes the problem here, rather than the oddball
> taxa?

Gruiformes per se is one of the major crown avian
problems regarding monophyly or lack thereof
(Procellariiformes and Pelecaniformes too but as noted
above, that ought to be sorted out in the next year or
two. Ciconiiformes is a major mess also, not for there
being *serious* doubts about monophyly
(shoebill/hammerkop in or out is a minor issue) but
because the branching pattern is far from clear.
Presently I'd consider "Gruiformes" being
2(3?)-lineage polyphyletic a good working hypothesis.

> Thirdly, the phylogenetic affinities of many large
> flightless birds are highly uncertain.  For example,
> although _Aptornis_ is 
> usually regarded as a gruiform, another view holds
> that it is actually a member of the Galloanserae
> (Weber and Hesse, 1995; 
> Courier Forschungsinstitut Senckenberg 181:

Presently, I'd rather see it as allied with the
"Gondwanan oddballs" (mesites, kagu, sunbittern),
Metaves or not, and somewhat to very distinct from
Grui+Ralli. Impossible to tell without thorough
analysis, but worth looking into. It certainly makes
sense from a biogeographical perspective; _Aptornis_'
affiliations are probably among lineages with a
Wallacea/Sahul/Gondwanan center of diversity, which
Grui aren't and Ralli are apparently only secondarily
(in which case, were _Aptornis_ allied with them, its
distinctness is hard to explain).

> AFAIK _Sylviornis_ is always allied with galliforms
> (especially megapodes), so it's weird that it pops
> up in the Gruiformes; but 
> again, has this actually been tested?

As _Sylviornis_ is peculiarly autapomorphic, the
analysis happening to over-weight such characters
might be an explanation.

If you include _Sylviornis_, you probably also want to
include the flightless (or nearly flightless) giant
true megapodes, such as _Leipoa gallinacea_
(=Progura). If one does not, it's unlikely that a
suitable "attractor" exists anywhere in the
Galliformes. For _Sylviornis_ being as weird as it is,
it certainly won't clade with anything if not baited
into it. Using a sample of crown Galliformes only for
example, I'd be surprised if _Sylviornis_ would clade
with them at all; they're just too different. It was
allied with the megapodes based on a qualitative
analysis, but in a huge quantitative study, such a
relationship is liable to get obscured, especially if
flightless giant birds outside Galliformes are


PS: for all that's being said, I wouldn't have
quantitative analyses overrule qualitative ones in
every case. The superiority of rigorous analyses is,
after all, a statistical thing (no matter
philosophical considerations - if it doesn't pass the
empiricism test, away with it!), so it need not be
correct in some particular case.

Thus, when a quantitative analysis contradicts a
qualitatively-based consensus, I wouldn't dismiss the
latter simply because it's got "no numbers". Rather, I
think, both analyses ought to be checked whether they
actually analyze the same thing, i.e. whether the
groupings proposed by each are actually testable by
the other. 
How the flamingo-grebe clade was stumbled upon, after
being simply untested-for properly in qualitative and
quantitative analyses alike, only to receive nice
support by both when it *was* analysed and tested, is
a nice case in point. That it is possible to make
Hesperornithes, grebes and loons clade together is
still the prime avian example of how not to do it.

PPS: As regards the thread-title question, a
resounding "of course - did anyone expect otherwise?"
from me. Note that this requires "plagued with" (or
"riddled with") but not "invalid due to" to be the
coice of words: as regards the actual question
(whether it makes sense at all to do it) - judging
from Mayr's work of the last years, an equally
resounding "of course, but it's not *that* easy"

Or in shorthand:
- "Cnemial crest present/absent" is not advisable as a
- Good taxon choice will not necessarily increase
robustness (and might even decrease it), but overall
reliability (as testability of hypotheses hinges upon

PPPS: I had a little dispute last week with a
co-student who claimed PAUP* was for molecular
analyses ONLY, "the Prof. said so". Unfortunately I
forgot the backronym of P.A.U.P. she gave me.
Something with "Alignment [of DNA sequences]" I think.
We actually had a brief shouting match (which was
shocking by itself because we are generally quite
close friends) as she stood by what I found to be an
entirely absurd "WTF doesn't EVERYBODY know this?!"
kind of thing.

      Wissenswertes für Bastler und Hobby Handwerker. BE A BETTER HEIMWERKER!