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

Re: Moa-Tinamou Clade Found Within Ratites



Tim Williams <tijawi@gmail.com> wrote:

> I admit to being confused on this point.  If neognaths were present at
> the end of the Cretaceous (e.g., _Vegavis_ and _Austinornis_, both
> regarded as galloanserans), how can palaeognaths be a "young" clade?
> Palaeognathae must be at least as old as these.

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

> Perhaps what is meant is crown-group palaeognaths, rather than the total 
> group?

I'm sorry, I was being imprecise. Yes, I was referring to the age of
the paleognath crown clade, which might be significantly younger than
the age of the respective total clade and -- unlike the latter --
can't be constrained using neognath fossils. It's true that
pan-paleognaths likely originated somewhere in the middle Cretaceous,
but most divergence dating analyses put the origin of crown
paleognaths closer to 60 to 70 Mya. That's what I meant by "young":
too young for their present-day distribution to be explicable by the
breakup of Gondwana.

Tim Williams <tijawi@gmail.com> wrote:

> I think we have to be careful here.  I'm reluctant to use words like
> "correct" and "right" in the context of any phylogeny - including
> molecular-based phylogenies.  We cannot "know" the true phylogeny - we
> can only use the evidence available to try and discern the
> phylogenetic signal.

I agree. It may be more accurate to say that Johnston's hypothesis
relies on a position of elephantbirds that is contradicted by other
morphological datasets (e.g., Worthy & Scofield 2012) and
mitochondrial DNA (Mitchell et al. 2014), and on a position of moa
that is contradicted by mitochondrial, nuclear sequence, and
retroposon data.

> On that score, molecular-based analyses are no silver bullet.  They
> have their own problems, and are underpinned by assumptions that
> in the future may be shown to be wanting.  This applies both to the
> gross topology and to the divergence dates ('molecular clock') of a
> phylogeny.

That's definitely true, but it applies to morphology as well, perhaps
even to a greater extent. Morphological characters are far more
complex than nucleotides, so it's not unreasonable to assume that
their evolution may be affected by an even greater number of factors
that can sometimes distort or obscure phylogenetic signal
(heterochrony being one example of this that is often invoked in the
context of paleognath phylogeny). One advantage of sequence data is
that they lend themselves to statistical modeling more easily than
morphology, so that biases may be countered with novel models that
correct for their effects. The near-universal incorporation of the
gamma rate parameter into substitution models is a good example of
this, as is the use of Dirichlet mixture models to accommodate
across-site process heterogeneity.

In the absence of any silver bullets, perhaps the best we can do is to
look at as many sources of data as possible and determine the degree
of congruence (or discordance) both among and within them, while
filtering out all known biases. This approach strongly favors the
Notopalaeognathae hypothesis.

> Long story short: I don't think palaeognath phylogeny is set in stone.

It's not, but the best current estimate (with the ostrich as the
sister group to the rest of the clade) is supported about as well as
the monophyly of passerines, even though the latter hypothesis has
been around for a far longer time.

>  This includes the position of tinamous vis-a-vis ratites.  I have to
> say that, based on the corpus of evidence, that it looks increasingly
> like multiple losses of flight occurred across Palaeognathae.  But one
> or two key fossils could confirm or overturn this hypothesis.

I think it may be useful to separate the issue of paleognath phylogeny
from that of evolution of flightlessness. However, I don't see how the
current consensus on either problem could be overturned with one or
two fossil finds. In the former case, that would probably require
identifying a previously unknown bias capable of skewing phylogenetic
signal across the whole nuclear genome, as well as a mechanism that
would explain why retroposons overwhelmingly favor the "wrong"
topology in the absence of any obvious traces of incomplete lineage
sorting. What information could a fossil provide that would circumvent
the need for this?

> That's okay if the high degree of shared similarity is synapomorphic.
> If _Lithornis_+Tinamidae are nested inside the ratites, then it makes
> it less likely that the shared similarities are plesiomorphic for
> Palaeognathae.

Actually, if we agree that several independent losses of flight in
"ratites" are more likely than tinamous being secondarily volant, it's
possible for tinamous to be primitive despite being deeply nested
within crown paleognaths. We would expect them to be more similar to
the volant MRCA of crown paleognaths than the flightless, cursorial
"ratites", at least in postcranial characters. The fact that losing
the ability to fly is far more likely than gaining it makes the
distribution of flightlessness-related characters highly
"unparsimonious", so I wouldn't trust the ability of equal-weighted
parsimony analyses to correctly optimize their ancestral states (i.e.,
to determine which similarities are derived and which are primitive).
I'm not sure Mitchell et al.'s (2014) Bayesian analysis is any better,
since the model they use for morphological data also assumes that
changes in both directions are equiprobable. Then again, it's
difficult to relax that assumption when character states are labeled
arbitrarily.

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

> Quite so. The easiest test of this in the world is the observed fact that 
> different molecular analyses can produce discordant results, and single 
> analyses produce greatly more than 1 most parsimonious tree. Both of these 
> indicate that the data are not 100% decisive, and that for a variety of 
> reasons there are ambiguities in the record. This is true for bones, it is 
> true for soft tissues, it is true for genes, it is true for genomes. Those 
> who are trapped in the "my technique yields the TRUE answer" mentality have 
> to grow out of the 1990s and move on...

I don't think I made the claim that there is a technique guaranteed to
give the True Answer (and I couldn't claim it to be "my" technique
anyway). In my argument for the Notopalaeognathae topology, I tried to
rely on the fact that in paleognath phylogenetics, different molecular
analyses do _not_ produce discordant results as long as they take into
account certain well-known biases. I also tried to emphasize that the
hypothesis is well-supported because of the congruence between
multiple lines of evidence, _not_ because it's based on one source of
data that is somehow superior to all others.

That said, my use of "incorrect" in reference to Johnston's placement
of moa and elephantbirds should definitely have been qualified with
"most likely".

-- 
David Černý