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

AW: Fwd: Greg Paul is right (again); or "Archie's not a birdy"



> But I agree with you that there is not good evidence that
> ribbon like feathers are primitive.
> 
> My point is precisely as you have suggested: Xu's new
> topology leaves it unclear whether the last common ancestor
> between Archaeopteryx and birds was volant or engaged in any
> aerodynamic behaviors at all. When you suggested that
> Epidexipteryx looks secondarily flightless I wanted to point
> out that it could look primitively flightless as well, given
> a different hypothesis.

Given that fully self-powered flight outside the ground-effect envelope was 
(for all we know nowadays) asomething that evolved in the late Jurassic only, 
this is quite trivial. If you are a "semi-volant" theropod (i.e. basically a 
gliding flight mode from some elevated location and possibly supported by 
ground effect, with feathered tail and the occasional wingbeat providing 
additional but not crucial lift), your ecological niche ist somewhat unstable: 
you'll be outcompeted in the long run.

With all those maniraptoran exaptations to "birdness", it's not surprising that 
we have as many dinos that are missing a few aspects to becoming fully 
self-powered fliers as we do. And of course, their descendants could easily 
evolve in either direction - towards self-powered flight, or away from it 
again. 

In the late J/early K, eastern Laurasia seems to have been an "incubator" for 
theropods that had evolved to a level where in many small-bodied lineages only 
a few apomorphies needed for self-powered flight were missing. Some lineages 
acquired these apomorphies, some didn't and were pressured back into 
(secondary) flightlessness. It simply confirms the old notion that transitional 
stages are rarely evolutionarily stable.


Someone might want to reanalyze all the pertinent taxa (the long-tailed 
"dinobirds"/"saururans" and the more advanced lineages) and choose not true 
outgroups (use a nice basal tyrannosauroid or compsognathid something similar 
to root the tree). Rather, choose "attractors" that represent the robustly 
known avetheropod lineages as follows:

3 taxa per major lineage, of which:
1 is basal (as determined by comprehensive analysis of each lineage), 
preferrably Jurasic and eastern Laurasian
2 are advanced, preferrably Cretaceous and eastern Laurasian

This way, you can spot misrooting: if within each lineage, the basal taxon 
clades with an advanced taxon to the exclusion of the other advanced taxon, 
this is due to incorrect ancestral state reconstruction; you'll have to change 
your taxon choice but if that does not work, try pruning those branches to see 
what effect this has.

To that mix of "attractors", add ALL the "saururans" as well as representatives 
of short-tailed avian lineages. The former must be analyzed without 
preconceptions on their relationships, while of the latter, Mesozoic taxa 
should be chosen, with the emphasis on basal forms as well as representatives 
of major-lineage diversity. Biogeography is less important here, because at 
least part of the Euenantiornithes as well as much of the ornithurans was 
technically able to overcome essentially all geographic boundaries. Do not use 
highly (aut)apomorphic taxa though; _Ichthyornis_ is probably OK if you include 
_Gansus_, but _Patagopteryx_ is not. Whether it is at all necessary to go 
beyond _Gansus_ is unclear; if _Gansus_, _Apsaravis_ etc resolve correctly, 
there is no need to include anyhting beyond that.


The key question here is: 
given that we know that in theory *almost any* maniraptoran could evolve into 
something bird-like, and 
given that we know a whole lot of taxa in the first stages of this process, and 
given that we know several more advanced lineages - 
is it already possible to tie the latter to nonvolant ancestors by means of the 
intermediate stages ("saururans"), or do we need still more material?

The case of Archie and _Xiaotingia_ is illustrative in another aspect: humans 
are natural-born verificationists. We like to seek "proof", "confirmation", 
"truth". Falsificationism is something that must be learned and trained and 
remembered. Many cladistic analyses (in and outside paleontology) seem to 
follow a verificationist pattern, trying to verify the authors' particular 
preconceptions by choice of taxa. Since it is too cumbersome to run a complete 
analysis of *all* material, we need to pick and choose taxa. But in the end, 
the best proof is that which is arrived by refutation of all alternative 
hypotheses. So the best cladistic analysis is one which tries as hard as it can 
to demolish the phylogenetic hypothesis it tests, and fails miserably in that.

In the present case, the physical constraints of flight capability introduce 
bias towards "flight monophyly" (which certainly is erroneous). It is not that 
hard to "prove" pygostylian monophyly, or even pygostylian + confuciusornithid 
+ sapeornithid monophyly. But in the absence of *known* long-tailed ancestors 
of each of these lineages, these assumptions of monophyly are in need of 
hardcore testing. 
Maybe true self-powered flight evolved only once. But the fact that each of the 
4 short-tailed "bird" lineages mentioned (and such studies as the Mayr et al 
analysis discussed before) has many (powered-)flight-related apomorphies 
realized in a  *different* way hints that their LCA may have been nonvolant or 
only barely volant (WAIRer/flutter-hopper or ground-effect glider).

In short, I am suspicious of the grass one can all too easily get for 
"saururans" and the monophyly of "ornithurans s.l.". 

In studies of crown diversity with good taxon sampling (~75% or better) a 
"savanna" (few trees surrounded by a lot of grass) is not unusual. E.g. 
Red-shouldered Hawk is part of such "grassland" (centered around Central 
America) basal to the "tree" - the bulk of true buteos which expanded to the 
Afrotropics and across the Holarctic and southernmost America via a 
multipronged radiation (2 panamerican, 1 C/N American that then independently 
yielded Rough-legged Hawk and the other Old World taxa).
But in dinosaur paleontology the fossil record should be too patchy to contain 
such radiations (which occur during a few Ma at most) in complete. It may be 
different for mammals or ostracods... in any case, to break up dense radiations 
more taxa (or precisely: more robust phylogenetic diversity) are from what I 
have seen superior to more characters at the point where science is at (>100 
chars). 
It is always good to remember that you can cause almost anything to clade with 
something *perforce*. But you shouldn't force, the phylogeny should more or 
less fall into place (resolve with high support values) if you use a 
more-than-modest number of characters.

If three or more taxa are too similar to each other for the analysis to discern 
any structure, at a depth-of-analysis where *most* taxa clade nicely, it is 
better to throw in "attractors" than to try and read from the coffee grounds. 
But one or two OTUs per "attractor" lineage are insufficient - you need three 
to recognize erroneous ancestral-state assumptions (= misrooting). In a pinch, 
you can create a hypothetical ancestor (see 
http://www.molekularesystematik.uni-oldenburg.de/download/Publications/Monophyly.pdf)
 to set the root for an "attractor" subtree, but that borders on cheating.

If you want a snappy title for that, call it "'Bird Is The Word': avian 
ancestors analyzed". 


Regards,

Eike