What makes a bird a bird? Everyone knows, it’s not feathers any more…)

The line between birds and theropod dinosaurs
has become increasingly fuzzy now that so many non-birds have feathers and other former bird-only traits.

This is a good sign
that evolutionary theory embraces: small changes and a gradual accumulation of traits in derived taxa.

Ultimately
it may come down to a single defining trait (like mammary glands in mammals, or alternatively a squamosal/dentary jaw joint when soft tissue is missing) when you have lots of taxa near the base of a new major clade. So what is that trait? Or what are those traits as recovered by the large reptile tree?

The basal bird and its proximal outgroup
At present the last common ancestor of all extant birds, scansoriopterygids and enantiornithes in the large reptile tree. is the Thermopolis specimen of Archaeopteryx (Fig. 1). The original authors (Mayr et al. 2007; Rauhut 2013) did not employ a phylogenetic analysis, so perhaps did not realize what they had.

For now
the pre-bird theropod, Eosinopteryx (Fig.1) nests just basal to the basal bird theropod, Archaeopteryx. You might find it interesting to see which traits differentiate the latter from the former in the large reptile tree. This list, short as it is, is by no means complete. It simply reflects the general characters used for all reptiles in the large reptile tree.

Archaeopteryx (Thermopolis) novelties vs. Eosinopteryx

1. Frontal/parietal suture straight and > than frontal/nasal suture
2. Metacarpals 2-3 subequal
3. Pubis and ischium oriented posteriorly (convergent with some deinonychosaurs)
4. Pedal 4 subequal to metatarsal 4  (convergent with some deinonychosaurs)
5. Pedal 2.1 not > p2.2
6. Metatarsal 5 shorter than pedal digit 5 (all vestigial, of course)

Unfortunately
none of these traits are unique to the bird clade.

I thought, perhaps
that an elongate and locked down coracoid (the key to the origin of flapping) would prove to be present in all basal birds. Such a coracoid is indeed present in other specimens of Solnhofen birds, but not in the Thermopolis specimen (Fig. 2), the basalmost example. 

So what we are seeing
in these six Solnhofen birds are discrete steps in the evolution of the flapping behavior, necessary for creating thrust and ultimately flight, as in many living birds. Just as in Late Jurassic pterosaurs, the island/lagoon environment of Solnhofen was as powerful an agent as the Galapagos islands at splitting basal birds into various clades.

From the Mayr et al. abstract on the Thermopolis specimen:
“We describe the tenth skeletal specimen of the Upper Jurassic Archaeopterygidae. The almost complete and well-preserved skeleton is assigned to  Archaeopteryx siemensii
 Dames, 1897 and provides significant new information on the osteology of the Archaeopterygidae. As is evident from the new specimen, the palatine of Archaeopteryx
 was tetra-radiate as in non-avian theropods, and not triradiate as in other avians. Also with respect to the position of the ectopterygoid, the data obtained from the new specimen lead to a revision of a previous reconstruction of the palate of Archaeopteryx. The morphology of the coracoid and that of the proximal tarsals is, for the first time, clearly visible in the new specimen. The new specimen demonstrates the presence of a hyperextendible second toe in Archaeopteryx*.  This feature is otherwise known only from the basal avian Rahonavis and deinonychosaurs (Dromaeosauridae and Troodontidae), and its presence in Archaeopteryx provides additional evidence for a close relationship between deinonychosaurs and avians**. The new specimen also shows that the first toe of Archaeopteryx was not fully reversed but spread medially, supporting previous  assumptions that Archaeopteryx was only facultatively arboreal*. Finally,we comment on the taxonomic composition of the Archaeopterygidae and conclude that Archaeopteryx bavarica Wellnhofer, 1993 is likely to be a junior synonym of  A. siemensii****, and Wellnhoferia grandis Elzanowski, 2001 a junior synonym of  A. lithographica***** von Meyer, 1861.”

* Actually not as prominent as in deinonychosaurs. Such a toe works just as well at climbing tree trunks as climbing dinosaur flanks.

**This may be a convergence as the two clades are separated by taxa without a hyper extensible pedal 2.

*** Perhaps facultatively able to perch, but arboreality would have been a precursor behavior.

**** These two are sisters in the large reptile tree.

***** These two are not sisters.

Other traits in the Theromopolis specimen 
visible in Figure 1 not present in the large reptile tree include the following:

1. Smaller antorbital fenestra
2. Longer attenuate tail
3. Slightly narrower coracoids
4. Slightly larger forelimb
5. Bowed gap between ulna and radius
6. More gracile pubis, posteriorly oriented

Mayr et al. looked at pedal digit 2
and noticed it was capable of hyperextension (Fig. 3). They likened it to pedal digit 2 in deinonychosaurs (Fig. 4) which is famous for its ability to elevate the ‘killer claw’.

The large reptile tree
does not nest birds with deinonychosaurs. Rather Xiaotingia and Eosinopteryx nest between these clades. And Xiaotingia also has a similar pedal 2.1 (Fig. 5).

On a final note:
Mayr et al. (2007) report four premaxillary teeth in the Thermopolis specimen. I think they might have missed counting the anteriormost premaxillary tooth (Fig. 6) bringing the total to five.

References
Rauhut OWM 2013. New observations on the skull of Archaeopteryx. Paläontologische Zeitschrift 88(2)211-221.
Mayr G, Pohl, B, Hartmann S and Peters DS 2007. The tenth skeletal specimen of Archaeopteryx. Zoological Journal of the Linnean Society 149:97-116.

Source: https://pterosaurheresies.wordpress.com/2015/11/12/what-makes-a-bird-a-bird-everyone-knows-its-not-feathers-any-more/