Sanz and Bonaparte, 1992. A new order of birds (Class Aves) from the Lower Cretaceous of Spain. in Campbell (ed.). Papers in Avian Paleontology Honoring Pierce Brodkorb. Science Series 36, Natural History Museum of Los Angeles County, Los Angeles. pp 38-49.

This paper described Iberomesornis, a new bird that was seen as intermediate between Archaeopteryx and more derived taxa. It included a small phylogenetic analysis thats sole purpose seemed to be to establish that fact.

Phylogeny-

|--coelurosaurs and deinonychosaurs
`--+--Archaeopteryx
   `--+--Iberomesornis
      `--Ornithurae

Taxon Issues

Coelurosaurs and deinonychosaurs- As recognized now, Coelurosauria includes Deinonychosauria. The authors never specify which taxa they use to code this OTU from, so all coelurosaurs recognized as such before the 90's are used.
Ornithurae- Sanz and Bonaparte use Ornithurae in the sense of Cracraft (1986), so presumably include Hesperornithes, Ichthyornis and Aves. Whether they also include Enantiornithes is never explicitly stated, but they seem to since Lecho enantiornithines have their ornithurine characters when known. They are here excluded for simplicity's sake.

Character Issues

1. Ichthyornis and other ornithurines do have neural spines on their cervicals (e.g. Clarke, 2004), though they are very low in some vertebrae. Similarly, the first three preserved cervicals in Iberomesornis probably have low spines that are simply not exposed. Many coelurosaurs such as Oviraptor have low neural spines as well, so that clade is recoded as polymorphic. The spines of Archaeopteryx are also low (Wellnhofer, 1974), so it is recoded as derived.

2. This character (tall dorsal neural arches) was intended as an autapomorphy of Iberomesornis, making it useless for the analysis. It seems primarily true of the first two dorsals, as the others do not differ noticeably from Archaeopteryx in neural arch height.

3. Fused cervical ribs are known in several coelurosaurs (e.g. ornithomimids- Ostrom, 1916; Avimimus- Kurzanov, 1987; Velociraptor- Norell et al., 2006), so that OTU is recoded as polymorphic. While Sereno (2000) reported no actual cervical ribs are preserved in Iberomesornis, he did state the broken anterior area of the cervical vertebrae may indicate fused ribs were present.

4. Cervicodorsal hypapophyses are present in several coelurosaurs (e.g. Ornitholestes- Makovicky, 1995; Avimimus- Kurzanov, 1987; Deinonychus- Norell and Makovicky, 1999), so they are recoded as polymorphic. Iberomesornis also has large hypapophyses (Sereno, 2000).

5. Some coelurosaurs such as oviraptorid IGM 100/42 have as few as ten dorsal vertebrae (pers. obs.), so they are recoded as polymorphic (0+2).

6. This character (sacral number) is problematic because Iberomesornis has eight sacrals (Sereno, 2000), not five as originally supposed. Thus it is intermediate between coelurosaurs and Archaeopteryx (5-7) and ornithurines (10-23). It is here coded as apomorphic because defining the character as "8 or more sacrals" will make it informative, whereas if it is defined as "10 or more sacrals" it becomes a useless autapomorphy of Ornithurae (as originally designed).

7. This character (sacral hypapophyses) was originally a useless autapomorphy of Iberomesornis, but also based on a mistake. The anterior sacrals of the specimen are in lateral view but incorrectly thought to be in ventral view by Sanz and Bonaparte, so what they saw as hypapophyses are actually sacral ribs (Sereno, 2000). Iberomesornis is thus recoded 0.

9. Ornithurae is polymorphic for the presence of caudal zygapophyses, since Ichthyornis has them (Clarke, 2004).

10. Several coelurosaurs have ossified uncinate processes (e.g. IGM 100/42- Clark et al., 2002; Deinonychus- Paul, 2002; Velociraptor- Paul, 2002), so the OTU is recoded as polymorphic.

11. Iberomesornis is here considered too incomplete to rule out the possibility it had gastralia.

13. This is a composite character combining interclavicular angle with the presence of a hypocleidium. Among Ornithurae, Hesperornis has a wide interclavicular angle and no hypocleidium (Marsh, 1880), Ichthyornis has a narrow angle and no hypocleidium (Clarke, 2004), and Aves generally have narrow angles but vary in their hypocleidium size. Similarly, coelurosaurs always have broad furculae, but vary in their hypocleidium size. Thus both taxa are recoded as polymorphic.

14. Ornithurae is recoded as polymorphic for ulnohumeral ratio, as Baptornis has a ratio of 18% (Martin and Tate, 1976) and Ichthyornis one of 94% (Clarke, 2004) (compare to 87-96% for Archaeopteryx).

15. Coelurosaurs are coded as polymorphic for pelvic fusion, as some exhibit it (e.g. Ornithomimus- Parks, 1926; Avimimus- Kurzanov, 1987). The condition in Iberomesornis is uncertain. Sanz et al. (2002) and Chiappe (1996) state no puboischial suture is visible, but Sereno (2000) illustrates such a suture. It is here recoded as unknown.

16. While the pubis of Iberomesornis is not propubic as hypothesized by Sanz and Bonaparte, it is not subparallel to the ilium either (Sereno, 2000). It is thus retained as plesiomorphic.

17. Iberomesornis has fused distal tarsals, the free tarsals identified by Sanz and Bonaparte being part of the tarsometatarsus that was separated from the rest by a film of matrix (Sereno, 2000). Similarly, the astragalus and calcaneum are fused to each other and the tibia (Sereno, 2000), Sanz and Bonaparte's description of separate proximal tarsals being due to breakage. Iberomesornis is thus recoded. This is also true of Avimimus (Kurzanov, 1987), so coelurosaurs are recoded as polymorphic.

18. Sereno (2000) showed the proximal tarsometatarsus of Iberomesornis is actually fused anteriorly, contra Sanz and Bonaparte. The Maxburg and Solnhofen Archaeopteryx specimens show metatarsal fusion (Wellnhofer, 1992), so Archaeopteryx is recoded. Coelurosaurs are again recoded as polymorphic based on Avimimus (Kurzanov, 1987).

19. Middleton (2003) finds that Archaeopteryx lacks the twisted metatarsal I needed for a reversed hallux, and finds Iberomesornis to be too distorted for comparison.

General analysis conclusions- This analysis is mostly problematic due to its small size. With only four taxa included, it's more of a demonstration of the characters uniting Iberomesornis with more derived birds than a test of anything's relationships. This is proven by the perfect (1.00) consistancy index, which shows every single character is consistant with the tree they present (amusingly, this is even true of the recoded matrix). Most of the other issues are due to misinterpretation of Iberomesornis, or assuming coelurosaurs have the primitive state when they are polymorphic for 8 (42%) of the 19 characters. All together, 23/76 (30%) of the characters are miscoded. When corrected, the tree is unsurprisingly basically the same-

|--other coelurosaurs
|--Archaeopteryx
`--Ornithothoraces
   |--Iberomesornis
   `--Ornithuromorpha

Phylogenetic conclusions- The table shows the number of extra steps needed to accomodate each rearrangement using Sanz and Bonaparte's original matrix, and his recoded matrix. A negative number means the arrangement is already most parsimonious, but that many steps are needed to undo it.

rearrangement original recoded
(coelurosaurs,Ornithurae(Archaeopteryx,Iberomesornis)) (Kurochkin, 1995) 8 3

The only other possible arrangement of the matrix is to support Sauriurae, which was moderately rejected by the original matrix but only weakly rejected by the recoded version. The matrix is too small to usefully test any hypothesis, despite its bias.