Perez-Moreno, Sanz, Buscalioni, Moratalla, Ortega and Rasskin-Gutman, 1994. A unique multitoothed ornithomimosaur dinosaur from the Lower Cretaceous of Spain. Nature. 370, 363-367.

This was a small analysis in the description of the basal ornithomimosaur Pelecanimimus which focused on supporting the ornithomimosaur-troodontid clade.

Phylogeny

|--Allosaurus
`--+--Albertosaurus
   `--+--Deinonychus
      `--+--Troodontidae
         `--Ornithomimosauria
            |--Pelecanimimus
            `--+--Garudimimus
               `--Gallimimus

Taxon Issues

Albertosaurus- This is based on Lambe's (1917) description of Gorgosaurus libratus.
Troodontidae- The authors cited Currie (1987) and Russell and Dong (1994) for this clade, which indicates it includes at least IGM 100/44, Sinornithoides, Borogovia, Saurornithoides, Zanabazar and Troodon.

Coding Issues

2. Garudimimus and Gallimimus should both be coded as unknown for "number of teeth more than 100", since they lack teeth, as coded by character 1. The character itself is a composite, based on the tooth counts of three pairs of elements.

3. Most theropods, including Allosaurus (Gilmore, 1920), Gorgosaurus (Currie, 2003) and Deinonychus (Paul, unpublished) have dentary teeth smaller than maxillary teeth.

5. Garudimimus (Kobayashi and Barsbold, 2005) and Gallimimus (Hurum, 2001) lack interdental plates.

6. The character "skull narrow and shallow, with elongated facial part" is a composite of skull height and snout length.

7. Troodontids should be coded as polymorphic, since Zanabazar (Barsbold, 1974) and probably Saurornithoides (Norell et al., 2009) have orbits shorter than their antorbital fenestrae.

8. Deinonychus has broad contact between its premaxilla and nasal due to an elongate subnarial process (Paul, unpublished).

10. As coded, this is an autapomorphy of Gallimimus, making it useless in the analysis. However, Garudimimus (Kobayashi and Barsbold, 2005) also has this character.

12. Garudimimus has a neck over twice skull length based on its preserved cervical vertebrae (Kobayashi and Barsbold, 2005).

13. The presence of a "conspicuous and well developed" coracoid tubercle is difficult to code due to its subjectivity. However, Gallimimus' tubercle (Osmolska et al., 1972; Kobayashi, 2004) does not seem larger than Allosaurus' (Madsen, 1976). Faced with changing Gallimimus to plesiomorphic or Allosaurus to apomorphic, the latter seems a better option since Pelecanimimus' tubercle size has not been reported precisely. In any case, troodontids (Sinornithoides- Paul, 2002) have large tubercles similar in size to Deinonychus.

14. Deinonychus (Ostrom, 1974) has a longer posterior coracoid process than Gallimimus, so should be coded as apomorphic.

15. Troodontids (Sinornithoides- photo in Russell and Dong, 1994; contra the reconstruction) have straight humeri in lateral view.

16. Gorgosaurus has a manus 63% of its humerus+radius length (Lambe, 1917). Sinornithoides' is 75% (Russell and Dong, 1994), so Troodontidae should also be recoded.

17. Gallimimus does have closely joined radius and ulna distally (coding in Kobayashi, 2004).

19. Allosaurus has metacarpal I 59-64% of metacarpal IIs length (Gilmore, 1920), so should be recoded as having metacarpal I over half the length of metacarpal II. Gorgosaurus' varies between 49-67% (Lambe, 1917; Matthew and Brown, 1923) while Deinonychus' ratio is 46-52% (Ostrom, 1969), so both should be polymorphic.

21. Gorgosaurus' manual phalanx I-1 is 100-132% the length of metacarpal II (Lambe, 1917; Matthew and Brown, 1923) and Allosaurus' is 109-113% (Gilmore, 1920), so both should be coded as derived.

22. Gorgosaurus has manual flexor tubercles placed as far distally as ornithomimosaurs (Lambe, 1917). The character is a composite of both flexor tubercle depth and proximodistal placement.

General analysis conclusions- The analysis is hindered by its small size and the fact it only uses characters which are codable in Pelecanimimus, therefore excluding dorsal, pelvic and hindlimb characters. Despite being so small, two of the characters were coded as autapomorphies in the matrix. 14% (3/22) of the characters are composites, though none are correlated with others. A relatively low 16% (24/154) of the characters were miscoded, which result in the following tree once corrected. 27 28

|--Allosaurus
|--Gorgosaurus
|--Deinonychus
`--Bullatosauria
   |--Troodontidae
   `--Ornithomimosauria
      |--Pelecanimimus
      `--+--Garudimimus
         `--Gallimimus

The only difference from the original tree is the lack of resolution among the non-bullatosaurs.

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

rearrangement original recoded
(Allosaurus,Deinonychus(Gorgosaurus,Troodontidae,Gallimimus)) (Holtz, 1994) 1 0
(Allosaurus,Deinonychus,Troodontidae(Gorgosaurus,Gallimimus)) (Paul, 1984) 6 2
(Allosaurus,Gorgosaurus,Gallimimus(Troodontidae,Deinonychus)) 6 2

The number of possible rearrangements is limited due to the low number of taxa. Enforcing Arctometatarsalia is basically ambiguous in the original matrix, and is ambiguous in the recoded version. Breaking up Bullatosauria either through a tyrannosaur-ornithomimosaur clade or a paravian clade is moderately rejected by the original matrix and very weakly rejected by the recoded version. This shows the recoded matrix is too small to reject any plausible topology and is not a good support of Bullatosauria.

Experiments with controversial taxa- As Bullatosauria and Ornithomimosauria are the main nodes of interest, adding basal tyrannosauroids, ornithomimosaurs and troodontids could affect the topology. Because of the emphasis on cranial and forelimb characters, Guanlong [added 12-5-2014 based on Choiniere, 2010], Yulong [added 12-5-2014], Deinocheirus [updated 12-5-2014 with data from more complete specimens in Lee et al., 2014], Harpymimus, Jinfengopteryx and Sinovenator could all be usefully coded. The resulting consensus tree is standard except for tyrannosaurids being arctometatarsalian while basal tyrannosauroids are maniraptorans. Due to the absence of tyrannosauroid characters and low number of characters, no intermediate taxa (e.g. a combined Dryptosaurus/Xiongguanlong OTU) succeed in uniting the clade.

|--Allosaurus
`--Coelurosauria
   |--Maniraptora
   |  |--Guanlong
   |  |--Yulong
   |  `--Eumaniraptora
   |     |--Deinonychus
   |     `--Troodontidae
   |        |--(other) Troodontidae
   |        `--+--Jinfengopteryx
   |           `--Sinovenator
   `--Arctometatarsalia
      |--Gorgosaurus
      `--Ornithomimosauria
         |--Deinocheirus
         `--+--Pelecanimimus
            `--+--Harpymimus
               `--+--Garudimimus
                  `--Gallimimus

The support for Bullatosauria was countered by including basal tyrannosauroids, ornithomimosaurs and troodontids.