dinosaur study 2

7/29/2019 Dinosaur Study 2

1/57

A NEW PHYLOGENY OF THE CARNIVOROUSDINOSAURS

Thomas R. HOLTZ, Jr.D e p a r t m e n t o f G e o l o g y , U n i v e r s i t y o f M a r y l a n d . C o l l e g e P a r k , M A R Y L A N D 2 0 7 3 5 . U S A

E - M a i l : t h o l t z @ g e o l . u m d . e d u

AB S T R A C T : The last several years have seen the discovery of many new theropod dinosaurtaxa. Data obtained from these and from fragmentary forms not previously utilized in cladis-tic analyses are examined. An analysis of forty one primary ingroup taxa and 386 charactersyielded a set of most parsimonious cladograms which preserves many previously discov-

ered relationships (e.g., a basal split between Ceratosauria and Tetanurae; a carnosaur-coelurosaur clade Avetheropoda outside of more primitive "megalosaur" - grade teta-nurines; Dromaeosauridae as the sister taxon to birds, and so forth). The Middle JurassicEnglishProceratosaurus was discovered to be a basal coelurosaur, as was (on less secureevidence) the Middle Jurassic Chinese Gasosaurus: these are among the oldest coeluro-saurs yet described. Several characters previously considered to be restricted to birds andother advanced coelurosaurs (e.g., furcula, semilunate carpal block) were found to be morebroadly distributed among tetanurines. Other characters, once considered synapomor-phies for Avetheropoda (e.g., loss of metacarpal IV, possession of a pubic obturator notch)were found to be convergent between advanced carnosaurs and advanced coelurosaurs,lacking in the basal members of both clades. At least three (and possibly four) separate ori-gins for the arctometatarsalian pes were supported in this study. The mosaic of derivedcharacter state distributions for troodontids relative to the dromaeosaurid-bird clade, the

tyrannosaurid-ornithomimosaur clade, and the therizinosauroid-oviraptorosaur clade sug-gests that relationships alternative to the most parsimonious found here may be supportedin future studies.

INTRODUCTION

Since the pioneering work of GAUTHIER (1986),there has been great scientific interest in the phy-logeny of the Theropoda MARSH, 1881. Much of thisinterest stems from the recognition that the origin ofbirds lies withinthe theropod dinosaurs, an hypothe-sisadvancedbyOSTROM (e.g.,1974,1975a,1975b,1976) primarily from his work on the dromaeosauridDeinonychus antirrhopus OSTROM, 1969a (see PA-DIAN & CHIAPPE, 1998 for a recent review of bird ori-gins). The results of Gauthier's phylogeneticanalysis are shown in Fig. 1A, B.

Numerous authors have proposed phylogenetichypotheses subsequent to Gauthier's initial 1986study: BAKKER, WILLIAMS & CURRIE, 1988; NOVAS,1992, 1997a; CURRIE & ZHAO, 1993a; RUSSELL &DONG,1993a,b;PREZ-MORENO et al.,1993,1994;HOLTZ, 1994, 1995a, 1996a; SERENO et al., 1994,1996,1998;SERENO,1997,1998;SUES,1997;HAR-RIS,1998;FORSTER etal.1998;MAKOVICKY & SUES,

1998. Results of some of these studies are pre-sented in Fig. 1.

The present analysis attempts to synthesize theproposed phylogenetic data from these studies, be-ginningasanupdateofpreviousworkbythepresentauthor (HOLTZ, 1994). Among the changes from thatworkincludecorrectionoftypographicalerrorsinthecharacter descriptions and elimination or modifica-tion of poorly coded characters (see CLARK, PERLE& NORELL, 1994; HUTT, MARTILL & BARKER, 1996;

CHARIG & MILNER, 1997; NORELL & MAKOVICKY,1997 for specific examples). Furthermore, manysignificant new theropod taxa and more completeremains of hitherto poorly known forms (therizino-sauroids, basal ornithomimosaurs, spinosaurids,sinraptorids, alvarezsaurids, etc.) which were notpreviously used have been incorporated into thenew analysis. Additionally, new characters drawnfrom the studies listed previously are included(some in a modified form) here, as are some previ-ously unused characters.

To fully describe each of the characters in thisstudy in detail will require a much longer work (inpreparation by thepresent author). This paper, how-

5

G A I A N 1 5 , L I S B O A / L I S B O N , D E Z E M B R O / D E C E M B E R 1 9 9 8 , p p . 5 - 6 1 ( I S S N : 0 8 7 1 - 5 4 2 4 )

artigos/papers


2/57

6

T.R. HOLTZ, JR .

Fig. 1 - Previously proposed phylogenies of theropod relationships. Taxonomy of listed forms revised to match thenames used here. A - Cladogram of comparatively well known theropods from G A U T H I E R (1986). B - Cladogram of alltheropods included in GA U T H I E R (1986), dashed line for Ornithomimosauria represents topology from GA U T H I E R (1986),solid line after W

I L K I N S O N

(1995) (differs from that presented in GA U T H I E R

(1986) due to incomplete computational analy-sis in that study: see W I L K I N S O N (1995) for details). C - BA K K E R , WI L L I A M S & CU R R I E (1988). D - NO V A S (1992).E - R U S S E L L & D O N G (1993a). F - P R E Z -MO R E N O et al. (1993). G - H O L T Z (1994). (Continued)

(Dromaeosauridae


3/57

To fully describe each of the characters in thisstudy in detail will require a much longer work (inpreparation by the present author). This paper, how-ever, will serve as an interim study pending thatmore detailed phylogenetic analysis.

METHODS AND MATERIALS

The operational taxonomic units (OTUs) em-ployedinthisstudyarelistedinTABLE I.Fortyonein-group taxa were used in the primary analysis. The

phylogenetic positions of three very fragmentaryforms (Deltadromeus agilis, "Megalosaurus" hes-

peris, and Unenlagia comahuensis) were examinedin subsequent analyses.

The sister taxon to those theropods used here isa matter of some recent debate. NOVAS (1994,1997b), SERENO (1997, 1998), and SERENO & NO-VAS (1992, 1994) have proposed that the Late Trias-sic taxa Eoraptor lunensis SERENO et al., 1993 of

Argentina and the more globally distributed Herre-rasauridae BENEDETTO, 1973 share a more recentcommon ancestor with the taxa used in this analysis

than do any other known forms (Fig. 2A). Under thisphylogeny, Eoraptor and the herrerasaurids would

7

A NEW PHYLOGENY OF THE CARNIVOROUS DINOSAURS

Fig.1(continued) - Previously proposed phylogenies of theropod relationships. Taxonomy of listed forms revised tomatch the names used here. H - P R E Z -MO R E N O etal. (1994). I - SE R E N O (1997, 1998). J - F O R S T E R et al. (1998).

K - MA K O V I C K Y

& SU E S

(1998).


4/57

8

T.R. HOLTZ, JR .

T A B L E I Theropod taxa used in this analysis.

OPERATIONAL TAXONOMIC UNITS

TAXA INCLUDED IN PRIMARY ANALYSIS

Abelisaurus comahuensis B O N A P A R T E & N O V A S , 1985Acrocanthosaurus atokensis S T O V A L L & L A N G S T O N , 1950Afrovenator abakensis SE R E N O , W I L S O N , L A R S S O N , D U T H E I L & S U E S , 1994Allosaurus spp. MA R S H , 1877

Alvarezsauridae B O N A P A R T E , 1991Archaeopteryxspp. ME Y E R , 1861

Bagaraatan ostromi OS M L S K A , 1996Caenagnathidae S T E R N B E R G , 1940Carcharodontosaurus saharicus (DE P R E T & S A V O R N I N , 1927)Carnotaurus sastrei BO N A P A R T E , 1985Ceratosaurus nasicornis MA R S H , 1884Coelophysidae WE L L E S , 1984Coelurus fragilis MA R S H , 1879bCompsognathidae CO P E , 1871Dilophosaurus wetherilli(W E L L E S , 1954)Dromaeosauridae RU S S E L L , 1969Dryptosaurus aquilunguis (CO P E , 1866)Elaphrosaurus bambergiJ A N E N S C H , 1920Eustreptospondylus oxoniensis W A L K E R , 1964Gasosaurus constructus DO N G & T A N G , 1985Giganotosaurus caroliniiC O R I A & S A L G A D O , 1995Megalosaurus bucklandi ME Y E R , 1832Microvenator celerOS T R O M , 1970

Monolophosaurus jiangiZH A O

& CU R R I E

, 1993Neovenator salierii HU T T , M A R T I L L & B A R K E R , 1996Ornitholestes hermanni OS B O R N , 1903Ornithomimidae MA R S H , 1890Ornithothoraces C H I A P P E & C A L V O , 1994Oviraptoridae BA R B O L D , 1976aPelecanimimus polyodon P R E Z -MO R E N O , S A N Z , B U S C A L I O N I , M O R A T A L L A , O R T G A & R A S S K I N -GU T M A N , 1994Piatnitzkysaurus floresi BO N A P A R T E , 1979Proceratosaurus bradleyi (WO O D W A R D , 1910)Rahonavis ostromi (FO R S T E R , S A M P S O N , C H I A P P E & K R A U S E , 1998)Scipionyx samniticus D A L S A S S O & S I G N O R E , 1998Sinraptor spp. C U R R I E & Z H A O , 1993a!

Spinosauridae S T R O M E R , 1915Therizinosauroidea RU S S E L L & D O N G , 1993a

Torvosaurus tanneri GA L T O N

& JE N S E N

, 1979Troodontidae GI L M O R E , 1924Tyrannosauridae OS B O R N , 1906Yangchuanosaurus spp. DO N G , C H A N G , L I & Z H O U , 1978 "

TAXA INCLUDED IN SUPPLEMENTARY ANALYSES

Deltadromeus agilis S E R E N O , D U T H E I L , I A R O C H E N E , L A R S S O N , L Y O N , M A G W E N E , S I D O R , V A R R I C C H I O & W I L S O N , 1996Megalosaurus hesperis WA L D M A N , 1974Unenlagia comahuensis NO V A S & P U E R T A , 1997

1 - P A U L ( 1 9 8 8 ) , S M I T H ( 1 9 9 8 ) , C H U R E ( 1 9 9 8 ) , H E N D E R S O N ( 1 9 9 8 ) a n d B A K K E R ( 1 9 9 8 ) a r g u e t h a t m o r e t h a n o n e s p e c i e s ( o r h i g h e r l e v e l

t a x a ) a r e p r e s e n t i n t h e M o r r i s o n F o r m a t i o n g e n u s A l l o s a u r u s . 2 - W

E L L N H O F E R

( 1 9 9 3 ) s u g g e s t s t h a t t w o s p e c i e s a r e p r e s e n t i n t h e

S o l n h o f e n L i t h o g r a p h i c L i m e s t o n e g e n u s A r c h a e o p t e r y x : A . l i t h o g r a p h i c a M E Y E R , 1 8 6 1 a n d A . b a v a r i c a W E L L N H O F E R , 1 9 9 3 . 3 - S i n -

r a p t o r i n c l u d e s t w o s p e c i e s , S . d o n g i C U R R I E & Z H A O , 1 9 9 3 a a n d S . h e p i n g e n s i s ( G A O , 1 9 9 2 ) . 4 - Y a n g c h u a n o s a u r u s i n c l u d e s t w o

s p e c i e s , Y . s h a n g y o u e n s i s D O N G , C H A N G , L I & Z H O U , 1 9 7 8 a n d Y . m a g n u s D O N G , Z H O U & Z H A N G , 1 9 8 3 .


5/57

phylogeny, Eoraptor and the herrerasaurids wouldbeconsideredtruetheropods(giventhedefinitionofTheropoda following GAUTHIER (1986): birds and alltaxa sharing a more recent common ancestor withbirds than with sauropodomorphs). Among the po-tential derived characters supporting such an hy-p o t h e s i s a r e p r o m i n e n t p o s t a x i a l c e r v i c a lepipophyses; greatly reduced manual digits IV andV; an intramandibular joint;and a distalenlargementofthepubis(NOVAS,1994,1997a;SERENO,1997).

Alternatively, HOLTZ & PADIAN (1995) and BONA-PARTE & PUMARES (1995) have argued that sauro-podomorphs share a more recent common ancestorwith the forms in this analysis than do Eoraptor orherrerasaurids (Fig. 2B). Derived characters sup-porting such an hypothesis include pollex unguallarger than other manual unguals; manual digit IIlongest digit in the hand; vertebrae 6-9 longest in thecervical column; and a distal expansion of the is-chium.

The question of the sister taxon tothe forms usedhereisthesubjectofaseparatestudybythepresentauthorandPADIAN, in preparation. For this analysis,

a compromise outgroup was used. Character statesshared in common in Eoraptor, herrerasaurids, and

basal sauropodomorphs (as well as those in com-moninallthreeoftheseandinmoredistantlyrelatedforms such as ornithischians, basal ornithodirans,

and non-ornithodiran archosauriforms) were con-sidered to the primitive relative to the taxa in the cur-rent study. Character states found in the taxa in thepresent analysis and in some (but not all three) ofEoraptor, herrerasaurids, and basal sauropodo-morphs were coded as derived for the ingroupforms: these are discussed below. Finally, characterstates found in some or all of the ingroup taxa butnone of the three potential sister taxa are coded asderived. Based on these codings, an "all zero" out-group with all primitive states was created to ap-proximate a compromise ancestral condition. Thecharacter states observed in herrerasaurids (pri-marily Herrerasaurus) and basal sauropodomorphs("prosauropods") are also included here, and use ofthese taxa rather than an "all zero" outgroup arebriefly discussed. In the longer study in preparationby the present author, differences between the re-sults using this methodand various arrangementsofthe known potential outgroups will be examined.

The characters employed in this analysis arelisted in APPENDIX I. 386 characters were used (135craniodental, 75 axial, 74 pectoral and forelimb, and102pelvicandhindlimbcharacters).301ofthechar-acters are coded as binary; 85 as multistate. All bi-nary characters were considered unordered;

multistate characters were considered unorderedunless otherwise indicated (see APPENDIX I). De-scriptors of facial pneumatic structures follow W IT-MER (1997). In APPENDIX I a brief description of theprimitive and derived state(s) is provided; in work inpreparation, each of these characters will be de-scribed in greater detail (as in GRANDE & BEMIS(1998) or WILSON & SERENO (1998)). However,some characters of particular phylogenetic signifi-cance in this or previous studies will be discussedbelow. APPENDIX II is the data matrix analyzed.

The resulting data matrix was analyzed usingPAUP 3.1.1 (SWOFFORD, 1993). The unwieldy size

of the data matrix required the use of the Heuristicsearch option, as the Branch-and-Bound and Ex-haustivesearchmethodswouldrequireprohibitivelylong run times given current computer calculationspeeds. In all cases, random branch addition wasrun with thirty replicates, to reduce the chance offalsely accepting a local rather than global minimum(thatis,atreeorsetoftreeswithlessthanmaximumparsimony). The set of most parsimonious trees dis-covered under these runs was analyzed using Mac-Clade3.07(MADDISON & MADDISON, 1997), in orderto determine character state distribution under ac-celerated and delayed transformation (ACCTRAN

and DELTRAN, respectively) optimizations and toexamine alternate topologies. The tree length, con-

9


Fig. 2 - Alternative phylogenies for sistertaxon to thero-pods used in this study. A - Herrerasaurids and Eoraptorshare a more recent common ancestor with (more ad-vanced) theropods than do sauropodomorphs, afterN O V A S (1994, 1997a), S E R E N O (1997, 1998), and SE R E N O & NO V A S (1992, 1994). B - Sauropodomorphs share amore recent common ancestor with theropods than do ei-ther herrerasaurids or Eoraptor, after HO L T Z & P A D I A N

(1995) and B O N A P A R T E & P U M A R E S (1995).


6/57

sistency index (CI), retention index (RI), and res-caled consistency index (RC) for each tree was cal-culated on MacClade, while the homoplasy index(HI) was calculated on PAUP: see pp. 364-368 ofMADDISON & MADDISON (1997) for the differences in

metric calculations between these two programs.Bremer support values were calculated by means ofthe AutoDecay program, version 4.0 (ERIKSSON,1998).

Names of clades found in this analysis are basedon the standardized phylogenetic definitions pro-vided in PADIAN, HUTCHINSON & HOLTZ (1999), andare summarized in TABLE II. See the referenceabove, SERENO (1998), and the references thereinfor a discussion of the principles of phylogenetictax-onomy. SERENO (1997, 1998) provides some differ-ent definitions for the same names, and somedifferent names for the same definitions as usedhere: see PADIAN, HUTCHINSON & HOLTZ (1999) fordiscussion of these taxonomic conflicts.

RESULTS

The primary analysis produced a set of 20equally parsimonious trees of 1404 steps.The CI forthese trees was 0.442,the RIwas 0.618,the RCwas0.273,andtheHIwas0.647.Thestrictconsensusofthese trees is presented in Fig. 3A. The normalizedconsensus fork index, a measure of consensus treeresolution (number of nodes in the strict consensustree over number of nodes in a fully resolved di-chotomous tree of the same number of taxa:COLLESS,1980)is0.825.UseofHerrerasauridaeasan outgroup does not alter tree topology, but insteadproduces the same 20 trees at a tree length of 1412,a CIof 0.439,an RI of 0.614, anRC of0.269,and anHI of 0.649. Use of basal sauropodomorphs ("pro-sauropods") yields 120 trees of tree length 1458, CI0.433,RI0.619,RC0.268,andHI0.658:thetreeto-pologies are identical to those analyses using the allzero outgroup or the herrerasaurid outgroup, exceptthat Megalosaurus, Torvosaurus, Eustreptospondy-

10

T.R. HOLTZ, JR .

TA B L E

IIPhylogenetictaxonomicdefinitionsusedinthisstudy,basedprimarilyonP

A D I A N

, HU T C H I N S O N

& HO L T Z

(1999)

TAXON TYPE A B

Theropoda MARSH, 1881 Stem Neornithes CetiosaurusNeotheropoda BAKKER, 1986 Node Ceratosaurus NeornithesCeratosauria MARSH, 1884 Stem Ceratosaurus NeornithesCoelophysoidea HOLTZ, 1994 Stem Coelophysis CeratosaurusNeoceratosauria NOVAS, 1992 Stem Ceratosaurus Coelophysis

Abelisauroidea NOVAS, 1992 Stem Carnotaurus CeratosaurusAbelisauridae BONAPARTE & NOVAS, 1985 Node Abelisaurus CarnotaurusTetanurae GAUTHIER, 1986 Stem Neornithes Ceratosaurus

Avetheropoda PAUL, 1986 Node Neornithes AllosaurusCarnosauria HUENE, 1920 Stem Allosaurus Neornithes

Allosauroidea CURRIE & ZHAO, 1993a Node Allosaurus Sinraptor

Allosauridae MARSH, 1879a Stem Allosaurus Sinraptor Sinraptoridae CURRIE & ZHAO, 1993a Stem Sinraptor AllosaurusCoelurosauria HUENE, 1914 Stem Neornithes AllosaurusManiraptoriformes HOLTZ, 1996b Node Ornithomimus Neornithes

Arctometatarsalia HOLTZ, 1994 Stem Ornithomimus NeornithesBullatosauria HOLTZ, 1994 Node Ornithomimus TroodonOrnithomimosauria BARSBOLD, 1976b Node Ornithomimus PelecanimimusManiraptora GAUTHIER, 1986 Stem Neornithes OrnithomimusOviraptorosauria BARSBOLD, 1976b Node Oviraptor ChirostenotesParaves SERENO, 1997 Stem Neornithes OviraptorEumaniraptora PADIAN, HUTCHINSON & HOLTZ, 1999 Node Deinonychus NeornithesDeinonychosauria COLBERT & RUSSELL, 1969 Stem Deinonychus Neornithes

Avialae GAUTHIER, 1986 Stem Neornithes Deinonychus

Aves LINNE, 1758 Node Archaeopteryx NeornithesMetornithes PERLE, NORELL, CHIAPPE & CLARK, 1993 Node Mononykus Neornithes

T a x o n , t a x o n t y p e , a n d r e f e r e n c e t a x a f o r p h y l o g e n e t i c d e f i n i t i o n s e m p l o y e d i n t h i s p a p e r . D e f i n i t i o n s a n d j u s t i f i c a t i o n s d i s c u s s e d i n P A -

D I A N , H U T C H I N S O N & H O L T Z ( 1 9 9 9 ) . T y p e : S t e m , s t e m - b a s e d ; N o d e , n o d e - b a s e d . S t e m - b a s e d t a x o n d e f i n i t i o n s a r e o f t h e f o r m " r e f e r -

e n c e t a x o n A a n d a l l t a x a s h a r i n g a m o r e r e c e n t c o m m o n a n c e s t o r w i t h r e f e r e n c e t a x o n A t h a n w i t h r e f e r e n c e t a x o n B " . N o d e - b a s e d

t a x o n d e f i n i t i o n s a r e o f t h e f o r m " a l l d e s c e n d a n t s o f t h e m o s t r e c e n t c o m m o n a n c e s t o r o f r e f e r e n c e t a x a A a n d B . "


7/57

lus, and Piatnitzkysaurus are equally parsimoni-ously placed in six different possible configurationsrelative to the Afrovenator-avetheropod clade. The

increase in tree length and decrease in CI, RI, RC,and HI is explained by homoplasy in some condi-tions found in herrerasaurids or prosauropods andvarious derived theropod clades, but not hypothe-sized for basal neotheropods under either acceler-ated or delayed transformation.

Examination of the resulting trees in the primaryanalysis (using the all zero outgroup) found that thepolytomies in the strict consensus cladogram couldbe decomposed as instability at three different re-gions. The variability at each these three regions isindependent of the variability at the other three. Thisindependence (five different possible topologies in

one region of the tree, two in the other two) results inthe 20 different equally parsimonious trees recov-ered.

In the first of these cases, the instability occurredbecause of the incompletely known taxon Procera-tosaurus bradleyi, known only from cranial remainsfrom the Middle Jurassic (Bathonian) of England,was found to occupy five different possible positionswith respect to the other basalmost coelurosaursGasosaurus and Dryptosaurus without change intree length, CI, or other tree metrics (Fig. 3B). Thesecond region of tree instability concerns two alter-native placements for the fragmentary Aptian-

Albian form Microvenator celer, as either the sistertaxontoOviraptoridaeorthesistertaxontothecladeOviraptoridae plus Caenagnathidae (Fig. 3C).

Thethirdinstabilityisofgreaterinterest,asitcon-cerns two alternative placements of the well knowntaxon Troodontidae as the sister group to two verydifferent clades: a sister group relationship with thedromaeosaurid-bird clade on the one hand, and asister group relationship with Ornithomimosauria onthe other. Bothtopologiesare equally parsimonious,and result in the apparent lack of resolution amongmaniraptoriform coelurosaurs shown in Fig. 3A. In

fact there is much greater structure than revealedunder strict consensus: all the other taxa have re-solved positions relative to each other, with the ex-ception of Troodontidae itself. The structure withinManiraptoriformes is presented in Fig. 4, whichshowsthetwoalternativepositionsfortroodontids.

A summary cladogram (Fig. 5) is used to discussthedistribution of characters in thepresent analysis.This represents one of the twenty most parsimoni-ous trees in the analysis. For this summary clado-gram, Proceratosaurus was placed betweenGasosaurus and Dryptosaurus, and Microvenatorplaced as outside of an oviraptorid-caenagnathid

clade (the position preferred by the lower stra-tigraphic position of this form). A tree in which Troo-

dontidae is more closely related to dromaeosauridsandbirdsthan to ornithomimosaurs wasselectedforthe summary cladogram, as this topology more

closely reflects the results of most other workers(e.g., GAUT HIE R, 1986; NOVAS, 1992; SERENO,1997, 1998; SUES, 1997; MAKOVICKY & SUES, 1998;FORSTER et al., 1998; NOVAS & POL, in press). How-ever, the analyses of HOLTZ (1994) and PREZ-MORENO et al. (1994) (and previously the non-numerical study of THULBORN (1984)) recovered atroodontid-ornithomimosaur clade to the exclusionof dromaeosaurids or birds,which represents the al-

11


Fig. 3 - Relationships among theropod dinosaursbased on maximum parsimony analysis of 386 morpho-logical characters. A - Tree represents the strict consen-sus of twenty equally most parsimonious trees (treelength=1404, CI=0.442, RI=0.618, RC=0.273, HI=0.647,normalized consensus fork index=0.825). Charactersused in analysis included in Appendix I. B - Dashed linesindicate the five equally parsimoniousalternative positionsof Proceratosaurus (Pr.) relative to Gasosaurus (Gaso.)and Dryptosaurus (Drypto.). C - Dashed lines indicate the

two equally parsimonious alternative positions ofMicrove-nator (Micro.) relative to Oviraptoridae (Ovir.) andCaenagnathidae (Caen.).


8/57

ternative topology of this study. Characters support-ing this position will be discussed below.

Note that this particular configuration (Fig. 5) is

used to facilitate discussion only, and is not pre-ferred by the data analysis over the nineteen otherpotential arrangements. Future analyses may helpto resolve the uncertainty with regards to the rela-tionships presented here (if not overturn some or allof those in the current analysis, pending the additionof new data).

Addition of the fragmentary Deltadromeus of theCenomanian of northern Africa results in 80 equallyparsimonious trees two steps longer than the pri-mary analysis (tree length 1406, CI 0.441, RI 0.618,RC 0.272, HI 0.627). This addition does not changethe tree topology: rather, Deltadromeus is equallyparsimoniously placed in five different positions.These positions are: as the sister taxon to Orni-tholestes; as the sister taxon to Coelurus, or as thesister taxon to nodes ff, gg, or jj. Similarly, inclusionof the poorly known Unenlagia of the Late Creta-ceous of Argentina does not alter overall tree struc-ture. Instead, its presence results in a total of 50trees eleven steps longer (tree length 1415, CI0.438, RI 0.619, RC 0.271, HI 0.649) than that in theprimary analysis. Unenlagia is equally parsimoni-ouslyplacedasthesistertaxonto Rahonavis,asthesistertaxontotheArchaeopteryx-Metornithes clade(node mm), or as the sister group to Ornithothora-ces. Inclusion of "Megalosaurus" hesperis (knownonly from cranial material of the Middle Jurassic of

England) results in 220 trees 1 step longer than theprimary analysis (TL 1405, other metrics identical tomain analysis). "M." hesperis is equally parsimoni-

ouslyplacedasthesistergrouptonodesG,H,I,J,K,orLorasthesistergrouptoSpinosauridae, Megalo-saurus, Eustreptospondylus, Torvosaurus, or Piat-nitzkysaurus.

In the following section, the character statesfound at each node are listed. The nodes are listedbythecorrespondingletterfromthesummaryclado-gram(Fig.5).Taxonnamesarelistedforsomeofthecladesmentionedhere,followingthedefinitionspro-posedinPADIAN, HUTCHINSON & HOLTZ (1999), andlisted in TABLE II. Stem-defined names are under-lined in the description headings, whereas node-defined names are not. Where two names are listed,

the first represents the stem-defined taxon nameandthesecondisthenode-definedtaxonname.Notall nodes are named.

For purposes of description of the nodal charac-ter state changes, ALL refers to those derived char-acter states present in all optimizations, ACCTRANrefers to those derived character states present atthat node under accelerated transformation (i.e., atthe basalmost point on the tree where this charactercould appear without requiring additional evolution-ary steps), and DELTRAN refers to those present atthat node under delayed transformation (i.e., at theterminal-most point on the tree where this charactercould appear without requiring additional evolution-ary steps). Characters are listed by the characternumber (before the period) and state (after the peri-od): for example, 17.1 refers to the condition "maxil-lary fenestra present." Characters listed with aminus sign (-) before them represent a reversal to astate found at a more basal position in the tree: forexample, -17.0 would indicate "maxillary fenestraabsent" for a form nested within a clade otherwisecharacterized by the state 17.1.

NODE A. THEROPODA - NEOTHEROPODA

ALL: 10.1 Premaxilla and nasal do not meet sub-narially; 26.1 Narial prominences present; 30.1 Lac-rimal broadly exposed on skull roof; 45.1 Orbit ovalor key-shaped, rounded dorsally, constricted ven-trally; 54.1 Postorbital frontal process about samelevel or slightly higherthan squamosal, producingT-shaped postorbital; 73.1 Vomera fused rostrally;110.1 Reduced overlap of dentary onto postdentarybones; 111.1 Intramandibular joint; 117.1 Rostralprong of angular penetrates the dentary-splenialcavity; 137.1 First intercentrum with large occipitalfossa (two or less times as wide as tall) and smallodontoid notch; 138.1 Second intercentrum cranial

articulation with first intercentrum with broad cres-centic fossa; 148.1 Postaxial cervical pleurocoels,one pair present; 152.1 Caudal cervical epipophy-

12

T.R. HOLTZ, JR .

Fig. 4 - Maniraptoriform theropod cladogram as recov-ered in this analysis. Solid lines indicate positions sharedby all twenty most parsimonious trees; dashed line indi-

cates two equally parsimoniouspositions for Troodontidae(as bullatosaurian arctometatarsalians or as paravianmaniraptorans). See text for discussion.


9/57

ses elongate; 164.1 Longest postaxial cervicals VI-IX; 180.1 Cranial and median dorsal pleurocoels,one pair present; 181.1 Presacral pleurocoels cam-erate; 185.3 Number of sacrals five; 197.1 Transi-tion point in distal half of tail; 219.1 Coracoid bicepstubercle conspicuous and well developed; 256.3Metacarpal V absent; 257.1 Metacarpal IV present,without ungual; 270.1 Digit II longest in manus;

271.1 Penultimate phalanx longest nonungual pha-lanx; 286.1 Ilium dolichoiliac; 290.1 Brevis fossadeep; 301.1 Acetabular height/craniocaudal length

about 50%; 312.1 Pubic blade at least six times aslong as broad; 329.1 Femur shape bowed in convexarc with less pronounced sigmoidality; 336.1 Proxi-malmost point of anterior trochanter below femoralhead; 372.1 Metatarsal V vestigial or absent; 378.1Metatarsal I reduced but retains phalanges; 379.1Metatarsal I placed near midpoint of metatarsal IIshaft.

ACCTRAN: 16.1 Promaxillary fenestra present,visible in lateral view; 24.1 Nasal participates in an-

13


Fig. 5 - Summary cladogram of theropod relationships, representing one of the twenty equally parsimonious trees in

this analysis. See text for character state changes at each node, listed by letter indicated. Note that the topology shownhere is not preferred by the data over other potential topologies, but was instead chose to facilitate discussion of characterdistribution. See text for details. A - Ceratosauria and non-maniraptoriform Tetanurae section of tree. B - Maniraptorifor-mes.


10/57

torbital cavity; 44.1 Orbit shorter than internal antor-bital fenestra length; 58.1 Jugal participates ininternal antorbital fenestra; 67.1 Quadrate foramen

reduced or absent; 81.1 Ventral ectopterygoid re-cess present and comma-shaped; 253.1 Distal car-pal I block does not overlap metacarpal II dorsally,but does so ventrally; 254.1 Distal carpal I fused todistal carpal II; 278.1 Pollex larger than other man-ual unguals; 317.2 Pubic boot rounded, angle be-tween shaft and caudal portion of boot acute; 319.1Pubic boot present, less than 30% as long as pubicshaft; 339.1 Muscle scar in craniodistal region of fe-mur present, non-elliptical in shape; 344.1 Ectocon-dylar tuber proximodistally long, pronounced, andextends almost to distal end of femur; 350.1 Cristafibularis present, not well developed.

DELTRAN: NoneAs defined by GAUTHIER (1986) Theropoda is a

stem-based taxon, comprised of birds and all taxasharing a more recent common ancestor with birdsthan with sauropodomorphs. The term Neothero-podaBAKKER,1986hasbeenusedbySERENO etal.(1993), SERENO (1997, 1998), and PADIA N ,HUTCHINSON & HOLTZ (1999) for the node-definedtaxon comprised of all descendants of the most re-cent common ancestor of Ceratosauria and Tetanu-rae (or, more explicit ly, of Ceratosaurus andNeornithes: see PADIAN, HUTCHINSON & HOLTZ(1999)).

How and if the diagnosis of Theropoda and Neo-theropoda differ hinges on the question of the imme-diate sister group of Neotheropoda (see above, andFig. 2). The following characters from the above listare shared by basal theropods and basal sauropo-domorphs, and would therefore diagnose a clademore inclusive than Theropoda if herrerasauridsand Eoraptor are not theropods themselves: 10.1,67.1, 117.1, 164.1, 270.1, 278.1, and 301.1. Thoseshared in common with Herrerasauridae and/orEoraptor (see below) would either have evolvedconvergently between these Triassic forms and true

theropods or have been present in the common an-cestor of these taxa, theropods, and sauropodo-morphs, and subsequently lost in the latter. Theremaining characters of the above list would be di-agnostic of Theropoda and Neotheropoda (whichwouldsharethesamediagnosis,asallknownthero-pods would be neotheropods).

If instead herrerasaurids and Eoraptor are truetheropods, then character197.1 (Transition point indistal half of tail) would be considered synapomor-phic for Theropoda, and the following for anherrerasaurid-neotheropod clade: 67.1, 111.1,117.1, 138.1, and 152.1. It should be noted, how-

ever, that the condition represented by characterstate 111.1 (Intramandibular joint) may not be ho-

mologous in Herrerasauridae and Neotheropoda,as SERENO & NOVAS (1994) noted the geometry ofthejointarereversedbetweenthetwotaxa:inherre-

rasaurids, the splenial has a concave surface whichslides against the convex ventral margin of the an-gular, while in neotheropods the splenial has a con-vex dorsal surface which slides against a concavedepression on the angular (p. 471). No characterswere observed which were present in Eoraptorandbasally within Neotheropoda that were not alsofound in herrerasaurids. Under this scheme, theseven derived characters shared by neotheropodsand basal sauropodomorphs mentioned previouslywould either have had to developed independentlyinthesetwolineages,orhavebeenpresentandsub-sequently lost in Eoraptor and Herrerasauridae.

There remain many characters found basally inCeratosauria and Tetanurae which are not alsofound in basal sauropodomorphs, Eoraptor, or Her-rerasauridae. These are diagnostic for Neothero-poda, regardless of either particular sister groupscenario.

NODE B. CERATOSAURIA

ALL: 31.2 Lacrimal prominences comprised ofridge continuous with raised surface of lateral edgeof nasals; 148.2 Postaxial cervical pleurocoels, twopairs present; 170.1 Dorsal transverse processesstrongly backturned caudally and triangular in dor-sal view; 180.2 Cranial and median dorsal pleuro-coels, two pairs present; 193.1 Ventral groove incranial caudals; 201.1 Shaft of cervical ribs ex-tremelylong(fourormoretimescentrumlength)andslender; 291.2 Brevis fossa distal end broad; 300.1Supracetabular shelf on ilium present; 309.1 Pubisorientation propubic, proximal portion of shaft ap-proximately 30 degrees from horizontal; 321.1 Is-chial antitrochanter large; 335.1 Anterior trochanterconical prominence; 338.1 Trochanteric shelf of fe-mur well developed; 340.1 Medial epicondyle (=mediodistal crest) of femur pronounced, extendsone quarter or more the length of the femoral shaft;342.1 Groove in lateral condyle of femur; 345.1 Sul-cus along medial side of base of crista tibiofibularis;356.1 Sulcus in proximomedial region of fibula;359.1 Anterior surface of distal fibula overlaps as-cending process of astragalus cranially; 375.1Metatarsal III dorsal surface area clearly larger thaneither metatarsal II or metatarsal IV

ACCTRAN: 189.1 Synsacrum present in adults;285.1 Pelvicgirdle sutures fused in adults; 366.1As-tragalocalcaneum(astragalusfusedtocalcaneum)

DELTRAN: 24.1 Nasal participates in antorbitalcavity; 44.1 Orbitshorter than internal antorbitalfen-

estra length; 67.1 Quadrate foramen reduced or ab-

14

T.R. HOLTZ, JR .


11/57

sent; 319.1 Pubis and ischium proximal shaftsnarrow

As first proposed by GAUTHIER (1986), Cerato-

sauria forms a major clade of theropods containingsuch forms as Ceratosaurus, Dilophosaurus, andCoelophysis.A s inHOLTZ (1994),Ceratosauriaisdi-vided into two primary branches, the relativelygracile Late Triassic and Early Jurassic Coelophy-soideaandthemorerobustLateJurassicandCreta-ceous Neoceratosauria.

ROWE (1989)andROWE & GAUTHIER (1990)con-s i d e r e d t h e E a r l y J u r a s s i c ( ? S i n e m u r i a n -Pliensbachian) Sarcosaurus woodiANDREWS 1921to be a ceratosaur. This taxon (known only from apartial pelvis) was not included in the present analy-sis, so its relationship with other ceratosaurs is notresolved here.

NODE C. COELOPHYSOIDEA

ALL: 11.1 Subnarial gap; 26.2 Narial promi-nences comprised of paired ridges along lateraledgesofnasals; 64.1 Dorsalramusofquadratojugaldoes not contact squamosal; 142.1 Axial parapo-physes reduced; 143.1 Axial diapophyses absent;145.1Axial pleurocoels absent; 151.1 Epipophyseson cervical vertebrae placed proximally; 234.1 Hu-meral torsion present; 235.1 Humeral shaftsigmoid;308.1 Pubic fenestra ventral to obturator foramen

AC CT RA N: 27.1 Paired crescentic crestsformed by nasal and lacrimal prominences; 127.1Dentary teeth more numerous and smaller thanmaxillary teeth; 331.1 Femoral head transverselyelongate

DELTRAN: -317.0 Pubic boot absent; -344.0 Ec-tocondylar tuber proximodistally short, proximallyplaced

As in ROWE (1989),NOVAS (1992),HOLTZ (1994),and SERENO (1997), a clade comprised ofDilopho-saurus and the Coelophysis-Syntarsus clade (Coe-lophysidae) to the exclusion of other ceratosaurs

was supported.Pairedcrescentic crestsformed by nasal andlac-

rimal prominences may be synapomorphic for Coe-lophysoidea (SERENO, 1997) as they are present inDilophosaurus and Syntarsus kayentakatae ROWE1989 (although not in Coelophysis nor in Syntarsusrhodesiensis). Such crests were the primary evi-dence for placing "Dilophosaurus" sinensis HU,1993 in that genus: since these structures are foundin other coelophysoids, and given certain other de-rived anatomical differences between these taxa(tooth row rostral to the orbit in D. wetherilli,fivepre-

maxillary teeth in "D." sinensis, among others), itmay be that the Chinese taxon does not share amore recent common ancestor with Dilophosaurus

than with some other coelophysoid genus. ROWE(1989) found evidence that Liliensternus liliensterni(H UENE, 1934) was a coelophysoid (see also

RAUHUT & HUNGENBHLER, 1998), as did SERENO(1997), who also included ProcompsognathusFRAAS, 1913 and Segisaurus CAMP, 1936 in thisclade. CARPENTER (1997) has described a largeLate Triassic North American coelophysoid Goji-rasaurus quayi. These taxa were not included in thisstudy: future analyses will hopefully clarify the rela-tionships of these taxa to each other and to otherceratosaurs.

The Late Jurassic African Elaphrosaurus bam-bergi does share numerous derived features withcoelophysoids in general, and coelophysids in par-ticular (e.g., 154.1, 160.1, 165.1, 376.2). Unlike

PAUL (1988) and NOVAS (1992), Elaphrosaurus wasnot found to be member of Coelophysoidea in thisanalysis, but rather was hypothesized to share amore recent common ancestry with abelisauroidsand Ceratosaurus, as in HOLTZ (1994) and SERENO(1997). However, moving this taxon to a sister groupposition with node C or with Coelophysidae requiresonly one additionalevolutionary step,and additionaldetails may reveal that Elaphrosaurus was a late-surviving coelophysoid. (See also Discussion).

The newly discovered taxon Genusaurus serusACCAIRE et al., 1995 was considered by thoseauthors to be a ceratosaur closer to Coelophysisthan to Ceratosaurus. If such a position were con-firmed, it would indicate the first known Cretaceous(middle Albian) coelophysoid. However, althoughthis form demonstrates some ceratosaurian fea-tures(pelvicgirdlesuturesfused,proximalportionofpubic shaft approximately 30 degrees from the hori-zontal, trochanteric shelf well developed, sulcusalong medial side of base of crista tibiofibularis), itdoes not show any unambiguously coelophysoidfeature. In fact, a well-excavated proximal region ofthe fibular medial face is not known in coelophy-soids, but is documented in the neoceratosaurCar-notaurus . Additional study may demonstrateGenusaurus to be a mid-Cretaceous European abe-lisauroid.

NODE D. NEOCERATOSAURIA

ALL: 149.1 Cervical epipophyses powerfully de-veloped and prong-shaped; 185.4 Six sacrals.

ACCTRAN: 4.1 Premaxillary symphyseal regionU-shaped in ventral view; 5.1 Premaxilla subnariallyvery deep, main body taller dorsoventrally than longrostrocaudally; 62.1 Infratemporal fenestra abouttwice as large as the area of the orbit in lateral view;66.1 Quadrate-quadratojugal suture fused; 68.1

Quadrate dorsal ramus greater than height of orbit;70.1 Quadrate articulation projects well caudal to

15



12/57

thecaudalpointoftheoccipitalcondyle;83.1 Nuchalcrestpronounced;84.1 Supraoccipitalwithverypro-nounced, strongly demarcated median ridge on oc-

cipital surface; 103.1 Occipital condyle constrictedneck; 116.1 Horizontal shelf on lateral surface ofsurangular, rostral and ventral to the mandibularcondyle, prominent and extends laterally; 144.1Ax-ial epipophyses prominent; 181.2 Presacral pleuro-coels camellate; 231.1 Ulna/femur length ratio lessthan 28%; 232.1 Radius/humerus length ratio lessthan 50%; 315.2 Pubic boot rounded, angle be-tween shaft and caudal portion of boot acute; 355.1Proximal region of fibular medial face shallow andnot conspicuous.

DELTRAN: 189.1 Synsacrum present in adults;285.1 Pelvicgirdle sutures fused in adults; 366.1As-

tragalocalcaneum (astragalus fused to calcaneum);376.2 MetatarsalIIIdorsalsurfacedumbbellshaped(cranial and (especially) plantar surfaces expandedto slightlyoverlap surfacesof metatarsalsII andIV).

NODE E.

ALL: 147.1 Cervical centra surfaces markedlyopisthocoelous; 150.1 Cervical epipophyses di-rected dorsolaterally and taller than neural spine;185.5 More than six sacrals; 186.1 Sacrals III-Vtransversely compressed; 302.1 Ilium about as longas femur; 334.1 Anterior trochanter present, sepa-rated from femoral head by cleft; 337.1 Fourth tro-chanter of femur present, but little developed.

ACCTRAN: 357.1 Cranialprotuberanceonfibulabelow expansion.

DELTRAN: 4.1 Premaxillary symphyseal regionU-shaped in ventral view; 5.1 Premaxilla subnariallyvery deep, main body taller dorsoventrally than longrostrocaudally; 58.1 Jugal participates in internalantorbital fenestra; 62.1 Infratemporal fenestraabout twice as large as the area of the orbit in lateralview; 68.1 Quadrate dorsal ramus greater thanheight of orbit; 70.1 Quadrate articulation projectswell caudal to the caudal point of the occipital con-

dyle; 83.1 Nuchal crest pronounced; 84.1 Supraoc-cipital with very pronounced, strongly demarcatedmedian ridge on occipital surface; 103.1 Occipitalcondyle constricted neck; 116.1 Horizontal shelf onlateral surface of surangular, rostral and ventral tothe mandibular condyle, prominent and extends lat-erally; 144.1 Axial epipophyses prominent; 160.2Midcervicalcentralengthlessthantwicediameterofcranial face; 315.2 Pubic boot rounded, angle be-tween shaft and caudal portion of boot acute; 317.2Pubic boot rounded, angle between shaft and cau-dal portion of boot acute; 339.1 Muscle scar incraniodistal region of femur present, non-elliptical in

shape.

NODE F. ABELISAUROIDEA - ABELISAURIDAE

ALL: 15.2 Maxillary antorbital fossa greatly re-duced in size, not extending much beyond rim of the

external antorbital fossa; 26.3 Narial prominencesknobby rugosities across dorsal and lateral surfaceof nasals, extending onto dorsalmost surface ofmaxillae; 35.2 Lacrimal dorsal (= rostral) ramus ab-sent; 37.1 Prefrontals reduced or absent; 41.1Frontal-frontal suture fused; 42.2 Frontal-parietalsuture on dorsal surface of skull fused, suture indis-tinguishable; 50.1 Postorbital-lacrimal contactbroad; 53.1 Postorbital suborbital flange

ACCTRAN: -117.0 Rostralprongofangulardoesnot penetrate the dentary-splenial cavity; 161.1 Mid-cervical centra greater than 20% broader than tall;176.1 Dorsal centrum transverse section wider thanhigh; 188.1 Sacral neural spines fused to form lam-ina; 211.1 Scapular blade long, slender (four timesor more longer than midshaft width) and strap-like;212.1 Distal expansion of scapula reduced, lessthan width of proximal end of scapula; 213.1 Acro-mion in scapula reduced; 229.1 Humerus/scapulalength ratio less than 65%; -234.0 Humeral torsionabsent; 237.1 Internal tuberosity on proximal end ofhumerus well differentiated and angular; 249.1 Ul-nar facet for radius transversely expanded and con-cave; 269.1 Metacarpal-phalangeal joints nothyperextensible, extensor pits on metacarpals I-IIIreduced; 304.1 Iliac-ischial articulation smaller thaniliac-pubic articulation; -309.0 Pubis orientation pro-pubic,shaftapproximately45degreesfromhorizon-tal; 330.1 Femoral head approximately 90 degreesfrom shaft (head directed horizontally); 355.2 Proxi-mal region of fibular medial face well excavated

DELTRAN: 16.1 Promaxillary fenestra present,visible in lateral view; 66.1 Quadrate-quadratojugalsuture fused

As in NOVAS (1992), HOLTZ (1994), and SERENO(1997, 1998), a clade comprised of Ceratosaurusandabelisauridswassupportedhere.Asinthestud-ies by the latter two authors, Elaphrosaurus was

found to be part of this clade. However, as notedabove, support for a coelophysoid placement of thistaxon is nearly as strong.

As HOLTZ (1994) noted, many features unitingneoceratosaurs are also found in tetanurines: underthe most parsimonious distributions of derived char-acter states, these are explainable either as conver-gences between Neoceratosauria and Tetanurae oras basal neotheropod characters subsequently lostin Coelophysoidea. Alternatively, Ceratosaurus and

Abelisauridae may share a more recent common an-cestor with tetanurines than with Coelophysoidea:

however, such a phylogenetic scenario requiresseveraladditionalstepsgiventhepresentdatabase(see Discussion below).

16

T.R. HOLTZ, JR .


13/57

Although this study did not examine various otherneoceratosaurs, CORIA & SALGADO (1998) describea new taxon assignable to this clade. That study de-

tails several forms that would belong to the stem-based taxon Abelisauroidea (all taxa sharing a morerecentcommonancestorwith Carnotaurus thanwithCeratosaurus)whichwerenotincludedinthisanaly -sis, as well as additional character evidence for rela-tionships within Neoceratosauria. SAMPSON et al.(1998) have described excellent, well-preserved re-mains of the abelisaurid Majungatholus atopusSUES & TAQUET, 1979, a form which they considerlikely to be the sister taxon to Carnotaurus within

Abelisauridae. This new material will greatly in-creaseourknowledgeofneoceratosaurosteology.

NOVAS (1997c) has suggested that Carcharo-

dontosaurus an d Giganotosaurus were abeli-saurids or abelisaurid relatives. Such a relationshipwas not supported here, but is discussed below.

NODE G. TETANURAE

ALL: 13.1 Rostral ramus of maxilla present, dra-matic change in curvature of rostrodorsal surface ofmaxilla rostral to dorsal ramus forming concave sur-face; 14.2 Rostral ramus as long or longer rostro-caudally as dorsoventrally; 34.1 Slot in ventralprocess of lacrimal for jugal; 35.1 Lacrimal dorsal(rostral) ramus dorsoventrally pinched and narrow;133.1 Caudalmost maxillary tooth position rostral toorbit; 147.1 Cervical centra surfaces opisthocoel-ous; 211.1 Scapular blade long, slender (four timesor more longer than midshaft width) and strap-like;212.1 Distal expansion of scapula reduced, lessthan width of proximal end of scapula; 240.1 Hu-meral ends well expanded, greater than 150% mid-shaft diameter; 304.1 Iliac-ischial articulationsmaller than iliac-pubic articulation; 305.1 Pubic pe-duncle of ilium more developed craniocaudally thanmediolaterally; 341.1 Extensor groove in craniodis-tal region of femur present, but shallow and not con-spicuous; 352.1 Tibia distal end expanded to backcalcaneum; 362.1 Astragalar ascending process

mediolaterally reduced, craniocaudally wide, andproximodistally low ("allosauroid condition")

AC CT RA N: 38.1 Prefrontal-frontal peg-in-socket suture; 48.1 Postorbital ventral processbroader transversely than rostrocaudally with U-shaped cross-section; 61.1 Jugal recess; 64.2Broad contact between dorsal ramus of quadratoju-gal and lateroventral ramus of squamosal; 76.1Palatine tetraradiate; 78.1 Palatine recesses; 123.1Retroarticular process of articular faces caudally;197.2 Transition point in proximal half of tail; 215.1Scapulacoracoid cranial margin with pronounced

notch between acromial process and coracoid;221.1 Sternal plates fused medially; 222.1 Sternumcarina present; 223.2 Sternum wider mediolaterally

than long craniocaudally; 225.1 Furcula; 233.1 Ma-nus/(humerus + radius) length ratio greater than66%; 255.1 Semilunate carpal block fullydeveloped

with transverse trochlea; 257.2 Metacarpal IV pres-ent, without phalanges; 261.1 Articular surface be-tween metacarpals I and II extends well intodiaphysis of metacarpal I; 263.1 Metacarpal IIIclearly shorter than metacarpal II; 268.1 MetacarpalIVlessthanhalflengthofmetacarpalII; 279.1 Pollexungualgreaterthanthreetimeslongerthanheightofarticular facet; 283.1 Manual ungual length ex-tremely long; 334.1 Anterior trochanter present,separated from femoral head by cleft; 346.1 Cne-mial process arises out of the lateral surface of tibialshaft; 351.1 Crista fibularis proximallyplaced; 353.1Fibula closely appressed to tibia throughout mainshaft; 360.1 Fibula distal end less than twice cranio-caudalwidthatmidshaft,andconsequentlyastraga-lar cup for fibula reduced; 364.1 Astragalar distalcondyles oriented cranioventrally

DELTRAN: 16.1 Promaxillary fenestra present,visible in lateral view; 160.2 Midcervical centralengthlessthantwicediameterofcranialface;291.1Brevis fossa distal end tapered; 338.2 Trochantericshelf of femur absent; 339.1 Muscle scar in cranio-distal region of femur present, non-elliptical inshape; 301.1 Crista fibularis present, not well devel-oped; 344.1 Ectocondylar tuber proximodistallylong, pronounced, and extends almost to distal end

of femur.

NODE H.

ALL: 234.1 Humeral torsion present; 249.1 Ulnarfacet for radius transversely expanded and con-cave; 330.1 Femoral head approximately 90 de-grees from shaft (head directed horizontally).

ACCTRAN: 22.1 Pneumatic excavation withoutfenestra in cranial portion of maxillary antorbitalfossa; 120.1 Splenial with notch for rostral margin ofinternal mandibular fenestra; 132.1 Premaxillarytoothcrownsasymmetrical(stronglyconvexlabially,

relatively flattened lingually); 149.1 Cervical epipo-physes powerfully developed and prong-shaped;205.1 Paired caudal and cranial chevron bases;315.2 Pubic boot shape rounded, angle betweenshaft and caudal portion of boot acute.

DELTRAN: 334.1 Anterior trochanter present,separatedfromfemoralheadbycleft;335.2Anteriortrochanter of femur alariform; 346.1 Cnemial pro-cess arises out of the lateral surface of tibial shaft;350.1 Crista fibularis present, not well developed;351.1 Crista fibularis proximallyplaced; 376.1 Meta-tarsal III dorsal surface hourglass shaped.

NODE I.ALL: 331.1 Femoralheadtransverselyelongate.

17



14/57

ACCTRAN: None.

DELTRAN: 22.1 Pneumatic excavation withoutfenestra in cranial portion of maxillary antorbital

fossa; 48.1 Postorbital ventral process broadertransversely than rostrocaudally with U-shapedcross-section; 149.1 Cervical epipophyses power-fully developed and prong-shaped; 315.2 Pubicboot shape rounded, angle between shaft and cau-dal portion of boot acute; 317.1 Pubic boot present,lessthan30%aslongaspubicshaft; 341.1 Extensorgroove in craniodistal region of femur shallow andnot conspicuous; 353.1 Fibula closely appressed totibia throughout main shaft; 360.1 Fibula distal endless than twice craniocaudal width at midshaft, andconsequently astragalar cup for fibula reduced.

NODE J.ALL: 31.2 Lacrimal prominences comprised of

ridge continuous with raised surface of lateral edgeof nasals; 336.2 Anterior trochanter of femur proxi-mal most point above distal margin of femoral head;355.1 Proximal region of fibular medial face slightlyconcave; 365.1 Pronounced horizontal grooveacross cranial face of astragalar condyles.

ACCTRAN: -212.0 Distal expansion of scapulabroad (subequal in width to proximal end of scap-ula); 339.1 Muscle scar in craniodistal region of fe-mur present, non-elliptical in shape.

DELTRAN: 41.1 Orbitshorterthaninternalantor-bital fenestra length; 61.1 Jugal recess; 64.2 Broadcontact between dorsal ramus of quadratojugal andlateroventral ramus of squamosal; 205.1 Pairedcaudal and cranial chevron bases; 215.1 Scapula-coracoid cranial margin with pronounced notch be-tween acromial process and coracoid; 261.1

Articular surface between metacarpals I and II ex-tends well into diaphysis of metacarpal I.

NODE K.

ALL: 139.1 Axial "spine table" (expanded distal

endofneuralspine); 146.1 Ventralkeelonaxialcen-trum absent; 155.1 Cervical zygapophyses dis-placed laterally away from centrum in dorsal view;235.1 Humeral shaft sigmoid.

ACCTRAN: 4.1 Premaxillary symphyseal regionU-shaped in ventral view; -10.0 Premaxilla and na-sal meet subnarially; 60.2 Jugal ventral quadratoju-gal process extends further caudally than dorsalquadratojugal process; 260.1 Metacarpal I one halfto one third metacarpal II length; 264.1 MetacarpalIII very much narrower (less than 50%) than meta-carpal II; 266.1 Base of metacarpal III set on palmarsurface of hand below base of metacarpal II; 267.1

Proximal articulation of metacarpal III triangular;126.1 Cranialcervicalsbroaderthandeeponcranial

surface, with kidney-shaped articular surfaces thatare taller laterally than at midline; 287.2Anterior tro-chanter of femur proximalmost point above distal

margin of femoral head.DELTRAN: 319.1 Pubis and ischium proximal

shafts narrow.

As in GAUTHIER (1986), NOVAS (1992), HOLTZ(1994), SERENO (1997), and most other recent stud-ies of theropod phylogeny, a robustly supportedclade of birds and theropods more closely related tobirds than to Ceratosaurus was discovered. Thisclade, GAUTHIER's (1986) Tetanurae, comprises pri-marily the subdivisions Carnosauria and Coeluro-sauria,discussedbelow.However,thereareseveralforms of theropod which (in the present analysis) lieoutside the carnosaur-coelurosaur clade Avethero-poda, yet were found to share a more recent com-mon ancestor with birds than with Ceratosaurus.

The relationships among the basal tetanurines,informally referred to as "megalosaurs" (in, for ex-ample,GAUTHIER (1986):p.10)havebeenproblem-atic in most recent studies (HOLTZ,1994;SERENO etal., 1994, 1996; SERENO, 1997). Unfortunately, thepresent analysis does not provide strong support forany particular scenario of "megalosaur" phylogeny.This uncertainty seems to stem from a number ofsources, the most important being: a) the fairly largenumber of missing data from some of these taxa,

representing our inadequate knowledge of the os-teology of these forms at present; b) the lack of spe-cializations in many of these taxa beyond thoseshared by all tetanurines, particularly in the case ofthe non-spinosaurid "megalosaurs"; and c) alterna-tively, the highly apomorphic nature of the skulls ofspinosaurids (CHARIG & MILNER, 1997; SERENO etal., 1998), in which the rostrum, dentition, palate,and basicranium are uniquely modified amongtheropods.

As previously noted, the fragmentary taxon"Megalosaurus" hesperis falls within this sector ofthe cladogram, but no particular sister group rela-

tionship was better supported than the others.

NOVAS (1992)proposedthename"Avipoda"foraclade comprised of Eustreptospondylus, Piatnitz-kysaurus,andmoreadvancedtetanurines.Thiscor-responds to node I in this study. If additional work ontheropod phylogenetics continues to support a sub-group excluding some basal tetanurines but unitingEustreptospondylus, Piatnitzkysaurus, and morederived taxa, this name would serve as a useful la-bel.

SERENO et al. (1994, 1996; 1998) and SERENO(1997) presented evidence that several of these ba-

sal tetanurines (in particular, Torvosaurus, Eustrep-tospondylus, and Spinosauridae) represented a

18

T.R. HOLTZ, JR .


15/57

distinct clade of theropod, exclusive of other car-nivorous dinosaurs. CHARIG & MILNER (1997) dem-onstrated that several of the "synapomorphies" for

this postulated "Torvosauroidea" (later changed to"Spinosauroidea") are absent in Spinosauridae. Inthe present study, some of the other characters sug-gested to support a monophyletic "Spinosauroidea"(13.1, 14.1, 22.1, 35.1, and 279.1) are explained inthe most parsimonious distribution of states in thisanalysis to be basal tetanurine features lost in someor all avetheropod taxa.

Of particular note are conditions related to therostral ramus ("anterior ramus" in BAKKER et al.(1992) and SERENO et al. (1994, 1996)) of the max-illa. The maxillae of basal tetanurines differ fromthose of ceratosaurs in the shape of the rostrodorsal

margin (Fig. 6). In ceratosaurs and herrerasauridsthis line is a simple curve, convex dorsally, from thedorsal ramus to the tooth line. In tetanurines primi-tivelythissurfaceisamorecomplexcurve,withade-pression ventral to the external naris. This producesarostralramustothemaxilla,rostraltothedorsalra-mus(13.1).Thepresenceofthisstructureisfoundinall non-avetheropods for which skull material isknown (althoughthe conditionin spinosauridsis dis-tinctfromtheothertaxa,giventheirelongatesnouts:CHARIG & MILNER (1997)), and is also present in theskullsofmostcarnosaurs(thenotableabsencesbe-ing Sinraptorand Yangchuanosaurus) and in basal

coelurosaurs for which the skull is known (Procera-tosaurus and Ornitholestes).Aswithsinraptoridcar-nosaurs, the clade of coelurosaurs comprised ofScipionyxand Maniraptoriformes have the primitivestate with a simple dorsally convex curvature of therostral portion of themaxilla.This distributionis mostparsimoniously explained as independent reversalsin Sinraptoridae and advanced Coelurosauria.

SERENO et al. (1994, 1996) specifically recog-nized a derived condition in which the rostral ramuswas longer rostrocaudally than tall dorsoventrally. Inthe present study two different derived states wererecognized to describe the relative proportions of

the rostral ramus: those forms for which the struc-ture is present, but shorter rostrocaudally than talldorsoventrally (14.1) and the condition recognizedby SERENO et al. (1994, 1996) (14.2). Rather thanuniting Spinosauridae, Torvosaurus, Eustrepto-spondylus, and Afrovenator outside of other thero-pods, however, this condition was found to be thestate at the base of Tetanurae, and subsequentlyshortened in various tetanurine taxa.

Curiously, all known basal tetanurines representfairlylarge sized taxa (approximately6 m or longer).

NODE L.

ALL: 17.1 Maxillary fenestra; -22.0 Pneumaticexcavation without fenestra in cranial portion ofmaxillary antorbital fossa absent; 322.1 Obturator

process of ischium separate, trapezoidal; 323.1 Ob-turator process of ischium proximally placed; 341.2Extensor groovein craniodistalregion of femur deepand conspicuous

ACCTRAN: 181.2 Presacral pleurocoels camel-late; 217.1 Ventralcoracoidprocesswelldeveloped;355.2 Proximal region of fibular medial face well ex-cavated

DELTRAN: 58.1 Jugal participates in internal an-torbital fenestra; 197.2 Transition point in proximalhalf of tail; 253.1 Distal carpal I block does not over-lap metacarpal II dorsally, but does so ventrally;254.1 Distal carpal I fused to distal carpal II; 255.1Semilunate carpal block fully developed with trans-verse trochlea; 260.1 Metacarpal I one half to onethird metacarpal II length; -279.0 Pollex ungual lessthan three times longer than height of articular facet;-283.0 Manual ungual moderate length; 339.2 Mus-cle scar in craniodistal region of femur present, ellip-tical in shape; 364.1 Astragalar distal condylesoriented cranioventrally

NODE M. AVETHEROPODA

ALL: -35.0 Lacrimal dorsal (rostral) ramus dor-soventrally deep; -48.0 Postorbital ventral process

broader rostrocaudally than transversely; 58.1Squamosal constriction of lateral temporal fenestra;169.1 Scars for interspinous ligaments terminatebelow apex of neural spine; 208.1 Middle chevronwith dramatic bend in distal portion ("L-shaped");274.1 First phalanx of pollex greater than length ofmetacarpal II; 288.1 Iliac preacetabular fossa for M.cuppedicus

ACCTRAN: 330.2 Femoral head greater than 90degrees from shaft (head directed dorsally)

DELTRAN: 38.1 Prefrontal-frontal peg-in-socketsuture; 76.1 Palatine tetraradiate; 78.1 Palatine re-

cesses; 120.1 Splenial with notch for rostral marginof internal mandibular fenestra; 123.1 Retroarticularprocess of articular faces caudally; 132.1 Premaxil-lary tooth crowns asymmetrical (strongly convex la-bially, relatively flattened lingually); 212.1 Distalexpansionof scapula reduced or absent; 217.1 Ven-tralcoracoidprocesswelldeveloped; 225.1 Furcula;233.1 Manus/(humerus + radius) length ratiogreaterthan66%;-249.0 Ulnarfacetforradiussmalland flat; 264.1 Metacarpal III very much narrower(less than 50%) than metacarpal II; 266.1 Base ofmetacarpal III set on palmar surface of hand belowbase of metacarpal II; 267.1 Proximal articulation of

metacarpal III triangular; 268.1 Metacarpal IV lessthan half length of metacarpal II; 278.1 Pollex larger

19



16/57

thanothermanualunguals; 355.2 Proximalregionoffibular medial face well excavated

As in NOVAS (1992), HOLTZ (1994),SERENO etal.

(1994, 1996), and SERENO (1997), a clade was rec-ognized containing Allosaurus and those taxa clos-est to it as one branch, and Aves and those taxaclosesttoitastheother,outsideofthemoreprimitivetetanurines. Following HOLTZ (1994), and using aname for this clade first published in PAUL (1988),this clade comprised of all descendants of the mostrecent common ancestor ofAllosaurus and Neorni-thes is termed Avetheropoda. SERENO et al. (1994,1996) and SERENO (1997, 1998) suggested the al-ternative name "Neotetanurae": indeed, SERENO(1997, 1998) uses the same definition as the above,rendering this term a junior objective synonym of

Avetheropoda (see also PADIAN , HUTCHINSON &HOLTZ (1999)). Perhaps the name "Neotetanurae"might be preserved as the name for a more inclusivetaxon (for example, Neornithes and all taxa sharinga more recent common ancestor with Neornithesthan with Spinosauridae).

The Early Cretaceous (?Barremian)Afrovenatorabakensis was found to share several derived char-acterswithAvetheropodalackinginotherbasalteta-nurines: in particular, unquestionable presence of amaxillary fenestra (17.1) and an obturator processontheischium(322.1, 323.1).Additionally,thesemi-lunate carpal block (255.1) ispresent inAfrovenator

as it is in avetheropods, but as the carpus is un-known in other non-avetheropod tetanurines, this

derived feature may be synapomorphic for a moreinclusive group (see Discussion below).

The two divisions of Avetheropoda are termed

Carnosauria (Allosaurus andalltaxasharingamorerecent common ancestor with Allosaurus than withNeornithes) and Coelurosauria (Neornithes and alltaxa sharing a more recent common ancestor withNeornithes than with Allosaurus): see PADI AN,HUTCHINSON & HOLTZ (1999) for discussion of thehistoryoftheseterms.Bothcarnosaursandcoeluro-saurs share the derived character state of fusedclavicles (furculae): however, the lack of preserva-tion of clavicles in more basal tetanurines allows forthe possibility of this character state being synapo-morphic for an even more inclusive group (see Dis-cussion below). Twenty four other derived character

states unite carnosaurs and coelurosaurs to the ex-clusion of other theropod clades, but sixteen ofthese cannot be fairly assessed at present in basaltetanurines, as the skeletal elements concerned arenot recovered for these taxa at present.

NODE N. CARNOSAURIA

ALL: 23.1 Lateral surface of nasal participates inantorbital cavity, forming a nasal antorbital fossa;29.1 Nasal recesses; 75.1 Palatines meet medially;83.1 Nuchal crest pronounced; 97.1 Distanceacrossbasaltuberalessthanthetransversewidthofoccipital condyle; 100.1 Basipterygoid processes

short, but not fused to pterygoids; 166.1 Ventral pro-

20

T.R. HOLTZ, JR .

Fig. 6 - Cladogram comparing the left maxillae (in left lateral view) of the ceratosaur Dilophosaurus (modified fromW

E L L E S

, 1984), the basal tetanurine Afrovenator(modified from S E R E N O et al., 1994), and the avetheropod carnosaurMonolophosaurus (modified from Z H A O & CU R R I E , 1993). Arrow indicates thepresence of the rostral ramus of themaxilla,a tetanurine synapomorphy (character 13.1). Not to scale.


17/57

cesses (hypapophyses) on cervicodorsal vertebraepresent as small protrusions.

ACCTRAN: 67.2 Quadrate foramen small and

enclosed within dorsal ramus of quadrate; 113.1Rostral surangular foramen large, in rostrally-oriented depression; 229.1 Humerus/scapulalengthratiolessthan65%;314.1 Pubicforamenper-forating pubic apron in distal half of shaft; 350.2Crista fibularis well developed; 382.1 Pedal digit Iphalanges1+2subequaltopedaldigitIIIphalanx1.

DELTRAN: 24.1 Nasal participates in antorbitalcavity; 60.2 Jugal ventral quadratojugal process ex-tends further caudally than dorsal quadratojugalprocess; 70.1 Quadrate articulation projects wellcaudal to the caudalmost point of the occipital con-dyle.

NODE O. ALLOSAUROIDEA

ALL: 21.1 Pneumatic excavation of the ascend-ing ramus of the maxilla; 26.2 Narial prominencescomprised of paired ridges along lateral edges ofnasals; 33.1 Lacrimal recess, single opening pres-ent; 47.1 Postorbital prominences present; 69.1Quadrate articulation projects well ventral of theventral surface of the maxilla; 116.2 Horizontal shelfon lateral surface of surangular, rostral and ventralto the mandibular condyle, prominent and pendant;179.1 Caudal dorsal neural spines oriented crani-

ally; 315.4 Pubic boot shape triangular (apex cau-dal) in ventral view and angle between shaft andcaudal portion of boot acute.

ACCTRAN: 118.1 External mandibular fenestrareduced; 175.1 Dorsal centrum "hourglass" shaped,central section depth less than 60% height of cranialface; 307.1 Obturator foramen of pubis open ven-trally to form obturator notch.

DELTRAN: 4.1 Premaxillary symphyseal regionU-shaped in ventral view; 67.2 Quadrate foramensmall and enclosed within dorsal ramus of quadrate;77.1 Jugal process of palatine expanded; 81.1 Ven-

tral ectopterygoid recess present and comma-shaped; 181.2 Presacral pleurocoels camellate;257.2 Metacarpal IV present, without phalanges;314.1 Pubic foramen perforating pubic apron in dis-tal half of shaft; 316.1 Caudal portion of pubic bootlonger than cranial portion, but latter present; 350.2Crista fibularis well developed; 382.1 Pedal digit Iphalanges1+2subequaltopedaldigitIIIphalanx1.

NODE P. SINRAPTORIDAE

ALL: 9.1 External nares with marked inset of thecaudal margin; -13.0 Rostral ramus of maxilla ab-sent, rostrodorsal surface of maxilla forms convex

surface from dorsal ramus to ventral margin; 20.1Promaxillary fenestra larger than maxillary fenestra;

56.1 Squamosal flange covering quadrate head inlateral view; 167.1 Neural spines of dorsals equal totwice centrum height; 357.1 Cranial protuberance

on fibula below expansion.ACCTRAN: -57.0 Squamosal does not constrict

lateral temporal fenestra.

DELTRAN: -10.0 Premaxilla and nasal meetsubnarially; -14.0 Rostral ramus absent; 113.1 Ros-tral surangular foramen large, in rostrally-orienteddepression; 175.1 Dorsal centrum "hourglass"shaped, central section thickness less than 60%height of cranial face; -330.1 Femoral head approxi-mately 90 degrees from shaft (head directed hori-zontally).

NODE Q. ALLOSAURIDAE

ALL: -58.0 Jugal does not participate in internalantorbital fenestra; 103.1 Occipital condyle con-stricted neck; 115.1 Rostral ramus of surangulardeep; 178.1 Dorsal column subequal to femurlength; 257.3 Metacarpal IV absent; 317.2 Pubicboot present, greater than 30% as long as pubicshaft; 318.1 Pubic-ischial contact only narrow re-gion; 349.1 Lateroproximal condyle of tibia with con-spicuous waisting between body of condyle andmain body of tibia in proximal view.

ACCTRAN: 16.2 Promaxillary fenestra present,obscuredinlateralviewbyascendingramusofmax-illa; 87.2 Paroccipital process curving ventrally andpendant; 112.1 Dentary caudal depth 150-200%depth of dentary symphysis; -170.0Apices of dorsalneural spines unexpanded; 214.1 Caudal margin ofacromial process of scapula forms abrupt change,perpendicular to blade.

DELTRAN: -113.0 Rostral surangular foramenabsent or very small pit; 118.1 External mandibularfenestra reduced; 229.1 Humerus/scapula lengthratio less than 65%; 307.1 Obturator foramen of pu-bis open ventrally to form obturator notch; 330.2Femoral head greater than 90 degrees from shaft

(head directed dorsally).

NODE R.

ALL: 3.2 Five premaxillary teeth; 199.1 Distalcaudal vertebrae with moderate interlocking, prezy-gapophyses extend more than one half, but lessthanone,centrumlength; -212.0 Distalexpansionofscapula broad,subequalin width to proximal end.

ACCTRAN: 31.2 Lacrimal prominences triangu-lar hornlets; -78.0 Palatine recesses absent; 96.1Basioccipital excluded from basal tuber; 195.1Proximal caudal zygapophyses elongate.

DELTRAN: 10.1 Premaxilla and nasal do notmeet subnarially; 87.2 Paroccipital process curving

21



18/57

ventrally and pendant; -169.0 Scarsfor interspinousligaments terminate at apex of neural spine; 214.1Caudalmargin of acromial process of scapula forms

abrupt change, perpendicular to blade.NODE S

ALL: 36.1 Lacrimal suborbital bar; 41.1 Frontal-frontal suture fused; 50.1 Postorbital-lacrimal con-tactbroad; 53.1 Postorbital suborbitalflange; -100.0Basipterygoid processes moderately long; 159.1Elevationofcranialfaceofmidcervicalcentra; 191.1Caudal pleurocoels present in centra.

ACCTRAN: -166.0 Ventral processes (hypapo-physes) on cervicodorsal vertebrae absent; 183.1Sacral pleurocoels; 236.1 Humeral head offset andemarginated ventrally by groove; 241.1 Deltapecto-ral crest on humerus expanded and offset from hu-meral shaft.

DELTRAN: -14.1 Rostral ramus of maxillashorter rostrocaudally than dorsoventrally.

NODE T.

ALL: -10.0 Premaxilla and nasal meet subnari-ally; 15.2 Maxillary antorbital fossa greatly reducedin size, not extending much beyond rim of the exter-nal antorbital fossa; -21.0 Pneumatic excavation ofascending ramus of maxilla absent; 26.3 Narialprominences knobby rugosities across dorsal andlateral surface of nasals, extending onto dorsalmostsurface of maxillae; 52.1 Postorbital bulbous ros-trally projecting rugosity; 88.1 Basisphenoid, but notparasphenoid rostrum, strongly expanded andpneumatized; 131.1 Lateral surface of teeth withwrinkles in enamel internal to serrations.

ACCTRAN: 5.1 Premaxilla subnarially verydeep, main body taller dorsoventrally than long ros-trocaudally; 8.1 Maxillary process of premaxilla re-duced, maxilla participates broadly in ventralsurface of external naris; -17.0 Maxillary fenestraabsent; 37.1 Prefrontals reduced or absent; 42.2

Frontal-parietal suture on dorsal surface of skullfused, suture indistinguishable; -57.0 Squamosaldoes not constrict lateral temporal fenestra; -60.0Jugal dorsal and ventral quadratojugal processessubequal in caudalmost extension; 62.1 Jugal re-cesses; 68.1 Quadrate dorsal ramus height greaterthan height of orbit; 105.1 Dentary end squared withexpanded tip; 157.1 Cranial cervicals broader thandeep on cranial surface, with reniform (kidney-shaped) articular surfaces that are taller laterallythan at midline.

DELTRAN: -87.0 Paroccipital process orientedmore caudally than dorsally.

As argued by PADIAN, HUTCHINSON & HOLTZ(1999) (see also HOLTZ & BRETT-SURMAN, 1997),

the long established name Carnosauria HUENE,1920 may be conserved as the clade comprised ofAllosaurus and all taxa sharing a more recent com-

mon ancestor with this taxon than with Neornithes.In the present study, the oldest and most basal formin this clade is the Middle Jurassic Monolophosau-rus jiangiZHAO & CURRIE, 1993 of China. However,SERENO et al. (1994, 1996) and SERENO (1997)have proposed that Cryolophosaurus ellioti HAM-MER & HICKERSON,1994isamemberofthisclade.Ifthis hypothesis is correct, the age of this form wouldindicate not only a minimum Early Jurassic date fortheoriginofCarnosauriabutalsooftheCoelurosau-ria (see also discussion of maniraptoriforms below),as wellas suchan early date for the origin of eachofthe various basal tetanurine lineages.

HARRIS (1998) observed that palatines that meetmedially are present in both Sinraptorand Allosau-rus. Such a geometry appears to be present inMonolophosaurus (ZHAO & CURRIE, 1993), but isabsent in basal sauropodomorphs, Herrerasaurus,ceratosaurs, tyrannosaurids, ornithomimids, dro-maeosaurids,andothercoelurosaurs.Inthesetaxa,the palatines remain separated medially by the ros-tral processes of the pterygoids.

Allosauroidea CURRIE & ZHAO, 1993a has beenproposed as the name for the clade comprised of alldescendants of the most recent common ancestorof Allosaurus and Sinraptor (see also PADIAN,HUTCHINSON & HOLTZ, 1999). (SERENO (1997,1998) uses the same name but with the definitionemployed here for Carnosauria). In the presentstudy,thiscladewouldcontainallcarnosaursexceptforMonolophosaurus. HARRIS (1998) observed thatallosauroids were characterized by caudal dorsalvertebrae in which the neural spines were orientedcranially rather than vertically (as in other thero-pods). This orientation is not present in Monolopho-saurus, but is found in Sinraptor, Allosaurus, andAcrocanthosaurus. As such, it is supported here asa synapomorphy of Allosauroidea.

Within the allosauroids, a sister group relation-ship for Sinraptor and Yangchuanosaurus, pro-posed inthe first study ofthe formertaxon (CURRIE &ZHAO, 1993a), is supported here. Similarly, HUTT,MARTILL & BARKER (1996) suggested that theirnewly described Neovenator(oftheWealdenGroupof the Isle of Wight) was closely related to the LateJurassic North American genus Allosaurus, a posi-tion also retained in the present analysis.

The union of Acrocanthosaurus, Giganotosau-rus, and Carcharodontosaurus, first proposed (witha different topology) by SERENO et al. (1996) wasfound in the present analysis. This clade of gigantic

mid-Cretaceous (Aptian-Cenomanian) carnosaurswas called "Carcharodontosauridae" by SERENO et

22

T.R. HOLTZ, JR .


19/57

al. (1996), although (under the phylogenetic taxon-omy used here) this clade is part of Allosauridae(i.e., Allosaurus and all taxa sharing a more recent

common ancestor withAllosaurus than with Sinrap-tor). Although the analysis of CURRIE & CARPENTER(in press) found that Acrocanthosaurus was moreclosely related to Allosaurus than to Giganotosau-rus orCarcharodontosaurus, this study agrees withSERENO etal.(1996)andHARRIS (1998) in groupingthe three giant mid-Cretaceous taxa exclusiveof theLate Jurassic Allosaurus.

ExaminingthecharacterstatesatnodeTrevealsmany derived features also demonstrated by abeli-sauroids.IndeedNOVAS (1997c)hassuggestedthatGiganotosaurus and Carcharodontosaurus areclosely related to the abelisauroid neoceratosaurs,

a very different phylogenetic position from thatfound here. Although there are several cranial syna-pomorphies potentially uniting Carcharodontosau-rus andAbelisauridae (in general) orAbelisaurus (inparticular), additional features unite the African di-nosaur with Giganotosaurus, while further cranialand postcranial character states group this South

American dinosaur with the unquestioned teta-nurine Acrocanthosaurus (see also the Discussionbelow).

In the current phylogeny, Carcharodontosaurus(of Cenomanian age) is the latest known carnosaur.No fossil evidence presently known indicates thesurvival of Carnosauria into the last twenty eight mil-lion years of the Cretaceous.

NODE U. COELUROSAURIA

ALL: -147.0 Cervical centra surfaces amphiplat-yan; 315.3 Pubic boot boat-shaped (pointed crani-ally and caudally) in ventral view and angle betweenshaft and caudal portion of boot acute; 354.1 Fibulaproximalend75%ormoreproximalwidthoftibia.

ACCT RAN: 15.1 Maxillary antorbital fossagreater than 40% of the rostrocaudal length of theantorbital cavity; -31.0 Lacrimal prominences ab-

sent; -45.0 Orbitshaperound; -54.0 Postorbitalfron-tal process sharply upturned; 60.1 Jugal dorsalquadratojugalprocessextendsfurthercaudallythanventral quadratojugal process; -61.0 Jugalrecessesabsent; 70.2 Quadrate articulation rostral to caudal-most point of occipital condyle; 80.1 Subsidiary fen-estra between pterygoid and palat ine; 82.1Endocranial cavity enlarged relative to other dino-saurs, but temporal musculature extends onto fron-tals; 86.1 Paroccipital process with hollow proximalportion; 88.1 Basisphenoid, but not parasphenoidrostrum, strongly expanded and pneumatized; 90.1Three tympanic recesses; 91.1 Branches of internal

carotid artery enter hypoglosseal fossa through sin-gle common foramen; 115.1 Rostralramusofsuran-

gular deep; 121.1 Coronoid extremely reduced orabsent; 153.1 Cervical prezygapophyses flexed;157.1 Cranialcervicalsbroaderthandeeponcranial

surface, with kidney-shaped articular surfaces thataretallerlaterallythanatmidline;-160.0 Midcervicalcentra length about twice diameter of cranial face;190.1 Number of caudals between 30 and 44; 196.1Caudal transverse processes only on caudals I-XVorfewer; 209.1 Distalchevronswithcranialandcau-dal projections, and more than twice as long cranio-caudally as tall dorsoventrally ("boat-shaped"); -215.0 Scapulacoracoid cranial margin smooth;226.1 Forelimb (humerus+radius+manus)/hindlimb(femur+tibia+pes) length ratio greater than 50% butless than 120%; 227.1 Forelimb/presacral vertebralseries length ratio greater than 75% but much lessthan 200%; 232.2 Radius/humerus length ratiogreater than 76%; 245.1 Ulnar shaft bowed cau-dally; 272.1 Length of phalanx 3 of manual digitIII/(sum of lengths of phalanges 1+2 of digit III)greater than 100%; 330.2 Femoral head greaterthan 90 degrees from shaft (head directed dorsally);332.1 Greater trochanter cleft from femoral head;347.1 Incisura tibialis cranialis occupies more than66% of medial surface of proximal tibia; 362.2Astra-galar ascending process craniocaudally reducedand proximodistally tall, with dorsal margin sigmoid("ornithomimoid/albertosauroid condition"); -365.0No pronounced horizontal groove across cranialface of astragalar condyles; 380.1 Metatarsal I plan-tar to medial side of metatarsal II; 386.1 Pedal un-gual II significantly longer than pedal ungual III.

DELTRAN: None.

NODE V.

ALL: None

ACCTRAN: 307.1 Obturator foramen of pubisopen ventrally to form obturator notch; 322.2 Obtu-rator process of ischium separate, triangularshaped; -341.1 Extensor groove in craniodistal re-gion of femur present, but shallow and not conspicu-

ous; 349.1 Lateroproximal condyle of tibia withconspicuous waisting between body of condyle andmain body of tibia in proximal view.

DELTRAN: 4.1 Premaxillary symphyseal U-shaped in ventral view; -10.0 Premaxilla and nasalmeet subnarially; 15.1 Maxillary antorbital fossagreater than 40% of the rostrocaudal length of theantorbital cavity; -45.0 Orbit shape round; -61.0 Ju-gal recesses absent; 70.2 Quadrate articulation ros-tral to caudalmost point of occipital condyle.

NODE W.

ALL: None.

ACCTRAN: 7.1 Medial alae from premaxillaemeetinfrontofvomera; -44.0 Orbitlongerthaninter-

23



20/57

nal antorbital fenestra length; 116.1 Horizontal shelfon lateral surface of surangular, rostral and ventralto the mandibular condyle, prominent and extends

laterally.DELTRAN: 307.1 Obturator foramen of pubis

open ventrally to form obturator notch; 318.1 Pubic-ischial contact only narrow region; 337.1 Fourth tro-chanter of femur little developed; -341.1 Extensorgroove in craniodistal region of femur present, butshallow and not conspicuous; 349.1 Lateroproximalcondyle of tibia with conspicuous waisting betweenbody of condyle and main body of tibia in proximalview; 362.2 Astragalar ascending process cranio-caudally reduced and proximodistally tall, with dor-sal margin sigmoid ("ornithomimoid/albertosauroidcondition"); -365.0 No pronounced horizontal

grooveacrosscranialfaceofastragalarcondyles.

NODE X.

ALL: 333.1 Femoralheadtransverselyelongate.

ACCTRAN: 60.1 Jugal dorsal quadratojugal pro-cess extends further caudally than ventral quadrato-

jugal process; 81.2 Ventral ectopterygoid recesspresent and subcircular; 277.1 Manual ungual re-gion palmar to ungual groove subequal in width toregion dorsal to ungual groove; 282.1 Manual un-gual cross section blade-like, more than three timesas deep as wide; 357.1 Cranial protuberance on fib-

ula below expansion.DELTRAN: -31.0 Lacrimal prominences absent;

-44.0 Orbit longer than internal antorbital fenestralength; 82.1 Endocranial cavity enlarged relative toother dinosaurs, but temporal musculature extendsonto frontals; 115.1 Rostral ramus of surangulardeep; -160.0 Midcervical centra length about twicediameter of cranial face; 181.2 Presacral pleuro-coels camellate; 190.1 Number of caudals between30 and 44; 196.1 Caudal transverse processes onlyon caudals I-XV or fewer; 209.1 Distal chevrons withcranial and caudal projections, and more than twiceas long craniocaudally as tall dorsoventrally ("boat-

shaped"); 227.1 Forelimb/presacral vertebral serieslength ratio greater than 75% but much less than200%; 322.2 Obturator process of ischium sepa-rate, triangular shaped; 332.1 Greater trochanter offemur cleft from femoral head.

NODE Y.

ALL: -149.0 Cervical epipophyses rugosities oncaudal zygapophyses; 154.1 Cervical neural spineslow and craniocaudally short; 186.2 Sacrals III-Vdorsoventrally flattened.

ACCTRAN: -13.0, -14.0 Rostral ramus of maxilla

absent, rostrodorsal surface of maxilla forms con-vex surface from dorsal ramus to ventral margin; -

26.0 Narial prominences absent; 71.1 Quadratepneumaticity well developed; 72.1 Secondary pal-ate well ossified from premaxilla through one-half

the length of the ventral surface of the maxilla; 103.1Occipital condyle constricted neck; 114.1 Caudalsurangular foramen large opening; 369.1 Metatar-sus proportions elongate relative to most othertheropods of same femur length; -376.0 MetatarsalIII dorsal surface shape elliptical.

DELTRAN: 263.1 Metacarpal III clearly shorterthan metacarpal II; 272.1 Length of phalanx 3 ofmanual digit III/(sum of lengths of phalanges 1 + 2 ofdigit III) greater than 100%; 277.1 Manual ungualpalmar and dorsal regions subequal in width; 282.1Manual ungual cross section blade-like, more thanthree times as deep as wide.

NODE Z.

ALL: -235.0 Humeral shaft straight; 257.3 Meta-carpal IV absent.

ACCTRAN: 157.1 Cranial cervicals broader thandeep on cranial surface, with reniform (kidney-shaped) articular surfaces that are taller laterallythan at midline.

DELTRAN: -13.0, -14.0 Rostral ramus of maxillaabsent, rostrodorsal surface of maxilla forms con-vex surface from dorsal ramus to ventral margin; -26.0 Narial prominences absent; 153.1 Cervicalprezygapophyses flexed; -215.0 Scapulacoracoidcranial margin smooth; -221.0 Sternal plates un-fused; -222.0 Sternum carina absent; -245.0 Ulnarshaft bowed caudally.

NODE aa.

ALL: 195.1 Proximal caudal zygapophyses elon-gate.

ACCTRAN: 8.1 Maxillary process of premaxillareduced, maxilla participates broadly in ventral sur-face of external naris; 37.1 Prefrontals reduced orabsent; 118.1 External mandibular fenestra re-

duced; 158.1 Cranial cervical centra extend beyondcaudal extend of neural arch; 168.1Apices of dorsalneural spines expanded transversely to form "spinetable"; 188.1 Sacral neural spines fuse to form lam-ina; -240.0 Humeral ends little or not expanded;284.2 Manual unguals straight.

DELTRAN: 114.1 Caudal surangular foramen alarge opening.

Coelurosauria is a well-supported clade of teta-nurine dinosaurs. Most of the traditional coeluro-saurs of previous studies (GAUTHIER, 1986; HOLTZ,1994;SERENO,1997;SUES,1997),knownfromrela-

tivelycompletematerial,compriseacladeofderivedforms, the Maniraptoriformes HOLTZ, 1996b: these

24

T.R. HOLTZ, JR .


21/57

taxa are discussed below. As in the case of basalTetanurae, however, there exist a number of frag-mentary forms which demonstrate some shared de-

rived characters with Maniraptoriformes comparedto other theropods, but which were found to lie out-side that clade..

Among these, Proceratosaurus bradleyi is theoldest currently known (Bathonian age, Middle Ju-rassic). An alleged therizinosauroid maniraptori-form from the Sinemurian age (EarlyJurassic) of theLower Lufeng Formation, Yunnan, China (ZHAO &XU, 1998) is even older. If substantiated, this wouldindicate that Maniraptoriformes (and indeed, thevariouslineagesoftheoviraptorosaur-Microvenatorclade, Paraves, Compsognathidae, and Arctometa-tarsalia, the ancestors of the basal coelurosaurs,

and Carnosauria) would date back to at least the Si-nemurian. However, the specimen in question is anisolated dentary, and given the resemblance of thedentary of therizinosauroids and basal sauropodo-morphs (the latter common to the Early Jurassic di-nosaurian fauna), this intriguing discovery isgreeted with some caution.

Regardless of the phylogenetic identity of theYunnan specimen, Proceratosaurus sharesuniquely with other coelurosaurs several derivedcharacteristics. The maxillary antorbital fossa (thewall of bone on the rostral portion of the antorbitalfenestra: WITMER, 1997) forms 40% of the total ros-trocaudal length of the antorbital fenestra, a derivedcondition first hypothesized as a coelurosauriansynapomorphy by SERENO et al. (1994, 1996) andsupported here (Fig. 7). In carnosaurs, basal teta-nurines, and ceratosaurs, the maxillary antorbitalfossa represents a much smaller fraction of thislength (and, correspondingly, the internal antorbitalfenestra represents a larger fraction of the totalstructure). Furthermore, the articulation betweenthequadratesandthemandiblelierostraltothecau-dalmost point of the occipital condyle in Procerato-saurus andmostothercoelurosaurs.Incontrast,thequadrate articulation lies at the same point or rostral

to the caudalmost point of the occipital condyle inceratosaurs, basal tetanurines, and carnosaurs.Even in the largest skulled of coelurosaurs, large ty-rannosaurids such as Tyrannosaurus rex(MOLNAR,1991)thearticulationofthequadratesisslightlyros-tral to the end of the occipital condyle. Coupled withthe flexion of the cervical prezygapophyses and thedevelopment of kidney-shaped articular surfaces inthe cranial cervicals, the forward placement of themandibular joint may indicate specializations to-wards greater lateral mobility in the necks of coelu-rosaurs relative to other theropods. In othertheropods, the posterior placement of the mandibu-

larjointmayhaveinterferedwithlateralmotionofthe

neck. Unfortunately, the cervicals ofProceratosau-rus are presently unknown.

The inclusion ofGasosaurus constructus in Coe-

lurosauria is novel to this study. This poorly knownMiddle Jurassic Chinese form is admittedly frag-mentary. Its pelvis retains several primitive featurestransformed in other, later coelurosaurs: the obtura-tor foramen is present, the ischium is footed, and theobturator process appears to be trapezoidal. Never-theless, the hindlimb possesses several coeluro-saur characteristics: a femoral head at an anglegreater than 90 degrees to the femoral shaft; alesser trochanter cleft from the femoral head, and afibula whose proximal portion is greater than 75%the proximal width of the tibia. Given the incompletenature of this taxon, additional character evidence

may reveal it does not belong to Coelurosauria.Pendingsuchdiscovery,however,thecurrentanaly-sissuggeststhatthisformdoesindeedshareamorerecent common ancestor with birds than with Allo-saurus. (P. Currie, pers. comm. 1998, indicates thatas yet undescribed specimens sugge

dinosaur study 2

Documents