Saurischian Monophyly and the Origin of Birds

JacquesGauthier
Museum
of
Zoology.
Vniyersily
of
Michigan,
Ann
Arbor,
Michigan
48109

Andifthewholehindquartersfromtheiliumtothetoes,ofahalf-hatchedchickencouldbesuddenlyenlarged,ossified,andfossilizedasthey
are,
they wouldfurnishus
with
thelast
step of
the
transition between Birds
andReptiles;
for
there
wouldbenothingmtheir
characters to
prevent
us
from refemng themto
the
Dinosauria.
T.H.Huxley1870a

Introduction because they violate a basic tenet of evolutionary theory; as

Michael Ghiselin put it (pers. comm.), "only species speciate,
The origin of birds has long been of interest to evolutionary
generadonotgenerate."Leavingasidetheobjectiontotheidea
biologists. Darwin and his colleagues were well aware of the
that higher taxa can act as ancestors in the usual sense, these
problempresentedbythe"embarrassinggap"betweenbirds
hypothesessufferforotherreasons.
and other amniotes (Desmond 1984). Indeed, because birds
Thepterosaur-birdhypothesisneverreceivedmuchattention,
appearsodifferentfromothertetrapods,itcouldbearguedthat
andthebird-"lizard"propositiondeservedlyreceivedevenless
ne.xt to the question of the origin of man. that of the origin of
in that it was based entirely on plesiomorphic resemblances
birds was one of the most serious impediments to a general
(suchassharingalongtail).Inlightofcurrentknowledge,many
acceptanceofDarwin'sconceptofthetransmutationofspecies.
apomorphies shared by birds and pterosaurs are most parsi-
The precise relationships of birds remain controversial, al-
moniouslyinterpretedas
convergences, which
explains
why
the
thoughthehypothesisthatbirdsarearchosaursisnowwidely
similaritiesbetweenbirdsandpterosaursareseldomeitherde-
accepted. One of the goals of this report is to show that this
tailedorgeneraltobothtaxa.However,theanalysisbelow
controversy has much less to do with the evidence than it does
revealssomeevidencethatcouldbeinterpretedassupporting
withphilosophicalissuesregardinginterpretationsoftheevi-
asomewhatmodifiedversionofOwenandSeeley'shypothesis:
dence.Ibelievethatthesimplestandmostinterestinginter-
namely, that pterosaurs might be most closely related to the-
pretationofthelimitedevidenceavailabletoT.H.Huxleyand
ropods as a whole and thus to the subgroup of theropods in-
hiscontemporarieswas that birds are part of Dinosauria. One
cludingbirds.Martin(1983a.i11cited
) Scleromocli/usasshar-
wouldhavehopedthatworksubsequenttoHuxley'swouldhave
ing"aremarkablyhighpercentageofthefeaturessuggestedto
beendirectedtowardsdeterminingtheprecisepositionofbirds
relatebirdstocoelurosaurs."Martinevidentlywasunawarethat
within dinosaurs. However, for reasons that will be discussed
Huene (1914a, 1948, 1956) thought that Sderomochlus might
below, this was not to be the case. In fact, Huxley's views on
berelatedtopterosaurs.AmoreexplicitrestatementofHuene's
birdoriginswerelargelyabandonedformuchofthetwentieth
hypothesis,includingadditionalcorroboratingevidence,may
century, and the phylogenetic relationships of birds remain a
be found in Gauthier (1984) and Padian (1984).
persistent
controversy.
Amongcurrenthypothesesfortherelationshipsofbirdsamong
amniotes,onlythreehavebeensupportedbysharedapomor- TheThecodont-BirdHypothesis
phies and thus can be considered legitimate hypotheses in a Oneoftheoldestandmostpopularhypothesesfortheorigin
phylogeneticcontext.Simplystated,thesehypothesesare:1) of birds has been that they evolved from "pseudosuchians"
birds are the sister-group of mammals; 2) birds are the sister- (hereinafterreferredtoasthe"thecodont"ancestryhypothesis).
group of crocodiles; and 3) birds are dinosaurs. The last two "Pseudosuchians"compriseabasalgroupwithinthebasalgroup
hypothesesagreethatbirdsarearchosaurs.Thethirdcouldbe of archosaurs, the "thecodonts," from which all other archo-
dividedintocompetingsubhypothesesregardingtheprecisepo- saurs,includingbirds,aresupposedtohaveevolved.Thishy-
sitionofbirdswithinDinosauria,butthisneednotconcernus. pothesishasbeenthechiefrivalofthedinosaurhypothesisfor
The first two hypotheses will be briefly reviewed below, along much of the twentieth century. The "thecodont" ancestry hy-
with a summary of the history of the third alternative, the prin- pothesiswas proposed byworkers whowere
impressed with
two
cipalfocus
of
this
paper. mainobjectionstothedinosaurhypothesis:1)"dinosaurs"were
Earlyworkerssuggestedhypotheses,suchasthatbirdswere eithermorphologicallytoospecializedorstratigraphicallytoo
derived from early pterosaurs (Owen 1875; Seeley 1881), or late to have given rise to birds; 2) convergence could explain
from"lizards"'(Vogt1879,1880),orfrom"thecodonts"(Broom equallywellthesimilaritiessharedby"dinosaurs"andbirds.
1913).Theseandotherhypothesesoftheirkindareill-advised The first objection is not without merit, but it addresses only
partoftheissue.Itemphasizesthedearthofevidencefordirect
The
' useofquotationmarksaroundagroupnamemeansthatinthiscontext filialrelationsofbirdsamongdinosaurs,butitignorestheabun-
that
groupparaphyletic.
is Paraphyleticgroupsincludean
ancestorandsome(but dantevidencesupporting common ancestry.(Justbecause
one's
not
all)
of
Its
descendants (e.g.,
"mammals" withouthumans, or
"lizards"without cousins are not one's ancestors does not mean that one is not
snakes,which are
moreclosely
related
to
some"lizards"than
other"lizards"
are).
Monophyletic groupsincludean
ancestorand
all
of
its
descendants, andthese
are relatedtothem.)
the
only
kindsof
groupsthat
are
consideredlegitimatefor
phylogenetic analysis. Thesecondobjectiontothedinosaurhypothesisismoresub-

111

tie than the first but is equally misguided. It is possible that
birdsand"dinosaurs"areconvergentononeanotherbecause
of their bipedal habits, but this is not the same as having evi-
dencetosupportsuchaclaim.Simplyinvokingthepossibility
of convergence is not enough (e.g., Charig 1976a, b). Both ho-
mologyand homoplasy are
deductiveconcepts
contingent upon
acceptingoneofseveralphylogenetichypotheses(Rieppel1980);
theapomorphiessupportingthepreferredhypothesisarecon-
sideredsynapomorphies and
those
supportingalternativesare
considerednonhomologous(e.g.,convergent).Thus,withoutan
alternative hypothesis of relationship, it is not possible to rec-
ognizeconvergence.Moreover,onemustacceptasharedapo-
morphyasapotentialsynapomorphy;toassumeotherwiseat
theoutsetsimplymakestheconvergenceargumentinvulnerable
to
test.
For example, one is able to reject the proposition that two
taxapossessinganapomorphic,mesotarsalanklejointshared
amorerecentcommonancestorwithinArchosauria;oneneed
onlyobservethatanewlydiscoveredsister-taxonofeitherone
ofthesemesotarsalarchosaursdisplaystheancestralcondition
(assumingnoreversals).Indeed,boththeconvergenceandho-
mologypropositions would agreeon
the
significanceof
finding
amemberoftheingroupwiththeancestralcondition(i.e.,con-
vergence).However, one
if simply
asserts
that
themesotarsal
condition arose convergently, this proposition would be im-
munetonewdata;thatistosay,nomatterhowmanytimes
newlydiscoveredmembersoftheingroupwerefoundtopossess
mesotarsal joints, one could still say that the crucial fossil (i.e.,
the unknown common ancestor) with the ancestral condition
has yet to be found. There is only one simple and informative
explanationforthemesotarsaljointsofthebirdlikearchosaurs,
and that is homology. There is, however, no limit to how com-
plexanalternativeexplanationmightbe;takentoitslogical
extreme, it might as well be argued that all 9,000 species of
birdsacquiredtheirmesotarsalanklesconvergently.
By the early twentieth century an apparent alternative, the
hypothesisof"thecodont"origins,becamefashionable,andac-
ceptingthis
thesis
required accepting
the
convergence argument.
The two objections listed above may have set the stage for
acceptance of the "thecodont" origins hypothesis, but broad
acceptance of this idea arose with the description of one of the
earliestandmostgeneralizedarchosaurs,Euparkcnacapensis.
Broom(1913)andHeilmann(1926)thenhadanarchosaurthat
was so ancient and primitive that it enabled them to derive
birdsdirectlyfromsuchastageinarchosaurevolutionwithout
havingtopassthroughdinosaurs.
If in fact birds were so related, then one would expect to find
evidencethatalldinosaurssharedamorerecentcommonances-
torwithoneanotherthananyofthemdidwithbirds.Heilmann
(1926) may have recognized this necessity, because he noted
thatthedinosaursthenknownlackedclavicles.Thissuggested
thatbirds,whichhaveclavicles,wereplesiomorphiccompared
todinosaursinthisrespect,andmusthavedivergedfromother,
more primitive archosaurs that retained clavicles. The single
pieceofevidencesupportingthe"thecodont"originshypothesis
was refuted in 1936 when clavicles were found in nonavian
dinosaurs(seePartIV,character58).
Moreimportantly,thegreatweightofevidencemarshalled
byHeilmann(1926)providesclearevidencethatbirdsarethe-
ropod dinosaurs. That is to say, even if Heilmann were correct
inconcludingthatnonaviandinosaurslackedclavicles,theweight
ORIGINOFBIRDSANDEVOLUTIONOFFLIGHT
of his evidence still favored the thesis that birds are dinosaurs.
Heilmannultimatelyrejectedtheevidencefavoringthedinosaur
hypothesisandconcludedinsteadthatbirdsdivergedfromother
archosaurspnortotheoriginofdinosaurs.Heilmann'sdecision
appears to have been predicated on a strict interpretation of
Dollo's law of the irreversibility of evolution. Granted, one
wouldnotexpectahummingbirdtoevolvebackintoArchaeop-
teryx. especially by exactly reversing the historical sequence
leadingfromthelattertotheformer,butonestretchesthepoint
by concluding that if clavicles were lost then they could never
be
regained.
Theinadequaciesofthe"thecodont"ancestryhypothesisare
mostclearlyseeninitsclaimthatbirdsarosefromanunknown
member of "Thecodontia." From a phylogenetic perspective,
"Thecodontia"andArchosauriaarediagnosedbythesamesyn-
apomorphies.Thus, these
taxa
are
redundant,and
when one
says that birds evolved from "thecodonts" one is simply reit-
eratingthatbirdsarepartofArchosauria.Tosaythatbirdsare
archosaurswouldbetrueregardlessofwhetherornottheywere
mostcloselyrelatedtocrocodiles,pterosaurs,omithischians,or
any other monophyletic group within archosaurs. Thus, the
"thecodont"ancestryhypothesiscannotbeconsideredalegit-
imatealternativetothemammal-bird,crocodile-bird,ordi-
nosaurhypotheses. Moreover, the"thecodont" ancestryhy-
pothesisISaredherringthathasservedtodeflectinterestin
andexaminationofthelegitimatehypothesesofbirdorigins.
Forexample,asrecentlyas1980,TarsitanoandHechtargued
thatbirdswerederivedfrom"advancedthecodonts"thatlacked
dermal armor and had a birdlike ankle joint. In fact, these and
otherattributesapplytoLagosuchiis.pterosaurs,andalldino-
saurs,includingbirds(seeAppendixA).TarsitanoandHecht's
(1980:176, fig. 9) cladogram correctly depicted birds as part of
thisgroup,butitalsomadeotherclaimsthatwereunsupported
byevidence.Forexample,their"Theropoda"andother"Saur-
ischia"wereshowntobemostcloselyrelatedwithinthisgroup,
with Lagosuchiis as their sister-group, and birds as the sister-
taxonoftheLagositchus-"saur\sch\an"group.Withtheexcep-
tionofbirds,however,theyfailedtodiagnosethesetaxaand
thus support their conclusions. A more accurate depiction of
theirresultswouldhavebeenacladogramwithanunresolved
multichotomy from the level at which the characters "meso-
tarsus"and"unarmored"arose.Thiswouldhaveestablished
that"thecodonts"areaparaphyleticgroup,andthatbirdsare
partoftheunarmored,mesotarsalgroupofarchosaurs.Oneis
still left with the possibility that birds are more closely related
to some members of this group than they are toothers.Rather
than address this issue, however, Tarsitano and Hecht (1980:
177)optedinsteadto"awaitbetterthecodontmaterialandstud-
ies."Insodoing,theyprovidedanexampleofhowtheuseof
paraphyleticgroupsasancestorscandeflectinterestfromevi-
dencerelevant to
hypotheses of
common ancestry.
Inconclusion,becausethe"thecodont"ancestryhypothesis
is at best redundant and at worst a red herring. I recommend
thatItbeignoredbyfutureworkers;discoveringphylogenetic
relationshipsamongthebirdlikearchosaursisdifficultenough
without the further obfuscation afforded by relying on para-
phyletic
groups
ancestors.
as
TheMammal-BirdHypothesis
AccordingtoDesmond(1975,1979,1984),Owen(1841)used
Lamarck'sthreecriteriaofnervous,respiratory,andvascular
GAUTHIER;
SAURISCHIAN
MONOPHYLI'
organizationtoarguethatdinosaurs,likebirdsandmammals,
ascendedtothephysiologicalheightsonNature'sladder.Owen
couldhardlyhavebeendriventosuchaconclusionbythescant
remainsofthethreedinosaursthenknown.Indeed,Desmond
argued that by raising these huge saurians to ordinal status,
Owencouldarguethatthe"reptiliantype"hadlongagoreached
itsapex,andthatithadsubsequently"degeneratedintoasorry
swarm of lizards" (Desmond 1984:1 19). Thus, Owen's Dino-
sauria depended less on evidence than on his abhorrence of
Progressivism;hisulteriormotiveinrecognizingthetaxonwas
"toaddonemorenailtothetransmutationistcoffin"(Desmond
1979:233).
Owen'shypothesisreceivedlittleattentioninthepost-Dar-
winianera,butithasbeenrevivedrecentlybyGardiner(1982).
I will not here offer an extensive criticism of Gardiner's work,
whichdealtwithtetrapodphylogenyingeneralandnotjustwith
the relationships of birds. Rather, I will confine myself to more
generalcriticismsandlettheevidencediscussedbelowspeak
to Gardiner's (1982:227) claim that the "detailed correspon-
dencebetween mammals and
birds
far
outweighs" synapomor-
physchemessupportingalternativehypotheses.
Gardiner intended to discover the relationships of "various
living tetrapod groups ... to one another" (1982:208) through
"consistentpatternsofderivedcharacters"(1982:228).Hisac-
tualeffortwassomewhatlessambitious,however,forhere-
viewedonlythatevidenceconsistentwiththeconclusionsof
somepre-Darwiniancomparativeanatomists.Moreover,heap-
pearstohaveoverlookedmuchoftheevidencepertinentto
tetrapodclassificationgatheredbycomparativebiologistsinthe
post-Darwinianera.Leavingasidetheseoversights,andcertain
misinterpretationsoftheevidence,Gardinermustbeapplauded
forsummarizingtheapomorphiessharedbymammalsandbirds
thatareabsentinotherextantamniotes;hishypothesisatleast
marshalledevidenceinanexplicitlyphylogeneticcontext,unlike
thecaseofthe"thecodont"ancestryhypotheses.
InviewofGardiner'ssedulouspursuitofapomorphiesshared
by mammals and birds, it is curious that he uncovered only
foursharedapomorphies—theformoftheheart,aorticarches,
occipitalcondyle,andpatternoftemporalfenestration—that
mightcontradicthishypothesizedcloserelationshipbetween
mammals and birds. As Gardiner's sources for this evidence
reveal, however, his search for contrary evidence came to a
virtualhaltatthebeginningofthetwentiethcentury.Moreover,
Gardiner claimed that he was unable to find a single unique
feature shared by dinosaurs and birds (1982:222). In view of
theevidencepresentedbelow,whichsummarizesonlypartof
the relevant information already in the literature, it is difficult
tocomprehendhisfailureinthisendeavor.Perhapstheproblem
residesinGardiner'suseoftheterm"unique"?Butthiscannot
bethecasebecauseifoneacceptsthemammal-birdhypothesis,
then endothermy is unique, but if one accepts an alternative,
suchasthecrocodile-birdhypothesis,thenendothermyisnot
unique. Thus, the term "unique" as used by Gardiner reflects
a conclusion, rather than an observation, and it cannot be a
completeexplanationofhisoversight.
PerhapsGardiner,likeOwen,pursuedthequestionofavian
relationshipswithanulteriormotive.Byvirtuallyignoringevi-
dencedeterminableinbothfossilandRecenttaxa,andbystress-
inginsteadonlypartoftheevidencedeterminableexclusively
inextantforms,Gardinermayhaverevealedhisintent:hewas
considerablylessinterestedinreviewingalltheevidencethan
inchastizingpost-Darwinianpaleontologyforwhatheviewed
as its excessive influence on our views of tetrapod phylogeny.
If Gardiner had been more faithful to his avowed goals as a
comparative biologist, he would have made a more complete
surveyoftheliteratureuponwhichtheevidentiarybasisofthe
traditionalviewrests.Theevidencesupportingthetraditional
view that birds are archosaurian diapsids is reason enough to
understand why comparative biologists have long held that,
although birds and mammals are "warm in the palm of the
hand,"theirimmediatecommonancestoramongamnioteswas
not.
TheCrocodile-BirdHypothesis
Fewhavedoubtedthatbirdsandcrocodilesareoneanother's
nearest relatives among extant amniotes. But Walker (1972,
1974, 1977) went further by claiming that birds and crocodiles
weremostcloselyrelatedevenwithinarchosaurs.Walker(pers.
comm.) subsequently rejected this hypothesis in favor of the
theropoddinosaurhypothesis,butthecrocodile-birdhypothesis
hasreceivedfurtherattentionfromWhetstoneandMartin(1979,
1981), Martin, Stewart, and Whetstone (1980), and Martin
(1983a). Martin (1983a) listed thirty characters that had been
usedtosupportthecrocodile-birdhypothesis,butheaccepted
TarsitanoandHecht's(1980)argumentthatnearlyhalfofthem
areplesiomorphicresemblances.Amongthecharactersthathe
didnotrejectowingtoobviousplesiomorphyarethefollowing.
1) Bipartite quadrate articulation, with apomorphic attach-
mentsanteriorlywiththeprooticandlaterosphenoidand
posteriorly with the prootic (in that the quadrate cotylus
liesattheanteriorbaseoftheparocciput);
2)Fenestrapseudorotundumcarryingtheperilymphaticduct;
3)Foramenaerosuminmandible;
4) Periotic pneumatic cavities in dorsal, central, and rostral
positions;
5) Two pneumatic cavities surrounding the cerebral carotid
arteries;
6)Pneumaticquadrate;
7)Unserratedteeth;
8) Tooth crowns short, bluntly conical, and with triangular
profile;
9)Constrictionbetweencrownsandrootsofteeth;
10) Expanded bony root covered with cementum and con-
nectedto
jaw
by
periodontal
ligaments;
11) Oval or round resorption pit;
12)Replacementtoothtiltslabiallyanditsmaindevelopment
takesplacewithinthepulpcavityofitspredecessor;
13) Teeth implanted in open groove at least in young individ-
and
uals;
14)Lingualwallsandseptaofmajortooth-bearingbonesformed
byextensionsofdensebone.
With the possible exception of a foramen aerosum. at least
some of the pneumatic sinuses in the skull, and characters 10,
13,and14amongthoserelatedtotooth-formandimplantation,
thesecharactersdonotappeartohavebeenpresentinArcho-
sauria ancestrally. I have observed a foramen aerosum in the
camosaurtheropodsTyrannosaunisandAlbertosaurusinaform
essentially identical to that of extant birds (and unlike that of
crocodilians). Other archosaurs are reported to have a croco-
dilelikeforamen aerosum. For
example,Wellnhofer (1985)
has

reponeditinpterosaursandJ.Clark(pers.comm.)hasobserved
this structure in the archaic archosaur Eiiparkena. Thus, the
distributionofthischaracteramongarchosaursindicatesthat
it may be the ancestral condition. However, L. Witmer (pers.
comm.)notesthatthisforamenisabsentinDeiuonychus.Fur-
therexaminationofotherarchosaursmustbeundertakento
determinethelevelofsynapomorphy,andsubsequenthistory,
ofthischaracter.
Martin(1983a)notedthatthefenestrapseudorotundumin
birds and crocodiles can be distinguished on developmental
noveltiesfromanalogousstructuresthatevolvedindependently
inmammalsontheonehand and squamates on the other (see
Gauthier, Estes, and de Queiroz, in prep.). Identification of a
fenestra pseudorotundum in fossils is complicated by the ab-
senceofsoftanatomicalanddevelopmentalevidence,andone
is left only with such evidence as can be inferred from the bony
craniumandendocasts.Unfortunately,therearefewwellpre-
paredbraincasespreservedinawaythatcouldprovideanun-
ambiguous conclusion
regardingits
presence
or
absence among
archosaurs. According to Walker (1972, 1974, 1977, pers.
comm.),fenestrapseudorotundaareabsentintheextinctrela-
tivesofcrocodylomorphs,theaetosaursandparasuchians,so
this fenestra appears to be absent in Archosauria ancestrally.
However,RaathandWalker(pers.comm.)identifiedafenestra
pseudorotunduminthetheropodSyntarsiis.Ihaveobserveda
small fenestra in the same part of the ear region that might be
thefenestrapseudorotunduminanothermemberoftheSyn-
/6;/ii«'clade,Z)/7op/;twa»n«(seeCeratosauriabelow).Thisstruc-
turealsoappearstobepresentinonecarnosaur,Acrocantho-
saunis. although it is said to be absent in the carnosaur
rv'ra/!«05ai(n«(WhetstoneandMartin1979,1981).However,
L. Witmer (pers. comm.) also examined Tyraiinosaiinis and
concludedthatthefenestrapseudorotundumwaspresent.Wit-
merandMartin(pers.comm.)useddifferentcriteriaforrec-
ognitionofthischaracter,sofurtherstudywillbenecessaryto
resolvethisissue.Fenestrapseudorotundaarealsoreportedin
troodontid,omithomimid.andcaenagnathidtheropods(e.g.,
Barsbold1983;Currie1986a.b),allofwhichappeartobecloser
tobirdsthanareeitherSyntarsus-Dilophosaumsorcamosaurs
(seeCoelurosauriabelow).WhetstoneandMartin(1981)ques-
tionedtheseidentifications,andnotwithoutreason,foritis
difficulttodistinguishthisfenestrafromothersconnectedtothe
more(e.g.,caenagnathid)orless(e.g.,omithomimid)extensive
periotic pneumatic cavities in the crania of these theropods.
Whetstone and Martin (1979, 1981) were unable to identify a
fenestrapseudorotunduminanankylosauromithischianorin
hadrosaurianomithopodomithischians.However,Sues(1980)
and Gallon (1983) claimed that it was present at least in or-
nithopodsancestrally.Thus,thereappearstobesomequestion
regardingthelevel(s)atwhichthissynapomorphyarosewithin
Archosauria. Finally, characters 1, 2, 4, 5, 9, and 13 do not
appeartosupportanexclusiverelationshipbetweencrocodiles
and birds within Archosauria, because Currie (1986a, b) has
recorded their presence in the theropod Troodon (=Stenony-
chosawus).
Martin(1983a)notedthatnoonehasarguedthattheancestry
of birds lies within crocodiles, only that they both share an
"unknowncommonancestor."Martin(1983a.-111)arguedthat
the crocodylomorph Sphenosuchus is not "as similar to birds
as is the Crocodilia"; the common ancestor of birds and croc-
ORIGINOFBIRDSANDEVOLUTIONOFFLIGHT
odiles,assoproposed,mustliewithinCrocodylomorpha.Un-
fortunately,this
relationship
hasnot
beenmadeexplicit;
birds
haveyettobeplacedwithinacladogramdepictingtheirprecise
relationshipamongcrocodylomorphs.Suchananalysismust
eventually be undertaken by the proponents of the crocodile-
bird hypothesis in order to make their hypothesis more ame-
nabletotest.Forexample,inapreliminaryphylogeneticanal-
ysisofcrocodilesandtheirextinctrelatives,Crocodylomorpha
wasdiagnosedby13synapomorphies(Gauthier1984).Ofthese,
only 7 are present in birds, indicating that if birds are croc-
odylomorphs,the former
diverged from
within
the
latter
after
theacquisitionofthefirstsevensharedapomorphiesbutbefore
theremaining6crocodylomorphcharactersappeared.More-
over,the14additionalcharactersdiagnosticoftruecrocodiles
within Crocodylomorpha must have evolved later still. This
raisesquestionsregardingMartin'sinterpretationsofthelevel
ofsynapomorphyofseveralcharacters,amongthembeingthe
characters in tooth form and implantation cited above. That is
to say, the tooth form unique to crocodiles appeared within
crocodylomorphsonlyaftertheevolutionofananterodorsally
inclined quadratojugal that extends to the skull roof, a partial
secondary palate formed by the maxillae, loss of the ventral
processofthesquamosal,lossofclavicles,ventromedialelon-
gationofthecoracoid,anddevelopmentofacharacteristicwrist
joint involving an elongate and columnar radiale and ulnare.
However,somecrocodylomorphs,suchasTerrestrisuchus(Crush
1984), possess all six of the characters listed above in addition
tothosesharedbybirdsandothercrocodylomorphs,indicating
that it is closer to true crocodiles than are birds. The problem
emerges in the presence of sharply pointed, serrated teeth in
Terrestrisuchusthatappeartobeimplantedandreplacedasin
Archosauriaancestrally.If,asthesedatasuggest,Terrestrisuchus
and true crocodiles are closer to one another than either is to
birds,thenthedentalapomorphiessharedbybirdsandpartof
Crocodylomorphamusthavebeenacquiredconvergently.This
classofproblemsfacesseveralofthesupposedbird-crocodile
characterslistedabove.Inordertorecognizeandcometogrips
withsuchcasesofcharacterdiscordance,however,onerequires
amorepreciselystatedhypothesisthanthatbirdsandcrocodiles
share"someunknowncommonancestor"(Martin1983a.111),
because this would be true in any case under the traditional
viewthatbirdsarearchosaurs.
TheDinosaurHypothesis
Huxley'sdinosaurhypothesisfounditsrootsinHaeckel(1866),
who considered birds to be basically "reptilian," and in earlier
work of his own (Huxley 1864, 1867) in which he concluded
thatbirdswerederivativesof"sauropsidreptiles."Huxleywas
joined in supporting a "dinosaurian" origin of birds by other
comparativeanatomistsincludingCope(1867),Schmidt(1874),
Marsh (1 877), Gegenbaur( 1878), Williston (1879), W. K. Par-
ker(1864), T. J. Parker (1882), Baur(1883, 1884, 1885, 1886),
and Darwin (1872). Indeed, Huxley's proposed "dinosaurian
affinities" of birds gained broad acceptance during the latter
quarterofthenineteenthcentury,evenamongtheranksofthose
who where "at best indifferent to Darwin" (Desmond 1984:
132).
In Huxley's 1869 address before the Geological Society, he
describedthe"ornithicpeculiarities"ofDinosauriaasopening
GAUTHIER:
SAURISCHIAN
MONOPHYLY
up "a very interesting field of inquiry" that inspired him to
devote"allmydisposable leisure during the winter of 1867-8"
todiscoveringcharacterssharedby"dinosaurs"andbirdsthat
arenotalsosharedby"lizards"andcrocodiles.Duringthistime
Huxley also set out to examme critically "the material in the
British Museum in order to ascertain how far the peculiarities
ofMegalosauruswerecommontotheDinosauriaingeneral."
Huxley(1868,1870a,ft)cited 35 characters as "evidence of the
affinity between dinosaurian reptiles and birds." Of these the
following 17 characters survive critical examination in light of
ourcurrentknowledge.
1) The skeleton is hollow and lightly constructed.
2)Thecervicalvertebraeareelongate.
3)Therearemorethantwosacralvertebrae.
4)Thescapulaiselongateandnarrow.
5) The coracoid is short and rounded.
6)Theiliumisprolongedanteroposteriorly.
7) The acetabulum is roofed above by a supracetabular but-
tressthe
of
ilium.
8) The bony contribution of the ilium to the acetabulum is
moreorlessreplacedbymembrane.
9) The ischium and pubis are much elongated.
10)Thefemurhasastronganteriortrochanter.
1 1) The femur has a crest on the ventral face of the outer con-
dylethatpassesbetweenthetibiaandthefibula.
12) The proximal end of the tibia is produced anteriorly into
a strong crest, which is bent outwardly, or towards the
fibular
side.
13) There is a crest on the lateral side of the proximal end of
thetibiaforattachmentofthefibula.
14) The tibia has a fossa distally for the reception of the as-
cendingprocess
the
astragalus.
of
15) The fibula is gracile compared to the tibia, and its distal
endISmuchsmallerthantheproximal.
16)Theastragalusiscompressed,itsarticulationwiththetibia
is concave proximally and it has a convex, pulleylike distal
surface,andthedisparityinsizebetweenthetibiaandfibula
is also reflected in the astragalus being much larger than
the
calcaneum.
17) An "ascending process" (the intermedium) more or less
tightlyconnectstheastragalusandtibia.
Onlysomeofthesecharacterscannowbeconsideredtohave
arisen in the immediate common ancestor of Dinosauria (see
Appendix A). Nevertheless, all of them appeared within the
subgroupofarchosaurscontainingbirdsandthustheyspeak
against the crocodile-bird hypothesis. At first glance, the ex-
planatorypowers of
the
crocodile-bird
versus
dinosaur
hypoth-
esesdonotappeartodiffergreatlyinthattheformerissupported
by 14 shared apomorphies and the latter by 17. One must bear
inmind,however,thatthedinosaurhypothesiswasformulated
over a century ago with a fraction of the evidence currently
available (see Appendix A). Although the crocodile-bird hy-
pothesisis
basedoncurrentknowledge, therewould
have been
no reason to accept it over the dinosaur hypothesis as it stood
in 1870, and there is even less reason to do so now. Huxley
(1870/1) mistakenly claimed that the absence of clavicles was
diagnostic of Dinosauria, but this did not deter him from hy-
pothesizingbird-"dinosaur"
a affinity.
doubt
I thatheconsid-
eredthematterinthisway,buthewascorrectinhisjudgement
to the extent that it would have been simpler to accept the
reappearanceofclaviclesinbirdsthantoacceptconvergence
as an explanation for each of the seventeen characters listed
above.
InthediscussionfollowingHuxley'spresentationbeforethe
Geological Society in 1869, Seeley remarked that he "thought
it possible that the peculiar structure of the hinder limbs of the
Dinosauriawasduetothefunctionstheyperformedratherthan
to any actual affinity with birds" (Huxley 1870a:31). With this
simpledeclarationarosetheissuethatwastobecomethebane
ofHuxley'shypothesis.Seeleysuggestedconvergenceasanal-
ternativeexplanation for
the
apomorphies shared by
"dino-
saurs"andbirdswithoutproposingtherequisitealternative
hypothesisofrelationship.
Theassertionofconvergencewastobeheardtimeandagain
intheensuingyears(e.g.,Mudge1879;Dollo1882,1883;Dames
1884;Parker1887;Furbringer1888;Osbom1900).Whatmade
mattersworsewastheriseofthe"thecodont"ancestryhypoth-
esisintheearlypartofthetwentiethcentury(e.g..Broom1913;
Heilmann 1926); this hypothesis appeared to dignify the oth-
erwisebaldassertionof convergence. The
error
was
compound-
edevenfurtherwhentheconvergenceargumentwasapplied
not only to bird origins but to the question of dinosaur mono-
phyly as well. Pandora's box had been opened; if an erect, bi-
pedalpostureandgaitaroseconvergentlyindinosaursandbirds,
thenwhatwastoforbidmultipleevolutionsofsuchadaptively
significantcharacters?Thedemiseofthedinosaur-birdhypoth-
esiswenthandinhandwiththedemiseofdinosaurmonophyly;
ifoneacceptedthenonparsimoniousreasoningbehindthehy-
pothesisthat
birds
were "derivedindependently"from"thec-
odonts,"thenwhynotaccepttheevenlessparsimonioushy-
pothesisthateachoftheremainingdinosauriangroupswere
"derivedindependently"from"thecodonts"aswell?Iseenoth-
ingthatwouldforbidthemultipleevolutionof"dinosaurian"
characters,butwhatevidenceindicatedthatthisdidinfacttake
place?Althoughafewresearchersadvocatedthedinosaurhy-
pothesis(e.g.,
Boas1930),the"unknown ancestor,"coupled
withtheconvergenceargument,enthralledmostsystematists
(e.g., Simpson 1946; de Beer 1954; Romer 1966).
Therewasrenewedinterestintheoriginofbirdsinthe1970's,
beginningwiththepublicationsofGalton(1970),Walker(1972),
andOstrom(1973).Galton(1970)stressedoneofthecharacters
initiallyemployedbyHuxley(1870a,b)—areversedpubis—to
suggestthatbirdswere"derived"fromomithischiandinosaurs.
UponintroductiontoOstrom's(1973)evidence,Galtonrejected
the omithischian hypothesis (e.g., Bakker and Galton 1974).
Martin(1983a)suggestedthatthepresenceofapredentarybone
inHesperomithesanditsinferredpresenceinIchthyornismay
be further evidence of an omithischian-bird relationship. The
absence of a predentary bone in Archaeopteryx and all other
birdsspeaksagainstMartin'sconclusion;evenifomithischians
werethesister-groupofbirds,itwouldstillbesimplertoaccept
apredentaryboneasdiagnosticofataxonincludingonlyIchthy-
ornis
Hesperomithes.
and
Just as Eiiparkeria spurred interest in the "thecodont" an-
cestryhypothesis,Ostrom'sfindsofanewandstrikinglybirdlike
theropod,DeinonychusaiUirrhopus{OstTom1969a.b.1974/),
1916b), revitalized the dinosaur hypothesis. Ostrom's (1973,
1974a,1975a,b.1976a)hypothesized"coelurosaurian"ances-
tryofbirdscouldbeviewedasanextensionofHuxley'spro-

posal,sincethetheropodMegalosaurusfiguredprominentlyin
Huxley'sarguments.Thetwohypothesesdiffered,however,in
thatHuxleylookedat"dinosaurs"ingeneralas"intermediate"
between"reptiles"andbirds,whileOstromsoughttheancestry
ofbirdsamong"coelurosaurtheropods"aloneandskirtedthe
questionofdinosaurmonophyly.
Early workers found favorable comparisons between birds
and"coelurosaurs,"andthesecomparisonswereablyreviewed
byHeilmann(1926).AlthoughHeilmannhasbeenconsidered
thechampionofthe"thecodont"hypothesis,hisclosingcom-
ment(1926:185)uponfinishinghiscomparisonsbetweenbirds
and "coelurosaurs" was: "We have therefore reasons to hope
that in a group of reptiles closely akin to the Coelurosaurs we
shallbeabletofindananimalwhollywithouttheshortcomings
hereindicatedforabirdancestor."
After Heilmann (1926), a few authors advocated a special
bird-"coelurosaur" connection (e.g., Lowe 1935, 1944a. b:
Holmgren1955).AsOstrom(1976a)noted,however,theirideas
receivedlittleattentionbecausetheysaddledthemselveswith
burdens—mostnotablyavianpolyphyly—thatinspiredothers
to reject their ideas out of hand (e.g., Simpson 1946; de Beer
1954).Ostrom'sresurrectionofthe"coelurosaur"hypothesis
metwithwideracceptance(e.g.,BakkerandGalton1974;Thul-
bom 1975, 1984; Thulbom and Hamley 1982), although his
hypothesiswasnotwithoutcritics(e.g..Walker1977;Tarsitano
and Hecht 1980; Martin 1983a). Many of the criticisms were
centeredontheinterpretationofcertainanatomicaldetailsin
imperfectlypreservedfossils.Perhapsmoretroublingwasthe
continued reliance on the part of Ostrom's critics on the "the-
codont" hypothesis and
its
henchman,convergence.
Ostrom considered birds to have "evolved from coeluro-
saurs."However, in
a
phylogenetic perspectiveOstrom'suse of
"coelurosaur" is no more informative than the name Therop-
oda,becausebothnamesarediagnosiblebythesamesynapo-
morphies(seeIntroductiontotheBasicTaxa,below).Thus,the
"coelurosaur"hypothesiswasnomorepreciselystatedthanwas
the crocodile-bird hypothesis, in that it merely claimed that
birdsand"theropods"sharedanunknowncommonancestor.
Infailingtolethishypothesisaccuratelyreflectthestructureof
his evidence, Ostrom opened himself to a variety of criticisms,
mostofwhichwereonlymisinterpretationsinvitedbytheobfus-
cationofparaphyly.Itisinappropriatetoconsidereachofthese
criticismsatthistime,andthereaderisreferredtothecharacter
discussionsbelowforexamplesoftheproblemsstemmingfrom
treatingparaphyleticandmonophyletictaxaasiftheypossessed
the same properties (i.e., a unique history involving origin, di-
versification,
extinction,
etc.).
AnotherresultofOstrom'svagueconclusionsregardingthe
relationshipsofbirdstoothertheropodswasthatsomeworkers
mistakenlyconcludedthatbirdsandtheropodsmightbesister-
groups (e.g., Tarsitano and Hecht 1980; Thulborn and Hamley
1982). The imprecision in Ostrom's proposition was rectified
by Padian (1982), who extracted the relevant evidence from
Ostrom's works and arrayed it in an explicitly phylogenetic
context.Padian'sprovisionalreanalysisdemonstratedthatbirds
aremorecloselyrelatedtosome"coelurosaurs"thantheyare
to others; the conclusion that birds and theropods are sister-
groupscouldnotbeextractedfromOstrom'sevidence.Onthe
contrary,birdsaredeeplyimbeddedwithinTheropoda,justas
ORIGINOFBIRDSANDEVOLUTIONOFFLIGHT
humansaredeeplyimbeddedwithinthephylogeneticnexusof
Mammalia.
Bakker and Galton (1974) integrated the hypotheses of Os-
tromandHuxleyandresurrectedtheconceptofdinosaurmono-
phyly.Theirmethodofanalysiswassomewhatmorerigorous
than the Simpsonian "evolutionary systematics" of Ostrom,
althoughtheywerealsohamperedbysuchparaphyletictaxaas
"thecodonts"and"prosauropods."TheessenceofBakkerand
Galton's(1974)argumentwasasfollows.
1) Dinosaurs (including birds) shared apomorphies not also
sharedby"thecodonts"andwerethereforemonophyletic.
2) "Prosauropods" were a primitive grade of dinosaurs that
bridged the gap between omithischian and "saurischian"
dinosaurs.
3) All dinosaurs, like modem birds, were active endotherms;
thebehavioralandphysiologicalapomorphiessharedbydi-
nosaursare important;
Dinosauriashouldtherefore
beac-
corded
status.
Class
BakkerandGalton'spaperinspiredconsiderablecontroversy
(e.g.,Thulborn1975;Chang1976/>).Themosttellingcriticisms
were directed to the second and third points. To be sure, the
linkbetweenBakkerandGalton'sevidenceandtheconclusion
of endothermy was a bit tenuous. The question of categorical
rank of Dinosauria is trivial; current methods of establishing
taxonomic rank rely entirely upon the authority of a system-
alist's subjective notion of what constitutes an ideal Class, Or-
der,etc.,andsuchPlatonicandtypologicalnotionsareanach-
ronisticatbest.Charig's(1976/i)criticismsofBakkerandGalton's
second point were for the most part trenchant and insightful.
However, although some of Thulborn's (1975) and Charig's
(1976/')criticismsoftheevidencesupportingdinosaurmono-
phylywerefactuallyaccurate,forthemostparttheydidnot
bearonthequestionofdinosaurmonophyly.
MostcriticismsoftheresurrectedDinosaunahypothesisfell
intooneofthreeclasses:1)saurischiansandomithischiansare
toodifferenttobemonophyletic,2)someoftheallegeddinosaur
characters were in fact present in some "thecodonts," and 3)
therestofthecharactersarefunctionallyrelatedandtheycould
havebeenacquiredconvergently.Thefirstclassofcriticismsis
beside the point. One may be different from one'ssiblings,yet
stillsharethesameparents;thedifferencesbetweenbatsand
whalesdonotprecludetheirbeingmammalsanymorethanthe
differencesbetweenthepostdentarybonesofOphiacodonand
Homo indicate nonhomology. Criticisms in the second class
only indicate that some "thecodonts" are closer to dinosaurs
(includingbirds)thanareothers;thus,"Thecodontia"ispara-
phyletic.Such evidence maynotsupport dinosaurmonophyly,
butitdoesnotspeakagainstiteither.Thethirdclassofcriticisms
leveled at Bakker and Galton's evidence for dinosaur mono-
phylyisexemplifiedbyaquotefromChang(1976/).'79):"Most
ofthesedinosauriancharacter-stateswereobviouslyadapta-
tionstothefullyimproved('fullyerect')postureandgaitofthe
dinosaurs . . . which could easily have evolved several times
over,inslightlydifferentways,inresponsetosimilarfunctional
requirements."
Wehavenowcomefullcircle;theexchangesbetweenHuxley
and Seeley. and between Bakker and Galton on the one hand
and Chang and Thulbom on the other, demonstrate that the
GAUTHIER:
SAURISCHIAN
MONOPHYLY
dinosaur controversy has not altered on this issue in over a
century. This controversy has nothing to do with evidence, al-
thoughitmustbeadmittedthatuntilHennig1966,itwasnot
widelyunderstoodthatancestry,ratherthanoverallsimilarity
ordissimilarity,mustbethebasisofphylogeneticclassification.
Consequently,thereislittletobegainedfromtakingtheposition
that the problems will be solved by finding more fossils. There
are, after all, many more fossils now available than there were
in Huxley's time, and they still have not forestalled the old
objections. The implications are clear: if our understanding of
bird origins is to progress, we ought to rid ourselves of typo-
logicalthought,namely,paraphyletic "Thecodontia";
reserve
hypothesesofconvergenceforcasesofcharacterdiscordance,
withoutwhichthereisnothingfortheconceptofconvergence
to explain; and remember that although the world has no ob-
ligationeithertobesimpleorinformative,ourhypotheseshad
better be both (Beatty and Fink 1979). By following these pre-
cepts,issuesinarchosaurphylogenycanbebroughtintosharper
focus.
Knowledgeoftheropodanatomyhasincreasedvastlysince
BucklanddescribedMegalosaunisin1824.Unfortunately,this
knowledgehasnotbeentranslatedintoadeeperunderstanding
oftheropodphylogeneticrelationships.Thiscircumstancere-
flects,inpart,thatTheropodahasyettobediagnosedonthe
basisofsynapomorphies.Indeed,mostworkershavebeencon-
tenttodefine"theropods"as"primitivedinosaurs"or"car-
nivoroussaurischians"; which
is
to
say
that
theropodsare
those
saurischiansthatarenotsauropodomorphs,andsaurischians
arethosedinosaursthatarenotomithischians.Inordertobring
thesetaxaintothephylogeneticsystem,andtherebyaddressthe
question of the phylogenetic relationships of birds, it will first
benecessarytodetermine which, if any, phylogenetic entities
withinDinosauriamightbepartsofSaurischiaandTheropoda.
Materials and Methods
For the most part, the specimens examined are listed in the
IntroductiontotheBasicTaxa.However,thematerialsconsti-
tutingthecoreoftheanalysiswereSegisaiinis.Dilophosaurus,
castsoftheskullofCeralosawus.andseveralundescribedspec-
imensinthecollectionsoftheUniversityofCaliforniaMuseum
of Paleontology (UCMP); Coelophysis in the collections of the
Museum of Northern Arizona (MNA), UCMP, and American
Museum of Natural History (AMNH); Procompsognathus in
thecollectionsoftheStaatlichesMuseumfiirNaturkunde,Stutt-
gart(SMNS);CompsognathusinthecollectionsoftheBayer-
ischeStaatssammlungfiirPalaontologieundhistorischeGeo-
logic,Munich(BSP);andthe
comparative osteological collections
ofextantbirdshousedintheUCMP,theMuseumofVertebrate
Zoology, University of California, Berkeley (MVZ), the Cali-
forniaAcademyofSciences(CAS),andtheUniversityofMich-
iganMuseumofZoology(UMMZ).Thesedataweresupple-
mentedbyphotographsandnotesonvirtuallyeverytaxon
referredtononaviantheropodscompiledbySamuelWellesand
RobertLong.Informationoncharacterontogenieswasderived
fromtheliteratureandfromclearedanddouble-stainedseries
oi Alligator. Gallus. and Podiceps occidentalis (UMMZ), in ad-
ditiontolateembryosandjuvenilesofafewpasserines(CAS
and pers. coll.), and the hindlimbs of three tinamou embryos
(MVZ). These data were supplemented by that derived from
skeletons of a hatchling Plerocnemia (MVZ) and a juvenile Ap-
teryx(UMMZ).
Forthepurposesofthiswork,ontogenieswillbedividedinto
fourstages:embryos,juveniles,subadults,andfullyadult/ma-
tureindividuals.Embryosrefertoprehatchingindividuals;ju-
venilesinclude
stagesfromhatchlings to
nearly
full-grownspec-
imens(i.e.,subadults);subadultsarethoseindividualsthatare
near to maximum size as is indicated by some, but not all, of
thedevelopmentaleventsmarkingthecessationofgrowth(i.e.,
senility).Asusedhere,theterm"fullyadult"makesnoreference
to sexual maturity, which may or may not be coincident with
this stage in skeletal ontogeny. Individuals that have reached
theterminalstageofontogenyarereferredtovariouslyasfully
adult, fully mature, or as having attained maximum adult size.
Subadultsmayalsobenearoratmaximumadultsize,butthey
do not display the full suite of developmental events in the
skeleton that mark the cessation of growth. The cessation of
growthintheropodsmayberecognizedbythefollowingevents
in the skeleton: fusion between the axial intercentrum and at-
lantal centrum, and fusion of this compound structure to the
axial centrum; fusion of neural arches to centra; fusion of the
vertebral components of the sacrum; full ossification of distal
tarsalII;and,atleastinnonaviantheropodsasidefromratites,
fusion of the scapula and coracoid. Some of these events may
precedeothers;however,anyspecimeninwhichallofthemare
presentishereconsideredtohaveattainedmaximumadultsize.
Bythisdefinition,therearefewnonaviantheropodfossilsthat
haveachievedmaximumadultsize(e.g..thetypespecimensof
SyntarsiisrhodesiensisandCemtosawusnasiconus).Thereare
alsofewspecimensthatcouldbeconsideredjuveniles,andeven
thesefewrepresentapproximatelyhalf-grownindividuals(e.g.,
thetypespecimenofCompsognathuslongipesand.contraryto
Howgate's1984view,theEichstattspecimenoiArchaeopteryx
lithographica).
ThemethodsemployedhereareessentiallythoseofGauthier
et al. (in prep.). Following the procedure of Maddison et al.
(1984). at least two outgroups were employed to identify 84
apomorphiccharactersdistributedamong2ormoreofthe17
basictaxa(18includingoutgroup)thatwerethesubjectsofthis
analysis (see Introduction to the Basic Taxa, Fig. 7, and Ap-
pendixB).ThesedatawerethenanalyzedwithSwofford'sPhy-
logeneticAnalysis
UsingParsimony (PAUP)program installed
in the University of Michigan's Terminal System. The analysis
wastwo-part;thefirstrunincludedonlythecomparativelywell
knowntaxawhoseinterrelationshipsweretheprincipalfociof
thisanalysis(viz.,Omithischia.Sauropodomorpha,Ceratosau-
ria, Camosauria, Omithomimidae, Deinonychosauria, and
birds),andthesecondrunincludedalltaxa,includingthosefor
whichwehaverelativelylittleinformationowingtoincomplete
preservation.Thefirstrunyieldedasinglecladogramdepicting
themostparsimoniousinterpretationofthephylogenyofthe
seven well-known dinosaur taxa (Fig. 8). This cladogram re-
quires94evolutionaryeventstoaccountforthedistributionof
84apomorphiesamongtheseventaxa,thusyieldingaconsis-
tencyindexof89%;thatistosay,thesingletreesoobtained
represents a highly corroborated hypothesis of relationship.
However,thesecondrun,whichincludedboththebetterknown
andlesswellknowntheropods,resultedinnumerous,equally

parsimonioustrees.Nevertheless,thesister-grouprelationship
among the seven well-known taxa were consistent across all
possible trees. The multiple trees obtained in the second run
resulted from the missing data in the 10 less well known taxa.
For example, let us say that Coelurosauria includes only three
taxa:deinonychosaurs,birds,andHulsanpesperlei.Onlythree
of the 84 characters under review were determinable for Hul-
sanpes,andalthoughthesedataindicatethatitisacoelurosaur,
they are moot on the point of its precise affinities within this
taxon. The PAUP analysis is so constructed that it considers all
possiblepositionsthatHulsanpescouldhaveamongcoeluro-
saurs (viz., it could be the sister-group of birds, deinonycho-
saurs,oradeinonychosaur-birdgroup)andthusyieldsthree
equallyparsimoniouscladogramsforthesethreetaxa.Ofcourse,
noneofthesecladogramsisactuallysupportedbyanyevidence
observableinHulsanpes.Thus,whenoneconsiderstheadded
possibilities allowed by the very incomplete remains of the 10
lesswellknowntaxa,thesourceofthenumerouspossibletrees
becomesapparent.However,twocriticalpointsmustbeborne
inmind:first,therelationshipsamongtheseventaxaforwhich
thereismorecompleteinformationremainedunchanged,and
second,theoverwhelmingmajorityofpossibletreeswereinfact
uninformative.Consequently,theywerecollapsedintoasingle
consensustreeincorporatingmultichotomiesstemmingfrom
thelevelssupportedbyobservablecharacters(Fig.9).Thecon-
sensustreerequired99evolutionaryeventstoaccountforthe
distnbution of 84 apomorphies among the 17 basic taxa, thus
yieldingaconsistencyindexof85%.
Following Gauthier et al. (in prep.), the classification used in
this work was constructed according to the following five con-
ventions.
1) Only monophyletic taxa including an ancestor and all of
its descendants are recognized, and in no case will a demon-
strablyparaphyletictaxonbeconsideredinthisanalysis.An-
cestry,ratherthanoverallsimilarity,mustbethebasisfora
phylogeneticsystem.Thesingleexceptiontothisconventionis
the
metataxon.
2) A new category, the metataxon, is employed for taxa for
which there is no positive evidence for or against recency of
commonancestry.FollowingthesuggestionofM.Donoghue,
metataxaareprovisionallyallowedintheclassificationandtheir
uncertainstatusisdenotedbyanasteriskfollowingthename.
For example, as will be argued below, the monophyly or para-
phylyofthefivefossilskeletonsandasinglefeatherimpression
referredtotheearliestbird.Archaeopteryxlitliographica*.has
yet to be firmly established, and it is therefore accorded meta-
status.
taxon
3)Certainwidelyusednamesarestandardizedbyrestricting
themtotaxawhosemonophylyamongextantamniotesisfirmly
established. Accordingly, Archosauria is standardized by lim-
itingthistaxontoallthedescendantsofthemostrecentcommon
ancestor of extant birds and crocodiles. And Aves is likewise
restricted to all the descendants of the most recent common
ancestorofRatitae,Tinami,andNeognathae.
4) Although the spelling of current taxonomic names is re-
tained,noformalcategoncalranksarerecognizedandhierar-
chicalrelationships within
taxa
are expressed insteadby
branch-
ingdiagrams.CategoricalrankssuchasClass,Order,Family,
and Genus, will not be recognized in this work.
5)Excepttopreservebinomials,noredundantnameswillbe
ORIGINOFBIRDSANDEVOLUTIONOFFLIGHT
recognized.Thus,althoughArchaeopteryxlithographica*isre-
tained,redundanttaxaconveyingnothingfurtheraboutphy-
logenetic
relationships,
such
Archaeopterygidae,
as Archaeop-
terygiformes,orSaururaewillbeignored.
Introduction to the Basic Taxa
Thissectiondefinesanddiagnosesthetheropodtaxathatare
the subjects of the present analysis (For diagnoses and defini-
tionsoftheotherbasictaxa,SauropodomorphaandOmithi-
schia,andtheoutgrouptaxa,seeAppendixA).Thesediagnoses
are not definitive; the basic taxa are assumed to be monophy-
letic,andthecharactersarelistedmerelytoshowthatthereis
at least some basis for the assumption of monophyly in each
case.
Severalworkershavedescribedtheconventionalgroupings
of"Camosauria"forlargetheropodsand"Coelurosauria"for
smalltheropodsasinadequateinviewoftheobservedvariation
among Theropoda (e.g., Colbert and Russell 1969; Ostrom
1969ft).Accordingly,modemclassificationsemphasizelessin-
clusivetaxa,typicallyrankedas"families."andtheseunitswill,
for the most part, serve as the basic taxa of this analysis. In a
phylogenetic context, "Coelurosauria" and "Theropoda" are
redundantinthattheyhavetraditionallybeendiagnosedbythe
samesynapomorphies.Huene(\9\Ab)originallydefined"Coe-
lurosauria"onthe
basis
of
plesiomorphicresemblances, such
as their small size and long necks. In other words, they are
theropods that are not camosaurs. However, because Huene
basedtheconcepton"coelurids,"someofwhichhavesynapo-
morphiesof particular
a subgroupof
Theropoda that
includes
birds, Coelurosauria will be retained in a modified form (see
below).
A principal goal of this work is an examination of the phy-
logeneticrelationshipswithin
Theropoda, with
particular
atten-
tiontotherelationshipsamongthebetterknowntheropodshere
included in Ceratosauria (n. comb.), Camosauria (n. comb.),
Ornithomimidae(n.comb.),Deinonychosauria,andbirds.For
thesakeofcompleteness,thedistributionofthecharactersdis-
cussedwillalsobenotedinlesswellknowntaxasuchasPro-
compsognathusiriasicus*.Liliensternusliliensterni*,Ornitho-
lesteshermanni*.Coelurusfragilis*.Compsognathuslongipes.
Microvenatorceler*,Saurormtholesteslangstoni*,Hulsanpes
perlei. Elmisauridae*, and Caenagnathidae. As the use of the
asteriskindicates,severalofthesetaxahavenotbeenadequately
diagnosed;theymayindeedbedifferentfromothertheropods,
butuntiltheyarediagnosedonthebasisofcharactersrelevant
tothequestionofmonophyly(i.e.,synapomorphies),theirstatus
as phylogenetic entities must remain suspect. The position of
thesegenerallypoorlyknowntaxaarenotsocriticaltothegoals
of this analysis, and the reader is referred to the literature for
more information (Marsh 188 la,- Osborn 1903, 1917; Ostrom
1970, 1978, 1981; Sues 1978; Osmolska 1981, 1982; Barsbold
1983;Welles1984).
The more inclusive basic taxa may be newly recognized, or
they may differ in diagnosis and content from concepts em-
ployedbyotherresearchers.SomeofthesynapomorphiesI
considerdiagnosticofthebasictaxacouldonlyhavebeenrec-
ognizedassuchafterthecompletionoftheanalysis;theyare
added now for the sake of completeness and in no case would
the monophyly of any of the basic taxa depend on such deter-
minations.
GAUTHIER:
SAURISCHIAN
MONOPH'lLV

Elmisauridae*Osmolska,1981 son'sobservations
indicate
that
Microvenator*,
elmisaurids*,
andcaenagnathidsmightbemonophyletic;suchpossibilities
Temporal Range.
—late
Cretaceous.
Included
fAA.
J Chiroslenoles
— pregmcilis*.
Matrophalangia
canadensis*,
and shouldalwaysbebomeinmindwhendealingwithmetataxa.
Elmisaunts
rants*.
CeratosauriaMarsh,1884ft(n.comb.)
Diagnosis. —Based on personal observation of Macropha-
langia*andCliirostenotes*,andonpublisheddescriptionsof Temporal Range.—late Tnassic
lo
late
Jurassic.
these taxa in Gilmore (1924), Sternberg (1932), and Osmolska Included
JAXA.
Ceratosaurm
— nasicorms.
Synlarsus
rhodesiensis.
Coelophysis
bauri.
Scgisaurus
halh*.
Sarcosauniswoodi*.
Ddopliosauniswelherelli
(including
(1981).Thesetaxaarerepresentedbyscantandoftennoncom- UCMP37302,37303,
and77270),andsome undescnbed formsrepresentedby
plementaryremains.Indeed,noneofthereferredtaxahavebeen UCMP 129618(referred lo
Coelophysis by
Padian,m
press).
UCMP 128659,
and
adequately diagnosed, and they are so poorly known that Os- MNAV.2623(referredtoSyntarsusbyT.Rowe,pers.comm.).
molska(1981)suggestedthattheymightbesynonymous.Ac- Diagnosis. — The initial basis for recognition of the mono-
cordingtoCume(pers.comm.),however,Elmisaunts*and phylyofthistaxonstemmedfromWelles's(1984)observation
Cliirostenotes*aresympatricinAlbertainthelateCretaceous.
that one specimen referred to Dilophosaiirus (UCMP 77270)
Currie and Russell are in the process of describing a partial
possessed a uniquely modified trochanteric shelf (=modified
skeleton,includinghandsandfeet,ofChirostenotes*.Prelimi-
anteriortrochanter:seephotographofSarcosauniswoodi*in
naryresultsindicate that
Chirostenotes*
and
Macrophalangia*
Charig1976ft).T.RowelaterobservedthisapomorphyinSegi-
mightbebasedonthesamespecies.Thisconclusionistentative
saurus*,andwehavesinceobservedthisandothersharedapo-
becausethereappeartobetwo"morphs,"anditisnotyetclear morphiesin all
taxa
here
included in
Ceratosauria.
ifthesemorphsresultfromsexualortaxonomicdifferences.To
Thepresenceofthetrochantericshelfinonlysomeceratosaur
furthercomplicatematters,eachoftheapomorphiessharedby
specimens is perplexing. However. Colbert, Rowe, and Raath
elmisaurids* is either matched in some other theropods, or it
(pers. comm.) have separately observed the presence of two
couldbeconsideredpartofatransformationseriesthatistaken
femoral types among the large series of Coelophysis and Syn-
to extreme in some group of theropods (e.g., proportions of
tarsus.arobustforminwhichthetrochantericshelfisdeveloped
manaldigitIapproachthoseofomithomimids).Moreevidence
in the form characteristic of ceratosaurs, and a gracile form in
will be necessary to address these issues, but for the present whichthetrochantericshelfislessmodifiedandmorelikethat
Osmolska'sElmisauridae*willbeacceptedasametataxon.Fol-
seen in dinosaurs ancestrally. Dimorphism in femoral form,
lowingisalistofapomorphiessharedbyelmisaurids*;assug- alongwithotherdifferencesinproportions,havebeenattnbuted
gestedabove,theseapomorphiesmayormaynotprovetobe tosexualdimorphism.
synapomorphies.
Although it has appeared elsewhere in theropods, another
Metacarpal I elongate and slender, relatively elongate first synapomorphyofCeratosauriaisthefusionbetweendistaltar-
andsecondphalangesofmanaldigitIII;metatarsuselongate
sals2and3andtheirrespectivemetatarsals(T.Rowe,pers.
andnarrow;metatarsalIIIpinchedbetweenmetatarsalsIIand
comm.; Raath 1969). Rowe (pers. comm.) has discovered ad-
IV, the latter two contacting one another proximally in front of
ditionalsynapomorphies of
Ceratosauria,
including the
shape
III(similarmodificationsofthehandsandtarsusarepresentin
andprominenceofthesupracetabularshelf,afibulargrooveon
omithomimids,troodontids,andomithurinebirds).
the proximal end of the lateral side of the fibula (e.g., Gilmore
CaenagnathidaeSternberg.1940 1920, fig. 65C), and a prominent groove on the ventrolateral
side of the fibular condyle of the femur. Rowe (pers. comm.)
Temporal
Range.
— late
Cretaceous. alsonotedthat,withthepossibleexceptionofSegisaurus*,all
Included
J^XA.
—Caenagnathus collinsi.
C.
sternbergi.
andOvirapioi
philo- ceratosaurshaveanarrowerpubisthanisseeninotherthero-
ceratops.
podsasidefrombirds.BecauseCoelophysisisoneoftheearliest
Diagnosis.— No one doubts the monophyly of these pecu- theropods, its narrow pubis was thought to be diagnostic of
liarlyspecializedtheropods(Osbom1924^;Osmolska1976; Theropoda. However, this apomorphy is diagnostic of most, if
Barsbold1983).Caenagnathidswereoncethoughttoberelated not all, ceratosaurs, and a relatively broader pubis appears to
toomithomimidsbecausebothshareedentulous,beaklikesnouts, betheancestralconditionforTheropoda(e.g.,.■illosaurus.Mad-
butmorerecentworksuggestsotherwise(Barsbold1983,and sen1976).
seebelow).Caenagnathidshavehighlymodifiedskulls,andthere A more complete discussion of the evidence supporting
isverylittleinformationregardingtheirpostcranialskeletons. monophylyofCeratosauriawillbepresentedelsewhere(Rowe,
Severalnewspecimenshavebeendiscovered,buttheyhaveyet in
prep.).
to be completely described and illustrated (Osmolska 1976;
Barsbold 1983). Barsbold is currently engaged in a revision of CarnosauriaHuene,1920(n.comb.)
this taxon based on new material including as many as three Temporal Range.—late Jurassic
to
late
Cretaceous.
species of Oviraptor. and an adequate diagnosis of this taxon Includedkx\.—.
J -Illosaurus fragdis.
.icrocamhosaunis alokcnsis.
Indosaurus
mustawaithisfindings. matleyi.
.4leclrosaiirus
olseni.
Drypiosaurus aqudungnis.
.ilhenosaunis
sarcoplia-
Currie (pers. comm.) has informed me that Caenagnathus giis.
hlnatus.
-i .-I.
lancensis..ilioranms remotiis.
Dasplelosaurustorosus,
Indo-
(knownonlyfromcranialmaterial)andChirostenotes*(known suchiis
raplorms. Tarhosaiirus hataar.
and
Tvrannnsauntsre.x.
onlyfrompostcranialmaterial)mightrepresentthesamespecies. Diagnosis. — Based primarily upon the series of Allosaunis
Moreover,WilsonandCurrie(pers.comm.)havesuggestedthat fragilisasdescribedbyMadsen(1976)and.-llheriosaunislibra-
Microvenator*mightbeacaenagnathid.Thesehypothesesare tits as described by Lambe (1917) and Russell (1970). These
only tentative, but they are included because Currie and Wil- dataweresupplementedbypersonalexaminationofAllosaiints.
10
Albertosaurus,andTyrannosaunis.andthedescriptionsofcar-
nosaurs published in Marsh (1896), Osborn( 1905, 1906, 1912,
1917), Gilmore (1920), Matthew and Brown (1922), Janensch
(1925), Sternberg (1932), Stovall and Langston (1950), Rozh-
destvensky (1958, 1965), Walker (1964), Colbert and Russell
(1969),Ostrom(1969a),Steel(1970),GallonandJensen(1979),
andBarsbold(1983).
Themedium-tolarge-sizedtheropodssuchasMegalosaiirits*
andEuslrcptospondyliis*possesssomecamosaurlikeattributes.
These taxa are examples of a pervasive problem in theropod
phylogeny,namely,the"megalosaur"problem.Megalosaums*
wasthefirstdinosaurdescribed,butitisrepresentedbylimited
materialwithnodiagnosticfeaturesdistinguishingitfromother
largetheropods.Asthenameimplies,"megalosaurs"arelarger
theropods,andseveraloftheirapomorphiesareprobablysize-
related in that they are also seen in large omithischians and
sauropodomorphs (e.g., femur longer than tibia). In view of
profoundcharacterdiscordance,itismoreparsimonioustoac-
ceptthesesharedapomorphiesasexamplesofconvergencebe-
tween"megalosaurs" and largeomithischians or
sauropodo-
morphs.When considering the
"megalosaurs" and
Camosauria,
however, the problem of distinguishing homology from con-
vergenceismoredifficult.
Camosauria sharesmany apomor-
phieswithaportionofTheropodathatincludesextantbirds,
andthesecanhardlybeconsideredsizerelated(seebelow).The
problem with "megalosaurs" is that they either do not have
theseapomorphies,ortheappropriateportionsoftheirskeletons
are unknown. S. P. Welles is currently involved in a revision of
the"megalosaurs,"anduntilhehasrevisedthealphataxonomy
of this confusing group of fossils, there is little point in consid-
ering
them
further.
Severalcamosaurapomorphieslistedbelowarealsopresent
inothermediumtolargetheropodssuchasDilophosawusand
Ceratosaunis.Amongthesize-relatedapomorphiesareopistho-
coelouscervicals,thegreaterlengthofthefemurrelativetothe
tibia, a robust skeleton, and enlarged neural spines and trans-
verseprocessesinthetrunkvertebrae.Theseattributesareseen
inalllargesaurischians.Nevertheless,thetaxahereincludedin
Camosauriapossesscorroboratingsynapomorphiesinaddition
tothoserelatedtotheirsize,andotherlargetheropods,suchas
CeratosaunisandDilophosaurus.donot.
CamosauriaisdistinguishedfromotherTheropodaconsid-
eredinthisanalysisinthatitpossessesthefollowingsynapo-
morphies:orbit dorsoventrally elongate and roughlykey-
hole-shaped(Fig.IG,H);
supraorbital crestsin
fully
mature
individuals (Fig. IG, H); frontals and parietals narrow and very
short; reduction of mandibular fenestra (Fig. IG, H); further
reductionofdentaryoverlapontopostdentarybonesandman-
dibularsymphysis (indicatingimproved intramandibular joint,
Romer 1956); pronounced development of bony shelf below
mandibularcondyleonlateralsurfaceofsurangular,presumably
associatedwithinsertionofenlargedpterygoideusmusculature
(Fig. IG, H); ilium expanded anterodorsally (Fig. 5D); strongly
opisthocoelouscervicalandanteriortrunkvertebrae(conver-
gentinpenguins);digitsIIandIIIreducedinhand,especially
the latter, which is shorter than digit I (Fig. 4L; analogous con-
ditioninoraithurinebirds,butresultingfromlossofphalanges);
very robust postcranial skeleton with stout, relatively thick-
walledlongbones,shortenedandstoutlyconstructedtmnkand
ORIGINOFBIRDSANDEVOLUTIONOFFLIGHT
cervicalvertebrae(especiallyintyrannosaurids),andlargeneu-
ralspinesandtransverseprocessesthroughoutvertebralcol-
umn.
Tyrannosauridae,includingAlbertosaurus.Tarhosaurus.and
Tyrannosaurus.arefurtherderivedwithinthisassemblagein
thattheyhavethefollowingsynapomorphies:lacrimalexcludes
frontal from orbit (Currie. in press a. b)\ enlarged surangular
fenestra and pteo'goideus shelf (Fig. IH); ventral process of
squamosal nearly honzontally oriented (Fig. IH); postorbital
andjugalmassiveandanteriorlydirectedpostorbitalreentrant
into orbit (Fig. IH); tooth row fails to reach posterior to antor-
bital fenestra (Fig. IH); forelimb less than one-quarter of hind-
limb length; wrist bones very reduced (convergent in omitho-
mimids);thirdmanaldigitreducedtonomorethanmetacarpal
splint;ascendingprocessverybroad,extendsdorsallyfornearly
one-third height of astragalus + tibia (convergent in coeluro-
saurs);calcaneumveryreduced;andproximalendofmetatarsal
IIIstronglyconstrictedbetweenmetatarsalsIIandIV(conver-
gentin
omithomimids, elmisaurids*, and
Hulsanpes).
OrnithomimidaeMarsh,1890
(n.comb.:includesDeinocheiridaeof
OsmolskaandRoniewicz,1970)
TemporalRange.
—late
Jurassic
to
late
Cretaceous.
Included
Taxa.—Elaphwsauriis
bambergi.
Archacornithommms asiaticus.
Ornithomiinusedmonticus,velo.x.
O sedens.
O Strulhiomimus
alius.
Dromi-
cciominmsbrevitcrtnis.
samiieli.
D Gallimimiis
hullalus.
Ingenta
yanshim.Ga-
ludimimia
brcnpcs.
and
DcituniiciiKs
ininficis.
Diagnosis. — Based primarily upon Gallimunus bullalus as
described by Osmolska et al. (1972) and data derived from
Russell(1972).Additionalevidencederivedfrompersonalob-
servationof
Strulhiomimus and
descriptionsin
Marsh
(1890,
1896),Osbom(1917),Parks(1928,1933),Gilmore(1920,1933),
Janensch (1925, 1929), Stemberg (1932, 1933, 1934). Ostrom
(1969u, b. 1970, 1974a, 1976^)), and Barsbold (1983). The di-
agnosisbelowisbasedonUpperCretaceousomithomimids,
althoughmorecompleteknowledgeofLowerCretaceousand
Upper Jurassic taxa may alter it.
Comified beak as indicated by form of unworn margins of
edentulous, beaklike jaws (Fig. II; convergent in Caenagnath-
idae and modem birds); premaxilla enlarged and beaklike,
broadlycontactingnasaltoexcludemaxillafromextemalnaris
(Fig.11);secondarypalateformedbypremaxillaeandmaxillae;
reducedjugalandventrallyelongatepostorbital;reducedlower
temporalfenestra;quadratestronglyinclinedsothatdistalend
liesfarforwardofproximalend;bulbousparasphenoid(alsoin
Troodontidae, and to a lesser extent in birds); metacarpal I
elongateandalldigitsofsubequallength(Fig.4M);carpusvery
reducedwithpoorlydefinedarticularfacetsonindividualcar-
pals;terminalungualslesstrenchantandrecurved,withreduced
basaltubera(suggestinglossofraptorialfunctionforthehand);
humeruslightlyconstructedanddeltopectoralcrestreduced;
ischiumventrodistallyrecurved(Fig.5E);metatarsusnartow,
andelongatecomparedtotibialength;metatarsalIIIstrongly
pinchedbetweenmetatarsalIIandIV,barelycontactingdistal
tarsals(analogousmodificationsofthemetatarsusarosecon-
vergentlyintyrannosauridcamosaurs,elmisaurids*,omithu-
rinebirds,andtroodontids);pedaldigitsshortandstout.
GAUTHIER:
SAURISCHIAN
MONOPHYLY
DeinonychosauriaColbertandRussell,1969
Temporal
Range—early
to
late
Cretaceous.
Included
Taxa.—
Troodontidae
and
Dromaeosauridae.
Diagnosis. — Modifications of the foot in general, and the
secondpedaldigitinparticular,indicatearaptorialfunctionfor
thepesinDeinonychosauria(ColbertandRussell1969;Ostrom
1969a.b).Accordingtotheseauthors,thesubequallengthsof
pedal digits III and IV, together with modification of the rap-
torialpedaldigitII,indicatefunctionaldidactylyduringloco-
motion.TheungualonpedaldigitIIbearsalarge,compressed,
trenchant,stronglyrecurved,scimitarlikeclaw.Thesecondpha-
lanxisshortenedandsubequaltothefirstphalanxinlength.
Moreover, the second phalanx has a prominent heel postero-
ventrally,anditsanteriorandposteriorarticularsurfacesallow
increaseddigitalexcursion.Troodontidsmaynotbethesister-
group of dromaeosaurs (see Section V). This point is not clear,
however,andfollowingpreviousauthors,thistaxonisaccepted
on the basis of the shared apomorphic resemblances in their
feet.
ItisinterestingtonotethatOsmolska(1982)arguedthatthe
formofthemetatarso-phalangealjointindicatedthatthesecond
pedal digit functioned differently in troodontids (=sauromi-
thoidids)anddromaeosaurs.Thisobservationalonecannotbe
takentoindicatenonhomology,becausethemorphologyofone
could be a transformation of that seen in the other, or both
couldbetransformationsofsomemoregeneralconditionshared
by their common ancestor. Troodontids vary in the degree to
whichthesecondpedaldigitismodified(Russell,pers.comm.).
Forexample,Troodonhasamorespecializedraptorialsecond
pedaldigitandismorelikedromaeosaursinthisrespect.How-
ever,thesecondpedaldigitislessmodifiedinothertroodontids
(Osmolska 1982; Barsbold 1977). The possible effects of age,
size,andsexonthedegreeofdevelopmentofthesecharacters
hasyettobedetermined.Informationonthepossibleinfluence
ofthesefactorsmightbegainedfromextantcariamids,Chunga
andCariama.thathaveanalogouslymodifiedsecondpedaldig-
its;examinationofthebiologicalrolesoftheirfeetmayalso
providesomeinsightintothefunctionofthesecondpedaldigit
inDeinonychosauria.(Althoughtheinformationwasreceived
too late to include in this analysis, Currie has informed me that
thereissomeevidenceforapossibletroodontid-omithomimid
group; in light of this it would have been more appropriate to
considerthreeseparatebasictaxa—1Dromaeosaurus.2Dei-
nonychus-Velociraplor.and3Troodontidae—ratherthanone,
viz.,
Deinonychosauria.)
TroodontidaeGilmore,1924
(=SaurornithoididaeBarsbold,1974)
Temporal
Range.
—late
Cretaceous.
Included
Ja\a.—Saurorniihoides
mongoliensis.
junior,
S. and Troodon
for-
inosiis
[=.Stenonychosaiiriis
inequalis
and
Pecunodon
bakkeri
following
Cume
(in
press
h).
Diagnosis. — Based upon personal observation of casts of
Stenonychosaurusinequalisanddescriptionsandcomparisons
ofboththistaxonandSaurorniihoidesinOsbom(1924ft),Stem-
berg (1932), Colbert and Russell (1969), Russell (1969), Bars-
bold (1974, 1977, 1979, 1983), Sues (1978), Osmolska (1981,
II
1982), Russell and Seguin (1982), Currie (in press a. b), and
WilsonandCurrie(inpress).
Barsbold(1974)separated5.juniorfromS.mongoliensisbe-
causetheformeris1.3timeslarger,hasafewmoreteeth,and
thespecimensderivefromdifferentstratigraphicformations.
Morespecimensmayindeedrevealthattheyaredifferenttaxa.
However,thedilTerencesbetweenthesespecimenscouldreflect
size and age; current evidence cannot exclude the possibility
that S. junior is merely an adult of the smaller S. mongoliensis
(Currie, in press a. b notes a similar size range for Troodon).
Accordingly, these taxa will be considered synonymous in the
following analysis.
Anteromediallyinclinedorbits,suggestingbroadlyoverlap-
pingvisualfields;deepdepressioninbraincaseinregionof
middle ear cavity (see Currie, in press a); bulbous parasphe-
noidal rostrum (also present in omithomimids and in a less
modified form in some birds); small, closely spaced teeth with
enlarged(alsoseeninsomedromaeosaurs),distallyhookedden-
ticlesonposteriormargin,anteriordenticlesreducedorabsent
at least in the lower jaw; deep, narrow Meckelian fossa of den-
tary;additionalcaudalvertebraeincorporatedintosacrum(six
sacralvertebrae);rodlikemetatarsuswithproximallyattenuate
metatarsalIIIwedgedbetweenslendermetatarsalIIandrobust
metatarsal IV; metatarsals II and IV in contact anteriorly in
frontofproximalendofmetatarsalIII(alsoinelmisaurids*and
omithomimids);andtonguelikedistalarticularsurfaceofmeta-
III.
tarsal
DromaeosauridaeMatthewandBrown,1922
Temporal
Range.—early
late
toCretaceous.
Included
Ta\a.
Dminaeosaiinis
— alberiensis.
Dciiwnychus
annrrhopus.
I'e-
locirapror
mongoliensis,
and
Adasaurus
mongoliensis.
Diagnosis.—BasedprimarilyuponDeinonychusasdescribed
by Ostrom (1969a. b. 1974ft, 1976/'). Supplementary data de-
rivedfromdescriptionsinMatthewandBrown(1922),Osbom
(1924ft),ColbertandRussell(1969),Barsbold(1976,1977,1979,
1983), Sues (1977, 1978), Bonaparte and Powell (1980), and
casts of the plastotype of Deinonychus. Sues (1978) included
Saurornitholestes* in Dromaeosauridae. but he did so on the
basis of plesiomorphic resemblances; until the evidence sup-
portingthisplacementismadeexplicit,thistaxonwillbecon-
separately.
sidered
Prezygapophysesandhaemalarchesexceedlengthofcaudal
vertebrae(Fig.2G);shortmetatarsuscomparedtofemurlength;
peculiarginglymoidstructureofthedistalendsofmetatarsals
II and III; deeply grooved distal ginglymus of metatarsal II.
Avialae(n.txn.)
(L.:
avis,
bird;alae.
wings)
Temporal Range.
—late
Jurassic
to
Recent.
Included
axa.—
J .Archaeoplery.x
lithographica*
plus
omithurine
birds.
Diagnosis.—BasedprimarilyuponArchaeopteryxlithogra-
phica*asdescribedbyHeilmann(1926),deBeer(1954),Ostrom
(1972, 1973, 1974a. ft, 1975a. ft, 1976a), Wellnhofer (1974),
TarsitanoandHecht(1980),Martin(1983a,ft).Whetstone(1983)
anduponpersonalobservationoftheEichstattspecimenand
castsoftheLondonandBerlinspecimens.
This new taxon, Avialae, is named so as not to violate the
12
classificatory conventions of this work, in which widely used
names hke Aves are restricted to living taxa in order to maxi-
mizestabilityandphylogeneticinformativeness.Becauseof
feathers and the presumed ability to fly. Archaeopteryx* has
always been considered a bird. This informal usage has been
maintained above, and use of the informal term "bird" for this
taxon will be continued in the following discussion. In a formal
sense,however,"birds"andAveswillnotbesynonymous.The
"wmgediheropods"includedinAvialaepossessthefollowing
synapomorphiesdistinguishingthemfromotherTheropoda.
Premaxillaeelongate,narrow,andmorepointedanteriorly,
with longer nasal processes; maxillary process of premaxilla
reduced so that maxilla participates broadly in external naris
(alsointroodontids;Currie,inpressa);enlargedbrain/basicra-
nium(temporalmusculaturefailstoextendoriginontofrontal
bones);double-condyledquadratedisplacedfromdistalposition
on opisthotic to more anteromedial position in contact with
prootic(Currie,pers.comm.andWalker,pers.comm.,disagree
withWhetstone'sinterpretationofthequadrate;Currienotes
theanteriordisplacement of the quadrate in troodontids, and
Walkerdoesnotconsiderthequadratetobedouble-condyled
inArcliacnpiery.x*):maxillaryanddentaryteethreducedinsize
andnumber(orlost),withunserratedcrownsandenlargedroots
that completely enclose replacement teeth within them (see
Howgate 1984, for an alternative view); robust furcula for hy-
pertrophiedflightmusculature(OlsonandFeduccia1979);scap-
ulawithmoreorlessprominentacromionprocessforligamen-
tousconnectiontoclavicle(seeMartin1983/).foralternative
view);length/breadthratioofscapulaatmidlengthexceedsnine
(notinpenguins)andscapulatapersdistally;acrocoracoidtuber-
ositylargerthaninothercoelurosaurs;coracoidenlargedand
inflectedposteromediallymoresothaninothercoelurosaurs;
verylongforelimbsandhands(e.g.,inArchaeopteryx*forelimb
is 120-140% of hindlimb length, and more than twice as long
asdistancebetweenglenoidandacetabulum),withforearmmore
than87%ofhumeruslengthandmetacarpalIIapproachingor
exceedingone-halfofhumeruslength;ischiumcompressedand
dorsoventrallydeep;comparedtoothertheropods,tibia,fibula,
andmetatarsalsrelativelymoreelongatewithrespecttofemur,
regardlessofbodysize(metatarsalsshortinpenguinsandsome
other birds, J. Cracraft, pers. comm.); fibula attenuate distally,
and may not extend to end of tibia; proximal tarsals fused to
tibia-fibulaandtooneanotherinadults;distaltarsalsandmeta-
tarsalsfusedatleastdistallyinfullyadultindividuals(conver-
gentinsomeceratosaurs,elmisaurids*.andHulsanpes);first
pedal digit elongate and reversed (may be reversed in some
extantbirds,R.Storer,pers.comm.),metatarsal1attachedon
distal quarter of metatarsal II; tail reduced to no more than 23
freecaudalvertebrae;featherscoverlimbsandtail,featherson
lateralmarginsoftailandposteriormarginsofarmsenlarged,
curved, and asymmetrically vaned, indicating aerodynamic
function(e.g.,FeducciaandTordofT1979).
It is not certain that feathers are confined only to avialans
amongcoelurosaurs.Compsognalhusapparentlylacksthem
(Ostrom1978),sofeathersappeartohavebeenabsentincoe-
lurosaursancestrally.
However, Conipsognathiisthe
isonly
non-
avialan theropod that is preserved in an environment of de-
positionconducive to
the
preservation of
featherimpressions.
Thus,futurefindsmaydemonstratethatfeathersarosepriorto
theoriginofbirds.
ORIGINOFBIRDSANDEVOLUTIONOFFLIGHT
As the use of the asterisk indicates, Archaeopteryxlithogra-
phica*ishereconsideredtopossessnoapomorphiesthatwould
not be expected in the common ancestor of all birds. Thus, the
specimensreferredtothistaxonareplacedherebecauseoftheir
geographicandstratigraphicoccurrenceandoverallsimilarity.
Thesespecimensprobablyrepresentasinglespecies,butsuch
opinionsshouldalwaysbedistinguishedfromthosebasedon
appropriateevidence.Notwithstandingtheinterestingpossibil-
itiessuggestedbyMartin(1983/»)andHowgate(1984),thereis
nounambiguousevidenceindicatingeitherparaphylyormono-
phyly, and these specimens will be referred to collectively as
themetataxonArchaeopteryxlithographica*.Becauseofitsgen-
eralizedmorphology andstratigraphic
position,
the
specimens
ofArchaeopteryx*couldbepartsofanancestralpopulationthat
gave rise to all later birds. Hypotheses of ancestral status can
only be weakly supported in that they are based on negative
evidence.Nevertheless,thereisnounequivocalevidencesup-
portingthealternativehypothesis,thatArchaeopteryx*is
monophyleticandthusnotanancestralbird.
Whetstone (1983) and Martin (1983a) suggested that, com-
paredtootherbirds,thesquamosaliseitherreducedorabsent
inArchaeopteryx*.Thisinterpretationisopentodoubtinview
ofthepreservationofthespecimenswithcranialmaterial;each
ofthespecimenswaspreservedsuchthatuponseparatingthe
slabs, the skulls fractured between the main body of the skull
andthelightlyconstructedelementssurroundingtheorbitand
temporalfenestra.Theseauthorscontendedthatthesquamosal
wasabsentbecausethereisnoevidenceforitssuturalconnection
to the skull. This contention loses much of its force because
thesesuturalsurfacesarealsoabsentintheropodsinwhichthe
squamosalisknowntobepresent(e.g.,Syntarsits:M.A.Raath,
pers. comm.). Under such circumstances, it is diflicult to dis-
tinguishbetween absence and
nonpreservation. further
If finds
corroboratetheWhetstone-Martinhypothesis,thenArchaeop-
teryx*mustberemovedfrommetataxonstatusanditshy-
pothesized ancestralposition
must
rejected.
be
In keeping with one of the goals of this work, namely to
providearelevantseriesofoutgroupsforphylogeneticanalyses
among the major groups of extant birds, two more taxa within
Avialaewillbedefinedanddiagnosedbelow.Theyarenotbasic
taxa in this analysis, but it is necessary to consider Omithurae
andAvesatthispointbecausetheconceptsrepresentedbythese
namesasusedinthisstudymaydifferfromthoseemployedby
others.
Omithurae is defined here in keeping with its original intent
as a taxon encompassing all extant birds, as well as all other
birds that are closer phylogenetically to extant birds than is
Archaeopteryx*.HavingbeensupplantedbyNeomithes(Ga-
dow1893),Omithurae(Haeckel1866)isseldomusedincurrent
ornithologicalliterature;theobscurityofthenamehassavedit
fromthediversityofmeaningsthatpossiblealtemativenames
have developed, and Omithurae is thus an appropriate name
for this taxon. As here defined, however, Omithurae is a more
inclusivetaxoncontainingAves,whichreversesthetraditional
hierarchicalrelationshipbetweenthesetaxa.ThetermsNeor-
nithesandCarinataeareavoidedbecausetheirambiguousand,
attimes,contradictorymeaningsinaviansystematicshavelad-
enthemwithtoomuchhistoncalbaggagetobeusefulinthis
work.
Omithurae is recognized by a host of flight-related modifi-
GAUTHIER:
SAURISCHIAN
MONOPHVLY
cationsintheskeletonthatdistinguishitfromothertheropods,
includingArchaeopteryx*.Themodificationsindicatethatthe
immediatecommonancestorofOmithuraepossessednotonly
an inherited ability to fly, but the capacity for sustained flight
approaching that seen in extant birds. It also appears that the
immediatecommonancestorofomithurinebirdshadalready
overcometheenergeticallymostdemandingaspectofflightin
modembirds—tobecomeairbornefromastandingstart.
About60speciesofbirdshavebeendescribedfromsediments
of Cretaceous age, but most of these species are too poorly
knowntocontributemuchtoourunderstandingofavianphy-
logeny (Elzanowski 1983; Thulbom 1984). To simplify the fol-
lowingdiscussion, onlyIchthyornis
and Hesperomithes will
be
considered.Moreover,becauseofcharacterdiscordance,and
ambiguitiesincharacterinterpretationsstemmingfromthespe-
cializedhesperomithmorphology,itisnotclearwhetherHes-
peromithesor Ichthyornis more
is closelyrelated
to
extant
birds.
Theevidencepresentedbelowindicatesthatneithertaxonbe-
longswithinanysubgroupofextantbirds.Indeed,thereissome
evidence,suchasthedetailedformoftheintramandibularjoint
andthepossiblepresenceofa"predentary"boneinbothtaxa,
indicatingthatIchthyornisandHesperomithescollectivelycon-
stituteasister-taxonofextantbirds(Martin1983a).Itisbeyond
thescopeofthisworktoaddressthesequestions,andOmithurae
willbediagnosedbysynapomorphiesthatcanbefoundinany
two of the following three taxa, Hesperomithes, Ichthyornis.
andextantbirds.Thisapproachassumesthattheflightless,foot-
propelleddiversofHesperomitheswerederivedfromanances-
torpossessingthefullsuiteofcharactersdiagnosticofOmith-
urae.Thismaynotbethecase,butatleastHesperomithesdo
notretainanunmodifiedancestralcondition.Forexample,as
diagnosedbelow,Omithurae possess a keeled stemum; Hes-
peromithesdonotpossessakeel,butthemorphologyofits
stemumiscertainlyderivedwithrespecttotheconditionpres-
entother
in dinosaurs.
OmithuraeHaeckel,1866
TemporalRange.—lowerCretaceous (at
least
Albian
Stage) to
Recent.
IncludedTaxa.—Extantbirdsandallothertaxa,suchasIchthyornisand
Hesperomithes.that
arc
closer
to
extant
birds than
Archaeopteryx*
is . ulnawithsemilunatearticularsurfacedistally(reversedinsome
Diagnosis.— This taxon is based on evidence derived from
Marsh(1880),MartinandTate(1976),Elzanowski(1977,1981),
Martin and Bonner (1977), Martin and Stewart (1977), Mc-
Dowell(1978),WhetstoneandMartin(1979,1981),Martin
(1980, 1983a. ^. 1984), Martin et al. (1980), Whetstone (1983),
andThulbom(1984).
Body of premaxillae fused, edentulous, and beaklike; nasal
process of premaxilla extends over nasal to closely approach
frontal; facial process of maxilla reduced and naris enlarged
(resultinginlossofmaxillaryfenestra);descendingprocessof
nasalcontactspremaxillatoexcludemaxillafromnarialmargin;
maxilla with prominent medial phalange in palate (=maxillo-
palatine);ectopterygoidabsent;palatineandpterygoidnarrow
andarticulatingnearlevelofbraincase(contraMcDowell's1978
interpretationofpalatineasanteriorpterygoid;pers.comm.L.
Witmer and K. Warheit); peg and socket articulation between
jugalandlateralcotylusofquadrate;extensivemesethmoidos-
sificationappearsearlyinontogenyandisexposedonskullroof
betweennasalsinlateembryos/neonates(thischaracterisre-
13
tainedintoadult
stages in
Hesperomithes [Martin,pers.
comm.]
andRatitae[Pycraft1900]);neckincludesmorethan9vertebrae
(convergentinOmithomimidae,whichhave10cervicals—or-
nithurine birds have at least 13 cervicals ancestrally); anterior
cervicalswithmoderatelydevelopedheterocoely(seebelow);
prominenthypapophysesinposteriorcervicalandanteriortrunk
vertebrae;lossofhyposphene-hypantraintervertebralarticu-
lations;sacrumincludesmorethan5vertebrae(convergentin
troodontids,whichhave6—omithurinebirdshaveatleast10);
freeportionoftailreducedtofewerthan16vertebrae;absence
ofcaudalzygapophyses;presenceofpygostyleincludingvariable
numberofcoossifiedvertebrae(secondarilylostinsomeHes-
peromithesand Tinami,
andin
most
Ratitae);
ossified
uncinate
processes (reversed in Anhimidae and megapods; R. Storer,
pers. comm.); ossified (rather than calcified) ventral ribs at-
tachedtostemum(thelasttwocharactersmayapplytoamore
inclusive taxon in that Oslrom [1969ft] and Paul [1984ft] iden-
tifiedossifiedventralribsanduncinateprocessesindromaeo-
saurs);absenceofgastralia;enlargedandposteriorlydisplaced
stemum,chondrogeniccellsofwhichproliferateandmigrateto
form keel; appearance of new stemal ossification center, the
lophosteon, arising in region of stemal keel; shoulderjointset
posteriorly and dorsal to center of gravity; hollowscapulaand
coracoidarticulateviascapularpegandcoracoidalsocketbelow
levelofcoracoidalportionofglenoid(reversedinsomeflightless
birds, such as Hesperomithes and Ratitae), and scapula and
coracoidfailtofuseinadults(reversedinRatitae);scapulalong
(exceedslengthof7trunkvertebrae),narrow(length/breadth
ratio at midlength exceeds 12), tapering distally, and scapula
lies near to and parallel with vertebral column and closely ap-
proachesilium posteriorly(these
charactersare
reversed
in
flightlessbirds);coracoidlong,slender,taperedatmidlength,
broad distally and in or near contact on midline, and articulate
in prominent grooves along anterior edges of stemum, with
prominentacromionprocessforarticulationofclavicle(=trios-
sealcanal;someaspectsofcoracoidformmayreverseinflight-
lessbirds,suchasHesperomithesandRatitae);prominent,con-
ical,intemaltuberosityonhumerusseparatedfromheadby
capitalgroove;ulnaapproximatelytwiceasthickasradiusand
withnobsalongposteriormarginforflightfeatherattachment;
flightlessbirdssuchasHesperomithesandRatitae);character-
isticcarpometacarpusinadultsformedfromcoossificationof
somedistalcarpalsandmetacarpals,withmetacarpalsI,II,and
III fused proximally, and II and III fused distally, second pha-
lanxofdigitIIbroadandflat,absenceoftwophalangesondigit
III, clawed unguals usually absent in adults (various flightless
birds have lost other manal elements, e.g., digit II is the only
fingerremaininginApteryxandCasuarius:presentinterpreta-
tioncontradictsdigitalhomologyproposedbyHinchliffeand
Hecht 1984); pelvis fused in adults; prominent antitrochanter
aboveacetabulum;preacetabularportionsofiliaelongate,closely
appressed on midline, and in contact with at least some sacral
neuralspines,butpostacetabularportionswidelyseparated(re-
versedinsomeomithurinessuchasHesperomithes,Ratitae,
Gaviidae); absence of bipartite distal moities of ischium (see
Fig.5H,I);pubesandischiawidelyseparatedonmidline;pubis
shortened,withoutexpandedfootdistally,andwithprepubic
processproximally(notinIchthyornis):femurwithdeeprotular
grooveanterodistallyandprominentfibularcondyle;tibiawith
14 ORIGINOFBIRDSANDEVOLUTIONOFFLIGHT

cnemialepiphysis;tibiawithprominenttendinalgrooveantero- Numerouscharactersfromlesspreservableportionsofthe
distally;fibulashort,notincontactwithproximaltarsals;prox- anatomy,togetherwithethologic,physiologic,genetic,andim-
imaltarsals,includingascendingprocess,fusedtooneanother munologicaldata attestingto
the
distinctiveness
of
Aves
within
andtotibiaearlyinpostnatalontogeny;distaltarsalsformmeta- extant Amniota could be cited at this point. However, because
tarsalcapwithintercondylarprominence(reducedorlostin no one has ever mistakenly placed an extant bird in any other
Ratitae), and this cap fuses to metatarsals early in postnatal extantamniotegroupasidefromAves,thereislittletobegained
ontogeny;proximalendofmetatarsalIIIposteriortoandmore frombelaboringtheissue.
or less compressed by metatarsals II and IV (as in troodontids,
omithomimids,tyrannosaurids,andelmisaurids*);coossifica-
tionamongmetatarsalsbeginsdistally(ratherthanproximally); Phylogenetic Analysis
smallforaminabetweenproximalendsofmetatarsals(notper-
I.PhylogeneticRelationshipswithinDinosauria
foratingmetatarsus in
Hesperomithes); loss
of
raptorial
modi-
ficationsofpedaldigitII(seebelow);lossofpedaldigitVin Inordertoaddressthequestionofthephylogeneticrelation-
adult (convergent in some omithomimids and perhaps cae- shipsamongthebasictaxa,itisfirstnecessarytodevelopa
nagnathids). moreinclusivehypothesisthatwouldprovidearelevantseries
of outgroups. Based on evidence presented in Appendix A, Di-
nosauriaconsidered
is monophyletic.Moreover, Herrerasau-
AvesLinne,1758
ridae*isconsideredthesister-group(s)ofallotherdinosaurs,
Temporal Range—late
CretaceousRecent,
to andV\tro%a.ur\di-Lagosuchus, Omithosuchidae, Eiiparkena*.and
IncludedTaxa.—Avesisherereslnctedtothetaxonencompassingallde- Pseudosuchiarepresentsuccessivelymoreremoteoutgroupsof
scendantsthe
ofmost
recent
common ancestor Ratitae,
of Tinami,
and
Neog- Dinosauria.
nathae,
asthese
taxa
arediagnosed in
Gauthier et
al.
(in
prep.).
Inviewoftheampleevidencesupportingthehypothesesthat
Theropoda(seebelow),Omithischia,andSauropodomorpha
Diagnosis. —Based on data derived from Huxley (1867), Py-
(including"prosauropods")areeachmonophyletic(seeAppen-
craft (1900), de Beer (1956), Bock (1963), Feduccia (1980), Cra-
dixA),noneofthesetaxacouldhave"givenrise"totheothers.
craft (1981), Elzanowski (I98I), Martin (1983/). 1984), and
Accordingly,BakkerandGallon's(1974)suggestionthatsome
Thulbom(1984),andreferencescitedtherein.Avesisdiagnosed
dinosaursevolvedfrom"prosauropods"mustberejected(see
withinomithurinebirdsbythepossessionofthefollowingsyn-
Charig \916b. for criticisms; see Bonaparte 1976, and Cooper
apomorphies;lossofteethonmaxillaanddentary,well-devel-
1981a.formorerecentrestatementsofthe"prosauropod"origins
opedbill;parietalsconfinedtoposteriormostportionofskull
hypothesis).Instead,"prosauropods"areconsideredparaphy-
roof;lossofcoronoidbone;presenceofbonymandibularsym-
leticbecausesomeareclosertosauropodsthanothers(Appen-
physis;presenceof
tricondylararticulation
betweenquadrate
dixA).GiventhemonophylyofTheropoda,Sauropodomorpha,
andmandible;narrow,fingerlikeodontoidprocessofaxis;sad-
andOmithischia,thereareonlythreephylogeneticrelationships
dle-shaped intervertebralarticulations
fully
developed
and
ex-
possibleamongthesetaxa:1)Omithischiacouldbethesister-
tendintoposteriortrunkvertebrae(convergentwithinHesper-
group of Theropoda; 2) Omithischia could be the sister-group
omithes);freeportionoftailcomposedoffewerthannine
of Sauropodomorpha; or 3) Sauropodomorpha could be the
vertebrae(someLamsandpenguinsdisplayvaryingdegreesof
sister-group of Theropoda. The evidence supporting each of
fusion of the first caudal, but the nine caudals so obtained are
thesealternativeswillbeconsideredbelow.
consideredreversals);large,dorsallyoriented,plowshare-shaped
Hypothesis I: Two apomorphies are shared by Theropoda
pygostylethatformsasingleelementinadults(pygostyleabsent
andOmithischiathatarenotalsopresentinSauropodomorpha
in some tinamous and neognaths, and in most ratites); fused
ancestrally.AncestralDinosaunapossessedthreesacralverte-
uncinateprocesses(reversedinloons,grebes,penguins,and
brae,ancestralSauropodomorpharetainedthisconditional-
Apteiyx.andoccasionallyunfusedinotherratites);glenoidsur-
thoughsubgroupswithinthistaxonhaveasmanyassixsacral
facenotperpendiculartoextemalsurfaceofscapula;single
vertebrae (Appendix A). In contrast, Theropoda and Omithis-
prominentarticularsurfaceofhumerusseparatedfromextemal
chiahavemorethanthreesacrals;atleastfivearepresentin
tuberosity;pneumaticskeleton,includingfossaandforamenin
Theropodaancestrally,andthelowestnumberreportedinOr-
humerus(reversedinsomedivingbirds;S.Hope,pers.comm.);
nithischiaisthefourinthe"juvenile"Scelidosaurus(giventhe
ulnar crest not in plane of long axis of humeral head; presence
sexual dimorphism in omithischian sacral number, the count
of lateral extension of intemal tuberosity of humerus (=crus
shouldbefourtofiveforomithischiansancestrally;seeGallon
lateraletuberculimedialis);deltopectoralcrestofhumeruswith
1974, 1982).
palmardeflection,andapexnotdistallyplaced,sodistalprofile
Reductionofthefifthpedaldigittoametatarsalspurisanother
does not curve abruptly to shaft at a steep angle; ilium further
apomorphic condition that is shared by Omithischia and The-
lengthened anteriorly so that it overlaps bases of at least one
ropoda,becausethefifthdigitretainsasinglephalanxinSau-
set of ribs; process on proximal portion of ischium contacts
ancestrally.
ropodomorpha
pubis; pubis thin; tibia with ossified supratendinal bridge in Hypothesis II: Cooper (I981d'.-8I9-829) reviewed the "pro-
adult(reversedinsomeratitesandowls);hypotarsusformsas sauropod"characters of
omithischians. Exceptfor
the
structure
outgrowthofdistaltarsalcap;smallforaminapiercetarsometa- oftheiliacprongandcheekteeth,thecharactershediscussed
tarsusproximally;fullyenclosedforamenbetweendistalends areuniformlyancestralconditionsatthislevelofanalysis(i.e.,
ofmetatarsalsIIIandIVforpassageofM.extensorbrevisdigiti theyapplytotheimmediatecommonancestorofDinosauria).
IV (this foramen is incompletely enclosed in Ichthyorms. and BakkerandGalton(1974)suggestedthattheelongateanterior
thedegreeofenclosuremaybevariableinsomeratites). process of the ilium (=iliac prong) present in two sauropodo-
GAUTHIER:
SAURISCHIAN
MONOPHYLY
morphs,AnchisauntsandAmmosaunis.indicatedamorerecent
commonancestrybetweenSauropodomorphaandOmithischia.
inthattheihacpronginthelattergroupisrelativelylongerthan
inotherarchosaurs.Cooper (1981a, fig. 39) noted that the iliac
prongmaylengthenduringontogenyinMassospondyhis,but
not to the degree seen in cither Anchisawus or Ammosaurus
(seeGallonandCluver1976).Becausetheanterioriliacprocess
iscomparativelyshortinallotherSauropodomorpha,including
themorphologicallygeneralizedmembersofthisgroupsuchas
Thecodontosaums*.thissynapomorphyisdiagnosticofAnchi-
saurusandAmmosaurusaloneamongSauropodomorpha.Coo-
per(1981(2)consideredAnchisawustobeajuvenileAmmo-
saurus.However, neither
taxon
displays
the
characteristic
skeletal
fusionsindicatingcessationofgrowth,soCooper'shypothesis
cannotyetbeevaluated.Moreover,Galton(pers.comm.)doubts
Cooper'sontogeneticargumentbecausethesetaxadiffernot
only in size but in the morphology of the ischium, pubis, pes,
and third sacral rib. In any case, this apomorphy cannot be
considered evidencesupporting sauropodomorph-omithischian
monophyly.
Closelypacked,leaf-shapedcheekteethareapomorphicfor
archosaurs.Analogoustooth-formsreflectherbivoroushabits
inextantlepidosaurs.Relativelywidelyspaced,sharplypointed
teeth with finely serrated margins are the ancestral condition
forarchosaurs(Romer1956).Insauropodomorphsandomith-
ischiansancestrally,however,thecheekteethdifferinthatthe
compressedcrownsaredistinctlysetoflTfromtheroot,theteeth
aremorecloselyspaced,andtherearefewerandlargerserrations
on the margins of the crowns than in archosaurs ancestrally.
Thecheekteethdiflferinthetwogroups:insauropodomorphs
thecrownsareelongate,andtheserrationsarefinerandmore
numerousthaninomithischians;butinomithischiansthecrowns
arenearlyaswideastall,andtheserrationsarefewerandlarger
than in sauropodomorphs. Chang (\9()lb) argued that these
differencesprecludederivationofonetooth-formfromtheoth-
er,andthatitwasequallylikelythatbothwerederivedfrom
themoregeneralconditionretainedbytheropodsancestrally.
Intheabsenceofpertinentdevelopmentalinformation,Charig's
first assertion is not testable. One must admit that the second
assertion is possible, but fewer assumptions are involved in
acceptingthattheapomorphicaspectsoftooth-formsharedby
OmithischiaandSauropodomorphaconstituteapotentialsyn-
apomorphy.
HypothesisIII:Hypothesesoftheropod-omithischianmono-
phylyor
omithischian-sauropodomorph monophyly are
each
supportedbyonlyoneortwopotentialsynapomorphies.Thus,
thereislittlebasisforpreferringoneofthesealternativesover
the other. However, neither hypothesis fares well against the
finalalternative,thetaxoncomposedofsauropodomorphsand
theropods,theSaurischia.
BecauseSauropodomorphaandTheropodaarelikelytobe
plesiomorphic with respect to synapomorphies diagnostic of
Omithischia, we have from Seeley (1887, 1888) to the present
dayconsidered"saurischians"tobe"primitivedinosaurs"(hence
Galton's 1977 description of Herrerasaurus* as a "primitive
saurischian,"eventhoughitwasdescribedasbeing"primitive"
comparedtoallotherdinosaurs).Inlightofevidencepresented
below,itwillbeapparentthat"saurischians"arenotthepara-
phyletic"stem-group"ofotherdinosaurs.Indeedthelineageof
dinosaursofwhichextantbirdsareapart,theSaurischia,isthe
monophyleticsister-taxonofOmithischiawithinDinosauria.
15
Saurischia
(n. comb.=Saurischia Seeley, 1887, plus Aves Linne, 1758)
TemporalRange.—late
Triassic
lo
Recent.
Included
Ta.xa.
—Sauropodomorpha and
Theropoda
(including
birds).
Diagnosis. —Saurischia is here defined to include birds and
all dinosaurs that are closer to birds than they are to Omithis-
chia.Intheensuinganalysis,Herrerasauridae*,Pterosauria-
Lagosuchus.Omithosuchidae,Euparkena*.andPseudosuchia
willbeusedassuccessivelymoreremoteoutgroups.Saurischia
possessesthefollowingsynapomorphiesdistinguishingitwithin
Dinosauria.
1)Contactbetweenmaxillaryprocessofpremaxillaandnasal
reduced or absent. The maxillary process of the premaxilla is
broadly in contact with the nasal at the posterodorsal end of
thelateralmarginoftheextemalnarisinarchosaursancestrally
(Benton 1983; Gauthier 1984). This condition is also ancestral
fordinosaursbecauseitisretainedbyallPseudosuchiaexcept
someaetosaurs(Sawin1947),anditisretainedbyEuparkeria*
(Fig. lA), Omithosuchidae (Fig. IB), and Herrerasauridae* (D.
Brinkman,pers.comm.).Comparedtotheancestralcondition,
Omithischiaisfurtherderivedinthepronouncedposteriorex-
tensionofthepremaxilla(Fig.IC,D),whichmaycompletely
separatethemaxillafromthenasalinsomeomithischians(Ro-
mer1956).RauisuchiaisalsodiagnosedamongPseudosuchia
by a long, but very thin, maxillary process of the premaxilla,
anditisthusconvergentonOmithischiainthisregard(Gauthier
1984).IncontrasttotheancestralconditioninDinosauria,the
maxillary process of the premaxilla is reduced and its contact
withthenasaliseithernarroworabsentinSauropodomorpha
(Fig. IE, F), Ceratosauria (Welles 1984), Camosauna (Fig. IG,
H), Compsognathus (Ostrom 1978), Ormtholestes* (Osbom
1917),Caenagnathidae(Barsbold1983),Deinonychosauria(Fig.
IJ, K), and birds (Fig. IL). Omithomimidae (Fig. II) is an
exception among Saurischia, in that it displays a prominent
maxillary process of the premaxilla. In the context of the evi-
dencepresented below,
however, omithomimidsare considered
to have reversed this character. In all birds except Archaeop-
teryx*.themaxillaisalsoexcludedfromtheexternalnaris.This
is not the ancestral condition, however, in that exclusion is
effectedbyanelongatedescendingprocessofthenasal,rather
thananascendingprocessofthepremaxilla(Marsh1880;Gin-
gerich1976).Pterosauriaalsohasareducedmaxillaryprocess
(Wellnhofer1978),andinthisdetailofpremaxillaryformptero-
saursareconsideredconvergent with
Saurischia.
2)Temporalmusculatureextendsontofrontal.Thetemporal
musculatureoriginatesonthedorsolateralsurfaceoftheparietal
in saurians ancestrally (Gauthier 1984). During postnatal on-
togeny,the temporalmusculature hypertrophies and
its
area
of
originextendsmediallyontotheparietaltable(Gauthieretal.,
inpress).ThisconditionisretainedbyPseudosuchia(Gauthier
i9U),Euparkeria*{Eyfjer1965),Omithosuchidae(Walker1964),
Pterosauria(Wellnhofer1978),andOmithischia(Galton1974),
soitappearstobeancestralforDinosauriaaswell.Incontrast,
thetemporalmusculatureextendsontotheposterodorsalsur-
faceofthefrontalboneinSauropodomorpha(e.g.,Huene1906,
1908,1932).Procompsognathus*(Fraas1913).Ceratosauria(T.
Rowe,pers.comm.;Gilmore1920),Camosauria(Madsen1976),
Saurornilholesles*(Sues1978),Caenagnathidae(Barsbold1983).
Omithomimidae(Osmolskaetal.1972),andDeinonychosauria
(ColbertandRussell1969).Thischaracterhasnotbeenreported
16
in any bird. In view of the evidence presented below, however,
it is simpler to accept that the failure of the temporal muscu-
laturetoreachthefrontalresultsfromexpansionofthebraincase
inbirds,ratherthanretentionofanancestralcondition.
3)Posteriorcervicalselongate.Exceptamongthelong-necked
protorosaurs, the neck constitutes approximately 33% of the
totallengthofthepresacralvertebralcolumninnonarchosaur
Archosauromorpha (Gauthier 1984). There are relatively few
complete and articulated vertebral columns known for basal
dinosauriantaxa,andcomparisonsmaybecomplicatedbydif-
feringnumbersofcervical,trunk,andsacralvertebrae,butthe
availableevidencesuggeststhattheneckconstitutedapproxi-
mately40%ofthepresacralvertebralcolumninthecommon
ancestor of Saurischia. For example, dividing the combined
lengthsofvertebrae2-9bythecombinedlengthsofvertebrae
2-23revealsthattheneckconstitutes33-34%ofthetotallength
incrocodiles(pers.obs.),34%inHctcrodontosaurus(SantaLuca
1980), and 33% in Hypsilophodon* (Gallon 1974). In contrast,
vertebrae 2-9 constitute 40% of the total length of 2-23 in
CoelophysisipeTs.obs.),38%inCompsognathits{OsUom1978),
41%-43% in Archaeoptery.x* (Wellnhofer 1974), and 41% in
Galliiuimus(Osmolskaetal.1972).AshasbeennotedbyGallon
(1976) and Gallon and Cluver (1976), the neck is at least 41%
in Sauropodomorpha, not only because of the length of the
individualcervicals,butbecauseatleastoneadditionalvertebra
has been added to the cervical series (see Appendix A). The
elongation of the neck appears to have been accomplished by
lengthening of the posterior cervicals in Saurischia. Leaving
aside the length of the axis, it is evident that vertebrae 3, 4, and
5 are the longest elements in the neck in Pscudosuchia (pers.
obs.), Euparkcna* (Ewer 1965), Omithosuchidae (Bonaparte
I975(?).Lagosuchiis*(Bonaparte1975/)),andPterosauriaan-
cestrally(the
neckbecomes longwithin
Pterosauriabuttheneck
is short in Scleromochlus*. the sister-taxon of all other ptero-
saurs[Gauthier1984]).ThisconditionisancestralforDino-
sauriainthatcervicals3-5arealsolongestinHerrerasauridae*
(Gallon 1977)andOmithischia (Gallon 1974, 1975, 1978; San-
taLuca1980;Colbert1981).Incontrast,thelongest(postaxial)
cervicals in saurischians are 6-9. For example, vertebrae 6 and
12 are subequal in length in crocodiles (pers. obs.) and such
omilhischians as Hcterodoutosawiis (Santa Luca 1980) and
Hypsilophodon* (Gallon 1974). In contrast, cervical 6 is 22%
longer than cervical 12 in Dilophosaurus (Welles 1984), 37%
in Coelophysis (pers. obs.), 35% in Coinpsognathus (Ostrom
1978), 62% in Gallimimus (Osmolska el al. 1972). and 45% in
Archaeopieryx*(Wellnhofer1974).Sauropodomorphsalsoshare
theapomorphicproportionaldifferencebetweenthelengthsof
vertebral centra 6 and 12 (e.g.. Gallon and Cluver 1976). Al-
thoughourknowledgeislessthancomplete,itappearsthat
elongation of the cervicals posterior to cervical 5 accounts for
the neck forming more than 33% of the length of the presacral
vertebralcolumninSauropodomorphaandTheropodaaside
fromtaxawithlargeleadsandconsequentlyshortnecks,such
asTyrannosaiirus(Charigetal.1965).
4)Axialpostzygapophysessetlateraltoprezygapophyses.In
anterior view, the pre- and post-zygapophyses are approxi-
matelyequidistantfromthemidlineoftheaxialcentrumin
Pseudosuchia (pers. obs.), Euparkcna* (Ewer 1965), Lagosu-
chiis*(Bonaparte 1975/)),and Pterosauriaancestrally
(Welln-
ORIGINOFBIRDSANDEVOLUTIONOFFLIGHT
hofer1975).ThisconditionappearstobeancestralforDino-
sauria in that it is retained in Ornithischia (e.g., Ostrom and
Mcintosh1966;Ostrom1970).Incontrast,theprezygapophyses
lieclosertothemidlinesothatthepostzygapophysesareentirely
lateraltotheprezygapophysesinanteriorviewinSauropodo-
morpha(Fig. 3D),
Ceratosauria (Fig.3E),
Camosauria (Madsen
1976), Omithomimidae (Osmolska et al. 1972), Deinonycho-
sauria (Fig. 3F). and Avialae (Fig. 3G).
5)Epipophysispresentonantenorcervicalpostzygapophyses.
Epipophyses(=processusdorsalisofBoas1929,oranapophysis
of Zusi and Storer 1969) on the anterior cervical vertebrae are
absent in Pseudosuchia (pers. obs.), Euparkcria (Ewer 1965),
Omithosuchidae(Bonaparte1975a),andLagosuchus(Bona-
parte1975/)),althoughtheymaybepresentatthebaseofthe
neck in fully adult archosaurs such as Alligator (pers. obs.).
EpipophysesarealsoabsentinPterosauriaancestrallyalthough
they are present in Pteranodontidae (Wellnhofer 1978). The
anterior cervicals lack epipophyses in Dinosauria ancestrally
becausetheseprocessesareabsentinHerrerasauridae*(Gallon
1977) and Ornithischia (e.g., Santa Luca 1980). In contrast to
theancestralconditioninDinosauna,epipophysesarepresent
in Sauropodomorpha (Fig. 3D; and see Hatcher 1901, 1903;
Huene 1908), Ceratosauna (Fig. 3E; and see Welles 1984), Car-
nosauria(Osbom1917),Coinpsognathus(pers.obs.),Coelurus*
(Marsh1881a,1884a),Omithomimidae(Osmolskaetal.1972),
Deinonychosauria (Fig. 3F), and birds (Fig. 3G). Epipophyses
arepresentonatleastthesecondthroughthefourthcervicals,
withthoseontheaxisbeingthemostprominent;lessprominent
epipophysesmaybepresentthroughoutthecervicalseriesin
largetheropods(e.g.,Allosaurus:Madsen1976).Epipophyses
extendcaudallyandsomewhatlaterallyfromthedorsalsurfaces
ofthecervicalpostzygapophyses;theyarelargestanteriorlyand
diminish in size posteriorly. In birds they are associated with
theinsertionsandoriginsofavarietyofdorsalcervicalmuscles,
such as the M. spleni colli, M. spinalis cervicis, M. ascendentes
cervicis,andM.inlcrcristales.Somemusclesinsertdirectlyonto
theepipophyses,andothersindirectlythroughinsertiononan
aponeurosisextendingbackfromtheepipophysis(ZusiandSto-
rer1969).Thecrocodilianhomologuesofthesemuscleshave
yettobedetermined.
6)Hyposphene-hypantmmaccessoryintervertebralarticu-
lationsin
trunkvertebrae.
Accessory intervertebral articulations
areabsentinallPseudosuchiaexceptsomeRauisuchia(Charig
1976/),- Bonaparte 1981) and Aelosauria (M. Parrish, pers.
comm.);sucharticulationsarealsoabsentinEuparkcria*(Ewer
1965),Omithosuchidae(Bonaparte,pers.comm.),Lagosuchus*
(Bonaparte 1915b). Pterosauria (Wellnhofer 1978), Herrera-
sauridae*(Galton 1977),and
Omithischia (Steel
1969).Thus,
accessoryintervertebralarticulationsareabsentinDinosauria
ancestrally. As noted most recently by Steel (1970) and Bakker
and Galton (1974), Theropoda and Sauropodomorpha alone
amongdinosaurspossesshyposphene-hypantraaccessoryin-
tervertebralarticulations in
the
trunk
region.
Hyposphene-
hypantraarticulationsarepresentinSauropodomorpha(e.g.,
Cooper 1981a), Lilicnsicrnus* (Huene 1934), Ceratosauria
(Welles1984),Camosauria(Madsen1976),Coelurus*(Marsh
1884a),Microvenator*(Ostrom1970),Omithomimidae(Os-
molskaetal.1972),andDeinonychosauria(Ostrom1969/)).
ThevertebraeofArchaeoptery.x*arenotexposedappropriately
GAUTHIER:
SAURISCHIAN
MONOPHYL\
todetermineifhyposphene-hypantraaccessoryintervertebral
articulationsarepresent,butotherbirdsdonotpossesshypo-
sphene-hypantra. The
vertebrae
birds
ofare
highly
modified,
however, and in the context of all the evidence it is simpler to
acceptthattheirintervertebralarticulationsarefurthermodi-
ficationsofthesaurischiancondition,nottheretentionofan
ancestralintervertebral articulation.
7) Manus more than 45% of length of humerus plus radius.
The longest digit in the manus (plus its metacarpal) is 28% of
the length of the humerus plus radius in the early crocodile
Protosuchus (Colbert and Mook 1951); this length is propor-
tionatelylongerin
extant
crocodiles
displayingnegative allom-
etryinlimblength(e.g.,37%inCrocodyluspowsusandCaiman
sclerops. pers. obs.). These size relationships appear to be an-
cestralfordinosaursbecauseinomithischianssuchasSciilel-
losaunis* (38%; Colbert 1981), Lesothosaums (26-29%; esti-
matesbasedonGalton1978),andHypsilophodon(34%;Galton
1974), the longest manal digit and its metacarpal is no more
than 38% of the length of the humerus plus radius. The single
exceptionappearstobeHeterodontosaurus.inwhichthemanus
is 56% of the length of the humerus plus radius. This attribute
isconsidereddiagnosticofthistaxonamongOmithischia(Santa
Luca 1980). In contrast, the longest digit and its metacarpal is
at least 45% of the length of the humerus plus radius in Saur-
ischia ancestrally. For example, the manus is 45% to 47% of
thelengthofthehumerusplusradiusinThecodontosaurus*.
60%inEfraasia*(estimatesbasedonHuene1914candGalton
1973a), 47% in Syntarsus (Raath 1969) and Coelophysis (Os-
trom 1969/)), 77% in Allosaums. 75% in Deinonychus. and 58%
inOrnithomimus(Ostrom1969^).
8)Manusmarkedlyasymmetrical.Inarchosaursancestrally,
theinnerdigitsofthemanusarestouterthanaretheouterdigits,
and the third digit is the longest digit in the manus (Fig. 4G).
Theasymmetryofthemanusbecomesmorepronouncedwithin
omithosuchianarchosaursandyieldsafurtherreductionofthe
outertwodigitsofthemanusindinosaursancestrally(Fig.4H-
O).PterosaursareexceptionalamongArchosauria;theenor-
mouslyenlarged fourthdigit
supporting the
wingmembrane can
hardlybeconsideredtheancestralconditionforarchosaurs,and
the hand is not modified this way in Scleromochlus (Huene
1914a).Nevertheless,thepterosaur'sthirddigitisthelongest
of the digits remaining unmodified. Ornithischia retains the
ancestral condition: digit three is longest (Fig. 4H). In contrast
totheancestralconditionindinosaurs,however,theinnertwo
digits(twoandthree)ofthemanusarefurtherenlarged,sothat
thesecond,ratherthanthethird,isnowthelongestdigitinthe
hand in Sauropodomorpha (Fig. 41, J), Ceratosauria (Fig. 4K),
Camosauria(Fig.4L),Ornitholestes*(Osbom1917),Elmisau-
ridae*(Osmolska1981),Caenagnathidae(Barsbold1983),Or-
nithomimidae (Fig. 4M), Deinonychosauria (Fig. 4N), and Avi-
alae(Fig.40).Asymmetricallydevelopedhandsarecharacteristic
of Archosauria, and this modification becomes more pro-
nouncedwithin Omithosuchia, culminatingin
themarkedly
asymmetricalhandsofbirds.Tobecomplete,aprocess-related
theoryoflimbdevelopmentmustaccountforthispeculiarity;
likewise, one must be cautious in developing a general theory
ofhandmorphogenesisbasedlargelyorexclusivelyonthehands
ofextantbirds.
9) Bases of metacarpals IV and V lie on palmar surfaces of
17
manaldigitsthreeandfourrespectively.Theancestralcondition
in Sauria (=extant diapsids; Gauthier 1984) is that the bases of
the medial metacarpals overlap the bases of the lateral meta-
carpals.ThisconditionisretainedbyPseudosuchia(Fig.4G),
Euparkeria* (Ewer 1965), Omithosuchidae (Walker 1964),
Pterosauria(Wellnhofer1978),andOmithischia(Fig.4H).Un-
likeotherarchosaurs.however,thebaseofmetacarpalIV,and
to a greater extent that of V, lie more on the palmar surface of
the hand in Sauropodomorpha (e.g., Janensch 1922; Cooper
1981a.-740, fig. 37, 38). The base of the fourth digit of early
theropodssuchasCeratosauriaalsoliesonthepalmarsurface
of the base of the third digit (e.g., Welles 1984:153, fig. 37; I
thankT.Roweforpointingoutthatthesauropodomorphcon-
ditionalsoappliestotheropods).Thefifthmanaldigitisabsent
from the ontogeny of extant Theropoda; however, Colbert's
unpublisheddrawingofanintacthandofCoelophysisreveals
a nubbin of bone lying on the palmar surface of the base of
metacarpal IV. Because this piece of bone lies in the same po-
sitionasthefifthdigitinsauropodomorphs,Colbert'sinterpre-
tationofthiselementasaremnantofthefifthdigitisprobably
correct. In more derived Theropoda the fourth manal digit is
lost, although it is retained in embryos of extant birds, and a
small remnant of this digit has been observed on the palmar
surface of the hand in Ornitholestes* (Osbom 1917; see Part
III,character45forfurtherdiscussion).
10)Saurischianpollex.InArchosauriaancestrally,metacarpal
I is only a little shorter than II, phalanx I of the first digit is
shorterthanmetacarpalI,andtheclaw-bearingungualisneither
verylargenorsharplypointed(Gauthier1984).Thiscondition
isretainedinPseudosuchia(Fig.4G).Asidefromtheoffsethead
ofmetacarpalIandarelativelylargerandmoresharplypointed
ungual, the first digit and its metacarpal in Omithosuchidae
(Bonaparte1975a)andOmithischia(Fig.4H)isasinarchosaurs
ancestrally.Manaldigitone,thepollex,ofTheropodaandSau-
ropodomorpha differs
from
that
other
of Archosauria
several
in
ways.First,thepollexismorerobustandbearsalargerungual
phalanx (Fig. 4I-0); this character is extreme within Sauropo-
domorpha(Fig.41,J).Second,metacarpalis Ionlyhalforless
the length of metacarpal II, and the distal condyles are more
markedlyasymmetrical(Fig.41.K;seeWelles1984).Third,the
firstphalanxismuchlongerthanmetacarpalI;thefirstphalanx
equals or exceeds the length of any other phalanx in the hand
(Fig. 4I-0). Galton (1971), Bakker and Galton ( 1 974), and Baird
(1980) discussed the grasping ability of dinosaur hands, and
commentedontherangeofmotionpossibleinthesaurischian
pollex, noting that the articular surfaces allow a fairly precise
reconstructionoftherangeofpossiblemovements.Galton(1971)
describedhowthearticularsurfaceswithinthesaurischianpol-
lexforcetheclawtodivergeandpointinwardduringextension
and to converge with the second and third digits and point
downwardsduringflexion;atmaximumextensiontheclawpoints
inward to a greater degree in Sauropodomorpha than in The-
ropoda.AsMassospondylusindicates(Fig.4J),themanuswas
modifiedtoplayagreaterroleinsupportandlocomotionearly
insauropodomorphhistory;theentirehandbecamebroader,
shorter,andmorerobust,andthefirstphalanxandmetacarpal
werelikewiseshortened.Asidefromthelargeclaw,scarcelyany
evidence of the grasping hand of saurischians remains in the
elephantinehandsofSauropoda(e.g.,Janensch1922).Although
18
the first digit is seldom used for grasping in omithurine birds,
its functional independence has been conserved. Indeed, the
dinosaurian modifications of the first digit toward the alula,
appearstohavebeenessentialtothedevelopmentofpowered
flight in birds (Bellairs and Jenkin 1960).
Saurischianmonophylyissupportedbysharedapomorphies
inconstructionofthesnout,patternofhypertrophyofthetem-
poralmusculature,elongationofthecervicalregion,modifi-
cationoftheaxialzygapophyses,presenceofepipophysesinthe
anteriorcervicals,accessoryintervertebraljoints,andavariety
of distinctly avian modifications of the manus. Possible alter-
nativehypothesesarelessabletoaccountforobservedpatterns
ofsharedapomorphies;accordingly,Sauropodomorphaand
Theropoda(includingbirds)arehypothesizedtobesister-groups
within Saurischia, and Omithischia is considered to be the sis-
ter-group Saurischia
of within
Dinosauria.
II.PhylogeneticRelationshipswithinTheropoda
Theropoda
(n. comb.="Theropoda" Marsh, 1881ft, plus Aves Linne, 1758)
Temporal Range—late Tnassicto
Recent,
Included
aw.
tProconipsognathus*,
— Lihcnstcnnts*.
Ceratosauna. Camosau-
Omithomimidae.
ria, Compsognathus. Cacnagnalhidae.
Elmisaundae*. Microve-
nalor*.
Coelwus*.
Saiirornilholesrcs*.
Hulsaupcs.
Ornitholesles*.
Deinonycho-
sauria.andAvialae.The
readeris
referred to
Olshevsky (1978)
andWelles (1^84)
foradditionalnonavian theropod taxa
that
were
notconsidered in
this
investi-
gation.
Diagnosis.—Theropodaisdefinedostensivelytoincludebirds
and all saurischians that are closer to birds than they are to
sauropodomorphs.Sauropodomorpha,Omithischia,Herrera-
sauridae*,Pterosauria-La^osHc/!;«,Omithosuchidae,Eupar-
keria*. and Pseudosuchia will be used as successively more
remoteoutgroupsinthefollowinganalysis.Theropodsaredis-
tinguishedfrom othersaurischiansby
the
following synapo-
morphies.
11)Reducedoverlapofdentaryontopostdentarybonesand
reducedmandibularsymphysis.Manyextantbirdsdisplaysome
degreeofintramandibularmobility,andseveralauthorshave
noted that the construction of nonavian theropod mandibles
alsoallowsintramandibularmobility(e.g.,Gingerich1976).Ex-
tantPseudosuchia have solidly
constructed mandibles thatshow
no indication of intramandibular mobility. That is to say, croc-
odiliansdisplayabroadoverlapbetweendentaryandpostden-
tarybones(withthedentaryextendingtotheleveloftheorbit),
a process of the dentary passing dorsal to the mandibular fe-
nestra,andaprominentmandibularsymphysis.Eachofthese
attributesindicatestheabsenceofintramandibularmobility,
andtheyareretainedinnontheropodomithosuchians,including
Omithischia (Fig. IC, D; Romer 1956) and Sauropodomorpha
(Fig.IE,F;Gallon1984).Incontrast,themandibularsymphysis
is reduced in Theropoda and the overlap of the dentary onto
the postdentary bones is very reduced, such that the antero-
dorsallyslopingposterolateralmarginofthedentaryterminates
antorbitally (Fig. IG-L). This region of the mandible is poorly
preservedinArchaeopteryx*.butGregory(1952)describedthe
similaritiesbetweentheintramandibularjointsofIchlhyorms
and Hesperomithes and those of some squamates with intra-
mandibularkinesis. Manyextantlineagesof
Aves retain
the
ancestral relations of theropod dentary and postdentary ele-
mentsandintramandibularmobility(e.g..Lams).However,
ORIGINOFBIRDSANDEVOLUTIONOFFLIGHT
birdswithnoncamivorousdiets,andshortandstoutjawswith
broadmandibularsymphyses,donot.
12) Lacrimal broadly exposed on skull roof The lacrimal
formstheposterodorsalandposteriormarginsoftheantorbital
fenestra, but does not participate in the formation of the skull
roof in Pseudosuchia (Bonaparte 1981), Euparkeha* (Fig. lA),
Omithosuchidae(thelacrimalgainssomeexposuredorsallyas
part of the supraorbital cornice; Fig. 1 B), Pterosauria (Welln-
hofer 1978), Omithischia (Fig. IC, D), and Sauropodomorpha
(Fig. IF). In contrast to the ancestral condition, the lacrimal
formsmuchoftheskullroofanteriorandlateraltotheprefrontal
abovetheorbitinPwcompsognathus*(pers.obs.),Ceratosauria
(Welles 1984), Camosauria (Fig. IG, H), Omithomimidae (Fig.
II),Deinonychosauria(Fig.IJ,K),Compsognathus{pers.obs.),
Ornillwlestes*(Osboml9l7),Caenagnathidae(Barsbold1983),
and Avialae (Fig. IL; see Part V, character 67).
13)Presenceofmaxillaryfenestra.Anaccessoryfenestrawith-
intheantorbitalfossaisabsentinallPseudosuchia(Krebs1976),
Euparkena* (Fig. lA), Omithosuchidae (Fig. IB), Pterosauria
(Wellnhofer 1978), Omithischia (Fig. IC, D), and Sauropodo-
morpha(Fig.IE,F).Thus,amaxillaryfenestraisabsentin
Saurischia ancestrally. In contrast, a small fenestra lies at the
anteriormarginoftheantorbitalfossainCeratosauria(T.Rowe,
pers. comm.), and a larger fenestra lies in a more posterior
position within the antorbital fossa in Camosauria (Fig. IG, H),
Ornitholestes*(Osbom1917),Compsognathus(Ostvom1978),
Caenagnathidae (Barsbold 1983), Omithomimidae (Fig. II).
Deinonychosauria (Fig. IJ, K), and Archaeopteryx* (Fig. IL).
This character is absent and has presumably been lost in Or-
nithurae(seePartIII,character37).
14) Vomers fused anteriorly. The vomers are short, narrow,
andpairedinArchosauriaancestrally(Gauthier1984);tojudge
fromtheconditionseeninPseudosuchia(asidefromthecom-
parativelylong
vomers of
aetosaurs.Walker
1961),
Euparkeria*
(Fig. 2A), Omithosuchidae (Fig. 2B), and Pterosauria (Welln-
hofer1978)thisconditionappearsancestralforOmithodira
(AppendixA).Incontrast,thevomersareelongate,extending
well posterior to the level of the anterior limit of the palatine
in Omithischia ancestrally (Galton 1974; Heaton 1972), Sau-
ropodomorpha ancestrally
(Huene
1906,
1908,
1932;
Galton
1984),andTheropodaasidefromNeognathae(pers.obs.),and
this appears to be the ancestral condition in Dinosauria. Few
nonavian theropod fossils have this region of the skull pre-
served.Theropoda differsfromdinosaurs,
otherthanthyreo-
phoran omithischians (P. Sereno, pers. comm.), in that the vo-
mersareindistinguishablyfusedanteriorly(althoughtheymay
be paired in a few neognaths, pers. obs., and inHesperornis,L,
Martin, pers. comm.). Nevertheless, the presence of the apo-
morphicconditionintheropodsasdiverseasProcompsogna-
thus* (Ostrom 1981), Ceratosaurus (T. Rowe, pers. comm.),
Dcmonychus (Ostrom \969b), Allosaurus (Fig. 2C), Tyranno-
saurus (Fig. 2D), Oviraptor (Osmolska 1976), Gobipteryx (El-
zanowski 1977, 1981 ), and Ratitae and Tinami (Huxley 1867),
suggeststhatthevomersarefusedanteriorlyinTheropodagen-
erally.Walker
(1964)
described the
enlarged,
diamond-shaped
anteriorendsofthevomersinOrmthosuchusasbeingsimilar
to the condition of the fused anterior third of the vomer in
Tyrannosaurus.However,thissimilarityobtainsbetweenTy-
rannosaurusandOrnithosuchusalone.Moreover,thepairedand
shortvomersofthelatter(Fig.2B)areotherwiseplesiomorphic
GAUTHIER:
SAURISCHIAN
MONOPHYLY
with respect to those of all other theropods. including Tyran-
nosaiinis(Fig.2D).
15)Expandedectopterygoidwithventralfossa.Colbertand
Russell(1969)arguedthatinearlytheropods(e.g.,Ceratosauria)
the ectopterygoid was "simple" and without a ventral fossa.
However, I have observed this fossa in the ceratosaur Coelo-
physis. I agree with the observation that there is an expanded
ectopterygoidwithamoreorlessprominentfossaventrallyin
themainbodyoftheelementinCamosauria,Omithomimidae,
andDeinonychosauria(ColbertandRussell1969).Saurornilho-
lestes* (Sues 1978), and in Caenagnathidae (Barsbold 1983).
This character cannot be determined in Archaeopleiyx*. and
theectopterygoidhasnotbeenidentifiedwithcertaintyinOr-
nithurae(seeMcDowell1978).Inthecontextofalltheevidence,
birds are considered to have attained their current condition
fromanectopterygoidlikethatseenintheropodsgenerally.
16) First intercentrum with large occipital fossa and small
odontoid notch. The occipital fossa on the anterior face of the
first intercentrum is more than three times as wide as it is tall
andtheodontoidnotchisconsequentlybroadanddeepinPseu-
dosuchia (pers. obs.), Euparkeria* (Ewer 1965). Omithischia
(Fig. 3A), and Sauropodomorpha (Hatcher 1901). In contrast,
the occipital fossa is only about twice as wide as it is tall and
theodontoidnotchisconsequentlysmallerinCeratosauria(Fig.
3B),Camosauria(Madsen1976),Omithomimidae(Osmolska
et al. 1972), Deinonychosauria (Ostrom 1969ft), and Avialae
(Fig.3C).
17)Secondintercentrumwithbroad,crescenticfossaante-
riorlyforreceptionoffirstintercentrum.Thearticularsurface
on the anteroventral margin of the axial (2nd) intercentmm is
convexinPseudosuchia(pers.obs.).Dinosaursdifferfromthe
ancestralconditioninArchosauriainthatthisarticularsurface
is at least partly concave (Fig. 3D). Compared to other dino-
saurs,however,thearticularsurfaceontheaxisforthefirst
intercentrumformsabroad, deep, and concave fossa in Cera-
tosauria(Fig.
3E).
Camosauria (Gilmore
1920),
Omithomim-
idae(Osmolskaetal.1972),Deinonychosauria(Fig.3F),and
Avialae(Fig.3G).
18)Pleurocoelouspresacralvertebrae,particularlyincervical
region.Fenestraleadingintohollowcentra(=pleurocoels)are
absentinPseudosuchia(Romer1956),Euparkeria*(Ewer1965),
Omithosuchidae(Walker1964),Lag05HcVn«*(Bonaparte1975ft),
Herrerasauridae(Galton1977),Omithischia(Romer1956),and
Sauropodomorphaancestrally(Cooper1981a).Thus,nonpleu-
rocoelousvertebraearetheancestralconditionforSaurischia.
In contrast, pleurocoels are present in all Theropoda; Ostrom
(1978) discussed the distribution of this character within the
group,notingregionalvariationsinthepresacralcolumn.Pleu-
rocoelsarepresentin
Ceratosauria (Gilmore
1920),
Camosauria
(Madsen1976),Omithomimidae(Osmolskaetal.1972),5aHr-
ornitholestes*(Sues1978),Coelwus*(Marsh1884a),Microve-
nator*(OsiTom1970),Deinonychosauria(Ostrom1969ft),and
Avialae(Ostrom1976a). Analogous modifications are known
inPterosauria(Wellnhofer1978)andSauropoda(Marsh1896),
butthisisconsideredconvergencebecausepleurocoelsareab-
sentintheirrespectiveoutgroupsamongsauropodomorphs,
omithischians, herrerasaurs*, andLagosuchus*.
19) At least two additional vertebrae incorporated into sac-
mm (including at least one from the caudal series and at least
onefromthetrunk).Asarguedabove,threesacralsarepresent
19
inDinosauriaancestrally.Incontrast,notheropodinwhichthe
sacmmiscoossifiedandintacthasfewerthanfivesacrals.Among
theropods,thecharacterhasbeenreportedinCeratosauria(Raath
1969), Camosauria (Steel 1970; Madsen 1976), Omithomimi-
dae(Osbom1917),Deinonychosauria(Barsbold1974),and
Avialae(Ostrom1976a).Omithuraehasmanymorevertebrae
in the sacrum, a character that is by no means unique to The-
ropoda.Asnotedabove,asacrumconsistingofatleastfourto
five vertebrae may also be ancestral for Omithischia. The Het-
erodontosaurus(SantaLuca1980)andomithopod(Galtonand
Jensen 1973) lineage of Sereno (1984) has at least five or six,
whichisalsothelowestnumberreportedinpachycephalosaurs
andceratopsians,withthelattergrouppossessingasmanyas
elevensacrals(Steel1969).Inaddition,subgroupswithinPtero-
sauria(Wellnhofer 1978)
andSauropoda (Berman and
Mcintosh
1978)havefiveormoresacrals.
20) Transition point in tail (sensu Russell 1972). In Dino-
saunaancestrally,thecaudalzygapophysesareshortandver-
ticallyoriented,andthetransverseprocessesarepresentpos-
teriortothemiddleofthecaudalseries(SantaLuca1980;Cooper
1981a). In contrast to the condition retained by Sauropodo-
morpha,however, in
Theropoda theneural arches
andthe
trans-
verseprocessesarereducedposteriorlysothattheyareabsent
in most of the posterior half of the tail. In addition, the caudal
prezygapophyses in the posterior half of the tail are elongate,
pointedanteriorly,andclasptheelongate,blocklikepostzyga-
pophysial moiety. Finally, the caudal haemal arches in at least
theposteriorthirdofthetailaredepressedandboat-shapedin
lateral outline (e.g.. Fig. 3E, F). The degree of transformation
in each of these aspects of caudal morphology is not precisely
correlated.Althoughthezygopophysesmaybeginelongation
prior to complete loss of transverse processes, with modified
haemalarchessubsequentlyappearingfurtherposteriorly,these
transformationstakeplacewithinafewvertebraeofoneanother;
accordingly, they are described collectively as the "transition
point."Theleast-modifiedtailsseenamongtheropodsarepres-
entinCeratosauria(Raath1969).However,thetransformation
incaudalformismoreprofoundandthetransitionpointbegins
closer to the base of the tail in Camosauria (Lambe 1917),
Compsognathus(OsXrom1978),Ormtholestes*(Osbom1917),
Omithomimidae(Osmolskaetal.1972),Caenagnathidae(Bars-
bold1983),Deinonychosauria(Ostrom1969ft),andArchaeop-
lery.x* (see Part III, character 40). The tail is further modified
in Omithurae, in that the proximal caudal zygapophyses are
lost, but the tail still retains mobile proximal and stiff" distal
portions(i.e.,thepygostyle).Moreover,inHesperomithesthat
possessamultisegmentedpygostyleintheadult,thesecaudal
vertebraehavelosttheneuralspinesandtransverseprocesses.
Althoughlackingtheventralkeel,boat-shapedhaemalarches
areretainedintheproximalcaudalsinHesperomithes(L.Mar-
tin,pers.comm.)andinseveralotheromithurinebirds,such
as Ichthyornis. penguins, and loons (Marsh 1880). A similarly
stiffened posterior part of the tail arose convergently in Stau-
rikosaitrus*(Galton1977);andvaguelysimilarmodifications
are present in ankylosaurs, in which the distal extremity of the
tail bears a club (Coombs 1978a). An analogous condition, al-
thoughhighlymodifiedinafashionsimilartodromaeosaurs,
arose convergently within Pterosauria (i.e., the tail in Scle-
romochlus*isnotstiffened).ThetailofStaurikosaimis*could
beconstruedasevidenceofarelationshiptotheropods.How-
20
ever, so far as Staurikosaurus* is preserved, it lacks the diag-
nosticcharactersofSaurischia,anditsharesanexpandeddistal
end of the pubis with Herrerasaums* (Gallon 1977; and see
Gauthier 1984).
21) Enlarged distal carpal I overlaps bases of metacarpals I
and II. Few early archosaurs have well-preserved hands, but
aetosaur(Sawin1947)andembryocrocodilianpseudosuchians
(pers. obs.) are Uke Omithischia (e.g.. Fig. 4H) in retaining the
ancestral saurian condition, in which the distal carpals are re-
strictedtothebasesoftheirrespectivemetacarpals.Thisregion
isnotpreservedinEfraasia*.InThecodontosaitnis*distalcarpal
I is also confined to the base of metacarpal I, but the base of
metacarpalIIanditsassociateddistalcarpalarenotpreserved
(Huene 1914c). The only reasonably complete hands of early
sauropodomorphsarethoseofMassospomiylusandsomepla-
teosaurs (e.g., Huene 1932; Young 1941, 1947, 1958; Galton
and Cluver 1976:135, fig. 7). Based on published illustrations,
personalcommunicationsfromP.Galton,andpersonalobser-
vationoiPlatcosaunis.itappearsthatdistalcarpal1isrestncted
to the base of metacarpal I in Sauropodomorpha ancestrally.
Massospondyhts may be an exception, in that distal carpal I
overlapsdistalcarpalII,althoughtheformerisseparatedfrom
metacarpal II by the latter (Fig. 4J; and see Cooper 1981^.737,
fig.32).Incontrast,Hallicosaiirus*(Huene1934),Ceratosauna
(Fig. 4K), Carnosauria (Fig. 4L), Coclunis* (Ostrom 19766(),
Caenagnathidae(Barsbold1983),Deinonychosauria(Fig.4N),
and Avialae (Fig. 40) are distinguished from other archosaurs
inthatanenlargeddistalcarpalIoverlapsthebasesoftheinner
twometacarpals,thusfunctionallyintegratingdigitsIandIIin
the wrist, just as in extant birds. In theropods in which distal
carpals I and II are present, a small distal carpal II lies distal
andlargelyposterolateraltodistalcarpalI;thusdistalcarpalII
lies at least partly between distal carpal I and metacarpal II
(Madsen 1976). Distal carpals I and II are fused in Synlarsus
rhodesiensis:theadultstatusofthetype-specimenisindicated
byseveralotherfusionsinthepostcranialskeletonmarkingthe
cessationofgrowth(Raath1969).AccordingtoMadsen(1976),
in Allosaurus fragilis the larger distal carpal I and the smaller
distal carpal II fuse to one another very late in postnatal de-
velopment.Omithomimidae is
anexception
in
thatits
wnst
is
composedofsmallandpoorlyossifiedcarpalsthatlackarticular
facets (Osbom 1917). Omithomimids are thus like tyranno-
sauridcamosaurs(Barsbold1983)inthatthecarpalsappearto
bearrestedatajuvenilestageofdevelopment.Noseparatedistal
carpalIIhasbeenreportedinanyadultdeinonychosaurorbird;
theelementsaresaidtoanseseparatelyinbirdembryos(Heil-
mann 1926), although Hinchliffe and Hecht (1984) have been
unabletoidentifymorethanasinglecondensationinGallus.
22) Manal digit V reduced to a vestige or absent. The fifth
manal digit is present in all Pseudosuchia (e.g.. Fig. 4G). Eii-
parkcna*(Ewer1965),andOmithosuchidae(Bonaparte1975a).
And, although the digit is reduced in Dinosauna ancestrally, it
is retained by Omithischia (Fig. 4H) and Sauropodomorpha
(Fig. 41, J). Thus, a fifth manal digit is present in Saunschia
ancestrally.Incontrast, all that remains of the fifth manal digit
in Ceratosauna is the small metacarpal splint lying at the base
ofthepalmarsurfaceofmetacarpalIVinCoelophysis(Co\htr\,
pers.comm.).Nootherceratosaursarereportedtohaveaves-
tigialfifthdigit,butCoelophysisistheonlyceratosaurthatis
wellenoughpreservedtobeabletodiscriminatebetweenab-
ORIGINOFBIRDSANDEVOLUTIONOFFLIGHT
senceandnonpreservation.Novestigeofthefifthmanaldigit
has been reported in Carnosauria (Fig. 4L), Compsognalhus
(Bidar et al. 1972), Saurornitholestes* (Sues 1978), Ornitho-
lestes*(Osbom1917),Caenagnathidae(Barsbold1983),Elmi-
sauridae* (Osmolska 1981), Omithomimidae (Fig. 4M), Dei-
nonychosauria(Fig. 4N),
orAvialae
(Fig.
40).
Thefifth
manal
digit is also missing in the ontogeny of extant birds (Heilmann
1926). Liliensternus* is described as retaining a reduced fifth
digit, but Huene's (1934) description and illustration of what
remainsofthehandspeaksagainstsuchaninterpretation;the
metacarpals he interpreted as being III and IV were capped by
anenlargedcarpal,thuscorrespondingtometacarpalsIandII
ofothertheropods,andthereisnoevidenceofafifthdigit.Loss
ofthefifthmanaldigitaroseconvergentlyinPterosauria(Welln-
hofer 1978).
23) Manal digit IV reduced or absent in adult. As discussed
above, manal digits IV and V are reduced in Dinosauria an-
cestrally.Theropodaisunique,however,inthatmanaldigitIV
isneverlongerthanmetacarpalIII,anditisneverrepresented
by more than a metacarpal with a vestigial phalanxonitsdistal
extremity. Ceratosauria (Fig. 4K) retains the vestigial manal
digit IV just described, but the fourth digit is reduced to a mere
nubbin of bone or is absent in postembryonic development in
Carnosauria (Fig. 4L), Ornitholesles* (Osbom 1917). Caenag-
nathidae(Barsbold 1983),
Cimipsognalhus (Bidaret
al.
1972),
Saurornitholestes*(Sues1978),Elmisauridae*(Osmolska1981),
Omithomimidae (Fig. 4M), Deinonychosauria (Fig. 4N), and
in all birds beyond embryonic stages (Fig. 40; Heilmann1926).
The ancestral theropod phalangeal formula is thus 2-3-4-1-0,
rather than 2-3-4-3-2 as it is in the hands of Dinosauria and
Saurischia ancestrally.
24)Manuswithelongatepenultimatephalanges.Thepenul-
timatephalangesarcshorterthanthemoreproximalelements
ineachdigitinPseudosuchia(Fig.4G)andOmithischia(Osbom
1924a;Heterodontosaunisisapomorphicamongomithischians
in this regard, see Fig. 4H). The same can be said for digits two
throughfiveinSauropodomorpha,exceptthatlikeothersaur-
ischians the first metacarpal is relatively short (e.g.. Fig. 4J).
Thus, a hand with the same intemal proportions as archaic
sauropodomorphssuchasThecodontosaurus*(Fig.41)appears
tobeancestralforSaurischia.Theropodsarefurtherderivedin
thatthepenultimatephalangesonthefunctionaldigitsareuni-
formlylonger thanthemore
proximal elements comprising their
respectivedigits.Thisapomorphyappearsinitsmostgeneral
form among Ceratosauria (Fig. 4K), but it is also present in
Camosauria (Fig. 4L); it is further developed in Saurornitho-
lestes*(Sues1978).Caenagnathidae(Barsbold1983),Elmi-
sauridae*(Osmolska 1981),
Omithomimidae (Fig.4M),Or-
mtholestes* (Osbom 1917), Deinonychosauria (Fig. 4N), and
Archaeopteryx* (Fig. 40). Omithurine birds retaining an un-
modifiedfirst
digit
usually
maintainits
ancestralproportions.
However, the first phalanx in manal digit II may be subequal
to or longer than the second phalanx, and the third digit often
retainsonlyonephalanx.Theomithurinecarpometacarpuscan-
notbeconsideredplesiomorphic,andthewayinwhichthe
omithunnemanusdiffersfromthatofothertheropodsiscon-
sideredto
have arisensecondarily.
Pterosaurs alsohaveelongate
penultimate phalanges in digits I through III, and this is con-
sideredyet
another exampleof
convergence between pterosaurs
and
theropods.
GAUTHIER:
SAURISCHIAN
MONOPHYLY
25)ManaldigitIIIwithshortfirstandsecondphalanges.The
thirdphalanxofdigitIIIisshorterthaneitherthefirstorsecond
inPseudosuchia(Fig.4G),Omithischia(Fig.4H),andSauropo-
domorpha (Fig. 41, J), so this condition is ancestral for Saur-
ischia.Incontrast,thefirstandsecondphalangesofmanaldigit
III are short, so that the third phalanx is the longest element in
Ceratosauria (Fig. 4K) and Camosauria (Fig. 4L), and this char-
acterisevenmoremarkedlydevelopedinOmithomimidae(Fig.
4M),Ornitholestes*(Osbom1917),Caenagnathidae(Barsbold
1983),Elmisauridae*(Osmolska1981),Sauromitholestes*(Sues
1978), and Deinonychosauria (Fig. 4N). This condition is an-
cestralforAvialae(Fig.40),butitisabsentinthediagnostically
modified carpometacarpus of Omithurae. in which only one
phalanx remains in manal digit III (Heilmann 1926). This char-
acterhasarisenconvergentlyinPterosauria,someofwhichmay
have very short first and second phalanges as in the subgroup
of theropods of which birds are a part (see below), but others
may have only the first phalanx shortened (Wellnhofer 1978);
the level at which these characters arose within Pterosauria is
unknown.
26)Manalungualsenlarged,compressed,sharplypointed,
stronglyrecurved,andwithenlargedflexortubercles.Asnoted
above,modificationofthepollexinOmithosuchidaeindicates
that the ability to grasp with the hands unites a more inclusive
groupofOmithosuchiathanDinosauriaalone.Nevertheless,
withthepossibleexceptionofPterosauria(Wellnhofer1978),
noarchosaursexcepttheropodshavemanalungualssomodified
as to indicate that they played an important role in securing
prey(Ostrom\9(>9b).Impressionsofclawsheathsarepreserved
onlyinCompsognathus,Archaeopteryx*(pers.obs.),andChi-
rostenotes*(Currie,pers.comm.)amongMesozoictheropods,
andclawsheathmorphologycorroboratesestimatesofclaw-
form extrapolated from ungual morphology (Ostrom 1978).
Basedonungualmorphology,raptorialclawsarepresentinthe
hands of Liliensterniis* (Huene 1934), Ceratosauria (Gilmore
1920;pers.obs.),Camosauria(Madsen1976),Coelurus*(Os-
trom1976a), Ornitholestes* (Osbom 1917),
Compsognathus
(Ostrom1978),Sauromitholestes*(Sues1978),Microvenator*
(Ostrom1970),Caenagnathidae(Barsbold1983),Elmisauridae*
(Osmolska 1981), Deinonychosauria (Russell 1969; Ostrom
1969^ 1974/7), and Archaeopteryx* (Ostrom 1976a). The ju-
veniles,andoccasionallytheadults,ofawidevarietyofextant
birds may retain clawed digits, but with the notable exception
ofjuvenileHoatzin(Opisthoconnts)theyarevirtuallynonfunc-
tionalinmodembirds(Heilmann1926).Asnotedabove,an
enlarged and sharply pointed first ungual is an ancestral con-
ditionforSaurischia;however,eventhisclawonlyapproaches
the level of specialization seen in Theropoda (compare Mas-
sospondyhisinCooper1981a.'748,fig.45,withDeinonychusin
Ostrom 1969/); 108, fig. 63). In the context of all the evidence,
the hands of Omithomimidae and Omithurae are considered
secondarilymodifiedinthisregard.
27) Preacetabular part of ilium enlarged and extending far
forward of acetabulum. A prominent iliac spine arose prior to
theoriginofArchosauriawithinArchosauromorpha(Gauthier
1984).TheancestralconditionisretainedbyPseudosuchia(Ro-
mer1956),Euparkena*(Ewer1965),Omithosuchidae(Bona-
parte1975a), Lagosuchus* (Bonaparte 1915b), and
Sauropo-
domorpha (Fig. 5B). Omithischians have a diagnostically
elongate iliac spine (Fig. 5A). In contrast, Colbert (1964) noted
21
that an enlarged preacetabular portion of the ilium obtains in
all theropods (e.g.. Fig. 5C-I). The increased length of the the-
ropod ilium is probably correlated in a general way with the
added number of sacrals. The character is not entirely redun-
dant,however,becausetheadditionofsacralvertebraehasnot
been accompanied by the same modifications of the ilium in
otherarchosaurs.Rowe(pers.comm.)pointedoutthattheM.
puboischiofemoralisintemus(2),whichoriginatesbeneaththe
posteriortransverseprocessesincrocodiles,hasmovedontothe
enlargedanteriorportionoftheiliuminTheropoda.BothWalk-
er(1977)andM.Parrish(pers.comm.)believethatthismuscle
originatedfromthemedialpartofthepubisandthatthedorsal
shift of origin of the pifi 2 (one of the so-called M. iliotro-
chantericus group found in Aves) to the transverse processes
appears to be correlated with reduction of the pubis in croco-
dilians. Their hypothesis strikes me as ad hoc. however, given
thatthismusclehasadorsalorigin,ratherthanaventralorigin
from the pubis, in the only archosaurs in which it can actually
be
observed.
28) Pronounced brevis fossa. The brevis fossa is a modified
area of origin for the M. caudofemoralis brevis on the ventral
surfaceofthepostacetabularportionoftheilium(Romer1923,
1927; Walker 1977). This muscle is largely a retractor of the
hindlimbinarchosaurswithlongilia(M.Parrish,pers.comm,).
Possession of a brevis fossa (or shelf) is an ancestral condition
inDinosauria,althoughitisalsopresentincursorialRauisuchia
(Bonaparte, pers. comm.; and see Appendix A). However, the
brevisfossaismostmarkedlydevelopedinTheropoda,inwhich
thereisabroad,deep,andelongatefossaontheposteroventral
marginoftheilium(seeMadsen1976:145,fig.46b).Theprom-
inentshelfformingpartofthebrevisfossaoftengivesthepos-
teriorextremityoftheiliacbladeasquared-off,trancatedprofile
in lateral view (Fig. 5C-E). An enlarged brevis fossa is present
in Liliensterniis* (Huene 1934), Ceratosauria (Welles 1984),
Camosauria(Madsen1976),Ornitholestes*(Osbom1917),Or-
nithomimidae (Osmolska et al. 1972), Elmisauridae* (Currie,
pers.comm.),Caenagnathidae(Barsbold1983),andDeinony-
chosauria(Ostrom \969b, 1976a).
Thedinosaurian origin
of
thismuscleisretainedinbirds(Romer1923),butabrevisfossa
as such appears to be absent in Avialae; in light of all the evi-
denceitissimplertoacceptreversalratherthanplesiomorphy
asanexplanationforitsabsenceinbirds.
29) Femur bowed in a convex arc and sigmoidal curvature
lessprominent.ThefemurhasasigmoidalcurvatureinArcho-
sauriaancestrally(Romer1956),inthattheshaftoftheelement
is S-shaped in two planes (Padian, in press). Dinosaurs retain
thiscondition,becauseasigmoidalfemurispresentinHerrera-
saundae* (Gahon 1977), and in Omithischia (Colbert 1981)
andSauropodomorphaancestrally(Cooper1981a),thusindi-
catingthatasigmoidalfemuristheancestralconditionforSaur-
ischia.Incontrast,thefemurisboweddorsallyandthedistal
end is inflected laterally so there is little or no sigmoidal cur-
vaturein
Procompsognathus* (Ostrom1981),
Liliensternus*
(Huene 1934), Ceratosauna (Gilmore 1920; Raath 1969), Coe-
lurus*(Ostrom 1976a), Compsognathus* (Ostrom 1978),
Or-
nitholestes* (Osbom 1917),
Microvenator* (Ostrom1970),
Or-
nithomimidae (Osmolska et al. 1972), Elmisauridae* (Currie,
pers.comm.),Caenagnathidae(Barsbold1983),Deinonycho-
sauria(Ostrom1976a),andAvialae(Ostrom1976/)).Thefemur
ISlessbowedinlargetheropods,especiallyinCamosauria,and
22
Ostrom(1976/))suggestedthatthisreflectsconstraintsimposed
by large size. The bowed and nonsigmoidal femur arose con-
vergently in Pterosauria (Padian 1983), and it appears to be
ancestralforomithopodomithischians(GaitonandJensen1973).
This character is evidently related to small size and highly de-
velopedcursorialhabitsinOmithosuchia(Coombs\97Sh:
Padian 1983).
30) Fibula closely appressed to tibia, and fibula attached to
crestonlateralsideofproximalendoftibia.Thefibulaisbroadly
separated from the tibia for most of its length, and there is no
fibular crest on the tibia, in Pseudosuchia (Krebs 1976). Eti-
parkena*(Ewer1965),Omithosuchidae(Bonaparte1975a),and
Lagflsuchus (Bonaparte \915b). This condition is retained in
Omithischia ancestrally (Romer 1956) and in Sauropodomor-
pha(Cooper1981a).Inlargedinosaurs,thetibiaandfibulaare
evenmorebroadlyseparatedfromoneanother.However,re-
gardlessofsize,thefibulaisalwayscloselyappressedagainst
the lateral face of the tibia in Theropoda. This character is
uniquetoTheropodaamongSaurischia(Ostrom1976a),butit
aroseconvergentlyinPterosauria(Wellnhofer1978)anditap-
pearstobetheancestralconditionforomithopodomithischians
(Gaiton and Jensen 1973). Unlike either of the last mentioned
taxa, however, only theropods possess the fibular crest on the
tibia. This synapomorphy is present in Liliensternus* (Huene
1934),Procompsognathus*(Ostrom1981),Compsognalhus
(Ostrom1978),Camosauria(Fig.6A),Microvenator*(Ostrom
1970),Oniitholeslcs*(Osbom1917),Caenagnathidae(Barsbold
1983), Omithomimidae (Fig. 6C). Deinonychosauria (Fig. 6B).
and Avialae (Bellairs and Jenkin 1960).
31) Metatarsus narrow and elongate. In Omithodira ances-
trallythemetatarsusisnarrowcomparedtoitslengthowingto
elongationofthemetatarsalsandreductionoftheouterdigits
(Fig. 6K-P). Within Sauropodomorpha. the manus and pes are
shortened,andtheyarebroadenedowingtoenlargementofdigit
I (Fig. 6N; this appears to be developmentally correlated with
modification of the pollex; Appendix A). In contrast, the the-
ropod metatarsus is relatively long and narrow, and it is thus
morelikethatofbirdsthanisthecaseinDinosauriaancestrally
(Fig, 60, P). The only other omithosuchian with such a narrow
andelongatemetatarsusisLagosiichus*(Fig.6K).Lagosuchus*
is considered to be convergent with theropods in this regard
becausethemetatarsusisnotsonarrowinSauropodomorpha
(Fig. 6N), Omithischia (Fig. 6M), or Herrerasauridae* (Fig. 6L),
all of which are closer to Theropoda than is Lagosuchus*. The
proportionsofthemetatarsusvarywithsize;thelargerdinosaurs
possessrelativelybroadmetatarsalscomparedtorelated,small-
erdinosaurs.Thisdifferenceappearstoreflectscalingeffects,
andappliestolargeversussmalltheropodsaswell.Nevertheless,
according to Ostrom (1976a), all theropods have a relatively
narrowermetatarsuswhencomparedtootherdinosaursofequal
size..Anarrow,elongatemetatarsushasbeenreportedinPro-
compsognathus* (Ostrom
1981),
Liliensiennis*
(Huene
1934),
Ceratosauria(Raath\969).Compsognalhus(Oslrom1978),Or-
nitholestes*(Osbom1917),Hulsanpes(Osmolska1982),Cae-
nagnathidae (Barsbold 1983),
Elmisauridae*
(Osmolska
1981),
Omithomimidae (Osmolska et al. 1972), Deinonychosauria
(Russell 1969), and Avialae (Ostrom 1976a).
32) Pes with pedal digit IV reduced and subequal to II in
length,thusmakingpessymmetricalaboutdigitIII.AsOstrom
(1981)argued,theropodfeetareunlikethoseofdinosaursgen-
ORIGINOFBIRDSANDEVOLUTIONOFFLIGHT
erally,inotherdinosaurspedaldigitIVislongerthanII(Fig.
6M, N), but in theropods pedal digit IV is reduced and ap-
proachesthelengthofpedaldigitII(Fig.60),thusmakingthe
pes even more symmetrical about digit III than in dinosaurs
ancestrally.ThesymmetricalpesispresentinProcompsogna-
ihus*(Os\mml981),Ceratosauna(Gilmore1920;Raath1969),
Carnosauria(Madsen1976),Elmisauridae*(Sternberg1932),
and Omithomimidae (Osmolska et al. 1972). In the context of
all the evidence, it is more parsimonious to accept that the
elongate pedal digit IV of Deinonychosauria (Ostrom 1969^)
(and to a lesser extent that of Ornitholestes* [Ostrom 1976a],
Compsognalhus [Ostrom 1978], Caenagnathidae [Barsbold
1983], and Avialae [Fig. 6P]) is a secondary modification (see
PartV,character84).
33) Fifth metatarsal reduced to no more than a spur of bone
in adult. Pedal digit five is reduced and bears only a vestigial
phalanx on the distal end of the fifth metatarsal in Dinosauria
ancestrally(Fig.6L,N).Incontrast,notheropodhasmorethan
a vestigial metatarsal spur (Fig. 60, P); the element has been
lost in adult Ornithurae, and it may also have been lost in some
omithomimids,caenagnathids,andelmisaurids*,althoughloss
andnonpreservationcannotbedistinguishedintheextinctthero-
pods. The fifth pedal digit is reduced to a spur of bone in
Proterochampsidae,thesister-groupofArchosauria(Gauthier
1984).WithinArchosauriathesameapomorphyappearedin-
dependently four
indifferent
groups:
Crocodylomorpha, ptero-
dactyloid Pterosauria, Omithischia, and Theropoda. The fifth
digit could have been reduced in the ancestral dinosaur and
subsequentlyreevolvedinSauropodomorpha,whichretainswhat
appears to be the ancestral condition. Alternatively, the fifth
digit could have been reduced twice, once in Omithischia and
once in Theropoda. Two evolutionary events are required in
eithercase;unlessoneiswillingtoassumethatconvergenceis
more likely than reversal in evolution (or vice versa), the level
ofsynapomorphyofthischaractermustbeconsideredambig-
uous.Fortunately,thischaracterisbutoneamongmanysup-
portingtheropodmonophyly,andregardlessofhowitisopti-
mized.Ithasnoefl^ectontheconclusionsofthisanalysis.In
this instance I accept convergence over reversal as an expla-
nationofthedistributionofthischaracteramongdinosaurs;
this conclusion predicts that we will one day find a fifth digit
likethatseeninsauropodomorphsincitheranomithischianor
atheropod(orboth).
34) Theropod first metatarsal. In Archosauria ancestrally,
metatarsal I articulates with the tarsus and it is proportioned
much like the other metatarsals (Fig. 6J). This condition is
retainedinLagosuchus*{F\g.6K),Pterosauna(Wellnhofer1978),
Herrerasauridae*(Fig.6L),Omithischiaancestrally(Fig.6M),
andSauropodomorpha(Fig.6N).Incontrast,theproximalpor-
tionofmetatarsalIfailstocontactthetarsusandthecompressed
andtnangularshaftoftheelementisboundbyconnectivetissue
to the medial side of metatarsal II in Liliensternus* (Huene
1934),Procompsognathus*(Ostrom1981),andCeratosauria
(Fig.60).MetatarsalIisevenmorebroadlyseparatedfromthe
tarsus in Camosauria (Osbom 1906; Lambe 1917; Gilmore
1920), Compsognalhus (Fig. 6Q), Elmisauridae* (Stemberg
1932), Caenagnathidae (Barsbold 1983), Omithomimidae
(Barsbold 1983), Deinonychosauria (Ostrom \969b), and Avi-
alae(Fig.6P).TarsitanoandHecht(1980)arguedthatinThero-
podawithanintactpes,metatarsalIisattachedabouthalf-
GAUTHIER;
SAURISCHIAN
MONOPHVL'i
way down metatarsal II (Fig. 60), and birds are further derived
in that the element attaches about three-quarters of the way
down metatarsal II (Fig. 6P). Ostrom (1976a) argued that a
reversedhalluxwasancestralforTheropoda.TarsitanoandHecht
(1980)tookexception,notingthatinthefewarticulatedthero-
pod feet, the hallux is short, unreversed, and metatarsal I lies
medial to metatarsal II. I agree with Tarsitano and Hecht that
the condition of the first digit as they describe it is indeed an-
cestralforTheropoda.However,thearticulatedfeetofComp-
sognathns show that metatarsal I in this taxon is like that of
birds, and unlike that of early theropods, in that it is short and
displaced to the posterior side of metatarsal II (Fig. 6Q). The
reversal,elongation,andposteriordisplacementofthehallux
presumablyrelatetotheperfectionofthegraspingfunctionof
the foot in Avialae (see Part V, character 84). Tarsitano and
Hecht(1980)arguedthatthisfunctionalcomplexisnotancestral
for Theropoda and I quite agree. Nevertheless, it is clear that
certainmodificationsthatareprerequisitetothereversedhallux
of Avialae apply to more inclusive groups of Theropoda than
to birds alone. In this regard, it is interesting to note that the
firstpedaldigitrecapitulatesitsphylogenetichistoryinthede-
velopmentofthepesinextantAves(Heilmann1926).The
position of the first pedal digit as preserved in early theropods
indicatesthatthisdigitwasunreversed,atleastatrest.However,
asThulbom(1984)hasnoted,Triassictheropodtrackwaysdis-
playingimpressionsofreversedfirstdigitsstandasmutetes-
tamentstoourinabilitytoinferfunctionfromstructurealone.
35)Thin-walledlongbones(=hollowskeleton).Thischaracter
haslongbeenrecognizedasatheropodsynapomorphy(seedi-
agnosisof'"Theropoda"inMarsh1881^,1884a)anditispres-
entinalltaxahereconsideredtobetheropodsexceptforHes-
peromithes and a few other diving birds. The walls of the long
bonesarethickerinDilophosauntsandCeratosaurus,andeven
thicker in Camosauria, as their larger size demands. Even so,
thelimbbonesofthelargesttheropod,Tyrannosaiinisrex,are
thin-walledincomparisontothesameelementsofcontempo-
raneouslarge dinosaurs, suchas
ThceralopsandAnatosaurus
(pers. obs.). The limb bones are hollow in Dinosauria ances-
trally,butasidefromPterosauria(BramwellandWhitfield1974),
nootherarchosaurshavelongbonesthatarequiteasthin-walled
asthoseofTheropoda(Romer1956).Moreover,T.Rowe(pers.
comm.) has pointed out that in theropods the entire skeleton,
includingthephalangesandcaudalvertebrae,arehollowand
lightly constructed in comparison to other dinosaurs of equal
size.Thismaypartlyexplainwhytheropodsingeneralandbirds
in particular are so uncommon in the fossil record.
III.PhylogeneticRelationshipswithinTetanurae
Tetanurae(n.txn.)
(Gr.:
letanos.
stiff;
ourae.
tails)
Temporal
Range.—late Jurassic
to
Recent.
Included
Taxa.—
Camosauna, Compsognathus. Ornilholcslcs*,
Coeliirus'.
Microvenator*.
Saitrornithotestes*
Hitlsanpes.
. Elmisaundae*,
Caenagnathidae.
Omithomimidae. Deinonychosauna,Avialae.
and
Diagnosis.—Tetanurae is defined here to include birds and
all other theropods that are closer to birds than they are to
Ceratosauria. In the following analysis, Ceratosauria, Sauro-
podomorpha, Ornithischia, Herrerasauridae*, Pterosauria-
Lagosuchiis*,Omithosuchidae,Euparkeria*.andPseudosu-
23
chia will be used as successively more remote outgroups. Te-
tanurinetheropodsarediagnosedbythefollowingsynapomor-
phies.
36) Absence of enlarged fanglike tooth in dentary.Gauthier
(1984) proposed that an unnamed taxon including erythro-
suchids,proterochampsids,andarchosaurs(n.comb.)canbe
distinguishedamongArchosauromorphabyanenlargedante-
riordentarytooth(e.g..Fig.1A,B).Thefangliketoothprojects
dorsallybetweenteethintheuppertooth-bearingbones,andit
may be received in a more or less pronounced notch between
thepremaxillaandmaxilla.Theancestralconditionisretained
by LiUenstemus* (Huene 1934) and, in a modified form, by
Ceratosauria. The fang is very large in ceratosaur theropods,
and it is received into the diagnostic subnarial gap in all cera-
tosaurs except Ceratosaurus (pers. obs.; Welles1984).Incon-
trast,anenlargeddentaryfangisabsentinCamosauria(Fig.
IG, H), Compsognathus (Ostrom 1978), Elmisauridae* (Gil-
more1924),Ornitholestes*(Osbom1917),Deinonychosauria
(Fig. IJ, K), and Avialae (Fig. IL). This character is considered
indeterminable in the toothless Omithomimidae (Fig. II) and
Caenagnathidae (Osmolska 1976).
37)Maxillaryfenestralargeandposteriorlyplaced.Anantor-
bital fenestra was present in the ancestral archosaur, and this
conditionwasretainedbytheancestraldinosaur(Romer1956).
Theropods differ in that they possess an additional fenestra
anterior to the antorbital fenestra, here termed the maxillary
fenestra (=second antorbital fenestra of various authors; see
Madsen 1976:65, plate 6). In contrast to the small, slitlike fe-
nestraconfinedtotheanteriormarginoftheantorbitalfossain
Ceratosauria(Welles1984),however,themaxillaryfenestrais
large,circular,andmoreposteriorlyplacedinCamosauria(Fig.
IG, H), Omithomimidae (Fig. II), Caenagnathidae (Barsbold
1983),Ornitholestes*(Osbom1917),Compsognathus(Ostrom
1978),Deinonychosauria(Fig.IJ,K),andArchaeopteryx*(Fig.
IL). The maxillary fenestra is absent in the highly modified
antorbitalregionofOmithurae.Asecondpre-antorbilalfenestra
IS present in some Rauisuchia (Sill 1974) and in a single eryth-
rosuchid,Shansisuchus(Young1964).However,thisfenestra
liesbetweenthepremaxillaandmaxilla,anditisnotconsidered
homologouswiththemaxillaryfenestrapresentinTetanurae.
38)Antorbitaltoothrow.InSaurischiaancestrallytheupper
and lower tooth rows terminate below the center of the orbit
(Fig. lA-F)- The ancestral condition is retained in Procomp-
sognathus*(pers.obs.)andCeratosauria(ColbertandRussell
1969).Incontrast,thetoothrowsareentirelyantorbitalinCar-
nosauria (Fig. IG, H), Deinonychosauria (Fig. IJ, K), Ornitho-
lestes*(Osbom 1903,
1917),Compsognathus* (Ostrom 1978),
and Avialae ancestrally (Fig. IL). In the context of all the evi-
dence,themostparsimonious explanationfor
theedentulous
Omithomimidae,Caenagnathidae,andAvesisthattheyeach
achievedtheirtoothlessconditionseparately,andthattheydid
soviatheconditionseeninothertetanurinetheropods.Antor-
bitaltoothrowsaroseindependentlyinseveralgroupsofdiap-
sids, most of which are carnivores (e.g., McDowell and Bogert
1954).However,someherbivores,forexamplesomesauropods
andceratopsianornithischians,havealsoacquiredthiscondi-
tion.Anantorbitaltoothrowalsooccursinavarietyofother
archosaursthatarethoughttohavebeencamivorous,suchas
someerythrosuchids(CharigandReig1970),rauisuchians(Bo-
naparte1981),andsomepterosaurs(Wellnhofer1978),butit
24
is absent in others, such as omithosuchids (Walker 1964) and
proterosuchids(Cruickshank1972).Thus,whilethecharacter
is not an infallible indicator, it appears to be correlated with
macropredaceoushabits(sensuMcDowellandBogert1954).
39) Spine table on the axis. Spine tables are present in the
posteriorcervicalsandanteriortrunkvertebraeinArchosauria
ancestrally(Gauthier1984).However,spinetablesareabsent
intheanteriorcervicalsincludingtheaxisinPseudosuchia(Ro-
mer1956),Euparkeria*(Ewer1965),Ornithosuchidae(Bona-
parte1975ft),Lagosuchits* (Bonaparte1975/'),Pterosauna
(Wellnhofer1978),Herrerasauridae*(Galton1977),Omithis-
chia (Santa Luca 1980), Sauropodomorpha (Fig. 3D), Lilicn-
sternus* (Huene 1934), and Ceratosauria (Fig. 3E). Thus, an-
teriorspinetablesare
absent in
theropods ancestrally.
Incontrast,
an expanded distal end of the neural spine (=spine table) is
present, at least on the axis, in Camosauria (Madsen 1976),
Omithomimidae(Osmolskaetal.1972),Deinonychosauria(Fig.
3F, H), and Avialae (Fig. 3G, I).
40) Transition point begins closer to proximal half of tail.
With the exception of a few birds lacking the pygostyle, all
theropods have tails that are mobile proximally and stiffened
distally. The transition between these regions is marked by
roughly coincident changes in neural spines, transverse pro-
cesses,haemalarches,and pre- and
postzygapophyses, which
are referred to collectively as the transition point. As noted
above, although ceratosaurs display the transition point, it is
lessmarked,andtheyretaintheancestralconditionofthecaudal
haemalarcheswellintotheposteriorhalfofthetail.Incontrast
toceratosaurs,however,thetransitionbetweenthestouterand
moremobileproximalandstiffenedandthinneddistalportions
of the tail is more pronounced and begins closer to the base of
the tail in tetanurine theropods. In addition, unlike the case in
Ceratosauria(Gilmore1920),thehaemalarchesinCamosauna
(Madsen 1976), Omithomimidae (Fig. 2E), Deinonychosauria
(Fig. 2G), and Avialae ancestrally (Fig. 2F) become progres-
sivelyshorter
dorsoventrally andelongate
anteroposteriorly from
the proximal to the distal end of the tail. Russell (1972) noted
somevariationintheposition of the transition point in the tails
of taxa that I include in Tetanurae, and detailed comparisons
of theropod caudal series may provide further characters for
phylogeneticanalysis.Analogousmodificationsofthecaudal
region are present in sand lizards, such as Callisaurus. Copho-
saunis.andHolhrookia(pers.obs.).Thezygapophysesarenot
deeplyimbricateasintheropods,buttheyareverticallydisposed
(as in most omithodiran archosaurs); the tail is stout basally,
but It tapers abruptly in a region analogous to the theropod
transition point, and the greater part of the tail is thin, with
short haemal arches that are expanded fore and aft. Thus, the
base of the sand lizard tail would correspond to the mobile
proximalportion,andtheremainderofthetailtothestiffdistal
portion, of the tetanurine tail. Sand lizards typically curl their
comparativelyshorttailsdorsally,particularlyduringbipedal
progression when the tail acts as a dynamic stabilizer. The tail
isalsousedfordisplay,withside-to-sidewaggingofthecurled
distal portion playing a role in social interaction. The stiffened
distalportionofthetheropodtailprobablycouldnothavebeen
curled as in sand lizards, owing to the deeply imbricate zyg-
apophyses.Although thetheropod
tail's
paramount role
may
have been as a dynamic stabilizer, one cannot overlook the
possibilityofaccessoryrolesinsocialdisplay.
ORIGINOFBIRDSANDEVOLUTIONOFFLIGHT
41)Scapulastraplike.AsinDiapsidagenerally,thedistalend
of the scapula is flared anteroposteriorly in Dinosauria ances-
trally(Fig.4A,B).ThisconditionisretainedbyCeratosauria
(Raath1969;Welles1984)andfromwhatremainsofthescapula
inLilicnsternus*(Huene1934),itappearstoretaintheancestral
condition as well. In contrast, the distal expansion is reduced
orabsentandthescapulaisthusstraplikeinCamosauria(Mad-
sen1976),Conipsognathus(OstTom1978),Oniilholeslcs*{Os-
trom 1976a), Omithomimidae (Fig. 4C), Deinonychosauria
(Ostrom 1976a), and Avialae (Fig. 4D). This modification is
further enhanced in Omithurae, in which the scapula tapers
distally(Elzanowski1981).ThecamosaurTyrannosaiirusisun-
usualamongtetanurinesinthepossessionofaflareddistalend
ofthescapula(Osbom1906);ratherthansymmetrical,however,
the scapula is flared mainly at its anterior margin, suggesting
thatthisconditionissecondary.
42) Coracoid tapers posteriorly in profile. In Omithodira an-
cestrallytheprofileofthecoracoidissubcircularbelowitsar-
ticulationwith the
scapula,
although the
element maytaperto
abluntedpointposteroventrally.Theancestralconditionisre-
tainedinOrnithischia(Fig.4A),Sauropodomorpha(Fig.4B),
and in Theropoda ancestrally in that the coracoid has a subcir-
cularprofileinLiliensternm*(Huene1934)andCeratosauria
(Welles 1984). In contrast, the pointed posteroventral margin
of the coracoid is more pronounced in that it extends well be-
yondtherimsoftheglenoidwhenthescapulocoracoidisori-
entedverticallyin
Camosauria (Madsen1976),
Compsognathus
(Bidar et al. 1972), and Omithomimidae (Fig. 4C). The rect-
angularprofile of
deinonychosaurs andavialans
is
hereconsid-
eredafurthermodificationoftheformofthecoracoidseenin
othertetanurinetheropods.Aposteriorlytapenngcoracoidap-
pearedindependently in
omithopod omithischians (e.g.,
Galton
1974).
43)Manusformsmorethantwo-thirdsofcombinedlengths
of radius plus humerus. Ostrom (1969ft) pointed out that the
manusislessthanhalfthelengthofthehumerusplusradiusin
Ceratosauria (47% in Coelophysis and Synlarsus). as it is in
Dinosaunaancestrally.Heterodontosaurustitckiisexceptional
among omithischians in that the manus is over half (i.e., 56%)
ofthehumerusplusradiuslength(SantaLuca1980).Incontrast,
Tetanuraepossessesconspicuouslyenlargedhands.Themanus
IS 77% of the length of the radius plus humerus in Camosauria
ancestrally(i.e.,Allosaunis).althoughtheforearmisgreatlyre-
ducedinTyrannosauridae(Lambe1917).Themanusismod-
ifiedinOmithomimidae,butitisstillatleast58%ofthelength
oftheradiusplushumerus.InCaenagnathidae(Osbom1924ft),
Ornirholestes* (Osborn 1917), Deinonychosauria (Ostrom
1969ft), and Avialae (Ostrom 1976a) ancestrally, the manus is
67-75%ofthehumerusplusradiuslength.Tetanuraeingeneral
andcoelurosaursinparticularareuniqueamongDinosauriain
therelati%esizeofthehand.AlthoughScU'roniochlus*doesnot
appeartobesomodified(Huene1914a),pterosaursaretheonly
otherarchosaurswhosehandsexceedtherelativesizeseenin
Tetanurae. Even the highly modified hands of Omithurae are
enormousrelativetothoseofmostotherarchosaurs.
44)BasalhalfofmetacarpalIcloselyappressedtometacarpal
II. As in Diapsida ancestrally, the metacarpals overiap one
anotherproximally,andthisconditionisretainedinPseudosu-
chia(Fig.4G),Ornithosuchidae(Walker1964),Pterosauria
(Wellnhofer 1978), Omithischia (Fig. 4H), Sauropodomorpha
GAUTHIER:
SAURISCHIAN
MONOPHYLY
(Fig, 41, J), and Ceratosauria (Welles 1984), In contrast, the
bases of metacarpals I and II are closely appressed for at least
half the length of metacarpal I in Camosauria (Fig, 4L), Or-
nithomimidae(Fig,4M),Ornitholestes*(Osbom1917),Caen-
agnathidae(Barsbold1983),Elmisauridae*(Osmolska1981),
Deinonychosauria (Fig, 4N), and Avialae (Fig, 40),
45)BaseofmetacarpalIIIsetonpalmarsurfaceofhandbelow
base of metacarpal II. As was argued in Part I. character nine
above, the bases of metacarpals IV and V (and thus their as-
sociateddigits),weresetonthepalmarsurfaceofthehandin
Saurischiaancestrally,Tetanurinetheropodsarefurtherspe-
cializedwithinthisassemblageinthatmetacarpalIIIisalso
displacedventrallywithrespecttometacarpalII,Ceratosauria
retains the ancestral condition (Fig, 4K). but the derived con-
ditionispresentinCamosauria(Fig,4L).Omithomimidae(Fig,
4M).Elimsauridae*(Osmolska1981).Caenagnathidae(Bars-
bold1983),Ornitholestes*(Osbom1917),Deinonychosauria
(Fig, 4N), and Avialae (Fig, 40), Even in disarticulated speci-
mens,thebeveledarticularsurfaceoftheproximalendofmeta-
carpalIIIissufficienttoidentifythischaracter,
46)Fourthmanaldigitabsentbeyondembryonicstages.As
arguedabove,thefourthmanaldigitisreducedinallTheropoda
and probably plays no role in the function of the hand (Galton
1971),UnlikeCeratosauria.however,thefourthdigitisabsent
inCamosauria(Fig.4L),Compsognalhiis(Ostrom1978),Saur-
ornilholestes* (Sues 1978), Elmisauridae* (Osmolska 1981),
Caenagnathidae (Barsbold 1983), Omithomimidae (Fig, 4M),
and Deinonychosauria (Fig, 4N), Osbom (1917) and Ostrom
(1969^)identifiedafragment of bone near the proximal end of
metacarpalIIIinOrnitholestes*thatmayrepresentavestigeof
metacarpal IV, The fourth digit is absent in all adult Avialae
(Fig, 40) but the precursor of the element is reported in em-
bryonicAves(Heilmann1926;seeHinchliffeandHecht1984,
foranaltemativeview),
47) Obturator process on ischium. An obturator process on
the ischium is absent in Pseudosuchia, Omithosuchidae, La-
gosiichus*.Pterosauria,andHerrerasauridae*(Gauthier1984),
An obturator process is also absent in all Ornithischia (Fig, 5A)
except for Lesothosaurus (Thulbom 1972) and Omithopoda
(sensuSantaLuca1980),anditisabsentinSauropodomorpha
(Fig.5B),Liliensternus*(Huene1934),Procompsognathus*(Os-
trom1981),andCeratosauria(Fig.5C).Incontrast,anobturator
process is present on the ischium of Camosauria (Fig. 5D),
Omithomimidae (Fig. 5E). Ornitholestes* (Osborn 1917),
Cotnpsognathiis (Ostrom 1978). Elmisauridae* (P. J. Currie.
pers.comm.),Caenagnathidae(Barsbold1983),andDeinony-
chosauria(Fig.5F.G).Ostrom(1976a.-129)maybecorrectin
interpreting the anterior of the two processes at the ventral
extremity of the ischium of Archaeopteryx* as an obturator
process(Fig.5H).Theischiumo^Archaeopteryx*is,however,
difficulttoassessbecausetheelementisinseveralwaysdifferent
fromthatseeninotherTheropoda,includingOmithurae(Fig.
51; Tarsitano and Hecht 1980). Unlike the pubis in Archaeop-
teryx*.themorphologyoftheischiumhasnotreceivedmuch
attention,andthiselementneedsfurtherstudy.Nevertheless,
evenifthelowerprocessontheischiumisnothomologouswith
the obturator process of other Tetanurae, it is still more parsi-
monioustoacceptthatbirdslost,ratherthanneverhad,an
obturatorprocess.
48) Expanded pubic foot. An expanded distal extremity of
25
thepubisispresentinHerrerasauridae*(Reig1963;Benedetto
1973;Cooper1981a.'Galton1977).However,theapomorphy
isunknownelsewhereamongDinosauria,includingLilienster-
nus*(Huene1934)andCeratosauria(Fig.5C).Thus,anex-
pandedpubic foot
applies
to
Herrerasauridae* onone hand, and
Tetanurae on the other, but not to a group including both. An
expanded foot at the distal end of the pubis is present in
Camosauria (Fig. 5D), Caenagnathidae (Barsbold 1983), Or-
nithomimidae (Fig. 5E), Microvenator* (Ostrom 1970),
Compsognathiis*(Ostrom\91S),Coelurus*(Marsh1896),Dei-
nonychosauria(Fig.5F,
G),
and
Archaeopteryx* (Fig,
5H),
Thus,
theabsenceofapubicfootinOmithuraeisconsideredsecond-
ary(Fig,
51),
49) Femur with winglike anterior (=lesser) trochanter. The
anteriortrochanterisaspikelikeridgeinDinosauriaancestrally,
and this condition is retained in Liliensternus* (Huene 1934)
and in a modified form in Ceratosauria (Raath 1969). In con-
trast,theanteriortrochanterisprominentandwinglikeinCar-
nosauria (Madsen 1976), Elmisauridae* (P. J. Currie, pers.
comm.). Omithomimidae (Osmolska et al. 1972), Microvena-
tor*(Ostrom1970),andinamodifiedforminDeinonycho-
sauriaandAvialae(Ostrom1976a.1976b;Padian1982).Itis
interesting to note at this point that a discrete, winglike lesser
trochanteroccasionallyappearsasavariantinafewextantbirds.
AwinglikeantenortrochanteraroseconvergentlyinOrnithis-
chia,anditisparticularlyprominentinOmithopoda(sensu
Santa Luca 1980; e.g., Galton 1974). Evidently, the dorsal mi-
grationandposteriordisplacementoftheareaofinsertionof
theso-calledM.iliotrochantericusgroupofAves(Cracraft1971),
isassociatedwithperfectionofbipedal,cursorialhabits(Coombs
l97Sb: see further discussion in Part V. character 82).
50)Ascendingprocessofastragalustall,broad,andsuperfi-
ciallyplaced.WellesandLong(1974)providedadetailedde-
scriptionof
the
theropodtarsus.Thedistributionof
characters
in their classification of theropod tarsi (p. 197) indicates that
theirfive"distinctkinds"arephenetic,ratherthanphylogenetic,
concepts.Forexample,theynotethatoneofthecharacters,the
"freemedialcomponent"(=superficialpartofascendingpro-
cess)isabsentbothindinosaursancestrallyandintheir"cer-
atosauroid"groupoftheropods(=Liliensternus*andCerato-
sauriaofthiswork).The"freemedialcomponent"is,however,
present in all members of their "allosauroid" (Fig. 6 A), "alber-
tosauroid,""tyrannosauroid."and"omithomimoid"(Fig.6B,
C) groups, as well as in Compsognathus, Ornitholestes*. and
Microvenator*.Moreover,anascendingprocessispresentinall
birds (Fig. 6D), and although its height and width may vary
with size and the presence of a tendinal groove in birds, it is
relativelylargerinbirdsthanitisintheropodsancestrally(pers.
obs.).Accordingly,thedevelopmentofalargerascendingpro-
cessisconsideredtobeasynapomorphyofTetanurae.Although
theascendingprocess,whicharisesasaseparateossification
centerinTheropoda(Heilmann1926;Welles1983;pers,obs,),
retains its ancestral relations with the tibia, it has shifted its
associationwiththeproximaltarsalsinNeognathae(pers,obs,),
51) Metatarsals II and IV with broader participation in ankle
andmetatarsalIIIcompressedbetweenthemtoavariablede-
gree.Using the
omithopod omithischian Hypsilophodon* (Fig.
6E),thesauropodomorphMassospondylus(Cooper1981a),the
earlytheropodLiliensternus*(Huene1934),andtheceratosaur
Dilophosaurus(Fig,6F)toassesstheancestralconditionofthe
26
metatarsusinSaurischia,neithermetatarsalIIorIVcontributes
asmuchtothesurfaceareaoftheanklejointasdoesmetatarsal
III in proximal view. In contrast, metatarsals II and IV partic-
ipatemorebroadlyintheanklerelativetometatarsalIII,and
metatarsal III is more or less compressed between II and IV.
The width of metatarsal III is more than 33% of the width of
metatarsals II-IV on the midline in Omithischia (Fig. 6E) and
Ceratosauria (Fig. 6F) and it is less than 26% of the width of
theproximalendsofthesemetatarsalsinCamosauria(Fig.6G),
Hiilsanpes(Osmohka1982),Elmisauridae*(Osmolska1981),
Omithomimidae(Osmolskaetal.1972),Deinonychosauna(Fig.
6H), and Avialae (Fig. 61; see Part IV, character 67, for further
discussion).
52)MetatarsalIshort.Asnotedabove,metatarsalIdoesnot
reachthetarsusintheropodsancestrally.Intheancestralcon-
dition,thecompressedshaftofmetatarsalIisrelativelylong,
as can be seen in Lilienstenms* (Huene 1934), Procompso-
gnallnts* (pers. obs.), and Ceratosauria (Fig. 60) with the pos-
sibleexceptionofCera/osai/n/s(seeGilmore1920).Bycontrast,
metatarsal I in Tetanurae is relatively short (e.g.. Fig. 6P). The
apomorphic tetanurine condition is reported in Camosauria
(Lambe 1917), Compsognalhm (Fig. 6Q), Caenagnathidae
(Barsbold 1983), Elmisaundae* (Sternberg 1932), Omitho-
mimidae (Barsbold
1983),
Deinonychosauria (Ostrom 1969/?),
and Avialae (Fig. 6P).
IV.PhylogeneticRelationshipswithinCoelurosauria
Coelurosauria(n.comb.)
Temporal Range.—late
Jurassic
lo
Recent.
Included
IfCHK.
Oniilhok'stcs*.
— dtmpsognathus.
Micrmcnator*.
Codurus*.
Saiirornilholestcs*.
Hulsanpes.
Elmisaundae*,
Caenagnathidae.
Omithomimidae.
Deinonychosauna,
.Avialae,
and
Diagnosis. —As defined here, Coelurosauria includes birds
and all other theropods that are closer to birds than they are to
Camosauria. This concept differs fundamentally from that of
previousworkers,allofwhomapplied'"coelurosaurs"toapara-
phyleticgroupincludingalltheropodsexceptforcamosaurs.In
thefollowinganalysis,Camosauria,Ceratosauria,Sauropodo-
morpha, Omithischia, and Herrerasauridae* will be used as
successivelymoreremoteoutgroups.Theremainingoutgroups,
Pterosauria-Lagosuchus*.Omithosuchidae,Euparkena*.and
Pseudosuchia, will be referred to as the "nondinosaurian ar-
chosaurs"or"otherarchosaurs"tofacilitatethefollowingdis-
cussion.Coelurosauria
possesses
the
following
synapomorphies
distinguishingitamongTheropoda.
53) Subsidiary fenestra between pterygoid and palatine. A
subsidiaryfenestraisabsentinnondinosaurianarchosaurs(Ro-
mer 1956), Omithischia (e.g., Heaton 1972), and Sauropodo-
morpha(Gallon1984).AccordingtoColbertandRussell(1969),
thischaracterisalsoabsentintaxaherereferredtoCeratosauria
andCamosauria.Incontrast,asubsidiaryfenestrabetweenthe
pterygoidandpalatineispresentinOmithomimidae(Osmolska
et al. 1972). Caenagnathidae (Osmolska 1976), and Deinony-
chosauria(Colbert andRussell
1969).Thepalate
is
notexposed
in Archaeopleryx*. and the palate of other birds is too trans-
formedtointerpretinthisregard(McDowell1978).Inthecon-
textofalltheevidence,birdsareconsideredtohavemodified
thepalatalelementsfromtheconditionseeninothercoeluro-
saurs.
ORIGINOFBIRDSANDEVOLUTIONOFFLIGHT
54)Deeplyexcavatedpocketonventralsurfaceofectopter-
ygoid flange. As noted above (Part II, character 15), a relatively
smaller excavation is present on the ventral surface of the ec-
topterygoid in Camosauria (Colbert and Russell 1969). Ac-
cordingtoSues(1978)thispocketisdeeperinSaurornitho-
lestes*. Omithomimidae, and Dromaeosauridae, and the
apomorphicconditionappearstoobtainaswellinTroodontidae
(Barsbold 1983) and Caenagnathidae (Osmolska 1976). As in
the case of the previous character, modification of the avian
palateprecludesasimpleconclusionregardingthepresenceor
absenceofthischaracter,especiallysincetheectopterygoidhas
not been identified in birds. Nevertheless, in view of their re-
lationshipswithinCoelurosauriaitwouldstillbesimplerto
acceptsecondarymodificationinbirds.Anothermodification
in this region of the palate may also be diagnostic of Coeluro-
sauria:aconspicuousovaldepression,subdividedbyaridge,
that lies on the dorsal surface of the pterygoid process of the
ectopterygoidinthedromaeosaurDeuwnychusandinSauror-
niiholestes*(Sues1978).Althoughatpresentthischaracterhas
been reported only in these taxa, it is absent in Camosauria,
indicatingthatitarosewithintheropodsafterthedivergenceof
camosaurs;moreinformationwillbenecessarytoestablishits
levelofsynapomorphy.
55)Cervicalribsfusedtocentrainadults.Cervicalribsremain
unfusedthroughoutontogenyinnondinosaurianarchosaursaside
from some Pterosauria (Romer 1956). Both Omithischia and
Sauropodomorpharetaintheancestralcondition,althoughfu-
sionbetweennbsandvertebraeinthecervicalregionarose
within Sauropoda. Fully adult ceratosaurs, such as the speci-
mensofSvfitarsiisandCeratosaiintsdescribedbyRaath(1969)
and Gilmore (1920) respectively, retain free ribs, and this ap-
pearstobethecaseinAllosawiisaswell(Madsen1976).Thus,
theancestralconditionintetanurinetheropodsappearstobe
the retention of free cervical ribs in adults. Fused cervical ribs
have so far been reported only in Coelwus* (Marsh 1881a),
fully adult specimens of omithomimids such as Gallimimus
(Osmolska et al. 1972) and Struthiomimus (Osbom 1917), and
in fully adult omithurine birds (Marsh 1880). Coeliirus* is too
poorlyknowntoallowplacementbeyondCoelurosauriaincertae
sedis. and will for that reason be ignored in the following dis-
cussion.
Atfirstglance,itseemsthatfusedcervicalribsarose(atleast)
twice,onceinOmithomimidaeandonceinOmithurae,because
neitherArchaeopleryx*nordeinonychosaursarereportedto
have fused cervical ribs. This interpretation itself depends on
assumingthattherearefullyadultexamplesamongtheknown
specimens of the latter two taxa, and according to the criteria
hereusedtodeterminecessationofgrowth,thisdoesnotappear
to be the case. The situation is further complicated by the fact
that Dcnionychm is the only deinonychosaur for which any
portion of the cervical region has been described, and these
specimensrepresentsubadultindividuals.Forexample,theax-
ialintercentrumremainssuturallydistinctfromboththefirst
andsecondcentmm(Fig.3F),whileinfullyadultreptiles(sensu
Gauthieretal.,inprep.)thesecondintercentrumisalwaysfused,
at least to the first centrum. Although this fusion takes place
earlier in ontogeny in some reptiles (e,g„ in extant crocodylo-
morphembryos),itmaybeanearterminaleventintheskeletal
ontogenyofothers(e.g.,mostsquamates).butbythecessation
ofgrowthitisinanycasefused.Fusionoftheaxialintercentrum
GAUTHIER;
SAURISCHIAN
MONOPHYLY
takes place prior to fusion of the cervical ribs in extant birds
(pers. obs.); thus the absence of fused cervical ribs in the cur-
rentlyknownDeinonychusmaysimplyreflecttherelativeim-
maturityofthespecimens,nottheretentionofanancestral
condition. The presence or absence of this character will be
scoredasunknowninDeinonychosauria.
For similar reasons, I also consider this character indeter-
minatein
Archaeopteryx*. If
thefive
Archaeopteryx* skeletons
arerankedaccordingtomeasuresofhomologousstructures,the
relative size from largest to smallest specimens would be the
subequal-sizedLondon.Maxberg.andHaarlemspecimens,fol-
lowedbytheBerlin,andthentheEichstattspecimens(seeWelln-
hofer 1974). The first three specimens are of comparable size,
the Berlin specimen is about 15% smaller than the London
specimen,andtheEichstattisabout45%smallerthantheLon-
donspecimen,basedonthelengthofthehumerus(datafrom
Wellnhofer 1974). The scapula and coracoid are unfused in all
specimens,althoughtheyarefirmlyattachedtooneanotherin
theLondonspecimen,indicatingbyextrapolationthattheLon-
don.Maxberg.andHaarlemspecimensareneartofullmaturity
(alsocorroboratedbyfusionoftarsometatarsus).Unfortunately,
thecervicalsareeitherpoorlypreservedorabsentintheseap-
parentlyfull-grown specimens. Free
cervical
ribs
arepresent
in
the smaller specimens from Berlin and Eichstatt. but this is to
be expected in view of the apparent immaturity of these spec-
imens.Thus,untilthischaractercanbedeterminedinnewand
fullyadultspecimensoiArchaeoptoyx*andDeinonychosauria.
it is simpler to accept that fusion of cervical ribs, at least by
maximumadultsize,issynapomorphicofCoelurosauriaamong
Theropoda.
56)Cervicalzygapophysesflexed.Innondinosaurianarcho-
saurs.thecervicalzygapophysesareplanar(pers.obs.).andthis
condition is retained in Omithischia (e.g.. Ostrom 1970). Sau-
ropodomorpha(Huene1908).Ceratosauria(Gilmore1920).and
Camosauria(Madsen1976).Thus,planarzygapophysialfacets
inthecervicalregionaretheancestralconditioninTetanurae.
Incontrast,thezygapophysialfacetsaresharplyflexedabouta
line dividing the facet into larger lateral and smaller medial
components in Omithomimidae (Osmolska et al. 1972). Dei-
nonychosauria (Ostrom 1969ft).
and
Avialae
(this
character
is
not determinable in Archaeopteryx*, but it is present in other
birds).
57)Anteriorcervicalvertebraebroaderthandeepanteriorly,
with kidney-shaped articular surfaces that are taller laterally
thanonthemidline.Amphicoelousintervertebraljointsarethe
ancestral condition for Archosauria (Romer 1956). Either the
ancestralcondition,oramoreorlessprominentopisthocoely
derived from it. is retained in Omithischia (Galton 1974). Sau-
ropodomorpha (Huene 1932). Lilienslernus* (Huene 1934).
Ceratosauria(Gilmore1920).andCamosauria(Madsen1976).
Unfortunately, the placement within the neck of isolated cer-
vicalsreferredtoCoelunis*(Marsh1881a)andMicrovenalor*
(Ostrom 1970) is unknown. In contrast to the ancestral condi-
tion,theanteriorarticularsurfacesintheanteriorcervicalver-
tebraeofOmithomimidae(Osmolskaetal.1972)andDeino-
nychosauria(Fig. 3H)
are
modified in
distinctly
a avian
manner,
althoughtheposteriorsurfacesremainamphicoelous(Ostrom
1969/7,-posterioramphicoelyleddeBeer1954,totheerroneous
conclusionthattheLondonArchaeopteryx*wasamphicoelous
throughoutthecolumn).Thekidney-shapedanteriorsurfaces
27
displaytheinitialstagesofavianheterocoely.Theyarenotmuch
lessmodifiedinthisrespectthanarehomologousvertebraeof
Ichthyornis (Marsh 1880). early Hesperomithes such as Bap-
tornis (Martin and Tate 1976). and embryos of extant birds
(pers. obs.). L. D. Martin (pers. comm.) notes that heterocoely
becomesmorepronouncedandextendsfurtherposteriorlyin
thevertebralcolumnwithinHesperomithes,andthatthesesame
modifications arose in parallel in Aves ancestrally. Thus, the
fullyheterocoelousconditionsocharacteristicofextantbirds
(e.g..Fig.31).arosetwicewithinOmithurae.
58) Furcula. In Dinosauria ancestrally the clavicles are re-
ducedandgracile,andtheyaredifficulttodistinguishfromribs
inallbutperfectlyarticulatedfossils(Ostrom1976a).Clavicles
are present in the ceratosaur Segisaunis* (Camp 1936; pers.
obs.).buttheyhaveyettobeidentifiedinCamosauria.inwhich
theforelimbsandgirdlesarereduced(Lambe1917).Fusedclav-
icles(=furcula)havebeenreportedinCaenagnathidaeandOr-
nithomimidae(Barsbold1983).andthesefurculaarelikethose
oiArchaeopteryx* in that they are very robust, unlike clavicles
in dinosaurs ancestrally, or for that matter in Omithurae of
equivalentsize(seefollowingcharacterforcommentsonfunc-
tionalimplicationsofarobustfurcula).Fusionoftheclavicles
takesplaceduringpostnatalontogenyinextantbirds(pers.obs.).
Thetimingofthiseventisunknownincoelurosaursancestrally,
but fusion of the clavicles would probably have occurred no
earlierintheontogenyofextinctthanextantcoelurosaurs.Clav-
icleshaveyettobedescribedinDeinonychosauria.although
they are said to be present in re/oarap/tv (Kielan-Jaworowska
and Barsbold 1972). Fused clavicles are also present in Avialae
ancestrally (Heilmann 1926). although the elements may be
reducedandunfused,orabsent,viapaedomorphosis,inflight-
lesstaxawithinthisgroup(Marsh1880;GlennyandFriedmann
1954; Van Tyne and Berger 1959). Based on the development
ofCoturnix.Lansdown(1968)suggestedthattheavianfurcula
maybeaneomorphbecauseatleastpartoftheelementforms
fromacartilaginousprecursor.ClavicledevelopmentinCotiir-
uix was, however, recently reconsidered by Russell and Joffe
(1985),andtheyreaflirmedthattheelementwasdermalrather
thanendochondralinorigin.
59)Bonystemalplatesfused,atleastinfullyadultindividuals.
Another character that may be functionally related to the de-
velopmentof thefurcula.
namely,
a
fused,bonystemum. may
haveoriginatedatalevelmoreinclusivethanOmithuraewithin
Coelurosauria. Olson and Feduccia (1979) argued that the ro-
bustfurcula of
Archaeopteryx*. togetherwith
thecoracoclavicu-
lar membrane, acted as the point of origin of the muscle pro-
vidingthepowerstrokeforthewing,ahypertrophiedM.
pectoralis.Theyfurthernotedthattheposteriorportionofthis
muscle also originates from the area of the stemum that is not
preempted by the underlying M. supracoracoideus in Omi-
thurae.TheypresumedthisportionoftheM.pectoralistobe
absentinArchaeopteryx*.becauseArchaeopteryx*wasthought
to lack a bony stemum. More recent finds allow us to modify
OlsonandFeduccia'sconclusions.
Although the stemum may calcify in fully adult specimens,
ossifiedsternalplatesareabsentinallnondinosaurianarcho-
saurs(exceptPterosauria).Pairedossificationswithinthecar-
tilaginousstemum homologous with
the
pleurosteons of
extant
AveshavebeenreportedinsomesubgroupsofOmithischiaand
Sauropodomorpha;butitisnotclearifsuchossificationswere
28
presentinthesegroupsancestrally(Romer1956).Ossifiedsterna
areunknowninearlytheropods,butmorecompleteknowledge
ofthisregioninwell-preservedceratosaurs,suchasCoelophysis.
isnecessarybeforeabsencecanbedistinguishedfromnonpres-
ervation.Thesternumbecomeswellossifiedfairlylatempost-
hatching ontogeny, it is superficially placed, and it is not at-
tachedtotheremainderoftheskeleton;thiscombinationof
factorsmakesthesternumapoorcandidateforpreservation.
Todate,pairedsternalplateshaveonlybeenreportedinasingle
specimenofCamosauria(Lambe1917)andinafewspecimens
ofCaenagnathidae,Omithomimidae,andDeinonychosauria
(Barsbold 1983). With the possible exception of the London
specimen (de Beer 1954), bony sternal plates are unknown in
Archaeopteryx*. However, they remain as pleurosteons in ju-
venileOrnithurae,andmaybeaccompaniedbyadditionalos-
sificationcenters,
the
lophosteonandmetosteon, in
the
keel
and
xiphoidprocessesrespectively(Bellairsand.lenkinI960).Cur-
rentknowledgeofthedistributionofthischaracteramongor-
nithosuchianarchosaursprecludesafirmdecisionregardingthe
levelofsynapomorphyofthischaracter.Theboldesthypothesis
wouldbethatbonysternalplatesaroseinancestralOmithodira.
Anossifiedsternumhaslongbeenconsideredsynapomorphic
forOrnithurae,especiallysinceabonysternumwasthoughtto
beabsentinArchaeopteryx*.However,inviewofthepresence
ofabonysternuminother tetanurines, if not all omithodirans.
thesternumwasprobablypresentinArchaeopteryx*:whether
itsabsenceresultsfromnonpreservationorimmaturityofthe
specimens, or a combination of both factors, cannot yet be
determined.Whetherornotfusionbetweenthesternalplates
issynapomorphicofOrnithuraeisalsoquestionable,because
somecacnagnathidsiOviraptor)andornithomimids{Ingeiiia)
alsohavefusedsternalplates,atleastinfullymaturespecimens
(Barsbold 1983). Thus, this character appears to have arisen
withinCoelurosauriaatamoreinclusivelevelthanbirdsalone.
Asternumhassofarbeenreportedinonlyonedeinonychosaur
(I'elociraptor: Barsbold 1983). The paired and apparently an-
cestralconditionofthesternuminthistaxonmaynotbecon-
clusive,however,becausethe
specimen hasnot
beendescribed
completelyenoughtodetermineitsstageofdevelopment.Full
ossification of the sternal plates in extant birds occurs late in
ontogeny,followedbycoossificationnearmaximumadultsize
(Bellairs and Jenkin 1960). Thus, the unfused sterna of I'elo-
ciraptormaysimplyreflectimmaturity.Inviewofthescant
evidence, it may be too early to consider fusion of the sternum
synapomorphicforCoelurosauria.Inthecontextofthepresent
hypothesis,however,Ipredictthatafusedsternumwillbefound
tohavearisenoutsideofOrnithuraewithinCoelurosauria.
Such a find would have interesting consequences in light of
argumentsmadebyOlsonandFeduccia(1979)regardingflight
capability in Archaeopteryx*. If, as seems to be the case, the
basicpatternofthepectoralapparatusofArchaeoptery.x*applies
to all Coelurosauria rather than to the immediate ancestor of
Avialae,thensomeaspectsofpectoralfunctionmustbeequally
general. That is to say, it would be inaccurate to consider the
modificationsofthepectoralgirdleof.Archaeoptery.x*solelyin
terms of flight. Thus, the presence of a hypertrophied M. pec-
toralis.asindicatedbyarobustfurcula,enlargedcoracoid,and
fusedsternum,suggeststhepresenceofpowerfulforelimbad-
ductorsinall
coelurosaurs.
Thesemodificationsmayhaveserved
initiallytoenhancetheraptorialcapabilitiesofcoelurosaurfore-
ORIGINOFBIRDSANDEVOLUTIONOFFLIGHT
limbsandonlylaterbeenconscnptedtoservethepowerstroke
in avian flight. Aside from Coelurosauria, the only other ar-
chosaurspossessing coossified,
a bonysternum are
pterosaurs
(Wellnhofer 1975, 1978), and this is hypothesized to be a case
convergence.
of
60) Elongate forelimb exceeds half the length of hindlimb
and/orpresacralvertebralcolumn.Ostrom(1969^)pointedout
that the forelimb is no more than 45% of the hindlimb length
in Theropoda ancestrally. In Camosauria, the forelimb varies
from 25-42% of the hindlimb length. In contrast, Omitho-
mimidae(Russell 1972),Caenagnathidae (Barsbold1983),
Or-
nitholestes*{Oslrom1969ft),Deinonychosauria(Ostrom1969ft),
and Avialae (Ostrom 1976a) are unique among Theropoda in
that the forelimb exceeds 51% of the hindlimb length (Ostrom
1969ft) and 52% of the presacral vertebral column (Cooper
1981a).TheforelimbsofOrnitholestes*andDeinonychosauria
are at least 66% of hindlimb length and 75% of the presacral
column,andtheextremelylongarmsseeninmanyflyingbirds
exceedthelengthsofboththehindlimbsandthepresacralver-
tebralcolumn(Ostrom1976a).Ostrom(1978)estimatedthe
forelimbsofCompsognathiistobeonly38%ofhindlimblength.
Yet in the context of all the evidence, the short forelimbs of
Compsognathiisaremostparsimoniouslyinterpretedasasec-
ondary modification
within
coelurosaurs.
61)Manusgracileandelongate,especiallydigitstwoandthree
andtheirmetacarpals,andmetacarpalIonlyone-thirdthelength
of metacarpal II. As argued above, in Saurischia ancestrally
metacarpalIisabouthalfthelengthofmetacarpalII.Moreover,
the length of its second digit and its metacarpal is always less
thanseventimesthewidthacrossthebasesofmetacarpalsIand
II.TheserelationshipsareretainedinCeratosauna(Fig.4K)and
Carnosauna ancestrally (Fig. 4L). In contrast, the hands are
enormouslyelongate,andthelengthoftheseconddigitandits
metacarpalisatleastninetimesthewidthofmetacarpalsIand
II in Ornitholestes* (estimated from Ostrom 1976a), Caenag-
nathidae (Barsbold1983),
Elmisauridae*(Osmolska 1981),
Dei-
nonychosauria(Fig. 4N),
andAvialaeancestrally
(Fig.
40).
The
characteristicallymodifiedmanusofOmithomimidaemakes
interpretation of the manus difficult. The length of the second
digit and its metacarpal is over seven times the basal width of
metacarpals I and II. However, metacarpal I is elongate and
subequal in length to metacarpals II and III (Fig. 4M). This
conditionisuniqueamongTheropodaingeneralandcoeluro-
saursinparticular,becausemetacarpalIisonlyone-thirdofthe
length of metacarpal II in all other coelurosaurs, and it is no
more than one-half of the length of metacarpal II in Saunschia
ancestrally. In the context of all the evidence, the diagnostic
modificationsoftheomithomimidhandarehypothesizedtobe
secondary.
62)Combinedlengthsoffirstandsecondphalangesofmanal
digitthreelessthanorequaltolengthofthirdphalanx.Asargued
above,inTetanuraeancestrallythefirstandsecondphalanges
are relatively short compared to the length of the third (e.g..
Fig. 4K, L). In contrast, the third phalanx equals or exceeds the
combinedlengthsofthefirstandsecondphalangesinOmith-
omimidae(Fig. 4M),
Caenagnathidae (Barsbold1983),
Elmi-
sauridae*(Osmolska1981),Ornitholestes*(Ostrom1969ft),
Deinonychosauria (Fig. 4N), and Avialae ancestrally (Fig. 40).
The loss of all but the first phalanx of digit three in Ornithurae
is considered secondary, but it is interesting to note that the
GAUTHIER:
SAURISCHIAN
MONOPHYLY
remaining phalanx is very short relative to the length of its
metacarpal,justasinothercoelurosaurs.Thisapomorphyarose
convergentlywithinPterosauria(Wellnhofer1978).
63) Fourth trochanter feebly developed or absent. A promi-
nentfourthtrochanteristheancestralconditionforArchosauria
(Romer 1956). Moreover, an aliform fourth trochanter is the
ancestral condition for Omithodira (see Appendix A), and this
condition is retained in Procompsognathus* (Ostrom 1981),
Ceratosauria (Welles 1984), and Camosauria (Madsen 1976).
Incontrast,thefourthtrochanterisrepresentedbyafeebleridge
in Omithomimidae (Osmolska et al. 1972), or it may be absent,
asinMicrovenator*(Ostrom1970),Deinonychosauria(Ostrom
1976ft),andAvialaeancestrally(Heilmann1926;Tarsitanoand
Hecht1980).ThisapomorphyaroseconvergentlyinPterosauria
(Wellnhofer 1978).
64)Moundlikegreatertrochanter(=posteriortrochanterof
Ostrom1976ft).Thegreatertrochanterisrepresentedbyaru-
goseareaontheposterolateralmarginoftheproximalendof
the femur (in vertical pose) in Archosauria ancestrally (Romer
1923).TheancestralconditionisretainedbyOrnithischia(Santa
Luca1980),Sauropodomorpha(Cooper1981a),Ceratosauria
(T. Rowe, pers. comm.), and Camosauria (Madsen 1976). In
contrast, a moundlike eminence develops within the greater
trochanterinOmithomimidae(pers.obs.),andOstrom(1976a)
hasnoteditspresenceinMicrovenator*,Deinonychosauria,and
Archaeopieryx*.Althoughlackinginmanysmallbirds,thegreater
trochanterappearstobemodifiedwithamoundlikeeminence
inatleastsomelargeratites,suchasDroniaeus(pers.obs.);thus
the development of this feature may in part be size-related in
extantbirds.
65)Ascendingprocessofastragalusenlargedbothinheight
andwidthtocovermostofanterodistalquarteroftibia.Welles
andLong(1974)pointedoutthesynapomorphicresemblance
among their "omithomimoid" group (e.g.. Fig. 6B, C), which
with the addition of Avialae is equivalent to Coelurosauria as
here defined. Birds present a problem in this regard, because
severalauthorshaveclaimedthattheavianascendingprocess
is not homologous with that seen in other Dinosauria (Whet-
stoneandMartin1979;TarsitanoandHecht1980;Martinet
al. 1980; Martin 1983a, ft). Tarsitano and Hecht (1980) even
claimedthatArchaeopieryx*didnothaveanascendingprocess.
My observations support the generally held view that an as-
cendingprocess present
is in
Archaeopieryx*.Moreover,Martin
etal.(1980)usedultravioletlighttodeterminethatthestructure
inquestionwasnotcalciteashadbeensuggestedbyTarsitano
andHecht(1980).
ThedevelopmentoftheascendingprocessinNeognathaewas
mostrecentlyreviewedbyMartinetal.(1980),whoarguedthat
the ascending process of birds is not homologous with that of
otherdinosaurs,becausetheascendingprocessaroseinbirds
fromaseparateossificationcenterthatfusedtothecalcaneum,
ratherthantotheastragalusasindinosaurs.Thisproposition
requiresreevaluationinlightoffurtherstudy.First,comparable
developmentalevidencefromfossiltheropodsislacking.Welles
(1983)has,however,describedasuturebetweentheascending
processandastragalusinDilophosaurus.Havingexaminedthe
DUophosaurustarsi,itisdifficulttodistinguishthesuturefrom
cracksinthespecimens.However,theorientationofthebone
fibers on one margin of the so called "suture" are distinctly
differentfromthoseintheunderlyingbone.Moreover,theas-
29
cendingprocess appears to
overiapthe
astragalusonthe
medial
side of the proximal astragalar surface. In addition, Welles's
interpretationoiDilophosaurushasbeensupportedbyexamples
ofothernonavialantheropodtarsi(pers.obs.).Moreobserva-
tionsofothertheropodfossilsarenecessarybeforewecancon-
fidentlydistinguishbetweencracksandsutures.Furtherex-
aminationof
the
ascendingprocesses of
other
dinosaursshould
beundertaken.Second,Martinetal.(1980)basedtheirconclu-
siononthedevelopmentofthetarsusinNeognathae,butthe
work of McGowan (1984) and my work with K. Warheit and
K. de Queiroz shows that the condition in Neognathae is not
ancestral for Aves. Indeed, all birds are like other dinosaurs in
the possession of an ascending process; Ratitae and Tinami
retain the ancestral condition in which the ascending process
fusestotheastragalus,butNeognathaeisspecializedinthatat
leastpartofitfusesinsteadwiththeprecociallydevelopedcal-
caneum.BothMartinetal.(1980)andMcGowan(1984)used
thepositionoftheascendingprocesswithrespecttotheprox-
imaltarsalsastheonlytestforhomology,andneglectedthe
fact that this process has maintained its phylogenetic and on-
togeneticassociationwiththe
anterolateral
margin of
thedistal
end of the tibia throughout Dinosauria. Based on the study of
avian tarsal ontogeny (Gauthier, Estes, and de Queiroz, in
prep.),itisclearthattheancestralavianascendingprocessdiffers
from that of other Coelurosauria only in that it is not so broad,
although it may be as tall (e.g., Siruthio).
Thedifferencesbetweentheascendingprocessesofavialans
(e.g..Fig.6D)andothercoelurosaursappeartodependontwo
factors. One is that the size of the ascending process may vary
withthesizeoftheorganisminquestion;theossificationcenter
appearsintheusualdinosaurianposition,andduringontogeny
thiscentergrowsbothdorsallyandmedially,sothatgivenenough
time it would eventually cover the entire distal end of the tibia.
This would account in part for the variation in size of the as-
cendingprocesses of
large
and
smallbirds.
The other factor affects the width but not the height of the
ascendingprocessinOmithurae;thatfactoristhedevelopment
of a prominent tendinal groove on the distal end of the tibia,
which would limit the medial spread of the ascending process
duringontogeny.Theonlyothertheropodswithascendingpro-
cesseswithproportionslikethoseofcoelurosaursaretyran-
nosauridcarnosaurs,andtheyarehereconsideredtohaveac-
quired this
condition
separately.
It must be emphasized that even if Martin et al. (1980) were
correct in assuming that the neognath condition was general
among Aves, this assumption would not preclude close rela-
tionshipbetween birdsandothercoelurosaurs.Toargue that
this region is "different" in birds, rather than ancestral to the
conditionoftheseelementsinTheropoda,doesnotremovethe
possibilitythatbirdspossessatransformationofthecondition
seenincoelurosaursgenerally.BirdsareunlikeotherDiapsida
in many details of the shape of the tibiotarsus and their ankles
are in no sense ancestral to the condition seen in other Thero-
poda.Inviewofthenumeroussynapomorphiessupportingthe
dinosaurianandcoelurosaurianrelationshipsofAves,andthe
dearthofdiscordantdata,itismoreparsimonioustoconclude
thatthesmallerascendingprocessanditsassociationwiththe
enlargedcalcaneuminNeognathaearesecondarymodifications.
66) Metatarsal I lies more on the posterior than the medial
sideofmetatarsalII.Asdescribedabove,coelurosaursarefur-
30 ORIGINOFBIRDSANDEVOLUTIONOFFLIGHT

ther derived within theropods (metatarsal I lies on the medial morpha will be used as successively more remote outgroups.
side of metatarsal II ancestrally) in that metatarsal I lies in a Maniraptorapossessesthefollowingsynapomorphiesdistin-
posteromedial position (e.g.. Fig. 6Q). Wellnhofer (1974), Os- guishing this
taxon
among
Theropoda.
trom (1976a), and Tarsitano and Hecht (1980) noted the pos- 68) Prefrontal reduced or absent. As in amniotes generally,
teriorpositionofmetatarsalIinAvialae,althoughtherewas boththelacrimalandprefrontalarepresentinSauropodomor-
some disagreement among them as to the level at which this pha(Galton1984).InTheropodaancestrallythelacrimalgains
transformation took place within Theropoda. Once again, the broad exposure on the skull roof thus partly replacing the pre-
controversystemsfromdifferentinterpretationsoftheonginal frontal.Theancestralconditionofthetheropodlacrimalisre-
positionsofdissociatedelementsinfossils.Thistransformation tainedbyCeratosauria(Gilmore1920)andCamosauria(Mad-
is here considered to have taken place in the ancestral coelur- sen 1976). This condition is also retained in Coelurosauria
osaur,becausetheapomorphicconditionisknowntobepresent ancestrally, as is indicated by the presence of both elements in
inarticulatedexamplesofConipsognalhusandAvialae.More- the growth series of the omithomimid Galliminnis (Osmolska
over,articulatedceratosaursdemonstratethatthisconditionis et al. 1972), and in all other ornithomimids in which this region
not ancestral for Theropoda (pers. obs. of Segisaunis*). The IS well preserved (Russell 1972). In contrast to the ancestral
position of this element, however, remains a matter of conjec- conditionincoelurosaurstheprefrontal,ifnotabsent,isatleast
tureincamosaurs,andfuturefindsmayshowthattheapo- reduced, its role in the construction of the skull roof having
morphicconditionappliestoamoreinclusivetaxonthanto beensupplantedbythelacrimal.Theprefrontalisabsentfrom
coelurosaurs alone among theropods. S. Hope (pers. comm.) all stages of ontogeny in Aves (Bellairs and Jenkin 1960), and
notesthatcorvidshavereversedtheancestralcoelurosaurcon- no separate element is reported from this region in Ornilho-
dition,becausemetatarsallies
I onthemedialsideofmetatarsal Icsies*(Osbom1917)orCaenagnathidae(Barsbold1983).
II, as it does in Theropoda ancestrally. Currie (in press a) may have identified a prefrontal, albeit
67) Proximal surface of metatarsal IV subequal to II in size reduced and no longer in contact with the nasal, in Troodon.
andmetatarsalIIImoreorlesspinchedbetweenthem.Asargued Henotesthatnoclearsutureremainsbecauseofcoossification
above, the enlargement of the fourth metatarsal and the con- withthefrontal.Nevertheless,heconcludesthatthestmcture
strictionoftheproximalendofthethirdistheancestralcon- ISaprefrontalbecauseitsrugosedorsalsurfacebearsshallow
ditionforTetanurae. In
contrast
to
Camosauna ancestrally
(Fig. channels for blood vessels and its so-called "sutural contact"
6G), however, metatarsal IV is subequal to HI in size and meta- with the frontal is marked by several foramina. Currie notes
tarsalIIformsnomorethan22%ofthewidthofthetransverse furtherthatatleastremnantsofsuturalscarsindicatethepres-
axis of the ankle joint in Omithomimidae (Osmolska et al. enceofasmallandnarrowprefrontalinDromaeosaurus.Saur-
1972).Hiilsanpes(Osmolska1982),Conipsognathiis(pers.obs.), onitholestes*.andperhapsinConipsognalhus.However,the
Elmisauridae*(Osmolska 1981). Deinonychosauna (Fig. 6H), criteriaCurrieusedtoidentifyaprefrontalinTroodonmayalso
and Avialae (Fig. 61). A further derived condition, in which the be observed along the anterolaterally in ratites with subtrian-
third metatarsal is so pinched between the second and fourth gular frontals. Moreover, the lateral margins of the frontal are
metatarsals that it barely contributes to the ankle joint, has similarly demarcated in tinamous; the dorsolateral surface is
ansen independently in tyrannosaurid carnosaurs, omitho- rugoseandanapparentsutureisformedbyaseriesofforamina.
mimids,Hiilsanpes.troodontids,andelmisaunds*.Omithunne The demarcated area often falls off when preparing the speci-
birds are unusual in that the proximal end of metatarsal II lies men,leavingtracesofwhatcouldbeconsideredasuturalsur-
behind the plane of metatarsals II and IV, but omithurines are face.Thereis,however,noevidencethatthisstructurerepre-
otherwisecoelurosaurlikeinsharingtheseapomorphiesinmeta- sentsaprefrontalbone.Inbothratitesandtinamous,the
morphology.
tarsal demarcationofthefrontalinthisregionappearstobeassociated
withtheexternalnasalgland.
Avesisdistinguishedamongextanttetrapodsbythesynap-
V.PhylogeneticRelationshipswithinManiraptora omorphicdisplacementoftheextemalnasalglandtoaposition
Maniraptora(n.txn.) above the orbit. In birds in which the frontal reaches broadly
(Gr.:
manm. hand;rapULs,to
sei/c) over the orbits, the extemal nasal gland appears to induce the
Temporal Range.
—lateJurassic to
Recenl. changesinfrontalformnotedabove,includingalineararrayof
Included Taxa.—This taxon is
erected to
emphasi/'c thai
all
the characters more or less complete foramina marking the gland's position
listedbelowweresharedby
common
a ancestor of
Avialae
andDeinonychosauna below the frontal margins (pers. obs.). (In some marine birds
(orat
leastdromaeosaurs) thatwas notalso
anancestor shared withOrnitho- the dorsal surface of the frontal may display a foramen-filled
mimidac. However, some of
these characters also
appear to
bepresentin
some trough that marks the dorsal displacement onto the frontal of
of
thelesswell
known coelurosaurs, such as
Coclunis*.Ornilholesres'. Mnro-
vciialor'.
Satiwrnilholestcs*.Hiilsanpes.
Caenagnathidae, Elmisaundae*. and ahypertrophied,salt-excretingextemalnasalgland.)Resolution
Coinpsiii;nalhi4s Unfortunately, these
taxa
are
too
incompletely preser\'edto
de- oftheproblemoftheidentityofthisstmcturewillrequiremore
terminethe
sequence
which
inthe
maniraptoran synapomorphies appeared(see evidence.Nevertheless,whetheroneconsidersthestmcturea
Fig.9).Amoreprecisedeterminationofthelevelofsynapomorphyofthese reducedprefrontal,ormodificationofthefrontalinducedbyan
characters must
await
future finds
(seeConclusions, part
for
Ifurthercomment).
avianlikepositionoftheextemalnasalglandinmaniraptorans,
Diagnosis.— Thh section will examine the phylogenetic re- thesetheropodsmustbeconsideredmorebirdlikeinthisrespect
lationshipsbetween Deinonychosauria and
Avialae.
As
in
pre- thanareothertheropods.
vioussections,evidencederivedfromlesswellknowncoelur- Wellnhofer(1974),likesomeearlierauthors,thoughtasep-
osaurslistedabovewillbeincluded.Inthefollowinganalysis, arateprefrontalbonemayhavebeenpresentintheEichstatt
Omithomimidae,Camosauria,Ceratosauria,andSauropodo- .4ichaeoptery.\*.Havingexaminedthisspecimen,Iamnotcon-
GAUTHIER:
SAURISCHIAN
MONOPHYLY
vinced of this interpretation, because the skull in general and
theprefrontalregioninparticulararecrushedandincompletely
preserved. I agree with other authors that the portion of an
elementintheantorbitalregiononthecounterslabistheremains
of the lower ramus of the lacrimal, and I see no evidence of a
separateprefrontal.Partsofthelacrimalremainintheantorbital
region of the main slab, and there is a division between a frag-
mentoftheantorbitalramusandanotherfragmentlyingonthe
skull roof The dispute centers on the identity of the latter frag-
ment,andtheconclusionthatthedivisionbetweenthefrag-
mentsonthemainslabrepresentsasutureratherthanabreak
inanoriginallyintactelement.Giventhedirectioninwhichthe
skullwascompressedduringpreservation,thepositionandori-
entationofthedivisioncorrespondstoanaturalfractureplane
betweentheskullroofandantorbitalmoietiesoftheusualman-
iraptoranlacrimal.Moreover,theshapeandrelationshipsofthe
so-calledprefrontalareunlikethoseofothertheropods,partic-
ularlysincethe"prefrontal"wouldexcludethelacrimalfrom
the skull roof unlike the case in any other theropod, including
allotherbirds.Althoughthereissomequestionaboutthischar-
acter,itappearssafetoconcludethatunlikeothertheropods.
theprefrontaliseitherreducedorlostinmaniraptorans.
69)Axialepipophysesprominent.Epipophysesarepresent
ontheanteriorcervicalsinSaurischiaancestrally.Ingeneralthe
prominenceoftheepipophysesvariesbothwiththeirposition
in the column and with the size of the saurischian in question.
Nevertheless, in contrast to the condition of the axial epipo-
physesinSaurischiaancestrally(Fig.3D,E),thesestructures
aremostprominentinDeinonychosauria(Fig.3F)andAvialae
(Fig.3G),regardlessofsize.
70)Hypapophysesonvertebraefromcervicothoracicregion.
In Archosauria ancestrally the cervical vertebrae bear a more
orlessprominentmidventralkeel.Thekeeliscommonlymost
prominentintheanteriorelementsofthecervicalseries,butit
has become reduced or lost several times within Archosauria
(Romer 1956). The keel is produced ventrally to form hypa-
pophyses eusuchian
in crocodylomorph pseudosuchians, but
hypapophysesareotherwiseabsentinnondinosaurianarcho-
saurs (Romer 1956). Hypapophyses are also absent in Sauro-
podomorpha(e.g.,GaltonandCluver1976),Ceratosauria(e.g.,
Gilmore 1920), and Camosauria (e.g., Madsen 1976). The an-
cestralconditionisretainedinCoelurosauriainthathypapo-
physesareabsentinOmithomimidae(Osmolskaetal.1972)
and Compsognathus (Ostrom 1978). In contrast, midventral
keelsintheanteriortrunkvertebrae(i.e.,cervicothoracicregion)
areproducedventrallytoformdiscretehypapophysesassociated
with the attachment of the M. longus colli ventralis in Deino-
nychosauria(Ostrom 1969/i).
Thecondition
not
isdeterminable
inanyoftheArchaeopteryx*specimens,butmoreorlessprom-
inenthypapophyses are
present throughoutOmithurae (Bellairs
and Jenkin 1960). In some groups of birds, such as loons and
grebes,thehypapophysesmaybelargeandelaborate,andthey
mayextendwellintothecervicalregion(ZusiandStorer1969).
However,inotherbirds,suchasApteiyx.thehypapophysesare
comparativelysimple,feeblydeveloped,andtheyareconfined
totheanteriorfewtrunkvertebraeasindeinonychosaurs(pers.
obs.).
71)Modificationsassociatedwithtransitionpointbeginclose
tobaseoftail.Asarguedabove,Tetanuraeisapomorphiccom-
paredtoTheropodaancestrallyinthatthemodificationsas-
31
sociatedwiththe
transition
point
aremoremarked. Compared
to Compsognathus (Ostrom 1978) and Omithomimidae (Fig.
2E),however,thetransitionpointismoreproximallyplacedin
that the transition takes place between caudals 7 and 1 1 in
Deinonychosauria (Fig. 2G) and Avialae (Fig. 2F). Fairly com-
pletecaudalseriesareknownfrombothDeinonychus(Ostrom
1969^))andtheEichstattArchaeopteryx*andtheydifferfrom
thetailsofomithomimidsandothertheropodsinseveralways.
First,theneuralspinesandtransverseprocessesareconfinedto
thefirstseventoninevertebrae,althoughlikeotherarchosaurs
bothstructuresaremoreprominentinthelargerDeinonychus
compared to the smaller Archaeopteryx*. Second, the first 5
caudalshaveshort,boxlikecentraandverticallyorientedzyg-
apophysialfacets.Andthird,bothhavemodifiedhaemalarches
that are longer than deep in lateral view extending nearly to
baseofthetail.Thus,maniraptoransarethemostderivedamong
Theropoda in the development of the dynamic stabilizer tail
(Ostrom1969*;Padian1982).Followingw'elInhofer(1974),L.D.
Martin(pers.comm.)agreesthatlong,taperedprezygapoph-
ysesarepresentposteriorly,butarguesthatonlythepostzyg-
apophyses are elongate in the anterior portion of the tail in
Archaeopteryx*(Eichstatt).Afterexaminingacastofthespec-
imeninMartin'slab,Ihave,however,concludedthatAr-
chaeopteryx* had
elongate
prezygapophysescaudals
on beyond
those in the base of the tail as in other maniraptorans for two
reasons.First,theprezygapophysesinthemiddleofthetailare
variable in length and shape as preserved; if these variations
arenatural,Archaeopteryx*wouldbeunlikeallothersaurians.
Second,elongatepostzygapophysialarticularfacetsareuniform-
lypresentinthisregion,andthismodificationseemssuperfluous
withoutconcomitantelongationoftheprezygapophyseswith
which the facets articulate. Taking these factors into account,
variationinprezygapophysialshapeandlengthinthemidcau-
dals is here attributed to removal of the counterslab and con-
sequentdamagetothenarrowandsuperficialportionsofthe
prezygapophyses.
Although the caudal series is quite modified in Omithurae,
principally by the reduction and fusion of the distal segments
andthelossofthezygapophyses,anumberofthemaniraptoran
synapomorphiesintailformhavebeenretained.Forexample,
the short and boxlike centra in the proximal series are wide-
spreadamongAves.Inaddition,althoughtheymaybemodified
and fused to the centra, boat-shaped haemal arches that are
longerthandeepandflat-bottomedinlateralviewareretained
in the caudal region of some groups, such as Hesperomithes
(Marsh1880),andloonsandpenguins(pers.obs.).Likeallbirds,
Archaeopteryx*hasenlargedfeathersalongthelateralmargins
of the tail (Ostrom 1976a), but then it is the only nonomithu-
rinemaniraptoranpreservedinanenvironmentofdeposition
inwhichfeatherimpressionscouldbepreserved.Theorienta-
tionofthefeathersandcaudalzygapophysesinArchaeopteryx*
indicatesthatthetailwasmostmobileinthedorsoventralplane,
asindinosaursgenerally.Evidently,thehorizontalorientation
of the tail feathers in extant birds cannot be accounted for in
functionaltermssolelyasanadaptationforflight.
72)Coracoidwithsubrectangularprofile.Thecoracoidissub-
circular in profile in Omithodira ancestrally (see Appendix A).
This condition is retained in Sauropodomorpha (Fig. 4B) and
Ceratosauria(Raath1969),andthisappearstobethecasejudg-
ingfromwhatremainsoftheelementinLiliensternus*(Huene
32
1934).Exceptforthetaperingposteroventralmargin(seePart
III. character 42), the coracoid is otherwise subcircular in Car-
nosauria(Madsen1976)andinthecoelurosaursCompsogna-
thus (Ostrom 1978), Omithomimidae (Fig. 4C), and Caenag-
nathidae (Barsbold 1983). The coracoid in Deinonychosauria
and Avialae is further derived relative to that seen in other
Coelurosauria in that it is enlarged, particularly at the ventro-
medialand posteriormargins; thesemodificationsimpartto
the
element a subrectangular profile (Fig. 4E). In addition, a pro-
nounced"coracoid tubercle" lies
just
anterior
to
the
glenoidrim
and immediately ventral to the coracoid foramen. During on-
togeny,theshouldergirdleofthechickentransformsfroma
subrectangulartoanelongatecoracoidcharacteristicofOrni-
thurae; I have been unable to determine if the coracoid was
subcircular in outline at its initial stages of development. In-
terestingly,
subrectangular
a coracoid
has
re-evolved
wilhin
Hesperornithes(comparecoracoidsofBaptotniswithHespewr-
nis in Martin and Tate 1976) and in some other flightless birds
such as the ratites (Feduccia 1980). This appears to be an ex-
ampleofpacdomorphosisinthatthesubrectangularshapeisa
transitory stage in the transformation of the coracoid in the
ontogeny of Aves. This is, however, the terminal stage in the
ontogeny of the element in the two successively more remote
outgroups of Omithurae, namely Deinonychosauria and Ar-
chaeopteiyx*.ContrarytoTarsitanoandHecht(1980),thecor-
acoidinArchaeopteryx*isnotancestralwithrespecttothe
condition seen in other theropods. On the contrary, the shape
andflexureoftheelementandtheenlarged"coracoidtubercle"
in Archaeopteryx* (Fig. 4D) are matched in kind, if not degree,
indeinonychosaurs(Fig.4F)andelmisaurids*(P.J.Currie,pers.
comm.). Ostrom (1974a. b. 1976c/. b) and Padian (1982) have
explored the significance of the transformations in coracoid
morphologyandtheirrelationshiptothefunctionoftheforearm
andtheoriginofflightinthetaxahereincludedinManiraptora
(seealsoGauthierandPadian1985).
73) Elongate forelimb nearly 75% of presacral vertebral col-
umnlength,andelongatehandnearlyequalsorexceedslength
offoot.Asarguedabove,coelurosaursarediagnosableinpart
onthebasisoftheirelongateforelimbsandhands,althoughthe
hands of omithomimids may be secondarily shortened. Ac-
cordingtoOstrom(1976a),theforelimbsofOniilholesles*and
Deinonychosauriaareabout75%ofthelengthofthepresacral
vertebral column, and those of Archaeopteryx* are 120% to
140% of the presacral column length. Moreover, the hand is
92% of the length of the foot in Deinonychosauria (Ostrom
19696, 1976a) and the hand exceeds the length of the foot in
Avialae ancestrally (Bellairs and Jenkin 1960). Ostrom (19746,
1976a), Padian (1982), and Gauthier and Padian (1985) have
notedthesynapomoi^phicresemblancebetweentheforelimbs
of deinonychosaurs and birds, and have explored the signifi-
canceofthesecharactersintheoriginofflightwithinCoeluro-
sauria.
74) Ulna bowed posteriorly. As in .\rchosauria ancestrally,
theposteriormarginisroughlystraightandtheelementisnot
bowedposteriorlyinSauropodomorpha(Cooper1981a.-736,
fig. 3 ID), Proconipsognalhiis* (Ostrom 1981), Lilienstermis*
(Huene1934),Ceratosauria(Welles1984),Camosauria(Mad-
sen 1976:137, fig. 42D), and Compsognathus (Ostrom 1978).
The ulna is only slightly bowed in Omithomimidae (Osmolska
et al. 1972). In contrast, the ulna is strongly bowed posteriorly
ORIGINOFBIRDSANDEVOLUTIONOFFLIGHT
in Caenagnathidae (Osbom 19246), Microvenator* (Ostrom
1970),troodontid(Russell1969:603,fig.9A)anddromaeosaur
(Ostrom19696.'95,fig.58A)deinonychosaurs,andArchaeop-
teryx*(Ostrom1976a).Theulnaisevenmoreprominently
bowed in Omithurae (Bellairs and Jenkin 1960). This apomor-
phyaroseconvergentlywithinPterosauria(Wellnhofer1978).
75) Semilunate carpal. As argued above, distal carpal I caps
thebasesofmetacarpalsIandIIinTheropodaancestrally(Fig.
4K), and this condition is retained in Camosauna ancestrally
(Fig, 4L). The carpus in Omithomimidae (Osbom 1917) and
Tyrannosauridae(Barsbold1983)iscomposedofsmall,poorly
formedcarpalslackingarticularfacets,andbothtaxaappearto
be paedomorphic in this respect. Although it is clear that the
tyrannosauridconditionderivedfromtheancestralcondition,
retainedbyothercamosaurssuchasAUosaitrus.thecondition
from which the omithomimid hand derived is unclear. As in
tyrannosaurids,thepaedomorphiccarpalsofomithomimids
couldhavederivedfromtheancestralcondition.However,be-
causeomithomimidsareclosertomaniraptoransthancamo-
saursare,itisalsopossiblethatthepaedomorphiccarpalscould
havederivedfromtheconditionhereconsidereddiagnosticof
Maniraptora.Iconsidertheformercasetobecorrect,although
I recognize that more information is needed to dispel the am-
biguityregardingthispoint;inanycase,thisdecisionhasno
influenceonthefinaloutcomeoftheanalysis.
In contrast to the ancestral theropod condition retained by
thetetanurineancestor,distalcarpalIinManiraptorawastrans-
formedinto depressed,
a semilunate-shaped carpal with
deep,
a
horizontal groove proximally for reception of a reciprocally
shapedridgeforarticulationwiththeproximalcarpal,andtwo
prominentfossadistallyforreceptionofthebasesofmetacarpals
I and II (see Russell 1969; Ostrom 1976a). A semilunate carpal
ISpresentinCoehirus*(Ostrom1976a),Caenagnathidae(Bars-
bold 1983), Deinonychosauria (Fig. 4N), and Avialae (Fig. 40).
The integration of manal digits I and II in the wrist was initiated
in Saurischia ancestrally. Functional integration was further
advancedinTheropodawhendistalcarpalIcappedbothmeta-
carpals.With the
development of
a
semilunatecarpal,the
carpus
reachedakindoffunctionalequivalencewiththatofOmithurae
in ancestral Maniraptora. As noted by Russell (1969) and Os-
trom(19696), the
resulting
mesocarpal joint
permittedincreased
mediolateral excursion of the hand relative to the forearm
(=190°), a prerequisite to the development of avian powered
flight(J.Rayner,pers.comm.).
76) Metacarpal III bowed laterally and very thin compared
to metacarpal II. Metacarpal III is neither very thin nor bowed
in Sauropodomorpha (Fig. 41, .1). Ceratosauria (Fig. 4K), and
Omithomimidae (Fig. 4M). The third digit and its metacarpal
are reduced, but never bowed, in Camosauna (Fig. 4L). In con-
trast,metacarpalIIIisverythincomparedtometacarpalII,
and it is bowed laterally in Deinonychosauria (Fig. 4N), Ar-
chaeopteryx*(40; basedon
pers.
obs.
of
Eichstatt
specimen)
and most Omithurae (Bellairs and Jenkin 1960). A gracile third
metacarpalisalsopresentinOrmtholestes*(Osbom1917)and
Caenagnathidae(Osbom19246).Iamunabletodeterminefrom
thepublishedfiguresiftheelementisbowed,andexamination
of the specimens will be necessary to decide if a thin and a
bowedthirdmetacarpalareoneandthesamecharacter.Asin
all tetanurines, the third metacarpal lies ventral to the level of
second,andthebowednatureofthethirdmetacarpalmayonly
GAUTHIER:
SAURISCHIAN
MONOPHVLY
be apparent in dorsal or ventral view; thus, the preserved ori-
entationoftheelementiscriticalinrecognizingthischaracter.
Differencesinorientationaspreservedmayaccountforwhythe
thirdmetacarpaloftheBerlinArchaeoplery.x*appearsbowed
in the left hand but not in the right.
77) Posterodorsal margin of ilium curves ventrally in lateral
view. The dorsal margin of the ilium is gently arched in profile
in Saurischia ancestrally (Fig. 5B, C), although very early in
sauropodomorphphylogenythearchingbecomesmoreprom-
inent.ThegentledorsalarcisretainedinTheropoda,however,
the posterior margin of the ilium is truncated vertically in Lil-
lensternus*(Huene1934),Ceratosauria(Fig.5C),Camosauria
(Fig. 5D), and Omithomimidae (Fig. 5E). In contrast, the dorsal
marginoftheiliumcurvesposteroventrallyinprofileinCaenag-
nathidae(Barsbold1983),Omitholestes*(Osbom1917),Dei-
nonychosauria (Fig. 5F, G), and Avialae (Fig. 5H, I).
78) Pubic peduncle of ilium extends ventrally beyond level
ofischiadicpeduncle,andpubisdirectedposteroventrally.The
pubicandischiadicpedunclesterminateataboutthesamelevel
andthepubispointsanteroventrallyinSauropodomorphaan-
cestrally(Fig.5B).TheancestralconditionisretainedinLil-
wnsternus*(Huene1934),Ceratosauria(Fig.5C),Camosauria
(Fig.5D),Caenagnathidae(Barsbold1983),Ornilholestes*(Os-
born 1917), and Omithomimidae (Fig. 5E). In contrast, the
pubic peduncle extends further ventrally and the pubis is di-
rectedmore or
lessposteroventrally in
Deinonychosauria (Fig.
5F, G) and Avialae (Fig. 5H, I). Ostrom (1976^) was the first
to note that both the morphology of the pubic peduncle of the
iliumandthepreservedorientationofthepubisinDeinonychus
indicated that the pubis could not have projected anteroven-
trallyasindinosaursancestrally.Thisobservationhasbeen
corroboratedbyfindsofarticulateddromaeosaurpelves(Bars-
bold 1977, 1979, 1983; Barsbold and Perle 1979). Of course,
in the absence of articulated pelves the precise orientation of
thepubisisdifficulttodetermine,asthevarietyofrestorations
of pubic orientations in Archaeopteryx* and dromaeosaurs
graphically illustrates (Fig. 5F, G). Tarsitano and Hecht (1980)
arguedthatOstrom's(1976a)restorationoftheArchaeopteryx*
pelviswasincorrect,andthatthepubesofArchaeopteryx*were
infactorientedmoreasinOmithurae.Regardlessofthedegree
ofreversal,however,thepointtobeemphasizedisthatDeino-
nychosauriaandAvialaearelikeoneanotherandunlikeall
otherTheropodainthegreatlengthofthepubicpeduncleand
that the pubis no longer points anteroventrally. If Archaeop-
teryx*isinfactmorebirdlikeinthisrespectthanotherthero-
pods, that is consistent with our belief that it is a bird, but this
observationhasnobearingonthequestionofrelationshipbe-
tweenbirds
and
deinonychosaurs.
Troodontidspresentapotentiallymoreseriousproblemwhen
tryingtooptimizethecharacter "reversed pubis" on the most
parsimonious tree for all the data. That is to say, an outline
drawing of a pelvis said to be from Saiirornithoides first ap-
pearedinBarsbold(1977),andthepubiswasfiguredinthe
unreversed,ancestralcondition.Subsequently,dashedlinesin-
cludedinanotheroutlinedrawingofthesamepelvisinBarsbold
(1983)revealedthatbothdrawingsrepresentedanincomplete
specimen.Aformaldescriptionofthisspecimen,includingad-
equateillustrationsandrationaleforitsreferraltoSaiirorni-
thoides.hasyettobepublished.Iftheancestralformofthe
pubis and its attachment to the ilium can be verified in Troo-
33
dontidae, then one might question the monophyly of Deino-
nychosauria, which
would
requireindependent origins
(i.e.,
con-
vergence)toexplainthesimilaritiesbetweentheraptorial
second pedal digits and didactyl pes in troodontids and dro-
maeosaurs.Alternatively, one
if accepts the
raptorial
second
pedal digit and didactyl pes of Deinonychosauria as synapo-
morphic, then reversal of the pubis took place twice among
maniraptorans:onceindromaeosaursandonceinavialans.Yet
anotherandequallyparsimonioushypothesiswouldbetheap-
pearanceof the
reversed pubisin
ancestral Maniraptora, with
itssubsequentreversaltotheancestralconditionintroodontids;
onceagainonemustassumethattroodontidsaremostclosely
relatedtodromaeosaursonthebasisofsharedapomorphiesin
theraptorialsecondpedaldigitanddidactylpes.
Eachoftheseinterpretationsrequiresthreeevolutionaryevents
to account for the distribution of these two characters among
these three taxa. There is thus no basis for choice among the
possiblealternatives.Accordingly,untilthereisfirmerevidence
to the contrary, I will consider the pubis to have been reversed
in the immediate common ancestor of deinonychosaurs and
birds. Assuming this to have been the case, a reversed and
posteroventrallyorientedpubismusthavearisenthreetimes
within Oraithosuchia, because this condition also obtains in
Omithischia(Fig.5A)andsegnosaurs(BarsboldandPede1979;
Barsbold 1983; segnosaurs are here considered sauropodo-
morphdinosaurs of
uncertain
relationships, see
Appendix A).
79) Pubic foot reduced anteriorly. As noted above, a pubic
footispresentinTetanuraeancestrally.However,itisexpanded
both fore and aft in Camosauria (Fig. 5D), Caenagnathidae
(Barsbold 1983), and Omithomimidae (Fig. 5E). In contrast,
the pubic foot is much less developed anteriorly than it is pos-
teriorlyin
Compsognathus (Ostrom 1978), Microvenator* (Os-
trom1970),Coeliirns*(Marsh1896),Deinonychosauria(Fig.
5F,G),andArchaeopteryx*(Fig.5H;Padian1982).Theabsence
of a pubic foot in Omithurae is considered a secondary modi-
fication(Fig.51).Thischaracter,liketheprecedingoneandthe
followingtwocharacters,hasyettobedeterminedintroodon-
tids.Theconditionofthepubicfootincaenagnathidspresents
certain problems; the profile of the structure differs in each of
Barsbold'sfigures(e.g.,1979,1983),butitisusuallyportrayed
as being longer anteriorly than posteriorly. In light of the evi-
denceunder consideration,caenagnathids arepart
of
Manirap-
tora;thustheirpubicfootmusthavebeenmodifiedfromthe
anteriorlyabbreviatedconditionseeninothermaniraptorans.
This may, however, simply reflect inaccuracies in Barsbold's
reconstmctions.
80) Ischium two-thirds or less of pubis length. The ischium
is at least three-quarters of the length of the pubis in Sauro-
podomorphaancestrally (Fig.5B),
in
Liliensternus*
is
as
it (Huene
1934), Ceratosauria (Fig. 5C), Camosauria (Fig. 5D), and Or-
mthomimidae (Fig. 5E). In contrast, the ischium is no more
than two-thirds of the length of the pubis in Compsognathus
(Ostrom 1978) and Caenagnathidae (Barsbold 1983). and al-
thoughthepubisisincompletedistally,thisappearstobethe
case in Omitholestes* as well. As noted by Ostrom (1976a, b)
and Padian (1982), the ischium is only half or less of the pubic
length in Deinonychosauria (Fig. 5F, G) and Archaeopteryx*
(Fig.5H).Omithuraeishypothesizedtobesecondarilymodified
owing to reduction in the length of the pubis (Fig. 51).
81)Obturatorprocessdistallyplacedonischium.Asargued
34
above, an obturator process on the ischium is the ancestral
condition for Tetanurae. In Camosauria the process is proxi-
mallyplaced(Fig.5D),andthisconditionisretainedinOmitho-
mimidae (Fig. 5E). Ornilholestes* (Osbom 1917) appears to
havealargeanddistallyplacedobturatorprocessbutitspelvis
isnotwellenoughpreservedtoallowfurtherinterpretation.The
obturatorprocessisdistallyplacedinCompsognathus(Ostrom
1978) and Caenagnathidae (Barsbold 1983), and it is both en-
largedanddistallyplacedinDeinonychosauria(Fig.5F,G).As
discussedabove,theconditioninArchacopleryx*isdifficultto
interpret; either an obturator process is absent, or it is repre-
sentedbytheventralofthetwoprocessesatthedistalendof
the ischium (Fig. 5H). This process is absent in Omithurae, but
inanycaseitsabsenceinbirdsisherehypothesizedtorepresent
a phylogenetic loss rather than retention of an ancestral con-
dition.
82)Anteriortrochanternearlyconfluentwithproximalhead
of femur. As noted above, the anterior (=lesser) trochanter is a
tall flange that is separated by a deep cleft from the femoral
shaftinCamosauria(Madsen1976)andCoelurosauriaances-
trally.Omithomimidae(Osmolskaetal.1972)andMicrove-
nator*(Ostrom1970)retaintheancestraltetanurinecondition.
The femoral head in Deinonychosauria (Ostrom 1976/i) and
Archaeopteryx* (Ostrom 1976a) is like that of other coeluro-
saursinthattheanteriortrochanterextendstothefemoralhead
(Padian1982).However,maniraptoransaredistinguishedfrom
othercoelurosaursinthattheanteriortrochanterisseparated
from the femoral head by no more than a shallow groove, and
the antenor trochanter is thus nearly to completely confluent
withthefemoralhead.Ostrom(1976a.124,125)describedand
figured the proximal end of the femur of Microvenator*, Ar-
chaeopteryx*,and Calhartes.and
arguedthat
thefemur
oi
Ar-
chacopleryx*intermediate
is betweenthat
"theropods"
of and
Recentbirds,inwhichthereisnoseparateanteriortrochanter
(exceptasanoccasionalvariantamongneognaths).Asisevident
fromRomer(1927),birdsrecapitulatethephylogenetichistory
of this transformation during ontogeny; the pifi 2 attaches in
theancestralarchosaurianpositionandsubsequentlymigrates
proximallythroughvarioustheropodattachmentpoints,even-
tuallyarrivingatthepositionseeninextantbirds.Thus,the
five (or four) muscles inserting on the avian major trochanter
yield a composite anterior plus greater trochanter (T. Rowe,
pers.
comm.).
83) Absence of fourth trochanter on femur. Tarsitano and
Hecht(1980)notedthat Archaeopteryx* and other nonswim-
ming birds are unusual among archosaurs in the loss of the
fourth trochanter. As noted above, however, the fourth tro-
chanterIS
feebly
developed among coelurosaurs ancestrally,
and
itisabsentinDeinonychosauriaandbirds.Pterosaursalsolost
thefourthtrochanter(Wellnhofer1978),andthisareaofmuscle
attachment is also reduced in a number of very large quadru-
pedaldinosaurs,suchasSauropoda(seeAppendixA).
84) Pedal digit IV longer than II and closer to III in length.
Pedal digit IV is longer than II and closer to III in length in
archosaursancestrally (Fig. 6J-N). As argued above, however,
theropods are apomorphic in that pedal digit II is only a little
shorter than IV, and the foot is thus more symmetrically de-
velopedaboutdigitIII(Fig.60).Thesymmetncaltheropod
foot is retained in Procompsognathus* (Ostrom 1981), Cera-
tosauria (Raath 1969). Camosauria (Madsen 1976), and
ORIGINOFBIRDSANDEVOLUTIONOFFLIGHT
Omithomimidae(Osmolskaetal.1972).However,theancestral
theropod condition appears to have reversedwithinCoeluro-
sauriainthatpedaldigitIVapproachesIIIinlength,andthus
islongerthandigitIIinOrnitholestes*(Ostrom1969/5),Comp-
sognathus (Ostrom1978),
Caenagnathidae (Barsbold
1983),
Deinonychosauria(Russell1969;Ostrom1969/)),andAvialae
(Fig. 6P). Tarsitano and Hecht (1980) argued that birds were
plesiomorphiccomparedtoothertheropodsowingtothegreater
length of pedal digit IV. However, in view of the data presented
here,thischaracterishypothesizedtobeanapomorphicreversal
rather than a plesiomorphic retention in birds and other man-
iraptorantheropods.
Another aspect of the construction of the pes that may be
related functionally to a lengthened fourth digit is the devel-
opmentofaraptorialsecondpedaldigit,anattributethatis
taken to extreme in troodontids and dromaeosaurs. No other
theropodsasidefromcariamidneognathsapproachthedegree
ofspecializationcharacteristicofdeinonychosaurs.However,a
few other extinct maniraptoran coelurosaurs appear to show
incipient specializations to this end. For example, an enlarged
articularsurfaceindicatingagreaterradiusofexcursionbetween
the second and third phalanx in the second pedal digit is ap-
parentinElnnsaurus* (Osmolska 1981)
andOrnitholestes*(Os-
bom1917);unfortunately,themoredistalportionofthisdigit,
includingtheungual,werenotpreservedineitherspecimen.As
pointed out by D. Yalden (pers. comm.), the claw impressions
encasingthepedalungualsofArchaeopteryx*resembletheclaws
of arboreal mammals more than they do the claws of raptors
using the foot to secure prey. The claw impressions of Ar-
chaeopteryx*are also
unlike
those
of
Compsognathus, in
that
theimpressionsofthelatterdisplayaflat-bottomedshapethat
would have been better suited to running than climbing. Yal-
den'sevidenceiscompelling;nevertheless,Archaeopteiyx*clearly
has an enlarged articular surface on the distal end of the first
phalanx of the second digit (pers. obs. of Eichstatt specimen),
accompanied by an enlarged claw (Wellnhofer 1974). This in-
dicatesthatArchaeopteryx*possessedsomeabilitytopinand
securepreywithitsfoot(orusethistoeasadefensiveweapon),
andthisattributemayhavebeenpresentinmaniraptoransan-
cestrally.Tobesure,noneofthesespecimensisasspecialized
asaredeinonychosaursinthisrespect.Nonetheless,thefootof
Archaeopteryx*appearstobeintermediatebetweenthefunc-
tionallytridactylfootoftheropodsancestrallyandthefunc-
tionallydidactylandraptorial
foot
of
deinonychosaurs.
If the use of the foot to grasp prey is ancestral for manirap-
torans,itmightexplaintheinitialadvantageofareversedfirst
digit (the hallux), especially in an organism whose formerly
raptorialforelimbshadbeenpreemptedtoservethedemands
ofpoweredflight.Becausethehalluxopposestheactionofthe
second digit, the pincerlike action of both digits would have
enhancedtheabilitytograspwiththefoot.Thisinterpretation
neednotcontradictYalden'sconclusionsregardingtheuseof
thefeetinArchaeopteryx*.Bythestageofavianevolutionrep-
resentedbyArchaeopteryx*,thepesmayhaveplayedacritical
role in its ability to climb. However, the ability to grasp with
the pes, so necessary to an arboreal biped, may simply have
been inherited from a more distant and nonavialan ancestor.
As noted above, a lengthened fourth toe is common among
extantbirds.However,asidefromthedeinonychosaurlikepes
of cariamids, I am not aware of anv other omithurine bird that
GAUTHIER;
SAURISCHIAN
MONOPHYLY
hasasimilarlymodifiedsecondpedaldigit(althoughanumber
ofratitesarenonethelessabletoinflictseveredamagewiththis
toe;Ostrom1969/)).Accordingly,thisattributeisconsideredto
havebeenlostinOmithuraeancestrally,andtohavere-evolved
inthecommonancestorofcariamidneognaths.
Conclusions
I.SummaryoftheMainPhylogeneticConclusionsofthisWork
As implied by their name "lizard-hipped" dinosaurs,
saurischians have always been grouped on the basis of their
plesiomorphicresemblances.However,comparedtoalternative
hypotheses, such as monophyly of Omithischia-Sauropodo-
morpha or Ornithischia-Theropoda, the taxon Saurischia
provides a better summary of the evidence pertinent to hy-
pothesesofmonophyly. Accordingly,Saurischia
is
hereinde-
finedostensivelytoincludebirdsandalldinosaursthatshare
a more recent common ancestor with birds than they do with
Omithischia.Themostobviousofthetensynapomorphiesunit-
ingSaurischiaistheirelongate,mobile,andS-shapednecks,a
characterthatdistinguishesbirdsamongextantamniotes.Other
birdlikeattributespresentintheancestralsaurischianinclude
a variety of modifications of the manus, such as digit II being
the main axis of the hand, a uniquely modified pollex, the pal-
mardisplacementofthelateraldigits,andtherelativelylarge
sizeofthehand.Tothesemaybeaddedothersynapomorphies
in the skull and postcranial skeleton, such as the modification
of the premaxilla, the spread onto the frontals of the area of
originofthetemporalmusculature,themodificationoftheat-
lanto-axialjoint,thedevelopmentofcervicalepipophyses,and
thepresenceofhyposphene-hypantraaccessoryintervertebral
articulations.
Saurischiaiscomposedoftwoprincipallineages,theextinct
SauropodomorphaandextantTheropoda.Twenty-fivesynapo-
morphiesunite Theropodawithin
Saurischia.
Intact
skulls
are
known from few extinct theropods, and the only synapomor-
phiessofaridentifiedfromthisregionoftheskeletonarethe
characteristicallymodifiedvomersandintramandibularjoint.
The vomers in Ratitae and Tinami are little changed from the
ancestralconditionintheropods.Mostextantbirdsretainthe
basictheropodshapeofthedentary-postdentaryarticulation
andsomedegreeofintramandibularmobility,althoughtheyare
unliketheropodsancestrallyinthattheypossessafusedman-
dibularsymphysis.Intramandibularmobilityhasbeenlostin
birds that have short, powerful jaws with broad mandibular
symphyses.Fromtheperspectiveoftheancestralconditionin
Theropoda, these birds invariably have unusual diets. Most
theropodsynapomorphiesarederivedfromthemorecomplete-
lyknownpostcranialskeleton.Theancestraltheropodwaswell
suited to cursorial habits and was distinctively birdlike, as is
indicated by a light, hollow skeleton, a broader area of origin
of muscles arising from the ilium, a variety of modifications of
thehindlimbsandfeet,andatailthatwasthinnedandstiffened
distallytoenhanceitsroleasadynamicstabilizer.Theancestral
theropodretainedthecarnivoroushabitsancestralforArcho-
sauria, but its raptorial hands indicate further specialization
towardthisend.Moreover,theintramandibularjointsuggests
macropredaceoushabitsinthatitwouldenabletheropodsto
ingestrelativelylargerprey.
ThelateTriassicPwcompsognathits*andHallicosaurus*are
35
ofuncertainrelationships,andtheyarethereforeplacedincertae
sediswithinTheropoda.Amongtheearlytheropodsareseveral
taxa,includingthewell-representedCoelophysis.Syntarsiis,and
Ceratosaunis. that are included here in a modified version of
Marsh'sCeratosauria.Asconstitutedhere,Ceratosauriaisthe
sister-group of a new taxon, Tetanurae, which includes birds
andalltheropodsthatareclosertobirdsthantheyaretoeither
Ceratosauria or to the other early theropods. The monophyly
of Tetanurae is supported by 17 synapomorphies, including a
profounddifferencebetweenthelarger,mobileproximaland
the narrow, stiffened distal parts of the tail. Other birdlike at-
tributesofTetanuraeincludethepresenceofalargemaxillary
fenestra,antorbitaltoothrow,lossofdentarycaniniformtooth,
spine tables in the anterior cervicals, enlarged hand, basal
appression of metacarpals I and II, palmar displacement of
metacarpalIII,lossofmanaldigitIVbeyondembryonicstages
ofdevelopment,pubicfoot,enlargedascendingprocess,and
shortened metatarsal I.
Coelurosauria is ostensively defined to include birds and all
theropods that are closer to birds than they are to camosaurs.
Coelurosauriaisthusremovedfromitsclassicalandparaphy-
leticstatus;andinkeepingwithevolutionarytheory,Coeluro-
sauriais appliedinstead to
a
monophyletic taxonencompassing
Ornitholesles*,Coelunis*.Compsognathus,Microvenator*,
Saurormlholestes* Hulsanpes.
. Elmisauridae*, Caenagnathidae.
Omithomimidae, Deinonychosauria, Avialae, and their im-
mediate common ancestor.
Fifteen
synapomorphies distinguish
Omithomimidae,Deinonychosauria,andbirdsfromCamosau-
ria, and at least some of these apomorphies are shared by the
remainingcoeiurosaurslistedabove.Severalcharactersdistin-
guishingextantbirdsamongarchosaursareseenintheskull
and hindlimb of all coeiurosaurs, but from the standpoint of
poweredflight,themostinterestingofthesearethemodifica-
tionsoftheforelimbandpectoralgirdle.Thesemodifications,
normallythoughttoberelatedtopoweredflight,includeafused
andbonysternum,furcula,elongateforelimbandhand,andthe
beginningsofmedialenlargementofthecoracoid.
A new taxon, Maniraptora, is proposed for the group of the-
ropodsincludingbirdsandallcoeiurosaursthatarecloserto
birds than they are to Omithomimidae. The immediate com-
monancestorofbirdsanddeinonychosaursmaybedistin-
guishedfrom omithomimids bypossession of
17
synapomor-
phies.Evennonavianmaniraptoransareessentiallybirdlikein
manydetailsoftheirmorphology,includingreduction(orloss)
oftheprefrontal,stiffeningandthinningofallbuttheproximal
partofthetail,acharacteristicallymodifiedcoracoid,verylong
forelimb,bowedulnaandthirdmetacarpal,severalmodifica-
tionsofthepelvisincludingareversedpubis(atleastindro-
maeosaursandbirds),absenceofdiscretefourthandanterior
trochanters on the femur, birdlike proportions of the relative
lengths of pedal digits II, III, and IV, and at least the incipient
developmentofaraptorialsecondpedaldigit.Variousspecial-
izationsindicatethattheraptorialfunctionoftheropodfore-
limbsreacheditsapexinancestralmaniraptorans.Thesemod-
ificationsplayed an
important role
in
theorigin
of
flight,
because
the essential elements of the flight stroke are realized in the
manneroffoldingandunfoldingthehands,andextendingand
retractingtheforelimbsandhandsduringpreycapture(Padian
1982;GauthierandPadian1985).
TheprecisediagnosisandcontentsofManiraptoraareprob-
36
lematic.Compsognalhuspresentsparticulardifficulties;itshares
maniraptoran characters 79, 80, 81, and 84 that are absent in
omithomimids, but it retains the ancestral condition, an un-
bowedulna, indicating
that
caenagnathids,Aficrovenalor*,
dei-
nonychosaurs,andbirdsaremorecloselyrelatedtooneanother
thananyofthemistoCompsognalhus.Theanomalouslyshort
forelimbsofCompsognalhuspresentfurtherproblems;thehand
in particular is incomplete and after examining the specimen I
donotfeelsecurewithOstrom's(1978)conclusionsregarding
the number of digits. Also, the ornithomimid tail appears in
some ways more maniraptoranlike than is that of Compsog-
nalhus.Thus,itappearsthatCompsognalhusisoutsidethe
remainingmaniraptorans,anditisalsopossiblethatitisoutside
apossibleomithomimid-maniraptorangroup.
Caenagnathidae,Coelurus*,Microvenalor*,Elmisauridae*.
andSaurornitholesies*,andespeciallyOrmiholesles*,arecloser
to birds and deinonychosaurs than are omithomimids. Al-
thoughthisconclusionseemssecure,thesetaxaaresoincom-
pletelyknown that
the
details
of
their
phylogeneticrelationships
remainobscure.Accordingly,thesetaxaareplacedmceriaesedis
within Maniraptora, and with some reservations, Compsog-
nalhusisalsoincludedinthistaxon.ComparedtoCompsog-
nalhus.itisclearthatdeinonychosaursandbirdsaremore
closely related in that they share characters 69, 71, 72, and 78,
but because of incomplete data it is not clear when these attri-
butesarosewithinManiraptora.Thedistinctlybirdlikepelvis
ofdromaeosaursappearstobeabsentinmostothermanirap-
torans,thus indicating
thatthis
character
arosewithin
thegroup.
If,asBarsbold(1983)suggests,thesemodificationsareabsent
introodontids,thenthemonophylyofDeinonychosaunawould
bebroughtintoquestion;inanycase,thepreciselevelatwhich
thesemodificationsaroseisatpresentindeterminable.Future
discoveriesandreanalyseswilldecidewhichamongthe17syn-
apomorphies listed for Maniraptora are diagnostic of all, as
opposed to some, members of this group. The cladograms in
Figures 8 and 9 depict the phylogenetic relationships among
Theropoda proposed here.
II. The Problem of the Definition of the Name Aves
Aves was initially applied to extant birds alone (Linne 1758).
Subsequentfindsofotherfeatheredtheropodsfromoutsidethe
immediateancestryofallextantbirds(i.e.,.-{rchaeoptery.x*,
Hesperornithes) resulted in the term Aves being applied to a
moreinclusivetaxon.Althoughthisdecisionaccuratelyreflects
phylogeneticrelationships,italsohastheunfortunatesideeffect
ofdiminishingthephylogeneticinformativenessandstability
of the name Aves. That is to say, Aves currently summarizes
only those synapomorphies diagnostic of what I call Avialae,
becauseitrestrictsthediagnosisofAvestothosecharactersthat
aredeterminableinthefossil.\rchacopteiyx*.
An example of the kind of problem that this decision has
generatedisevidentinWellnhoferandWhetstone'sdisagree-
mentovertheidentityofaparticularboneinArchaeopleryx*.
Wellnhofer(1974)consideredtheelementinquestionasqua-
mosal,whereasWhetstone(1983)claimedittobeanopisthotic;
surely,neitherwouldhavedifficultyidentifyingtheseelements
in the skulls of extant birds. This raises the issue of whether or
notsuchaneasilyrecognizabletaxonasextantAvesshouldbe
left to the vagaries too often inherent in interpreting the mor-
ORIGINOFBIRDSANDEVOLUTIONOFFLIGHT
phologyoffossils.Onemustbearinmindthatallsynapomor-
phiespresentin.Archaeoplcry.x*mustalsobepresentinextant
birds, although perhaps in a modified form. However, the re-
versecannotbethecase;notonlydoextantbirdsshareamore
recent common ancestor with one another than any of them
doeswith.Archaeopleryx*,butextantbirdssharenumerousapo-
morphies that are not preservable in fossils. Furthermore, it is
exceedinglyunlikelythatanyfossilwouldleadtotheconclusion
that Recent birds are not most closely related to each other
among extant amniotes. After all. Appendix A and the text
above record over 100 synapomorphies that distinguish the
skeletonsofextantbirdsfromtheirnearestlivingrelatives,the
crocodiles;andtherearenumerouscharactersinsoftanatomy,
behavior, physiology, and biochemistry that are diagnostic of
thistaxonaswell.
Another drawback of current practice is that it treats char-
actersasiftheyweredefining,inthesensethatifanorganism
has feathers, it must be a bird. Not only is this perspective
typologicalandthusantievolutionary,butonecanenvisionthe
difliculties resulting from the possible discovery of feathers in
deinonychosaurs, which most certainly did not fly. Moreover,
undercurrentpracticehowwouldoneclassifyagroupofbirds
that lost feathers during their evolution? In the phylogenetic
system, this presents no problem, for if an organism is bom to
a bird, it is a bird, regardless of the characters it may or may
not possess. Characters may aid recognition of ancestry, but
they do not define it (Gauthier et al, in prep.). Accordingly, it
IS recommended that the taxon Aves be standardized in the
phylogeneticsystembyapplyingthisnameonlytothatportion
ofTheropodathatincludesthemostrecentcommonancestor
ofRatitae,Tinami,andNeognathae,andallofitsdescendants.
Toreflectthemorecomplexrelationshipsthatareofinterest
mainly to paleontologists, new or less widely used names are
applied to taxa including Aves and one or more of its extinct
sister-groups. The name Omithurae is applied to Aves plus all
extinctmaniraptoransthatareclosertoAvesthanis.Archaeop-
tery.x*, and the name Avialae is applied to Omithurae plus all
extinct maniraptorans that are closer to Omithurae than they
aretoDeinonychosauria.Thus,althoughthemeaningofAvialae
and Omithurae is expected to vary according to the discovery
of new fossils, the contents and diagnosis of the taxon Aves in
thecontextofextantamnioteswillremainunchanged.
Birdsarepoorcandidatesforfossilizationandthepaucityof
their fossil record is widely recognized. This unfortunate cir-
cumstanceisnotconfinedtobirdsalone,butappliestoThe-
ropodagenerally, all
of
whichhavelightly
constructed skeletons
fortheirsize.Nevertheless,theropodstratigraphicoccurrences
areconcordantwiththephylogenetichypothesisofferedhere.
For example, Ceratosauria first appears in the late Triassic,
althoughmostceratosaursarereportedfromtheearlyJurassic,
and they persist until the late Jurassic. Continental deposits of
niid-Jurassic age are rare, so it is not until the late Jurassic that
theropodsonceagainbecomerelativelycommoninthefossil
record.BythelateJurassicmostofthemajortetanurinelineages
haveappeared:camosaurs,omithomimids,andmaniraptorans,
including birds, are reported from Upper Jurassic sediments.
Indeed, the only major coelurosaur lineage that is not yet rep-
resentedin the
late
Jurassic
is
Deinonychosauna, whose earliest
appearance in the fossil record is in the early Cretaceous. Or-
GAUTHIER:
SAURISCHIAN
MONOPHYLY 37

nithurinesarealsoreportedfromtheearlyCretaceous,andthe ymousreviewer,andDaphneFautinandthereviewcommittee
earliest record of Aves is in the late Cretaceous. of the California Academy of Sciences. I would like especially
to acknowledge the efforts of Kevin Padian, for sharing his
III. Implications for the Origin of Right knowledgeofpterosaurs,forlongdiscussionsofarchosaursys-
tematics,andforspendingconsiderabletimeandenergyincrit-
Theabove-hypothesizedrelationshipsamongTheropodahave
icallyreviewingandeditingthismanuscript.Ialsowishtothank
importantimplicationsforhypothesesconcerningtheoriginof
Donald Brinkman, Jose Bonaparte, James Clark, Edwin Col-
flight.MostoftheskeletalmodificationsenablingArchaeopter-
bert,PhilipCurrie,RobertLong,MichaelParrish,MichaelRaath,
V.v* to fly are present in all maniraptorans, although only avi-
Jeremy Rayner, Timothy Rowe, Dale Russell, Alick Walker,
alans are able to fly. This requires that any functional expla-
Michael Wilson, Derick Yalden, and John Zawiskie for allowing
nationfor the origin of flight must also account for the
metousetheirunpublishedobservationsand/orcitetheirpa-
apomorphicsimilaritiessharedbyArchaeopleryx*andcoelur-
persinpress.TheWelles-Longphotographiclibraryofthetype
osaursingeneralanddeinonychosaursinparticular(Gauthier
andreferredspecimensofnumerousarchosaurswasofinvalu-
and Padian 1985). One may not agree with the particular ex-
ableassistance.SamuelWellesgraciouslyallowedmetostudy
planationinitiallyoflferedby
Ostrom (1974a.i.e..
theinsect net
undescribed theropods in his care, and made available his li-
hypothesis),butonemustappreciatethatinsofarashesought
brary,notes,andphotographsofvirtuallyeverytheropodcon-
todevelopageneralexplanationforthefunctionoftheforelimbs
sideredinthiswork.SpecialthanksareextendedtoAmold
andpectoralgirdleinManiraptora,hisapproachwasbasically
Kluge and William Fink for their assistance with the PAUP
sound.Thosewhofavorotherhypothesesshouldbegintoana-
program,andtoLynnBarrettiforherillustrations.Iwouldalso
lyzetheevolutionofflightintheropodsinthecontextofthe
like to thank Kevin de Queiroz, Michael Donoghue, and Rich-
independentlineofevidenceaffordedbytheirphylogenetichis-
ardEstesfortheircontributionstothedevelopmentoftheclas-
tory(Gauthier and Padian1985).
sificatory conventions employed in this work. For providing
In view of the evidence placing birds within Maniraptora,
specimens and allowing access to their collections, I wish to
Coelurosauria,Tetanurae,Theropoda,Saurischia,Dinosauria,
thankthecuratorsofthefollowinginstitutionsanddepartments:
Omithodira,Omithosuchia,andArchosauria,itshouldbeclear
UniversityofCalifomiaMuseumsofPaleontologyandZoology;
that birds are neither the sister-group of mammals (Gardiner
DepartmentsofHerpetologyandOrnithology,CalifomiaAcad-
1982),noraretheythesister-groupofcrocodiles(Walker1972;
emyofSciences;DepartmentofPaleontology,AmericanMu-
Martin 1983a). Likewise, this evidence provides a new per-
seumofNaturalHistory;DepartmentsofPaleontology,Omi-
spectiveonWalker's (1964)suggestionthattheropods arepoly-
thology, and Herpetology, University of Michigan Museum of
phyletic.Walker'shypothesiscanberejectedontwocounts:(1)
Zoology; Staatliches Museum fiir Naturkunde, Stuttgart; Mu-
Ornithosuchus is not a camosaur, and (2) the alternative hy-
seumofNorthemArizona;BayerischeStaatssammlungfiirPa-
pothesisof
separateoriginsof
"camosaurs" and
"coelurosaurs"
laontologieundhistorischeGeologic,Munich;InstitutundMu-
from within "thecodonts" is uninformative, because it claims
seumfiirGeologicundPalaontologiederUniversitat,Ttibingen;
nomorethanthat"camosaurs"and"coelurosaurs"arearcho-
DepartmentofPaleontology,FieldMuseumofNaturalHistory.
saurs.AsimilarlineofreasoningcanbeappliedtoChatterjee's
I would like finally to thank the Califomia Academy of Sciences
(1985)postulatedconnectionbetweenthepoposauridrauisu-
for asking me to participate in this symposium and for sup-
chianpseudosuchianPostosuchusandtyrannosauridcamo-
portingthepreparationofthismanuscriptduringmytenureas
saurs;inthecontextofalltheevidence,theirsharedapomorphic
a Herpetology Fellow. This research was supported in part by
resemblancesbeyondthosecommontoallarchosaursmustbe
NSFgrantBSR-8304581.
considered convergent.
Apopularmisconceptionisthatalldinosaurswerehugelum-
beringbeaststhatlongagobecameextinct.Tobesure,onlyone Literature Cited
lineageofdinosaurssurvivedtheendoftheCretaceous.Never- Baird,
D.
1'580.prosauropod
a trackway fromtheNavajo Sandstone (Lower
theless,thislineagecurrentlyaccountsfornearlyhalfofthetotal Jurassic)Pp.219-2.10inAspectsofVertebrateHistory,EssaysinHonorof
speciesdiversityofextantAmniota.Birdsarelivingdinosaurs, Edwin Harris Colbert, L.
Jacobs, ed.
Mus. N-Anzona Press,Flagstaff,
Arizona.
andassuchtheyhaveextendedthepreeminenceofDinosauria Bakker,R.1971.Brontosaurs.Pp.179-181inMcGraw-HillYearbookSci.
Tech.
1971.
amongterrestrialvertebratesfromthelateTriassictothepresent Bakker,R.andP.Galton.1974.Dinosaurmonophylyandanewclassof
day. vertebrates.Nature 248:
169-172.
Barsbold.R.
1974.Sauromithoididae. new
a family of
smalltheropod dinosaurs
Acknowledgments from Central Asia andNorth .America. In
Resultsofthe
Polish-Mongolian
Palaeontological Expedition, Part
Kielan-Jaworowska,
Z.
V, ed.
Palaeontol.
Initial research on this topic was completed as part of my Polonica
30:5-22.
dissertationworkintheDepartmentofPaleontology,Univer- 1976.
.Kevolyuciisistematike
i pozdnemezozoyskikh khishchnykh di-
nozavrov. In
Paleontologiya biostratigrafiya
i Monogolii, N.
Kramarenko, ed.
sityofCalifornia,Berkeley.Theseresearcheswereextended Trans.Sovm. Sov.
Mongol. Paleonlol. Eksp.,
Nauka 3:68-75.
during postdoctoral studies at the California Academy of Sci- 1977.K
evolyucii khishchnykh dinozavrov. In
Fauna,Flora
Biostra-
i
encesandtheMuseumofZoology,UniversityofMichigan.The tigrafiyaMezozoya Kainozoya
i Mongolii.
Trofimov,
B. ed.
Trans.
Sovm. Sov.
quality of this manuscript was substantially improved by the Mongol.Paleontol. Eksp.,
Nauka 4:48-56.
critical reviews provided by Joel Cracraft, Philip Currie, Sylvia 1979.
. Opisthopubicpelvisinthecarnivorousdinosaurs.Nature279:
792-793.
Hope, Peter Galton, Larry Martin, Michael Parrish. Timothy 1983.
. Carnivorous dinosaurs from the
late
Cretaceous of
Monogoha (in
Rowe,RobertStorer,KennethWarheit,LarryWitmer,ananon- Russian). Acad. Nauk SSSR,Trudy Paleontol.Inst
19:1-117.
38 ORIGINOFBIRDSANDEVOLUTIONOFFLIGHT

Barsbold,R.anda.Perle.1979.Modifikalsiyalazazaunskhii parallel'noe
i Charig,A.J,andO.A.Reig.1970.TheclassificationoftheProterosuchia,
raxutic khishchnykh dinozavrov. In
Fauna,Flora
Biostratigratiya
] Mivozoya Biol. Linnean
J. Soc.Lond. 2(2):125-I71.
Kainozoya
1 Mongolii. B.
Trolimov, cd.
Trans.Sovm. Sov. Mongol. Palcontol. Chatterjee, 1985.
S. Poslosnchiis. a
new thecodontian reptile from theTnassic
Eksp., Nauka 8:39-44. ofTexas,andtheoriginoftyrannosaurs.Phil.Trans.Roy.Soc.Lond.309(B);
Baur.G.1883.DerTarsusdcrVogclundDinosaurierMorphJahrh.8:417- 395-460.
456. Colbert,E.H. 1964. Relationshipsofsaunschiandinosaurs.Am.Mus.Nat.
1884.
. DinosaurierundVogel.Morph.Jahrb.IU:446-454. Hist.
Novitates 2181:1-24.
1885.
Bcmerkungen ubcrdas
Becken der
Vogel und Dinosauner. Morph. 1981.
. A primitive omithischian dinosaur from theIvayenta Formation
Jahrh.
10:613-616. ofArizona.Mus.N.ArizonaPressBull.53:1-61.
1886.
. Zur Vogel-Dinosauncr-Frage. Zool. An/.8:441-443. Colbert.E.H.andC.C.Mook. 1951. TheancestralcrocodilianProtosuchus.
Beattv.J.ANDW.Fink.1979.(Reviewof)Sober,E.1975.Simplicity.Clar- Bull. Am. Mus. Nat.Hist. 94(3):143-182.
endon Press,
Oxford. 189
pp.
Syst.
Zool.
28(4)643-651. Colbert.E.H.andD.A.Russell. 1969.ThesmallCretaceousdinosaurDro-
Bellairs,A.d'A.andC.R.Jenkin.I960.Theskeletonofbirds.Pp.241-300 macosaunis Am.Mus. Nat.
Hist.Novitates 2380:1-49.
IIIBiologyandComparativePhysiologyofBirds,Vol.I..A.J.Marshall,ed. Coombs.W.P..Jr. 1978a. Thefamiliesoftheomithischiandinosaurorder
Academic Press, NewYork,London. Ankylosauna. Paleontology1):143-170.
2(
Benedetto, 1973.
J.L. Herrerasauridae.nucvafamiliadeSaunsquiosTrias- 1978/).
. Theoretical aspects of
cursorial adaptation in
dinosaurs. Q.Rev.
sicos.
Ameghiniana 10:89-102. Biol.
53:393-418,
Benton, M.
1983.
J. The Tnassic reptile Hypcrodapedon from Elgin: functional Cooper,M.R. 1980. Theprosauropodankleanddinosaurphylogeny.S.Afr.
morphology andrelationships. Phil.
Trans. Roy. Soc. Lond. 302(1112):605- J.Sci
76:176-178.
720. The
198
prosauropod
u1 dinosaur Massospondyltis cannaius Owen from
1985,
. Classification and phylogeny ol'the diapsid reptiles. Zool. .1.Lin- Zimbabwe: its
biology, mode of
life
and phylogenetic significance. Occas. Pap.
neanSoc.84:97-164. Nail. Mus. Rhodesia (B)Nat.
Sci.
6(IO):689-840.
Berman,D.S.andJ.S.McIntosh. 1978.Skullandrelationshipsoftheflpper 1981/).Archosaur ankles: interpretation ofthe
evidence (replyto
Cruick-
Jurassic sauropod Apalosaurus (Reptilia:
Saurischia). Bull.
Carnegie Mus. Nat. shankl.S.Afr.J.Sci.77:308-309.
Hist.,
Pittsburgh 8:1-35. Cope,E.D. 1867. Anaccountoftheextinctreptileswhichapproachedthebirds.
BiDAR, A.,
L.
Demay, andG. Thomel. 1972.
CompiOiiiuilhus coralleslns. nou- Proc. .Acad. Nat.Set.
Philadelphia 1867:234-235.
velleesp^ce de dinosaunen theropode du
Ponlandien deCanjuers. AnnMus. Cracraft, 1971.
J. Thefunctionalmorphologyofthehindlimbofthedomestic
d'Hist. Nat.Nice 1(0:3-34. pigeon, Coliimha livia
Bull.
Am. Mus. Nat.Hist.144:171-268.
Bo.As,J.E.1929.
V. Biologisch-AnatomischeStudieniibcrdenHalsderVogel. 1981.
. TowardaphylogeneticclassificationoftheRecentbirdsofthe
K..
danske Vidensk. Selk.Skr.,
Nalurvid. Math. Afd. 9(11:105-222. world (Class Aves). Auk98:681-714.
1930
Uberdas Verhaltnisder Dinosaurier zu
den Vogeln. Morph, Jahrb. Cruickshank,a.R.1972. I. Theprotcrosuchianthecodonts.Pp.88-179m
64:223-247. StudiesinVertebrateEvolution,K..A.JoyseyandT.S.Kemp,eds.Oliverand
Bock,W. 1963.
J. Thecranialevidenceforratileaffinities.Proc,Int.Ornithol. Boyd,
Edinburgh.
Congr.
13:39-54. 1979
, Theanklejointinsomeearlyarchosaurs.S.Afr.J.Set.75:168-
Bonaparte, 1975u.
F.
J. The family Omithosuchidae (Archosauria: Thecodon- 178.
tia).
Colloq. Intemat. CNRS 218:485-502. Crush,P. 1984.
J. AlateUpperTnassicsphenosuchidcrocodilianfromWales.
1975/1
. Nuevos materiales deLagosuchiis lalampayciuis Romer (The- Palaeontology 4(1):
1
3
1-157.
codontia Pscudosuchia) su
ysignil'icado enelongen delosSaunschia. Cha- Currie,p.J.InpressaCranialanatomyo(Swnonychosaiirmiiiequalis(Saur-
fiarense inferior, Tnasico medio deArgentina. Acta Geol. Lilloana 13(
):5-90.
I ischia.Thcropoda) and its
beanng onthe
originofbirds.Canad. Earth
J. Sci.
1976.
Pisanosaunis mcrtu Casamiquela and theongin oftheOmithis- In
press
. b.
Troodon formosiis (Saunschia, Troodontidae): senior syn-
chia
Paleontol.
J 50:808-820. onym SlcnoiiYchosaurus
of iitniiialis
and
Peciinodon bakkeri.
Vert.
J.Palcontol.
1981. Descripcion de t'asolLiMiLinis iciuix y
su significado en la
siste- Dames, W. 1884. Uber Aniiaeopteryx Palaeontol. Abh.Bd.2(31:119-198.
maticayevoluciondelosThccodontia. Mus.
ArgentinoCien. Natur
Bernardino Darwin,C. 1872. Theonginofspeciesbymeansofnaturalselection,6thed.
Ri\adavia Inst. Nacion. Invest. Cicn.
Nat.Paleontol. 3(2):55-l01. with additions andcorrections. John Munay. London.
Bonaparte,J.F.andJ.E.Powell.1980.Acontinentalassemblageoftctrapods Davis.D.D. 1964. Thegiantpanda,amorphologicalstudyofevolutionary
from theLIpper Cretaceous beds of
Brete.
El northwestern Argentina (Sauropo- incchanisnis. Fieldiana Zool. Mem. 3.
Chicago Nat.Hist.
Mus., Chicago. 339
da—Coelurosauna —Camosauria —Aves). Mem. Soc.Geol.France (Nat, Set.) PP
139:19-28. DEBeer, 1937.
G. The development ofthe vertebrate skull.Oxford, Clarendon
Bonaparte, F.
J.andM. Vince. 1979. El
hallazgodel primer nidodedinosaunos Pressxxiu+552pp.
tnasicos (Saunschia. Prosauropoda), Tnasico
superior Patagonia,
de Argentina. 1954. Archacoplcryx lithograpliiCLi slud>
.\ based upon theBntish Mu-
Ameghiniana I6(I-2):I73-I82. seumspecimen. Br.
Mus.
(Nat. Hist.),
London No.
224:1-68.
Bramwell,C.D,andG.R.Whitheld.1974.BiomechanicsoiPicranodon 1956.
. Theevolutionofratites.Bull.Br.Mus.(Nat.Hist.)4:59-70.
Phil.
Trans. Roy. Soc.Lond.267(B):503-58I- Desmond, A.
1975.
J. The hot-blooded dinosaurs. Blond and Bnggs, London,
Brinkman, D.
1981. Theoriginofthccrocodiloid tarsiand the
interrelationships 238pp.
thecodontian
of reptiles.
Breviora 464:1-23. 1979.
. Designing thedinosaur: Richard Owen's response to
Robert Ed-
Broom, R.
1913. On theSouth African pseudosuchian Eiiparkciia and allied mund Grant.
70:224-234.
Isis
genera. Proc. Zool.Soc.
Lond. 1913:619-633. 1984.
Archetypes and
ancestors: Paleontology Victonan
in London 1850-
Buckland, W. 1824. Notice on theMcgalosaiinis. or great fossillizard of
Stones- 1875 University of
Chicago Press. Chicago. 287pp.
field. Trans. Geol. Soc.
Lond. 2:1-390. Dollo,L.1882.PremierenotesurlesdinosauriensdeBemissart.Bull.Mus.
Camp,C.S. 1930.Astudyofthephytosaurs,Mem.LIniv.Calif10:1-161. Roy. Hist. Nat.Belgique 1:161-180.
1936. Anew typeof
small bipedal dinosaur from the Na\ajo Sandstone 1883.Note surle
presence chez les
oiseaux du"troisieme trochanter"
ofArizona.Univ.CalifPubl.Geol.,Sci.24:39-53. desdinosauriensetsurlafonctiondecelui-ci.Bull.Mus.Roy.Hisl.Nat.Bel-
Chario,.a.J.1976a-Archosauna.Part13:1-6,;//HandbuchofPalaohcrpe- 11:13-20.
gique
tologie, Thecodonlia, O.
Kuhn. ed.
Gustav Fischer Verlag. Stuttgart. Elzanowskj, .a.
1977. Skulls of
Gobipicryx (Aves)from theUpper Cretaceous
1976/'.
. "Dinosaurmonophylyandanewclassofvertebrates":acntical Mongolia
of Results
In ofthe Polish-Mongolian Palaeontological Expedition,
review. Pp. 65-104 in
Morphology and biology ofreptiles, A.
d'A.Bellairs and Vll,Z.
Kielan-Jawarowska, ed.
Palaeontol. Polo.
37:I5.VI65.
C.B.
Cox, eds. Academic Press. London. 1981.Embrxonic bird
skeletons from the
late
Cretaceous ofMongolia.
Charig.A.1979.Anewlookatthedinosaurs.WilliamHeinemannLtdLon- . In
Results of
thePolish-Mongolian Palaeontological Expedition, I.\,
Z.
K.ielan-
don.
pp.
160 Jawarowska, ed.
Palaeontol. Polo.
42:147-179.
Charig..A.J.,J.Attridge.andA.W.Crompton.1965.Ontheonginofthe 1983.
. BirdsinCretaceousecosystems.InSecondSymposiumonMe-
sauropods and theclassification of
theSaunschia Proc, Linnean Soc. Lond. sozoic Terrestnal Ecosystems, Jadwisin. 1981..Acta
Palaeontol. Pol.
28(1-2):
176:197-221. 75-92.
GAUTHIER:
SAURISCHIAN
MONOPHYLY 39

Ewer,R.1965.
F. TheanatomyofthethecodontreptilesEuparkeriacapensis Gingerich, P.
1976.
D. Evolutionary significance of
theMesozoic toothed birds.
Broom. Phil.Trans. Roy.See.Lond. 2488(751 ):379-435. Smithsonian Contrib.Paleobiol. 27:23-33.
Feduccia.a. 1980.TheAgeofBirds.HarvardUniversityPress,Cambndge. Glenn>-,F.H.ANDH.H.Friedmann. 1954. Reductionoftheclaviclesinthe
196
pp. Mesoenatidae, with
some remarks conceming the
relationship of
the
clavicles
Feduccia,A.andH.Tordoff. 1979.FeathersoiArchaeopleryx:asymmetric toflight-functioninbirds.OhioJ.Sci.54:111-113.
vanes indicate aerodynamic function. Science203:1021. Gregory,J.1952. T. ThejawsoftheCretaceoustoothedbirds,Ichlhyornisand
Fraas,1913.
E. Die neusten Dinosaurierfunde derschwabischen Tnas. Natur- Hesperornis. Condor
54:73-88.
wissenschaften, 45:1097-1
I 100. Haeckel, 1866.
E.Generelle Morphologic derOrganismen. Allgemeine Grund-
FuRBRiNGER, 1888.
M.Untcrsuchupgen zur
Morphologie und Systematik der zuge derorganischen Formen Wissenschaft, mechanisch bcgrundet durch
die
Vogel. Holkema, Amsterdam. 1751 pp. von
Charles Darwin reformierte Deszendenz-Theone. Allgemeine
II. Entwick-
Gadow,H. 1893. Vbgel.IISystematischerTheil.Dr.J.G.Bronn'sKJassenund lungsgeschicte der
Organismen. Kritische Grundzuge der
mechanischen Wis-
Ordnungen des
Their-Reichs. Winter'sche
F,
C, Verlagshandlung, Leipzig.340 senschaft von
dan
entstehenden Formender
Organismen, bcgrundet durch
die
pp. Deszendenz-Theone. Georg Reimer, Berlin,clx
462
+ pp.
Galton,P.M. 1970. Omithischiandinosaursandtheoriginofbirds.Evolution Hatcher, 1901.
B.
J. Diplodociis (Marsh): usosteology, taxonomy, and probable
24:448-462. habits, with restoration
a of
the skeleton. Mem. Camegie Mus. 1:1-63.
1971.
. Manus movements of
thecoelurosaurian dinosaur Syntarsus and 1903..Osteology of
Haplocanthosaurus. with description of
new species
opposability of
the
Iheropod hallux (sic).
Amoldia (Rhodesia) 5:1-8. andremarksonprobablehabitsoftheSauropoda,andtheageandonginof
1973a.
. Onthe anatomy andrelationships of
£/raai/a(/;ai?nc«(ica(Huene) theAtlaniosawus beds. Mem. Camegie Mus.2:1-75.
n.
gen., prosauropod
a dinosaur (Reptilia: Saurischia) from the
Upper Tnassic Heaton,M. 1972.
J. ThepalatalstructureofsomeCanadianHadrosaundae
of
Germany. Palaontol. 47:229-255.
Z, (Reptilia: Omithischia). Canad. Earth
J. Sci. 9:185-205.
1973/1.
. On thecheeks ofomithischian dinosaurs. Lethaia 6:67-89.
Heilmann,G. 1926.Theonginofbirds.Witherby,London,v+210pp.
1974.
. The omithischian dinosaur Hypsilophodon from the Wealden of
theIsleofWight.Bull.Br.Mus.Nat.Hist.(Geol.)25(1):1-152. Hennig, 1966.
W.Phylogenetic systematics. LIniversity of
IllinoisPress, Urbana.
1975.
,English hypsilophodonlid dinosaurs (Reptilia:Oraithischia). Pa- 263
pp.
18:741-752.
leontology Hinchliffe,J.R.ANDM.K.Hecht. 1984.Homologyofthebirdwingskeleton:
1976.
. Prosauropod dinosaurs (Reptilia: Saurischia) of
Nonh Amenca. embryological versus paleontological evidence. In
M.K.
Hecht, B.
Wallace,
Yale Peabody Mus. Nat.Hist.
Postilla 169:1-98. and G.Prance, eds.. Evol. Biol. 18:21-39.
1977.
. Slaunkosauruspricei.anearlysaunschiandinosaurfromthe Holmgren.N. 1955.Studiesonthephylogenyofbirds.ActaZool.36:243-328.
Tnassic of
Brazil, withnotes ontheHerrerasauridae andPoposaundae. Pa- Howgate.M.1984. E. TheteethoiArchaeopteryxaniareinterpretationofthe
51(3/4):234-245.
laontol.
Z. Eichstatt specimen. Zool.Linnean
J. Soc.82:159-175.
1978.
. Fabrosaundae, thebasal family ofomithischian dinosaurs (Rep- Huene. R.
von. 1906.
L'ber dieDinosaurien deraussereuropaischen Tnas.Geol.
tilia:
Omithopoda). Palaontol.
52:138-159.
Z. Palaontol. Abh. (N.F.)
9:99-156.
1982.
.The postcranial anatomy of
stegosaurian dinosaur Kivitrosaiinis 1908.
. Die Dinosauner der europaischen Tnasformation mit Bemck-
from the Upper Jurassic of
Tanzania, EastAfnca. Geol.Palacontol. 15:139- sichtigung der
aussereuropaischen Vorkommnisse. Geol.
Palaontol.Abh.
suppl-
160. Bd. 1,
Lfg. 6:345-419.
1983.
. The cranial anatomy o(Dryosaurus. hypsilophodontid
a dinosaur 1914a,
. Beitrage zurGeschichleder Archosauner. Geol.Palaontol. Abh.
fromtheUpperJurassicofNorthAmericaandeastAfnca.withareviewof (N.F.)13(l):l-53.
hypsilophodontids fromthe
Upper Jurassicof
Nonh Amenca. Geol.Palaeontol. 1914^.
. DasnaturlicheSystemderSaurischia.Clb.Mineral.Geol.Pa-
17:207-243. laontol. Abt.
Jg.
(B):
154-158.
1984.Cranialanatomy of
theprosauropod dinosaur Plaleosaurus from Ac-
191Nachtrage zu
meinen fruheren Beschreibungen tnassischer Saur-
the Kjiollenmergel (Middle Keuper. Upper Tnassic) of
Germany. Two
I. com- ischia. Geol.Palaontol. Abh.
(N.
13(3):69-122.
F.)
plete skulls
fromTrossingen/Wiirttemberg. with
comments on
the
diet.
Geol. 1920-
Bcmerkungen
, zur
Systematik undStammesgeschichte einiger Rep-
Palaeontol.
18:139-171. tilien.
Zeit.
Indukt. Abslamm.-Vererbungsl. 22:209-212.
Galton.P.M.andM.A.Cluver.1976.Anchisauruscapensis(Broom)anda 1923.
. Camivoroussaunschians in
Europe since theTnassic. Bull.
Geol.
revision of
the Anchisauridae (Reptilia: Saurischia). Ann. S.
Afr.Mus. 69:121- Soc.Am. 34:449-457.
159. 1932.
. DiefossileRcptilordnungSaunschia,ihreEntwicklungundGe-
Galton, P.
M.andJ. A.Jensen. 1973. Skeleton ofahypsilophodontid dinosaur schichte. Monogr. Geol.Pal
4(11:1-368.
(Nanosaiirus (?)
rex)from the Upper Jurassic of
Utah.Brigham \oung Univ. 1934.
. EinneuerCoelurosaunerinderthunngischenTnas.Palaontol.
Geol.
Studies 20:137-157. 16(3/41:145-170.
Zeit.
1979.
. AnewlargetheropoddinosaurfromtheUpperJurassicofCol- 1948.
. Shortreviewofthelowerletrapods.Roy.Soc.S.Afr.,spec.publ.
orado. Bngham Young Univ.Geol.Studies
26(2):
1-1
2. (Robert Broom commemorative volume):65-l06. Cape
Town.
Gardiner, 1982.
B. Tetrapodclassification.Zool.J.LinneanSoc.74:207-232. 1956..Palaontologie und Phylogenie der
niederen Tetrapoden. Gustav
Gauthier,J.1984.Acladisticanalysisofthehighersystematiccategonesof Fischer Verlag, Jena, Berlin. 716 pp.
the Diapsida. Ph.D. thesis.Department of
Paleontology, University of
Cali- Huxley,T.H. 1864.Lecturesontheelementsofcomparativeanatomy.Chur-
fornia, Berkeley.#85-12825, UniversityMicroftlms,Ann
Arbor,
Michigan. London.chill,
Gauthier, and
J. K.Padian. 1985.Phylogenetic, functional, andaerodynamic 1867.
. Ontheclassificationofbirdsandonthetaxonomicvalueofthe
analysesoftheonginofbirds.Pp.185-197inTheBeginningsofBirds:Pro- modifications of
certain of
the cranial bones observable in
thatclass.Proc.
Zool.
ceedings the
International
of Archaeopteryx Conference, Eichstatt
1984,
K.
M. Soc.
Lond. 1867:415-472.
Hecht,J. H.Ostrom,G. Viohl.andP. Wellnhofer. eds.
Eichstatt,WestGermany: 1868.
. Ontheanimalswhicharemostnearlvintermediatebetweenthe
Freunde desJura-Museums Eichstatt. 382
pp. birds and reptiles. Ann. Mag. Nat.Hist.Lond. 2(4):66-75.
Gauthier,J.,R.Estes,andK.deQueiroz.InPrep.Aphylogeneticanalysis 1870a.
. Furtherevidenceoftheaffinitybetweenthedinosaunanreptiles
of
Lepidosauromorpha. In
Thephylogenetic relationships of
the
lizard
families, andbirds.Q.J.Geol.Soc.Lond.26:12-31.
R.EstesandG.Pregill,eds. 1870ft.
. OntheclassificationoftheDinosaunawithobservationsonthe
Gegenbaur, C.
1878.Gnindnss dervergleichenden Anatomic. Englemann, Dinosauna of
the Tnas. I.
Theclassification and aflinities of
theDinosauna.
Leipzig. 655pp. Q.J.Geol.Soc.Lond.26:32-38.
Gilmore,C.W.1920.OsteologyofthecarnivorousDinosaunaintheUnited Jain,S.L.,T.S.Kutty,T.Roy-Chowdhury,andS.Chatterjee.1975.The
States National Museum, withspecial reference to
the
genera Amrodemus {Al- sauropod dinosaur from theLower JurassicKota Formation of
India.Proc.
/ojauriis) and Ceralosaurus. Bull.U.S.
Nat.Mus.110:1-159. Roy, Soc. Lond. (A)
188:221-228.
1924.
. AnewcoelunddinosaurfromtheBellyRiverCretaceousof 1979.. Some charactenstics of
Barapasaiinis lagorei. sauropod
a dinosaur
Alberta. Bull.Canad. Dep.Mines, Canad. Geol.Surv.
38(G.S.43):1-12. from the Lower Jurassic of
Deccan, India. Prov.IVthInt.Gondwana Symp.,
1933.
. OnthedinosaunanfaunaoftheIrenDabasuFormation.Bull. Calcutta 1977 1:204-216.
Am. Mus. Nat. Hist.
67:23-78. Janensch, 1922.
W. Das Handskelett vonGiganlosaurus rohustus und Brachio-
40 ORIGINOFBIRDSANDEVOLUTIONOFFLIGHT

saunis hrancai aus

denTendaguru-Schichten Deutsch-Ostafrikas. Centr.
Min. McDowell,S.andC.Bogert. 1954. ThesystematicpositionofLamhanotus
Geol. Palaontol. 5:464-480.
I andtheaffinitiesoftheanguinomorphanlizards.Bull.Am.Mus.Nat.Hist.
1925.
.DieCoelurosauner und Theropoden der
Tendaguru-Schichten 105:1-142.
Deutsch-Oslafnkas. Palaeontographica 7:1-99, McGowan, 1984.
C. Evolutionary relationships of
ratitesandcannates from
1929.
. Einaufgestellles und rckonstruicrtes SkclcllvonElaphiosuiirus ontogeny of
the
tarsus. Nature 307:733-735.
hambergi Palaeontographica 7:279-286. MurxjE,B.F. 1879. Arebirdsdenvedfromdinosaurs''KansasCityRev.Sci.
Kjelan-Jawarowska,Z.andR.Barsbold. 1972.NarrativeofthePolish- 3:224-226.
Mongolian Paleontological Expedition 1967-1971. Palaeontol.
Polo.
27:5-13. Norman, D.
1984.
B. A systematic reappraisal ofthereptileorder Omithischia.
Krebs,B. 1963. BauundFunktiondesTarsuseinesPseudosuchiersausder Pp.157-162inThirdSymposiumMesozoicEcosystems,W.-E.ReifandF.
Tnas des Monte SanGiorgio (Kanton Tessin. Schweiz). Palaontol. Z.
37(1/2): Westphal, eds.
Tubingen. 259
pp.
88-95, Olshevskv, 1978.
G. The archosaunan taxa(excluding Crocodylia). Mesozoic
1976.
. Pseudosuchia.Pp.40-98inThccodontia,HandbuchderPalao- Meandenngs 1(1):1-53.G.
andT.
Enterprises, Toronto.
herpetologie. part13.
O.
K.uhn. ed.
Gustav Fischer Verlag.Stuttgart. Olson,S.andA.Feduccia. 1979, Flightcapabilityandthepectoralgirdleof
Lambe, C.
1917.
M. The Cretaceous theropodous dinosaur Cor^oiuuriu Mem. Archacopleryx Nature
278:247-248.
Geol. Surv. Canada 100:1-84. OsBORN,H. 1900.
F. Reconsiderationoftheevidenceforacommondinosaur-
Lansdown,a.B.G. 1968.Theonginandearlydevelopmentoftheclaviclein avian stem in
thePermian. Am. Nat.34:777-799.
the quail Coturnix lapomca.
c J.
Zool.. Lond. 156:307-312. 1903.
. Oniitholeslcs hermanm. new
a compsognathid dinosaur from the
LiNNE, C.
VON.1758.Systema naturae perregna tna
naturae, secundum classes, Upper Jurassic. Bull.Am. Mus. Nat. Hist.19:459-464.
ordines, genera, species, cumcharactcnbus, differentus. synonymis. locis.
Ed. 1905.
rira/im«aHri« andother Cretaceous carnivorous dinosaurs. Bull.
10,Tom. 1-2.Holmiae, impensis direct.Laurentii Salvii.
1758-1759. Am. Mus. Nat.Hist.21(14):259-265.
Lowe,P.R. 1935. OntherelationshipsoftheStrulhionestothedinosaursand 1906.
Tvrannosauriis.
. LIpper Cretaceous carnivorous dinosaur (second
tothe rest of
the avian class,with special reference to
thepositionoi
Archaeop- communication). Bull.
Am. Mus.Nat.Hist,
22(161:281-296,
leryx.Ibis
13:398-432. 1912,Crania ofTyrannostiiinis and Allosaurus. Mem, Am, Mus, Nat.
1944a.
. SomeadditionalremarksonthephylogenyoftheStruthiones. Hist,
(new senes) 1:1-30.
Ibis
86:37-43. 1917.
. SkeletaladaptationsofOrnilholesles.Struthioinimus.andTy-
1944/1.
. An analysis of
the
characters of
.-l/v/Tuco/Jfcnv andArchaeornis rannosaiirus. Bull.Am.Mus. Nat.Hist.35(1531:733-771.
Were they reptiles orbirds? Ibis86:517-543. 1924a
, Psinacosaiirus and Pwtigiianodon: two Lower Cretaceous igua-
Maddison,W.p.,M.J.DoNOGHUE,ANDP.R.Maddison. 1984.Outgroup nodonts from Mongolia. Am. Mus. Nat.Hist.Novitates 127:1-16.
analysis and parsimony. Syst.
Zool.33(1):83-103. 1924/).
. Three new Theropoda, Prorocerarops Zone,
Central Mongolia.
Am- Mus. Nat.Hist. Novitates 144:3-12.
Madsen.J.A. 1976.Allosaurusfrugilis:arevisedosteology.Bull.UtahGeol.
Mm. Surv.109:1-163. OsMOLSKA, H.1976. New light of
theskullanatomy andsystematic position of
Marsh.O.C. 1877. IntroductionandsuccessionofvertebratelifeinAmenca. Oviraplor Nature262:68.3-684
Proc. Am. Assoc. Adv.Sci.
1877:211-258. 1981.
. Coossified tarsometatarsi in
theropod dinosaurs andtheir
beanng
onthe problem ofbird ongins. In
Results ofthePolish-Mongolian Palaeon-
1880.
. Odonlomithes: A
monograph on
theextincttoothed birdsofNorth tological Expedition, Part 1\,
Kielan-Jaworowska,
Z ed.Palaeontol. Pol.
42:
America. Rept. Geol. Expl.40th Parallel. 201pp. 79-95.
1881a,
. AneworderofextinctJurassicreptiles(Coeluria).Am,J,Sci, 1982.
. Hulsanpesperlein.g.,n.sp.(Deinonychosauria,Saunschia,Di-
21:339-340- nosauria) from theUpper Cretaceous Barun Goyot Formation of
Mongolia. N.
18816.
. Principal characters ofAmerican Jurassic dinosaurs Part V.
.Am. Jahrb.Geol.Palaeontol, Monatsh, 7:440-448,
Sci.
J. 21:417-423. OsMOLSKA,H,andE.RoNiEwicz.1970,Deinocheindae,anewfamilyofthe-
18840-
. PnncipalcharactersofAmencanJurassicdinosaurs.PartVIII ropod dinosaurs. Results
In the of
Polish-Mongolian Palaeontological Expe-
The order Theropoda. Am.Sci.
J. 27(38):329-341. dition, Part
Kielan-Jaworowska,
Z,
II, ed,
Palaeontol, Pol.
21:5-19.
1884/)
, Theclassificationandaffinitiesofdinosaurianreptiles.Nature Osmolska,H.,E.Roniewicz,andR.Barsbold.1972.AnewdinosaurGal-
31(784):68-69. limimiis hullalns n.
gen.,n.
sp.(Omithomimidae) fromthe
Uppermost Creta-
1890.
. Descnptionofnewdinosaurianreptiles.Am.J.Sci.39:81-86. ceous Mongolia.
of Results
In the ofPolish-Mongolian Palaeontological Ex-
1896.
. DinosaursofNorthAmerica.Ann.Rept,U.S.Geol.Surv.pt.I. pedition, Part
Kielan-Jaworoska,
IV,
Z. ed.
Palaeontol. Pol.
27:96-143.
16:133-244.
OsTROM.J.H.1969a.AnewtheropoddinosaurfromtheLowerCretaceousof
Martin,L. 1980. Thefoot-propelleddivmgbirdsoftheMesozoic.ActaXVII Montana. YalePeabody Mus. Natur. Hist. Postilla128:1-17.
Congressus Int.
Omithol2:1237-1242. 1969/1.
OsUo\og\ of
Dcinonvi-hus iinlirrhopiis. an
unusualtheropod from
1983a.
. Theoriginofbirdsandofavianflight/hCurrentOrnithology, theLowerCretaceousofMontana,Bull,YalePeabodyMus,Nat,Hist,30:1-
R.
Johnston, ed.
1:105-129. 165-
1983/1.The ongin and early radiation of
birds(withcommentanes by 1970,
, Stratigraphy and paleontology of
the CloverlyFormation (Lower
D.Steadman and P.
Rich).Pp.291-338 mPerspectives mOrnithology, Essays Cretaceous) oftheBighorn Basin area, Wyoming andMontana, Bull,Yale
Presented for
theCentennial of
theAmencan Ornithologists' Union, A.
Brush Peabody Mus, Nat,Hist,35:1-234,
and G.
Clark, eds.Cambridge University Press,Cambridge. 1972,Descnption of
theArchacoptcryx specimen in
the
Teyler
Museum,
Martin, 1984.
L. AnewhesperomithidandtherelationshipsoftheMesozoic Haarlem. Proc, Sect. Sci.
ned. Akad. Wet. (B)
75:289-305.
birds. Trans. Kansas Acad.Sci.
87(3-41:141-150. 1973-
. Theancestryofbirds-Nature242:136.
Martin,L.andO.Bonner. 1977.AnimmaturespecimenofBapiornisadvenus 1974a
. .■lrc/;ae()/)/m'.vandtheonginofflight.Q.Rev.Biol.49:27-47.
from theCretaceous of
Kansas. Auk 94(41:787-789. 1974/)
Thepectoral girdle andforelimb function of
Deinonychus (Rep-
Martin,L.andJ.Stewart. 1977. Teethinhinhyornts(Class:Aves).Science tilia,
Saurischia): A
correction. YalePeabody Mus. Nat.Hist.
Postilla 165:1-
195:1331-1332. II
1982.
. An ichthyomithiform birdfrom the
Campanian of
Canada, Canad, 1.975a Theonginofbirds.InAnn,Rev,EarthPlanet,Sci,,F,A,Donath,
Earth
J. Sci. 19:324-327. 3:55-57,
ed,
Martin,L.,J.Stewart,andK.Whetstone. 1980.Theonginofbirds:Structure 1975/1
, OntheoriginofArchaeopwryxandtheancestryofbirds,Proc.
ofthe tarsus and teeth. Auk97:86-93. C.N.R.S.,CoUoq.Int.Prob.Act.Paleontol.-Evol.Vert.218:519-532.
Marthm,L.andJ.Tate. 1976. TheskeletonofBapuwnisadveniis(Aves,Hes- 1976a.
. Archaeoplcrvxandtheoriginofbirds.Biol.J.LinneanSoc.
perornithiformes). In
Smithsonian Contnb Paleobiology, S.
L.
Olson, ed.
27: Lond,
8:91-182,
35-66. 19766 Onnew
a specimen of
theLower Cretaceous theropod dinosaur
Matthew,W.D.andB.Brown. 1922. ThefamilyDeinodontidae,withnotice Deinonychus antirrhopus YalePeabody Mus. Nat,Hist.
Breviora439:1-21.
ofanewgenusfromtheCretaceousofAlberta.Bull.Am.Mus.Nat.Hist.46: 1978.
Theosteology of
Compsognalhus longipes Wagener. Zitteliana 4:
367-385. 7.3-118.
McDowell, 1978.
S.
B. Homology mapping of
theprimitive archosaunan reptile 1981.
P/()a)»!/).s<).?)ia(/!(«—
. theropod thecodonf
or Palaeontographica
palate onthe palate ofbirds. Evol. Theory 4:81-94. (A)175:179-195.
GAUTHIER:
SAURISCHIAN
MONOPHYLY 41

OsTROM,J.H.ANDJ.S,McIntosh. 1966.Marsh'sdinosaurs,thecollections theropod Stenonvchosaurus inequalis

hypothetical
and
a dinosauroid,
Syllogeus
from Como Bluff.Yale University Press. NewHaven. London. 388 pp. 37,Natl.Mus.Nat.Sci.,Canada,43pp,
Owen,R. 1841. ReportonBritishfossilreptiles.Part11.Rep.Br.Assoc.Adv. SantaLuca,A, 1980.
P, ThepostcranialskeletonofHelerodonlosaurustuoki
Sci.
11th Meeting, Plymouth 1841:60-294. (Reptilia, Omithischia) from theStormberg of
South Afnca, Ann,S,
Afr,
Mus.
1875.
. MonographofthefossilreptilesoftheLiassicformations.II 79(7): 159-21 I,
Pterosauria. Palaeontol. Soc.
Monogr. 1875:41-81. Sawin, 1947,
H,J, The pseudosuchian reptile Typothorax meadei. new species,
Padian, 1982.
K. Macroevolutionandtheoriginofmajoradaptations:verte- Paleontol,
J, 21(3):201-238,
brateflight asa
paradigm fortheanalysis of
patterns. Proc.Third
N. Am. Schmidt,O, 1874, ThedoctrineofdescentandDarwinism.Appleton,New
Paleontol. Conv.
2:387-392. York.
1983.
.Osteology and functional morphology of
Diiiiorphodon macronyx Seelev,H.G. 1881.ProfCarlVogtontheArchaeoplervx.Geol.Mag.8(2):
(Buckland) (Pterosauna: Rhamphorynchoidea) basedon
new
material the
in 300-309,
Yale Peabody Museum. Yale Peabody Mus.Nat. Hist.Postilla
189:1-44. 1887.
, OntheclassificationofthefossilanimalscommonlynamedDi-
1984.
. Theonginofpterosaurs.Pp.163-168inThirdSymposiumon nosauna. Proc.Roy.Soc.
Lond.43(206):165-171.
Mesozoic Terrestrial Ecosystems, W.-E.Reifand Westphal,
F. eds.
Tubingen. 1888,
. TheclassificationoftheDinosauria.Rep.Br.Assoc.Adv.Sci.,
259
pp. 57th
Meeting, Manchester, 1887:698-699.
In
press.
. On the type matenal ofCoelophysis (Saurischia: Theropoda), Sereno,P.C. 1984. ThephylogenyofOmithischia,areappraisal,Pp,219-226
and new
a specimen from the Petrified Forestof
Arizona (LateTriassic,
Chinle in
Third Symposium onMesozoic Terrestrial Ecosystems, W,-E,Reifand F,
Formation). InTheBeginning of
theAgeof
Dinosaurs: Faunal Change Across Westphal, eds.Tubingen. 259
pp.
theTriassic-Jurassic Boundary, K.
Padian. ed.
Cambndge UniversityPress, 1986.
. Phylogeny of
the Bird-hipped Dinosaurs (OrderOmithischia).
NewYork. Nat.Geol.Research. 2(2):234-256.
Parker,T.J.1882.OntheskeletonofNotornisinanlelli.Trans.Proc.New Sill,W.D.
1974, TheanatomyofSaurosiicinisgalileiandtherelationshipsof
Zealand Inst.
14:245-258. therauisuchid thecodonts. Bull,
Mus, Comp. Zool.
Harvard 146:317-362.
Parker.W.K. 1864. RemarksontheskeletonofArchaeopleryxandonthe Simpson,G,G,1946,Fossilpenguins.Bull.Am.Mus.Nat.Hist.87:1-95.
relationsofthebirdtothereptile.Geol.Mag,(I)1:55-57. Steel. 1969.
R. Omithischia.InHandbuchderPalaoherpetologie,part15,O,
1887.
. Onthemorphologyofbirds.Proc.Roy.Soc.Lond.42:52-58. Kuhn, ed,Gustav Fischer Verlag, Stuttgart, 84pp.
Parks, W.1928.
A. Struthwmimus samueli. a
newspecies of
Omithomimidae 1970.
. Saunschia./n Handbuch derPalaoherpetologie. part13.0,
Kuhn,
from the Belly River Formation of
Alberta. Univ. Toronto Studies. Geol.Ser. ed.Gustav Fischer Verlag. Stuttgart. 88
pp.
26. Sternberg,C.M. 1932. TwonewtheropoddinosaursfromtheBellyRiver
1933,
. NewspeciesofdinosaursandturtlesfromtheLIpperCretaceous Formation of
Alberta. Can,Field-Nal, 46:99-105.
formations of
Alberta. Univ.Toronto Studies. Geol. Ser.
34. 1933.
. A new Onulhoniinuis with complete abdominal cuirass. Canad.
Parrish,J.M. 1984. LocomotorgradesinThecodontia.Pp.169-173(nThird Field-Nat.47:79-83,
Symposium onMesozoic Terrestrial Ecosystems, W,-E,
ReifandWestphal,
F. 1934.
, Notes on certain recently described dinosaurs. Canad. Field-Nat.
eds,Tubingen, 259pp, 48:7-8.
Paul,G.1984a.
S. Thesegnosauriandinosaurs:relicsoftheprosauropod- Sternberg,R.M. 1940. AtoothlessbirdfromtheCretaceousofAlberta.J.
omithischian transition?Vert.
J.Paleontol, 4(41:507-515, Paleontol.
14(l):81-85.
1984/)
, Thearchosaurs:aphylogeneticstudy.Pp.175-180inThird Stolpe, M.von.
1932.
Physiologische-anatomische Untersuchungen iiber
die
Symposium on
Mesozoic Terrestrial Ecosystems. W.-E.
ReifandWestphal.
F. hintere Extremitat derVogel.
Omilhologie
J. LXXX(2):163-247.
eds.Tubingen. 259pp. Stovall.W.J.andW.Langston. 1950. Acrocanlhosaunisatokensis.anew
Pycraft.W.p.1900.PartII.OnthemorphologyandphylogenyofthePa- genus and species of
Lower Cretaceous Theropoda fromOklahoma. Am.Mid.
laeognathae (Ratitae and
Cryptun) and
Neognathae (Carinatae).Trans.
Zool. 43(31:696-728.
Nat.
Soc.
Lond. 15:149-290. Sues.H.D. 1977.Theskullof\'elnciraplormongoliensis.asmallCretaceous
Raath,M. 1969. AnewcoelurosauriandinosaurfromtheForestSandstoneof theropod dinosaur fromMongolia. Palaeontol. Zeit.
51:173-184.
Rhodesia. Amoldia (Rhodesia) 4:1-25. 1978.
. AnewsmalltheropoddinosaurfromtheJudithRiverFormation
1972.
. FossilvertebratestudiesinRhodesia:anewdinosaur(Reptilia: (Campanian) of
Alberta, Canada. Zool.Linnean
J. Soc.Lond.62:381-400.
Saurischia) fromnear the
Trias-Jurassic boundary, Amoldia (Rhodesia) 30:1- 1980.
. Anatomy and relationships of
a
new hypsilophodontid dinosaur
37, from the Lower Cretaceous of
North America. Palaeontographica (A)I69:5I-
72.
Reig,O,
1963, Lapresenciadedinosauriossaurisquiosenlos"Estratosde
Ischigualasto" (Mesotriasico superior)de las
provencias de
SanJuan
La
yRioja Tarsitano,S.andM.Hecht. 1980. Areconsiderationofthereptilianrela-
(Republico Argentina). Ameghiniana 3(l):3-20. tionships Archaeopleryx.
of Zool.
Linnean
J. Soc.
Lond.
69(2):
149-182.
Thulborn,R.1972. a. Thepost-cranialskeletonoftheTriassicomithischian
RiEPPEL,O. 1980,Homology,adeductiveconcept^Zeit.Zool.Syst.Evol,18: dinoiauT Fabrosaiints auslralis.
Palaeontology 15:29-60.
315-319,
1975.
. Dinosaur polyphyly and the classification ofarchosaurs and birds.
RoMER, 1923,
A,
S, Crocodilian pelvic musculature and theiravianandreptilian Aust. Zool.
J. 23:249-270.
homologues. Bull,
Am. Mus. Nat.Hist.
48:533-552. 1977.
. Relationships of
theLower Jurassic dinosaur Scelidosaurus har-
1927.
. Thedevelopmentofthethighmusculatureofthechick.J.Mor- nsoniiPaleontol.
J. 5l(4):725-739.
phol.
Physiol. 43:347-385. 1982,
. Significance ofanklestructure in
archosaur phylogeny. Nature
1956.
. Osteologyofthereptiles.Univ.ChicagoPress.Chicago,xxi+ 299:657,
772
pp. 1984,
The avian relationships of
.-Irc/iaeo/'ffrv.v. andtheongin of
birds,
1966.
. Vertebrate paleontology. 3rded..
Univ. Chicago Press. Chicago, Zool,Linnean
J. Soc.
82:119-158.
368
pp, Thulborn,R.A.andT.L.Hamlev. 1982. Thereptilianrelationshipsof.-Ir-
Rozhdestvensky,a,K, 1958, AufDinosaunerjagdinderGobi.Leipzig.240 chaeopieryx. Aust.
Zool.
J. 30:611-634.
PP VanTyne,J,andA.J.Berger. 1959.Fundamentalsofornithology.Wiley.
1965.
. GrowthchangesinAsiandinosaursandsomeproblemsoftheir New York. 624 pp.
taxonomy. Int.
Geol.Rev.
8(1966):792-793. Vogt,1879,
C, ArchaeopWrvx. emZwischengleid zwischen denVogel und Rep-
Russell,A,P,andD.J.Joffe.1985,Theearlydevelopmentofthequail tihen.
Naturforscher. Beriin
1879:401-404.
(Coturmx japomca)
c. furcula reconsidered. Zool.
J. Lond. (A)206:69-81. 1880.
. Archacoptcryx macroiira. anintermediate form
between birdsand
Russell,D.A. 1969. AnewspecimenofSienonychosaurusfromtheOldman reptiles.Ibis
4(4):434-456.
Formation (Cretaceous) of
Alberta. Canad.Earth
J, Sci. 6:595-612. Walker,AD. 1961. TriassicreptilesfromtheElginarea:Slagonolepis.Das-
1970.
. Tyrannosaurs from thelateCretaceous ofwestern Canada. Natl. ygnathus and theirallies.
Phil.
Trans. Roy, Soc, Lond, (B)244(709):103-204,
Mus.Canada Publ,Palaeontol, 1:1-36, 1964.
. Triassic reptiles from the Elginarea: Orniihosucfnis andtheongin
1972,
, OstrichdinosaursfromthelateCretaceousofwesternCanada, ofcamosaurs. Phil.
Trans.Roy,
Soc,Lond,(B)248(744):53-134.
Canad, Earth
J, Sci, 9(4):375-402, 1972,
. Newlightontheoriginofbirdsandcrocodiles.Nature237:257-
Russell,D,andR,Seouin. 1982, ReconstructionofthesmallCretaceous 263.
42 ORIGINOFBIRDSANDEVOLUTIONOFFLIGHT

1974.Evolution, organic McGraw-Hill Yearbook Sci.

Tech.1974:177- say.bydefining Archosauna andits
major subgroups onthe basis oftheancestry
179. of
extant archosaurs. an extinctand diagnosible archosaur can have only
oneof
1977.
. Evolutionofthepelvisinbirdsanddinosaurs.Pp..M9-357in twopossiblerelationships:itcouldbemorecloselyrelatedtobirds,oritcould
Problems in
Vertebrate Evolution, S.
Andrews, R.
Miles,and A.
Walker,eds. bemore closely related tocrocodiles. Pseudosuchia (n.comb.) is
redefined to
Academic Press,London. include extant crocodiles andall
extinct archosaurs thatarecloser tocrocodiles
Welles,S.P. 1983.Twocentersofossificationinatheropodastragalus.J. than they are
tobirds. Likewise. Omithosuchia (n.
comb.) is
redefined toinclude
Paleontol.
57(2):401. extantbirdsandallextinctarchosaursthatareclosertobirdsthantheyareto
1984.
Dilophosaunis
. weiherdb (Dinosauria, Theropoda): osteology and crocodiles.
comparisons. Palaeontographica 185:85-180. realize
I thatitisnotenoughsimplytolistsupposed"synapomorphies"in
Welles,S.P.andR.Long.1974.Thetarsusoftheropoddinosaurs.Ann.S. support of
particular
a hypothesis. Tomake the hypothesis most vulnerable to
Afr.
Mus. 64:191-218. test,
oneshould proceed through the
entire system of
Hennigian argumentation
Wellnhofer, 1974.
P. Dasfunfte Skelettexemplar vonArchaeoplcryx. Palae- for
each character and array this
evidence in
taxon/character
a matnx. Further-
(A)147:lb9-:i6.
ontographica more.have
I had to consider many specimens oftaxa thathave
I not had
the
1975.
Die
. Rhamphorhynchoidea (Pterosauna) derOberjura-Platten- opportunity to
study first-hand, and
have
I relied instead on
published illustrations
kalke
Suddeutschlands. Allegemeine
I: Skelettmorphologie. Palaeontographica anddescriptions.Finally,have I nothadthetimetoconsiderseveralrecent
148(A):
1-33. publications onarchosaur phylogeny (e.g.. Paul1984j, b)in
thedetail that
they
1978.
. Pterosauna.InHandbuchderPalaoherpelologie,19,O.Kuhn, deserve. Accordingly, the
following can
only be
considered preliminary
a analysis
ed-Gustav Fischer Verlag, Stuttgart 82
pp. (orasR.Bakker[pers.comm.]putit,anexercisein"armchaircladistics").To
1985.
. Neue pterosaurier aus derSantana-Formation (APT)derChapada balance these caveats. should
I addthatthehypothesis presented below is
broadly
Aranpe.
do Brasilien. Palaeontographica (A)187: 105-182. concordant with those amved at
byother workers in
this
area (e.g.. Benton 1985;
Whetstone,K. 1983.
Braincaseofmesozoicbirds:I.Newpreparationofthe Pamsh 1984 and pers. comm).
"London" Archaeopteryx. Vert,
J. Paleontol.2(4):439-452. Asdefined here. Pseudosuchia and its
major subgroups can bediagnosed as
Whetstone,K,andL.Martin.1979.Newlookattheonginofbirdsand follows (cases of
homoplasy will
only belisted if
they occur between, rather than
crocodiles. Nature279:234-236. within,
basic
taxa).
198
. Common
1- ancestryforbirdsandcrocodiles'^Areply.Nature289:
98. Pseudosuchia
WiLLisTON,S.W. 1879.
Arebirdsderivedfromdinosaurs'^KansasCityRev. comb.
(n. =Parasuchia. Aetosauna. Rauisuchia.
and
Crocodylomorpha— includ-
3:457-460.
Sci. crocodiles)
ing
Wilson.M.C.andP.J.Currie, Inpress. Swnonychosaurustnfqualu(Saur-
ischia, Theropoda). from the
Judith River
(Oldman) Formation of
Alberta:
new Crocodile-normal
1) crurotarsalanklejoint,
in
whichthe
peg
is
ontheastragalus
andthe
socket is
onthe
calcaneum.
findings on metatarsal structure. Canad. Earth
J. Sci.
WiMAN. 1929.
C. Die Kreide-Dinosauner aus Shantung. Palaeontol. Sinica(C) Calcanear
2) tubercle enlarged(also
Omithosuchidae).
in
6(l):l-67. 3)
Cervical nbsshort andstout.
Young,C. 1C.
94,A
1complete
. osteologyofLulengosaunishuenei\'oung(gen. 4)
Discrete postpanetal confinedto
early
juvenile
or
prehatching ontogenetic
et
sp.nov). Palaeontol. Sinica(C) 7:1-53. stages(also
in
omithosuchians asidefromEuparkeria'—xt
discussion be-
1947. OnLufengosaiirusmagnm(sp.nov.)andadditionalfindsof low).
Lufengosaiinis hueneiYoung Palaeontol. Sinica
(C)l2:l-53. 5)
Palatalteeth
absent (also
in
omithosuchians aside
fromEuparkeria').
1958.
. The dinosaurian remains of
Laiyang, Shantung. Palaeontol. Sinica
(C)16:53-138. Unnamed TaxonIncluding
Aetosauria,
Rauisuchia,
and
Crocodylomorpha
1964,
. ThePseudosuchiainChina,Palaeontol.Sinica(0151:1-205. Septomaxilla
6) absent.
Zusi,R.L.ANDR.WStorer.1969.Osteologyandmyologyoftheheadand 7)
Noseparatepostpanetal at
anytime
in
posthatching ontogeny.
neck ofthe Pied-BiUed Grebes {Poddymhiis). Misc.Publ.Mus. Zool..
Univ. 8)
Fusionof
second intercentrum andfirst
centrumin
juvenile
or
earlier
stages
Michigan
139:1-49. ontogeny.
of
9)
Tnradiate pelvis
(also
in
omithosuchians; see
Pamsh 1984).
10)
"Screw-joint" tibio-astragalar articulation
(R.
Long.
pers.
comm.; also
in
Omithosuchidae. M.
Pamsh.pers.
comm).
AppendixA 11)
Fully
developed crocodile-normal crurotarsal
joint
(Krebs
1963;
Bnnkman
1981).
Archosaur Phylogeny: On the Relationships of 12)
Osteoderms onventralsurfaceof
tail.
Certain Extinct Archosaurs to
Extant Crocodiles and Birds Unnamed Taxon
Including Rauisuchia
and
Crocodylomorpha
13)
First
(atlantal) inlcrcentrum much longerthan
wide.
The
hypotheses of
relationship discussed belowdenve mainly fromGaulhier 14)
Axial
diapophysis reduced or
absent (or
correspondingprocess of
axial
rib
(1984 and references therein). They provide the
context in
which thepresent reduced: Clark,
J. pers. comm).
analysis of
saunschian phylogen> wasundertaken As
used here.Sauna, n.
comb., 15)
Enlarged,pneumatic, basipterygoid processes
Zawiskie,
(J. pers.
comm.).
is
restncted to
the
leastinclusive laxon encompassing extant diapsids. and the 16)Length of
pubis exceeds three timeswidth of
acetabulum(also
in
omitho-
nameDiapsidaisappliedtothetaxonincludingextantSaunaanditsextinct suchians
Euparkena*).
from
aside
sister-group. Araeoscelidia. Themore inclusive
groups of
Recent Amniota rec- 17)
Fewer thanfour phalanges in
pedal digit
five.
ognizedin this
work follow theusagein
Gauthier (1984) andGauthier et
al.
(in
prep.):Mammalia the
issister-group of
Reptilia
within
Amniota: Reptiliaincludes Crocodylomorpha
thesister-taxa Chelonia and Sauna: andSaunaincludes the
sister-taxa Lepido-
sauna
and
Archosauna. 18)
Quadratojugal extends to
dorsal
surface
of
skull.
Archosauna (n.
comb.) is
redefined to
encompass all
thedescendants of
the 19)
Quadrateand
quadratojugal inclined
anterodorsally.
most recent common ancestor ofcrocodiles andbirds.Sodefined, some taxa 20)
Quadratecontacts
prootic.
previously included within Archosauna are
now excluded from it.
Thus, Protero- 2Postfrontal
1) absent(alsoin
omithosuchians aside
from
Euparkeria'andOr-
champsidae, Erythrosuchidac, and
Proterosuchidae are
here
considered succes- nithosuchidae).
sivelymoreremotely related outgroups Archosauna
of (s.s.).
ratherthan"pro- 22)Maxillanes
contactto
form secondary palate
antenorly.
terosuchian-gradc archosaurs." Defining
Archosauna this
inwayprovides more
a 23)
Internal
jugular
vein
absent.
fruitfulperspective fromwhich to
examine archosaur phylogeny. Rather than 24)
Squamosalwithoutventralramus.
proceeding fromposition
a of
less
evidence—by trying
to
determine which group 25)
Post-temporal fenestravery'
small
(analogousmodification
achieved
by
sep-
of
"thecodonts" gave rise
tobirdsorcrocodiles—we may now regard more pre- arate
means
within
Omithosuchia).
ciselystated
hypotheses from the
position
affordedby
more secureknowledge 26)
Fenestra"pseudorotunda" present
(also
appeanngat
someunknown level
(i.e.,noonehaseverconfusedalivingcrocodilewithalivingbird).Thatisto withm
omithischiansand
theropods).
 GAUTHIER:
SAURISCHIAN
MONOPHYL1 43

27)
Entire
dellopectoralcrestdistallyplaced. tion,character
will
be
followed
by
L?
for
Lagosuchus*
and
Pt?
for
pterosaurs,
28)
Clavicles
absent. including
Scleromochlus*).
29)
Coracoid
ventromedially elongate (also
pterosaurs).
in
30)
Radialeand
ulnareelongate andcolumnar. 7)
Discrete postpanetal absent in
post-hatching ontogeny(L?;
also
in
pseudosu-
chians).
31)Pedal
digit
five
with
fewerthan four
phalanges (alsoin
omithosuchians). 8)
Palatalteethabsent(L?;also
in
pseudosuchians).
M.
Pamsh(pers.
comm.) noted some interesting characters indicating
closer
a 9)
Coracoid tubercle lies
closeto
glenoidfossaand
coracoidforamen.
relationshipbetweenaetosaurs and
rauisuchians than
suggested
is by
the
characters 10) First
metacarpal withoffsetdistal
condyles,and
pollex
directedmedially and
listed
above. Crocodylomorphs may turnout
to
bethesister-groupof
rauisu-
a beanngenlarged
ungual (L?;Pl?).
chian-aelosaur group,
or
they maybe
the
sister-group of
rauisuchians as
suggested 11)
Manus more asymmetncal thanin
pseudosuchians, withinner digits
much
above. Nevertheless, it
must beemphasized thattheanalysesof
Parrishand larger
than
outerdigits
(L?;Pt?).
Gauthier agreeon
twopoints:
PseuJosuchia
)I is
monophyletic, and
2)
Parasuchia Supra-acetabular
12) buttress.
the
sister-group
is the
of
crocodile-aetosaur-rauisuchian group. 13) Prominently tnradiate pelvis,
with
pubislength
at
leastthreetimes widthof
acetabulum (also
crocodylomorph-rauisuchian
in group,
and
lesser
a
to extent
Ornithosuchia aetosaurs).
(=Euparkena*l, Omithosuchidae, l.agosuchus' Pterosauria—
. including Scle- 14)Antenor trochanter on
femur appearsearly
in
posthatchingontogeny.
wmochltts" Herrerasaundae*,
. Omithischia, Sauropodomorpha, and
Theropo- Aliform
15) fourthtrochanter (Pt?;
characters13-15
correlatedwith
erectposture;
da— includingbirds) see
Pamsh 1984).
16)
Fifth
metatarsal gracile.
One might question whether or not Euparkeria* is
inside oroutside of
Archo-
sauria (s.s.).
Evidence presented below indicates that Euparkeria' closer
is to birds
than to
crocodiles. ThisearlyTnassic archosaur is,
however, exceedingly pnmi- Ornilhodira (n.
txn.)
tive,
and some evidence supports an
alternative hypothesis which
in Pseudosuchia (Gr.
ornilhos. bird; deire.
neck)
and Omithosuchia (excluding Euparkeria') are
most closely related. For
example, (=Eagosuchus'. Pterosauna,
Herrerasaundae*, Omithischia,Sauropodomorpha,
unlike Euparkeria' all
other
. archosaurs share theapomorphic absence of
palatal Theropoda)
and
teeth, absence ofdiscrete intercentra throughout thetrunk andinmost of
the The
possession of
the
followingsynapomorphies makes clear
it that
omithodiran
cervical region, andabsence of
discrete
a poslpanetal andexoccipitals beyond monophyly is
one of
themosthighly
corroborated hypotheses in
archosaur phy-
juvenile stages development.
of Thus, four potential synapomorphies could unite logeny. Indeed, the
immediate common ancestor of
Oraithodira was
fundamen-
all
archosaurs with respect to
Euparkeria' However,
. discrete trunk intercentra tally
morebirdlikethan
wereomithosuchids, Euparkeria', pseudosuchians.
or
are alsoabsent in
thetwo successively more remote outgroups of
Archosauna, Unfortunately, theprecise
level
at
which thefollowing characters arose
is
prob-
Proterochampsidaeand Erythrosuchidae; the
presence discrete
of trunk intercentra lematic some
in instances,
becauseLagosuchus' incompletely
is known
and
ptero-
inEuparkeria* must then beinterpreted aseither reversal,
a or
as
further evidence saursarc sospecialized.character
a
If listedas
diagnostic of
Omithodira is
only
ofthe
immaturity of
the
known specimens. Six
potential synapomorphies unite questionably present due
incomplete
to preservation profound
or modification,
Euparkeria' with the birdlike archosaurs. have
I been unable to
discover any thisfactwillbedenotedbyeither(L?)forLagosuchus'or(Pt?)forpterosaurs,
evidence thatEuparkeria' closer
is to crocodiles. following thecharacter inquestion.It
is
notyet
clear\{Lagosuchus' or
Pterosauria
Looking again at
the evidence, it
is
interesting tonote that intercentra, post- IStheimmediatesister-groupofdinosaurs,althoughthefemuris,atleastin
panetals, and
exoccipitals are
present juvenile
in archosaurs ancestrally (e.g..
Camp Dimorphodon. moredinosaurlike than
that
is of
Lagosuchus* (Gauthier1984).
1930;deBeer1937).ThatEuparkeria'couldberepresentedbyjuvenileorat
least incompletely grown specimens also
is indicated byunfinished articular sur- 17)
Postfrontal absent (L?;
alsocrocodylomorphs).
in
faceson longbones, an
unossified distaltarsal 11,
an
unfused scapulocoracoid, 18)
Atlantalintercentrum enlarged, completely surrounding odontoid ventrally
neural arches thatareunfused withtheir respective centra, and separate sacral andlaterally andfitting into
prominent recessed area below odontoid onaxis.
nbs. Thus, rather than character discordance ansmg from convergence, theap- 19)Axialintercentrum, and then odontoid, fuses to
axisat
cessation of
growth.
parent discordance may
anse
instead fromcompanng nonequivalent ontogenetic 20)Modification of
cervical centra and zygapophyses thatcombine to
yield an
stages. one
If disregards the
apparently age-related characters, andaccepts the S-shaped neck(compared to
dinosaurs, mdimentary in
both Lagosuchus'
evidence indicating that
Euparkeria' an
isomithosuchian archosaur, thenone andin
Pterosauria ancestrally; although within pterosaurs theneckmay be
need onlyinvoke a
single adhoc hypothesis of
convergence, namely, theinde- in
some ways more birdlike than is
theneckoiArchaeopleryx').
pendentloss of
palatalteeth in
Pseudosuchia on
theonehand, andin
Omitho- 21
Zygapophysial
) facets nearly vertically disposed in
all
butproximal part oftail
suchiaaside from Euparkeria' on
theother, to
account for
the
available evidence. (L?).
This problem should begiven further consideration, but forthepresent thefol- 22)
Interclavicle absent (L?).
lowing characters are
accepted diagnostic
as Omithosuchia.
of Fortunately, Eu- 23)
Clavicle reduced and gracile (L?;
enlarged in
coelurosaurs).
parkeria' plesiomorphic
sois that
ultimate
its placement makes little
difference 24)
Glenoid facet
on
scapulocoracoid facesposteroventrally (Pt?).
in
determining character polanty in
the following analysis. 25)Coracoid small, with subcircular profile,and lying in
nearly thesame plane
Squamosal
1) reduced and
descending ramus gracile(reversed large-headed,
in as
the
scapula (Pt?).
26)Forelimbslessthan55%ofhindlimblength(Pt?),andhindlimbverylong
carnivorous omithosuchians such
tyrannosaurs).
as relativeto
length of
trunk.
2)
Centra steeply inclined in
at
least the first
four postatlantal cervicals.
3)
Modifications in
the hindhmb and girdle correlated with semierect gait(also 27)Apex of
dellopectoral crestplaced distallyon
humenis (Pt?).
pseudosuchians
in asidefromParasuchia; seePamsh 1984). 28)Less than fivephalanges in
manal digit
four and lessthan three phalanges in
4)
Ventral flange of
astragalus absent. manal digit
five
(L'^).
29)At
leastthree vertebrae involved in
sacmm (L?;alsoin
Omithosuchidae?).
5)
Crocodile-reversed ankle joint, with peg on
calcaneum and socket on
astrag- 30)Brcvis shelfappears onventral surface of
postacetabular portion of
ilium
alus(including loss
of
perforating foramen; secCooper 1980,1981/»; Bnnk- (Pf).
man 1981,Thulbom 1982).
31)
Birdlikedistalendof
femur—prominent antenor and posterior intercondylar
6)
Pedal digitfivewithfewer than fourphalanges (also in
rauisuchian-crocody- grooves, withthelatterconstncted by
prominent external tibial
condyle, and
lomorph
group). appearance of
discrete
a fibular groove andcondyle—modifications in
the
Unnamed Taxon
IncludingOmithosuchidae, Lagosuchus', Pterosauria, knee jointplayed key roles in
enabling a
narrow-tracked, bipedal gaitand
Herrerasauridae*, Ornithischia,
Sauropodomorpha, Theropoda
and erect
stance(Stolpe 1932).
32)Tibia as
long orlonger than femur (reversed in
all
dinosaurs overa
fewmeters
Thistaxonmight more profitably
beconsidered Omithosuchia, in
that
the in
length, orlarger in
the case of
theropods).
distnbution of
characters amongarchosaurs leaves
little
doubt as
to
the
monophyly 33)
Fibula thin
and strongly tapered distallyandcalcaneum reduced.
of
this
taxon.Untiltheposition of
Euparkeria* is
better
understood, however, Astragalus
34) transversely widened.
this
taxon will
remain unnamed. Should Euparkeria' beshown to
beoutsideof 35)
Astragalus andcalcaneum withsmooth, rollerlikearticular surfaces abutting
Archosauna (s.s),
thenby
definition
the
name Omithosuchia wouldbe
restncted against
depressed distal
tarsals.
to
thistaxon.
This group ofarchosaurs is
diagnosed by thefollowing synapo- 36)
Metatarsals elongate and
closelyappressed.
morphies indeterminable
(if owing
nonpreservation
to profound
or transforma- 37)
Pes
digitigrade.
 44 ORIGINOFBIRDSANDEVOLUTIONOFFLIGHT

38)
Pesfunctionally tndactyl
(Pf). 1981). Accepting this
placement requires that
development of
more deeply inset
39)Pedaldigitfivereduced,doesnotexceedlengthofmetatarsalIV(Pt?),and tooth rows tookplacetwicewithinOmithischia: oncethe
inscelidosaur-stegosaur-
composed of
nomore
than
twophalanges. ankylosaur group, andoncein
theremaining omithischians. From evidence pre-
40)
Parasagittal rowsosteoderms
of absent. sented by
Thulbom (1977), the
so-called "juvenile scelidosaur" appears bndge
to
thegapbetween Sculcllosaunis' and
the
type-specimen of
Scchdosaurus (Gallon,
Dinosauria pers.comm.. considers the
scelidosaur specimens conspecific). Scchdosaurus.
comb.
=Herrerasauridae*.
(n. Omithischia.Sauropodomorpha.and Theropoda— stegosaurs, andankylosaurs appear to
bemore closely related in
that theyshare
Including
birds) anenlargedpalpebralcovenngtheorbitdorsally;ahmdlimbtotmnkratioof
0.85orless,atibiathatislessthan80%offemurlength;ametatarsalIIIthatis
Beforeproceedingwithdiagnosis
a of
Dinosauna. will
it be
necessaryto
bnefly lessthan35%offemurlength(ratiosfromThulbom1977);arelativelyshort,
consider
Herrerasaundae*. and
the
diagnoses
of,
and
phylogenetic relationships broad, and stoutly constructed manus and pes;andstout
a postcramal skeleton.
within,
Omithischia
and
Sauropodomorpha. Within thisassemblage, ankylosaurs and stegosaurs appear to
bemost closely
Herrerasaundae* represented
is by
three
very
incompletely knowntaxa,
Slaii- related, although acceptance of
this
conclusion must await full
preparation and
nkosaunis*.Ischtsaurus',
and
Herrcrasaunis*
that
.are
uncertain
of relationships descnption Scchdosaurus.
of Stegosaurs and
ankylosaurs share twoadditional
to
oneanother.Theyneverthelessappearto
bethesister-group(s)ofall
other osteoderms above theorbit;
the absence of
an
antorbital fossa; reduced
a or
absent
dinosaurs
(Gauthier
1984). upper temporal fenestra; aninclined quadrate, short
a neck and long torso; thick-
walled limb bones; a
more uniform width of
thescapular blade and anenlarged
Diagnosis Omithischia
of acromial region of
the
scapulocoracoid; absence supra-acetabular
of
a buttress;
Thediagnosis andcontents of
Omithischia areunproblematic, which might be femora with reduced fourth and anterior trochanters; a
tibiathat is
lessthan 70-
expected since themonophyly of
thistaxon has been accepted fornearly century
a 75%offemurlength;ametatarsusthatisonly25%oftibialength(ratiosfrom
(Seeley 1887). Thesame cannot be said for phylogenetic relationships within Thulbom 1977), andvery short, broad, and stout manus and pes bcanng hooflike
Omithischia. Llsing Saunschia sister-group,
as
a with herrcrasaurs* and Ptero- unguals. Most of
thesecharacters areassociated withgraviportal habits, and they
saundi-Liigosuchus* successively
as more remote outgroups, the
following char- appeared independently withinceratopsians andomithopods (s.s.).
Onecould
acters are
considered diagnostic Omithischia.
of The
evidence discussed below is thus argue thatthesecharacters could be
collapsed into
single
a character related
derived from Gauthier (1984) but modified in
light of
the studies of
Sereno (1984. tolarge sizeand quadrupedal habits. Nevertheless, whether few or
many, these
1986) and Norman (1984), The only examples of
relatively generalized omith- data areaccepted as
indicating recency of
common ancestry until there is
evidence
ischians that were available forstudy were specimens of
Sculcllosaunis' and a the
to contrary.
cast ofHelcrodontosaunis, and most ofthecharacters listed below are derived Diagnosesoftaxa3and4arelefttoPaulSereno(1986).AtthistimeIonly
from theliterature (see review in
Gauthier 1984). wish lo
point outthatboth might bemost closely related withm Omithischia, in
Predentary bone; mandibular condyle set
slightly to
deeply below tooth rows; thattheyshare asymmetrically enameled toothcrowns, more deeply insetmarginal
cheekteethwithdistinctcrownandroot,crownsofcheekteethlow,bulbous toothrows,atleastfivetosixsacrals,andtheabsenceofasupra-acetabular
basally, sublriangular profile,
in and bearenlarged accessory denticles onmargins; buttress. Moreover, these taxaare theonly omithischians in
which fully
a open
quadratojugal dorsal process reduced, not
in
contact with squamosal; ossified acetabulum couldbeconsidered theancestral condition. That is
to
say,although
palpebral cartilage; antorbital fenestra reduced and much smaller than orbit; max- the acetabulum completely
is perforate in
some stegosaurs, only
is
it semiperforate
illary processpremaxilla
of enlarged andlengthened posteriorly separate
to maxilla ankylosaurs,
in Scchdosaurus. .Sculcllosaurus*. and
Lesothosaurus- Thus,
semi-
a
from nasal for
half or
more of
their former sutural contact, quadrate elongate and pertbrate acetabulum with
supra-acetabular
a buttress the
isancestral condition
massive; dentary participates coronoid
in eminence; gastralia absent; ossified ten- for
Omithischia, andbecause this
condition also
is ancestral for
Sauropodomorpha
donspresent atleast above sacral region; fourto
fiveor
more sacral vertebrae and Herrerasaundae*, semiperforate
a acetabulum with
prominent
a supra-ace-
and sexual dimorphism insacral number (see Gallon 1974, 1982); pubis com- tabular buttressalso
appears ancestral
be
to for
Dinosauna.
pletely reorganized, with
shortanterior process and
long,rodlikeposterior ramus The analysis of
omithischian phylogeny is
notintended tobedefinitive. Never-
paralleling ischium; ilium with enormously elongate iliacspine; symphysis re- theless,following
the conclusions appear
secure.
stncted todistal end of
ischium; winghkc anterior trochanter; pendant fourth Omithischia
1) monophylelic.
is
trochanter; fifth
pedal digitreduced to
metatarsal spur. 2)
Lesothosaurus diagnosluus appears to
bethesister-taxon of
all
other omith-
For thesake ofdiscussion, recognize
I thefollowing basic taxa within Omith- ischians.
ischia: 3)
Although relationships among themajor groups of
omithischians remain un-
Lcsolhosiiiirus.
1) clear.
nonetheless
IS
It evident that
anycharacter shared by
Hypsdophodon-
Thyreophora,
2) taxon
a including Sciiwllosiiiinis*. Scclidnsniinis. stegosaurs, and Hclerodonlosaurus, pachycephalosaurs-psittacosaurs-A/i[Tocera(ops Scu-and
ankylosaurs. lellosaurus' and
Scchdosaurus that
also
ispresent Lesothosaurus.
m represents
3)Following Sereno (1984, 1986), taxon
a including llclcnHiiniltKdiinis and Or- the
ancestral condition for
Omithischia.
nilhopoda scnsu Santa Luca (1980).
4)Following Sereno (1984), and Osmolska andMaryanska (pers. comm.), taxon
a Diagnosis
Sauropodomorpha
of
including pachycephalosaurs and
ceratopsians (including psittacosaurs). Sauropodomorph monophyly more
is problematic than
thatof
Omithischia
Taxon Lcsothosaunis.
I, is
very incompletely known. Its
current placement in (e.g.,
Chang al.
el
1965). Nevertheless, usingherrcrasaurs*, and
Pterosauna-Z-a-
"Fabrosaundae" uninformative,
is as
conceived in
Gallon (1978) "fabrosaurs" gosuchus* as
successively moreremote outgroups. the
following characters are
are paraphyletic in
that some taxa included within thisgroup appear lobe more considered diagnostic Sauropodomorpha.
of includingsuch
taxa
Thecodonto-
as
closely related to
the
entity composed of
omithischian taxa 2-4
hereinafter
( termed saurus'.
Efraasia*.
.-tnchisaurus'.Ammosaurus. Plateosaurus.Lufengosaurus.
higher omithischians). than they are to
one another. Lfsiiih<isaiiriis is
saidto segnosaurs. Atassospondylus. Riojasaurus.I'ulcanodon.Barapasaurus. and
Sau-
possess anobturator process ontheischium (Thulbom 1972), which is
otherwise ropoda. Thecharacters discussed below aredenved pnmanly from the
literature
known only in
Omithopoda (Santa Luca 1980).However, other evidence indicates reviewed in
Gauthier 1984).
( and
weresupplemented by
examination of
Plateo-
that Lesothosciunis the
issister-group of
higher omithischians (Gauthier 1984; saurusspecimens
Tubingen.
in
Sereno1984,1986).Thatistosay,groups2-4shareamandibularfenestrathat EIraasia'
andThccodomosaurus* and
.lesser
to
a extent .4nchisaurus'. appear
IS
verysmall orabsent, spout-shaped
a mandibular symphysis, slightly todeeply to
be
the
mostpnmitive sauropodomorphs. Thesetaxa
correspond to
the
"narrow-
inset marginal tooth rows (i.e.,structures analogous to
"cheeks" are thought to footed prosauropods" of
Gallon
andCluver (1976).
Of
course,the
"broad-footed
have been present; Gallon 1973/)), a
distal condyle ofthe quadrate that is
wider prosauropods" of
Gallon andCluver1
(976)
are morecloselyrelated to
sauropods.
than themandibular condyle, anenlarged facial process ofthemaxilla that further thusdemonstrating the
paraphyly "Prosauropoda."
of Efraasia'
and Thccodon-
reduces thesizeof
theantorbital fossa and fenestra, robust
a jugal, fused panetals, losaurus' are
too
incompletely preserved to
be
very
informative; nonetheless they
and an
anlenor pubic process that is
at
least moderately developed. Thus, in
light sharecertain
apomorphies diagnosticSauropodomorpha.
of first
among them
ofcurrent knowledge it
appears that anobturator process arose twice within beingtherobustpollexanditsenlargedclawnotedabove(alsothehallux).In
Omithischia. addition,the
most recentcommon ancestor of
Efraasia'.
Thecodontosaurus',
Scutellnsuurm' is
very poorly known, but it
shares with theother members of .■inchisaurus*. and
the
morecompletelyknown sauropodomorphs. possessed lan-
taxon 2
raised osteoderms onthe dorsum, an ilium that exceeds the length of
the ceolate teeth
with
coarsely
serratedcrowns,
comparatively
a small
skull
suspended
femur, and longer forelimbs and trunk compared to
hindlimb length (Colbert, onalongneckcomposedofatleast10cervicals.eachofwhichisatleast25%
 GAUTHIER:
SAURISCHIAN
MONOPHVLV 45

longerthanaremostofthetrunkvertebrae,andahindhmbthatissubequalto caudals: greatly enlarged acromial region of

scapulocoracoid. resulting in
pro-
orshorterthanthetrunkandinwhichthetibiaisinvariablyshorterthanthe nounced demarcation between scapular blade
and
base
scapula;
of metacarpals
femur (reversed from ancestral condition Omithodira).
in stout and arranged in
hemisphencal
a colonnade below theforearm (Note: The
The
remaining sauropodomorphs, including Anchisaurus*. are
further special- fourth and fifthmetacarpals and digitsare seton the palmar surface ofthe hand
izedcompared to
Thecodomosaunis* in
thatthey possess an
even more robust inSaunschia ancestrally, thusinitiating thecupped metapodial unit that is
taken
first
metacarpal and digit, wider-based neural spines ontheantenor caudals, an toextreme insauropods; compare Massospondyliis in
Cooper 198 la.
fig.
37, with
arched dorsal margin of
theilium, and completely
a open acetabulum. Brachiosaurus in
Janensch 1922); manus elephantine, onlydigit
one retains the
Compared nchisaurus'.
-1
to andespecially Thecodonlosaurus'
to theremaining
. ancestral phalangeal formula andlarge,
a clawlike ungual, withthe
others being
sauropodomorphs arelarger animals (thistaxon will
hereinafter bereferred to
as reduced single
to phalanges supporting hooflike unguals (phalangeal formula 2-1-
broad-footed sauropodomorphs). The
broad-footed sauropodomorphs also
share 1-1-1, asopposed to2-3-4-3-2. the ancestral condition for Dinosauria); pubis
thefollowing apomorphies: themandibular condyle is
set below thetooth row ventrally directed and more massively developed than ischium; puboischiadic
(also in
EIraasia'. which according toGalton [pers. comm.) actually represents a junction robust; forelimbs at
least three-fourths of
hindlimb length (may beless
juvenile example of
Selhsaurus gracilis); theintcmasal process of
thepremaxilla in
diplodocids); limbs massive, vertically disposed, long bones nearly solid (sau-
is
compressed; the nares are very large owing totheprevious character and emar- ropodsappear to
beone of
the few dinosaur groups with truly
a vertical, para-
gination of
thenasals; the teeth in
theupper tooth row increase in
height antenorly sagittal, limb posture); tibiawithnanow descending flange, pes
stout, veryshort
and areespecially long in
the premaxilla (segnosaurs?); proximal caudal centra and broad, phalanges of
outer digits short, with digitthree losing three phalanges
relatively compressed anteroposteriorly andwith
broad-based neural spines; ro- (phalangeal formula 2-3-4-2-1 as
opposed to
2-3-4-5-1); gastralia absent.
bustforelimbs withrelatively
a shorter, broader, and stout manus, and to
lesser
a Sauropoda may be
divided intotwo principal groups, here refened to
informally
extent pes (i.e.,
broad-footed), withgreatly
a enlarged pollex; proximal carpals fail as
camarasaurs andtitanosaurs. Camarasaurs sharestrongly
a arched intemanal
toossify; acetabular fenestra is
much larger than the size of
the femoral head, and barofthepremaxilla,asnoutthatissharplydemarcatedfromtherestofthe
was
it presumably filledwith cartilage throughout life;
and theinitialdevelopment skull, a
relatively elongate ischium thatextends well postenor to the level ofthe
ofdescending
a flange onthe postero-distal end ofthetibia (Cooper 1981a). ilium while twisting tobecome more honzontally disposed distally.and relatively
a
Although segnosaurs {Erhkosaurus. Segnosaurus) have been
considered related very
deep
puboischiadic contact.
to"theropods" (e.g., Barsbold 1983), orto
represent relics of
transition
a between Euhclopus (Wiman 1929) may bethesister-taxon of
thetitanosaurs; both taxa
Omithischia and"prosauropods" (Paul 1
984a.b).
segnosaur relationships appear share a
quadrate that slants upand back from themandibular condyle, neural
tobe
withthis subgroup of
sauropodomorphs. Segnosaurs are remarkably spe- spines thatare slightly to
deeply bifurcate (convergent in
Camarasaunis). and the
cialized, with
retroverted pubes, edentulous
an and
apparently beaked premaxilla incorporation of
three or
more trunk vertebrae intothecervical senes.
likeomithischians, and very large ascending processes likecoelurosaur and Forthe sake ofdiscussion, titanosaurs willbedivided into two informal taxa:
tyrannosaur theropods. In
thecontext of
all
theevidence, however, these characters the
antarctosaurs for
suchtaxa
as
Antarclosaurus. Alamosaunis. Laplalasaurus.
arehere considered to
have been acquired separately, and thus theyare diagnostic and Tilanosaurm. thatshare caudal
a senes inwhich thefirst centrum is
biconvex
segnosaurs
of among sauropodomorphs. and subsequent caudals arc
procoelous; anddiplodocids for
Apalosaurus. Baro-
Within the
broad-footed sauropodomorphs, Riojasaurus, I'lilcanodon. Bara- saunis.Cetiosaiiriscus. Dicraeosaurus. Diplodocus, Mamcnchisaunis. and
Nemeg-
pasaurus. and Sauropoda appear tobe most closely related. For example, the tosaurus. thatshare diagnoslically modified haemal arches with
fore-and-aft pro-
descriptions of
\'ulcanodon (Raath 1972) and Barapasaurus (Jainet
al.1975, jections, midsacral neural spinesthat
are
deeply cleft,
andischia
that
are
expanded
1979) reveal that these taxa share with Sauropoda the following characters that distally.
are
absentSauropodomorpha
in ancestrally: spatulate teeth; strongly opisthocoe- Thephylogenelic analysis of
Berman and
Mcintosh (1978) established titano-
louscervical and antenor trunk vertebrae; cer\'ical centra at
least twice thelength saurmonophyly in
that antarctosaurs and
diplodocids possess the
following syn-
of
midtrunk centra; deep, oval,"pleurocoelous excavations" below transverse apomorphies: snout long,
broad,and
depressed; body
premaxilla
of and
supra-
processes; at
least one more vertebrae added to
sacrum (minimum of
four sacrals); antorbital process of
maxilla extend dorsally to
levelof
orbit,intemasal processes
elongate neural spines confluent with one another in
the sacrum; forelimb at
least of
premaxilla and nasals reduced orabsent, and nasals very short,thus external
two-thirds of
hindlimb length; iliumshort anteropostenorly andstrongly arched nares areconfluent onmidline and situated high onskull; lacnmal very reduced;
in
profile; brevis shelf reduced or
absent; texture of
articular surfaces of
long bones dorsal process of
quadratojugal very reduced and anterior process broadly in
indicates retention of
thick pads ofcartilage; distal tarsals fail
toossify; antenor contact with maxilla atanteroventral margin of
orbit; very long basipterygoid
trochanter reduced rugosity
to
a or
absent; founh trochanter reduced and displaced processes basisphenoid;
of elongate, pencil-like teethconfined antenormost
to part
distally tonear middle offemur; femur robust, ihick-walled, wider laterally than of
jaws; neural spines in
postenor cervical and antenor trunk vertebrae deeply
anteropostenorly, and with straight shaft; tibia to
femur ratio less than 0.63; tibia cleft
and V-shaped in
antenor view(convergent in
Camarasaunis); sacral neural
compressed side to
sideand cncmial crest reduced; enlarged descending flange on spines very tall;very long tail
forming distalwhip-lash consisting of
a
long senes
postenorfaceofdistalendoflibia;metatarsalIIIlessthan37%oftibialength; cylindncal
of caudals; relatively short forelimbs thatare
approximately two-thirds
metatarsals subequal inlength and arranged in
shallow
a arch, with broad,
a very of
hindlimb length; and metatarsal IV
exceeds the length ofmetatarsal III.
short and stoutly constructed pes with a
hooflike ungual on pedal digitthree; The
preceding discussion sauropodomorph
of phylogeny establishes the
follow-
fourth and fifth
pedal digits without clawlike unguals. points.
ing
As
thefollowing list
of
synapomorphies attests, sauropod monophyly highly
a
is Broad-footed
1) sauropodomorphs monophyletic.
are
corroborated
hypothesis. 2)
Evidence currently available indicates that Anchisaurus' could be thesister-
Relatively short postorbital region ofskull that
strongly
is inclined posteroven- group ofthe broad-footed group, and thatthe group composed ofboth these
trally. thus making the skull appear flexed about the braincase inlateral view; taxawould form anunresolved Inchotomy with EIraasia* and Thecodonlo-
frontals and panetals wider than long; very long nasal process of
prcmaxillae saurus*. Thus,
Sauropodomorpha appears
monophyletic.
beto
(absent in
many diplodocids); nasals deeply excavated to
form postenor margins Sauropoda
3) monophyletic.
is
of
retracted external nares; postorbital reduced to
thin,
a elongate bar extending 4)
Such asthey are known, Riojasaurus, and especially I'ulcanodon and Bara-
nearly toventral surface of
orbit; postenor and postorbital processes of
jugal pasaurus. appearmore closely
related
Sauropoda
to than
are
other
broad-footed
reduced; dorsal process of
quadratojugal reduced in
thatfails
it to
reach squamosal, sauropodomorphs.
butantenor process enlarged and nearly reaching maxilla; lacnmal thinand elon- 5)
Thus, any character thatis
present in
broad-footed sauropodomorphs, thatis
gate posterodorsally. forming most anterodorsal
of margin orbit;
of nanow process also
present .-Inchisaurus*.
in EIraasia*. and
Thecodontosaurus* considered
is .
of
maxilla extends over antorbital fenestra nearly to
anterodorsal margin of
orbit; have
to been present Sauropodomorpha
in ancestrally.
lower temporal fenestra inclined anteroventrally to
extend well below orbit;palate
abbreviate, with broad choana; pterygoid flanges blunt and wellantenor to
brain-
case; epipterygoid absent; tooth rows entirely antorbital and antenor teeth pro- Diagnosis
Dinosauria
of
cumbent; mandibular fenestra absent; least
attwo vertebrae added cervical
to
senes (minimum of12cervicals); cervical nbs fused tocentra in
fully adult spec- Giventhat
Omithischia,
1() Sauropodomorpha, and
Theropodaare
eachmono-
imens; vertebrae withhighlycancellous cavemously
to pleurocoelous centra sur- phyletic; and
(2)
accepting
that
any
character
shared
Lesolhosaurus
by and
higher
roundedby dense lamellar bone; tall
neural spines in
postenor pan of
trunk, over ornithischians,
Efraasia*
-Thecodonlosaurus*
and
broad-fooled
sauropodo-
sacrum, andin
antenor caudal vertebrae; at
least one vertebra incorporated into morphs, Procompsognaihus'-Liliensiernus'
and ceratosaurs
and represents
the
sacrum (=five or
sixsacrals); relatively broad sacrum with enlarged sacral trans- ancestral
conditionsfor
Omithischia,
Sauropodomorpha, and
Theropodarespec-
verseprocesses that arelevelwith lopof
deeply arched tlium; massive proximal tively;
accepting
and
(3) that
herrerasaurs*
and
Scleromochlus*-Lagosuchus* rep-
46 ORIGINOFBIRDSANDEVOLUTIONOFFLIGHT

resent successively more remote outgroups, Dinosauna canbe diagnosed by

the note that the foot is
gracile, elongate, and functionally tndactyl in
Lagosuchus*.
following
synapomorphies. herrerasaurs*, theropods, andomithischians
in ancestrally. Although the
fourth
toe(and finger) and fifthtoe areunusually long in
pterosaurs, the proportions and
41)
Vomers elongate, reachmg postenorly at
least tolevelof
antorbital fenestra lengthsofthemiddlethreetoesarelikethoseofbirdsandunlikethoseofeither
aetosaurs'!').
(also
in
pseudosuchians or
omilhosuchians ancestrally (Padian 1983,Gauthier 1984),
42)Scapulaatleastthreetimeslongerthanwidthatbase,andentirescapulo- Consequently, the
simplest explanation for
the
shorter, stouter, andtelradactylous
coracoid further inclined posterodorsally (also pterosaurs
in asidefromScle- foot
Sauropodomorpha
of that
isthese conditions represent apomorphic reversals.
roniochllis*).
In
fact,sauropodomorphs areunique among archosaurs in
the
sizeand shape
43)
Increased asymmetry of
hand, with small outer two digits having fewer pha- ofthefirstfingerandtoe.Thehandinparticularisstnkinglymodifiedevenin
langes (ancestral phalangeal formula for:
Archosauria =2-3-4-5-3; Omiiho- earlysauropodomorphs. Gallon (1971) and Baird (1980) discussed the
relations
dira^2-3-4-4-*'; and Dinosauna =2-3-4-3-2). between thearticular surfaces of
the metacarpal and firstphalanx in
manal digit
44)
Scmiperforate acetabulum and
prominent supra-acetabular buttress (T.
Rowc, (the
I pollex) in
early sauropodomorphs (="prosauropods"). They noted that
pers. comm., notes that theinner wall of
theacetabulum is
thin in
semi-erect adjacent articulations within thepollex were modified so
that upon maximum
archosaurs. Thus, it
is
not surpnsing to
seethe independent appearance of
this character infully erect omithosuchids, and in
some fully erectcroco- extension theenormous ungual phalanx was directed medially andthuswas nearly
honzontally onented. Upon flexion, however, theclaw was brought into thesame
dylomorphs and rauisuchians. Theacetabulum becomes fully
perforate at lineofactiontakenbytheotherdigits.
least three times within Dinosauna, once in
omithischians, once in
sauro-
Baird (1980) descnbed fossil
a trackway of
an
Ammosiiunis-\\)f.e sauropodo-
podomorphs. and once in
theropods. The degree to
which the acetabulum is morphwalking quadmpedally. most
In
a elegant test
functional
of
a hypothesis
open alsovanes with absolute size; with theexception ofankyiosaurs. the based onanextinct organism, Baird used themorphology of
the trackway to
show
larger the
dinosaur the
more open the
acetabulum). that the manus functioned as
predicted. Indeed, themanus was held in
an essen-
45)
Birdlike femur and antitrochanter: medial rotation about long axis
of
element tiallydigitigrade pose, with thedigits fully
extended so
thatthe enlarged clawwas
ofthatportionoffemurproximaltofourthtrochanter =(intumedheadof laidflatupon its
side against thesubstrate. Perhaps theunusual modifications of
femur); proportional elongation of
femoral shaftdistalto
antenor trochanter; thepollex reflect divergent functional demands placed on thehand in
the larger
dorsal arc
of
entirefemoral shaft;fore-and-aft compression of
femoral head sauropodomorphs (five
meters and
above). Stoutforelimbsand powerful, grasping
inproximalview;andfemoralheadmoredistinctlysetotffromshaftof hands may have supported the role of
the head and neck ascropping organs while
femur.Also,modificationsofthedistalendofthefemurnotedaboveare
in
bipedal
a (ortnpodal) feeding pose (Bakker 1971by
),providing secure purchase
more prominent in
dinosaurs. These modifications are
also accompanied by onthetrunksoftrees.Thisandotherpossiblerolesofthegraspinghandand
rconentation of
theantitrochanter, which faces mostly dorsally archosaurs
in enlargedfirstclawwouldhavetobebalancedagainsttheroleofthehandasa
ancestrally, but faces mainly ventrally in
dinosaurs (T.
Rowe, pers.
comm.). weight-beanng structure dunng quadrupedal locomotion. Perhaps thesefactors
The proximal end of
thefemur in
pterosaurs likeDi
morphodon (Padian 1183) combined to
yield the
unusually modified pollex of
sauropodomorphs.
ISin
some ways more birdlike than is
that of
Lagcsuchus* (Gauthier 1^84). Withregardtotheproblemofthehistoryofthetndactylfoot,itismost
46)
Antenor trochanter enlarged, interesting to
note that in
sauropodomorphs the modifications of
thefirstfinger
47)
Dinosaur tibia:
cnemial crestprominent and with weakly crescentic profile (pollex) apply as well,albeit less dramatically, tothefirst toe (hallux). The mod-
indorsalview(sizeandshapeofcnemialcrestvaneswithsi/cofanimaland ificationsshared by
thepollex andhallux cannot beaccounted for
by
invoking
styleof
locomotion). Distal end of
tibiabroadened mediolaterally, thusele- functional similanty. Rather, asDavis (1964) argued in
the case of
the panda's
mentappears twisted nearly 90°with respect to
proximal end.Andwith thumb, simultaneous modification of
the hallux and pollex suggests thatwas
it
prominent fossa on anterolateral face of
distal end oftibia forreception of simpler toalter the developmental program of
both limbs than to
change thatof
ascending
process, the hand alone. Ofcourse, whatever accessory roles the hand may have played
48)Birdlike ankle: proximal tarsals fit
caplike onto tibia and fibula; crurotarsal early in
sauropodomorph history, these functions were overwhelmed by
the
de-
joint
andcalcanear tubercle absent; ascending process of
astragalus present mands weight-hcanng
of Sauropoda.
in Thus,the
telradactylous impressions of
(i.e.,
intermedium moves dorsally)—thus, motion within the ankle confined hind feet attnbuted to
some sauropodomorphs donotrepresent anancestral
mainly tosimple,
a hingelike jointbetween the rollerlike proximal and the condition; onthecontrary themodified hallux as
is
diagnostic of
sauropodomorphs
compressed, distal tarsals. (Like thefemur, the ankle joint in
pterosaurs ap- as
IS
theuniquely modified pollex.
pearsto
be
more dinosaurlike than that
of
Lagosiichiis': these apomorphic In
conclusion, theancestral dinosaur was essentially birdlike. It
had mobile,
a
similanties suggest
monophyletic
a pterosaur-dinosaur group [Omithotarsi, S-shaped neck, short
a trunk, and long hind limbs with modifiedjoints that enabled
n.txn.],as
argued inGauthier 1984). narrow-tracked
a gait. Its
short forelimbs, together with other modifications in
49)PedaldigitfiveshorterthanmetatarsalI;thefootistndactylinthetypical thehands and girdles, indicate that theforelimbs were suited lessto
performing
dinosaunan
condition.
roles in
support and locomotion than they were tograsping and manipulating
One of
thekeyelements in
thedinosaur controversy is
the problem presented food. Asargued by Bakker and Gallon (1974), vanety
a ofstmctural modifications
by
Sauropodomorpha. Chang et
al.
(1965) and Chang (1976a. h)argued that
the indicate thatthe pnmary responsibility for
support and locomotion had shifted
erect,
quadrupedal pose of
Sauropoda wasan
ancestral condition for
dinosaurs, to
thehmdlimhs and girdles in
theancestral dinosaur. That theancestral dinosaur
and
thai
theropods and
omithischians acquired their
bipedalhabitsindependently wasabipedalcursorisespeciallyevidentinthemorphologyofitstail,inthe
(Chang accepts that
the
quadrupedal pose of
some omithischians secondary).
is elongate and nanow distal portions of
its
hindlimbs, and in
itsfunctionally tn-
AsCooper(198It!)pointedout,sometaxaareeasilyrelegatedtoparticularlo- dactyland digitigradc pes. By
thestandard of
typical Mesozoic dinosaurs, the
comotor categones. example,
For sauropods probably
are obligate quadrupeds, ancestral dinosaur wasquite small. Indeed, using other omilhodirans as
outgroups,
andtheropods obligate bipeds, butanumber of
taxa arenot so
readily placedin onemightinferthatitwascertainlyundertwometersandperhapsunderone
either
category (e.g.,
the
sauropodomorph MiissosponJvlus). Whether we
view
this meter in
total length. Although small, the ancestral dinosaur would have had a
problem from the
perspective of
observed morphology or
from thatof
inferred broadvisualhorizonbecauseofthebirdlikeposeofitsneckandhindlimbs.
functions, weare
still
left
with
Chang's problem; are
sauropodomorphs ancestral Considenng these modifications whole
as
a suggests that the
ancestral dinosaur
or
apomorphic in
these regards'* That is
to
say,theancestral dinosaur couldhave was small, cursonal, erect and normally bipedal. Its
divergently specialized fore-
beenacursonalbiped(orconsidenngtheroleofthetail,atnped),andsauro- and hindlimbs indicate thatthe ancestral dinosaur was capable of
more than the
podomorphs could
subsequently have
reacquired
quadrupedal habits.
Alterna- mere facultative bipedality inhented
it from itssaunan ancestors. On the other
tively,the
common ancestor of
dinosaurs couldhavebeenpachypodal
a quad- hand,trackwaysindicatethatforelimbscouldstillplayaroleinsupportand
ruped, with
omithischians and
theropods subsequently becoming bipedalcursors locomotionatlowspeed,sotheancestraldinosaurwasnotsolockedintothis
independently of
oneanother as
Chang suggested. Eitherhypothesis requiresthe mode of
progression asareits
obligately bipedal living relatives.
same number of
evolutionary events, andone
if is
unwilling to
make assumptions The main points of
theforegoing discussion of
archosaur phylogeny may be
about theprobability of
reversals versus convergences, these data alonedo
not summanzed as
follows (and see Fig.7).
affordbasis
a for choice. This question can, however, beevaluated by
reference
to
additional outgroups (Maddison et
al.
1984). Given thattaxa such as
Slaiin- Archosauna
1) composed
is the
ofsister-groupsPseudosuchia and
Omithosuchia.
kosaunts*. Scleromochlus* and
Lagosuchus'
. represent outgroups, appears
it likely 2)
Parasuchiais
thesister-group of
all
other
pseudosuchians, andEiiparkena'
that
the
ancestral dinosaur was
small,
a erect-postured, cursorial biped. appears
to
be
thesister-group of
all
otheromithosuchians.
The same kind of
analysis canbeused toaddress thequestion ofthe ancestral 3)
WithinOmithosuchia, Omithosuchidae the
issister-group
of
Omithodira
(n.
dinosaur morphology L'singthedinosaur footas
anexample, it
is
interesting to txn).
GAUTHIER:
SAURISCHIAN
MONOPHYLY 47

4)
Omithodira (n.
txn.)
is
erected
to
encompass an
unresolvedtrichotomy
com-
posed
Lagosuchus*,
of Pterosauna,
and
Dinosauna.
5)
Dinosaunamonophylelic
a
is taxon
composed of
themetataxonHerrerasau-
ndae*.
and
the
monophyletic
sister-groups
Omithischia
and
Saunschia.

AppendixB too
poorly preserved to
be
certain of
state assignment, or
that
theregion in
question
was
too
transformed allow
to unambiguous state
assignment (i.e.,
the
ectopterygoid
Taxon/Character Matrix for Saurischia Data Set hasyetto
beidentified inbirds,
sothecondition of
theelement cannot becoded).
Thereader is
referred to
discussions of
individualcharacters in
the
textfor
caveats,
The
followingmatnx consists
of
theeighteentaxa
and
eighty-four
characters vanation, and rationale for
state
assignments. The
characters are
listedin
vertical
that
were
subjected to
PAUP. Characterstates
coded
denote
0 ancestral
conditions columns bytaxon,beginning withcharacter on
1the
left
andending withcharacter
and
characterstalescoded
denote
1 apomorphic conditions.
Characters coded 9 84
onthenght.Tofacilitatereference withnumbered character discussions in
indicateeitherthattheregionoftheskeletonwasnotpreserved,orthatitwas text,
thecharacters are divided intogroups of
five.

000000000000000000000000000000000000000000000000000000000000000000000000000000000000outgroup
000000000000000000000000000000001000000000000000000000000000000000000000000000000000ORNITHISCHIA
11111mil 0000000000 0000000000 00000
1000000000 0000000000
000000000000000 00000000000000 SAUROPODOMORPHA
mil mil mil II11111IIIIIIII111110000000000000000000000000000000000000000000000000CERATOSAURIA
mil III9I mil III1111111111111111111IIII11IIIII111100000900000000000000000000000000CARNOSAURIA
Ollll II19I mil mil 91111 91111 mil 9191 1 mil mil mii mii mii 91000 09009 0ooooooooornithomimidae
II91I II19I mil mil mil mil 10111 uiii iiiii iiiii 11119 11991 mil 91111 iiiii 11111 1111 deinonychosauria
101110119111119mil mil 11011 10111 mil 1111 1 10110 11991 1111 1 iiiio iiiii 11111 1111 1 0111 avialae
1919919111911999999191111mil1099111199191111119199999999911199999199991911I09I1991ORNITHOLESTES*
999119999999999991991999919919999919999999999991999999199999999999999999991990199999COELURUS*
mil 9999911199991I9111911911lOm1119111999IIIII9199999990999901999909009990111991COMPSOGNATHUS
999919999999999991999999919911999919999999999991119999999999991119999999919990199019MICROVENATOR'
91999991999999199199 9111119999 99991
9999999999 1999999919
999999999999999 99999999999999 SAURORNITHOLESTES*
999999999999999999999999999999199999999999999999919999999999999999199999999999999999HULSANPES
11999III9I mil 99999milmil101119199999111111999111999111119999919999911110011991CAENAGNATHIDAE
99999911919999999999911II1911911111199999991I119191119999999119991199991999999999991ELMISAURIDAE*
919919999991919999999999999911111119009999999909999099999999990990999999909999999900PROCOMPSOGNATHUS*
999991999999999999911999911111199110990900999900000099990999999000099990900900000009LILIENSTERNUS*

Addendum 2)AsnotedinPanV,character78.above,thereisnounequivocalevidencein
Continuing
I) research onearlydinosaur phylogeny revealed an
additionalsyn- theliterature for
theonentation of
the pubisin
troodontids. Because troodontids
apomorphy for
Saunschia, thelateral
overlap of
thequadratojugal onto the
pos- and dromaeosaurs appeared to
be
sister-groups (Ostrom 1969a, h;
Colbert and
teriorprocess of
thejugal.Archosauromorpha is
diagnosible in
part
by
long
a Russell1969),tentatively
I accepted that,
as
in
dromaeosaurs andbirds,thepubis
postenor process of
the jugal thatextends nearly to
the
end of
the
lower temporal was reversed in
Troodontidae. Thisassumption nowappearsmistaken. Dunng a
fenestra to
overlap laterally the
antenor process of
thequadratojugal (Gauthier recent visit
to
the Amencan Museum of
Natural History
to
studySaurornilhoides
1984;e.g..Fig. A,
IB).Theomithodiran(seeFig.7)quadratenolongerdisplays mongohensis. discovered
I thatthe
pubis wasprobably not
reversed in
Troodon-
the
anterodorsal to
posteroventral slope
charactenstic of
early
archosaurs. Con- idae.
sequently,neither the
ventral
portionof
the
lower
temporal fenestranorthe Osbom mentioned, but
did
notfigure or
descnbe, bonc-bearing
a concretion
postenor process of
thejugal are
as
longas
in
archosaurs ancestrally.Nevertheless, lyingneartothatcontainingthetypeskull"whichmaybelongtothesamein-
the jugal
still
has prominent
a lateraloverlap ontothe
quadratojugal in
dinosaurs dividual" 1924^:3).
( This
concretion,now
completely
prepared,
containsportions
ancestrally (Fig.IC,
D).
Just theopposite relationsbetween thesebones obtain of
postcranial
a skeleton including partial,
a associated pelvis;
the
pubis is
incom-
in
saurischians, however, because the
quadratojugal wraps around the
ventral plete,but
cleariy
It retains the
ancestral, anteroventral orientation. This
fact,
marginofthejugal,extendingupthelateralfaceofthatelementtogainbroad together with new information ontroodontid anatomy (Cume in
pressa.
b),
casts
exposureonthelateralsurfaceoftheskull(Fig.lE-K).Thisapomorphyisalso further doubt onthe monophyly of
Deinonychosauna. At
this
point, theonly
shared by
Hesperonus (Gingench 1976),
so
the
saunschian condition appears theropods inwhich reversed
a pubis hasbeen identified
withcertainty areDei-
ancestral forbirdsas
well. nonychus. I'clociraplor.and
Avialae.
 48 ORIGINOFBIRDSANDEVOLUTIONOFFLIGHT

FigureSkulls
\. inlateral
view:
A)
Euparkena capemis*;
B)
Ornilhosiichiislont^idens(Omithosuchidae); C)
Lcsolhosaiirus
diagnosncus (Omilhischia);
D)
Hel-
cwduiuosaurus lucki
(Omithischia);
Plaleosaurus
E) engelhariin (Sauropodomorpha); MassospontJylus
F) cannalus
(Sauropodomorpha), Allosaunts
G) fragilis
(Car-
C3rnos3uni):l)DwnuceunnimusbreyUerniis{Ornahomim\d3c)J)DromaivsauriisalherU'nsntDromaeosaundae]:K)Saurornnhoi^^
nosauna);H)
nuingoliciisis(Troodontidae), L)
Archaeopleryxhlhographica' (Avialael,
Drawing A
after
Ewer
(1965),B
after
Walker
(1964);C
after
Galton(1978);
D
after
Chang
(1979);EafterGalton(1984);FafterCooper(1981<j);GafterMadscn(1976);HafterRomcr(1956),after
I Russell(1972);after
J ColbertandRussell(1969);K
after
Russell(1969);
after
L Ostrom (l976o).
GAUTHIER:
SAURISCHIAN
MONOPHYLY 49

Figure
2.
Skulls
in
ventral
view:
A)
Euparkena capensis*:
B)
Ormlhosuchus longidens
(Omithosuchidae);C)
Aliosaurus
fragilis
(Carnosauna); D)
Tyrannosaurus
rex
(Carnosauna). Caudal
senes
in
lateral
view:
E)
Dromiceiomimus breviierlius
(Ornithomimidae);F)
Archaeopleryxhthographica' (Avialae);
G)
Demonychus
anlinhopus(Dromaeosaundae). DrawingA
after
Krebs
(1976);
B
afterWalker(1964);
Cafter
Madsen(1976);
Dafter
Romer (1956);
E
after
Russell
(1972);
F
after
Ostrom(1976a);
after
G Ostrom(1976*).
50 ORIGINOFBIRDSANDEVOLUTIONOFFLIGHT

Figure3.Antenorviewsofatlas(A-C).axis(D-G).ccrvicals3-4(H,I):A)Siegosaunisslenops(Omithischia);B,E)Ceratosaurusnasicorms(Ceratosauria);C.
G,
Aplen'x
I) ausirahs (Aves),D)
Camarasaurus grandis(Sauropodomorpha); F.
H)
Deinonychusanlirrhopus (Dromaeosaundae). Drawing
A,
after
D Ostrom
and
Mcintosh(1966);B,EafterGilmore(1920);F,HafterOstrom(1969/)).
GAUTHIER:
SAURISCHIAN
MONOPHYLY 51

Figure
Scapulocoracoids
4. lateral
in view:A)
Heterodonlosaurus tucki
(Omithischia);B)
Massospondylus cannatus(Sauropodomorpha); C)
Galhmimus bullalus
(Ornilhomimidae); D)
Archaeopleryx lilhographtca' (Avialae);
E,
F)
Deinonychus antirrhopus(Dromaeosaundae). Left
manusin
dorsal view:
G)
Crocodylus sp.
(Crocodylomorpha); H)
Heterodonlosaurus lucki
(Omithischia);
based
I) on
Thecodonlosaurus anltquus"
and
Efraasia
diagnosuca'(Sauropodomorpha); Masso-
J)
spondylus carinalus
(Sauropodomorpha); Synlarsus
K) rhodesiensis
(Ceratosauna);
AHosaurus
L) fragihs
(Camosauna);
Stntthiomimus
M) alius
(Omithomimidae);
N)
Deinonychus antirrhopus {.Droms,to%z\inAaty.O) Archaeopteri'xlilhographica*(\\\a.\3.t)^
Drawingafter
A.
H SantaLuca(l980);
after
B.
J Cooper(1981a);
after
C
Osmolskaelal(1972);D,EafterOstrom(1976a);FafterOstrom(1974ft);GafterRomer(1956);partly
I afterBakkerandGalton(1974)andHuene(1932);Kafter
Raath(1969);LafterMadsen(1976);MafterOsbom(1917);NafterOstrom(1976a);OpartlyafterOstrom(1976a).
52 ORIGINOFBIRDSANDEVOLUTIONOFFLIGHT

Figure
5.
Pelvic
girdle
in
lateral
view:A)
Scelidosaurus (Omithischia), B)
Massospondylus carinalus
(Sauropodomorpha); C)
Coelophysis bauri(Ceratosauna);
D)
Allosaurus
fragihs
(Camosauna);Gatlimimus
E) hullatus(Omithomimidae), Adasaurns
F) nwngoliensis(DromaeosaundacI; Demonychus
G) anlirrhopus(Dromaeo-
saundae); H)
Archaeopleryx hthographtca' (Avialae),I)
Apalornis
ccler(Omilhurael.DrawingA
after
Chang (l976/i);
B
after
Cooper (l')8la):
C
after
Raath
(1969);
DafterMadsen(1976),EafterOsmolskaetal.(1972);FafterBarsbold(1983);GafterOstrom(19766);HafterOstrom(1976a);after
I Marsh(1880).

Figure
6-
Tibia,
fibula,
andproximaltarsals
in
antenor view:
A)
Allosaunis
fragihs
(Camosauna); B)
Demonychus anlirrhopus(Dromaeosaundae); C)
Galtimimus
i«//aiiis
(Omithomimidae)- Distal
end
oftibiotarsusin
antenor
view:D)
Baptornis advenusiOmwhutae). MetatarsalsII,
III,
andIV
in
proximal
view:
E)
Hypsilophodon
foxi'(Omithischia);
Dilophosaurus
F) welherdli
(Ceratosauna); G)
Allosaurus fragdis
(Camosauna); H)
Demonychus anlirrhopus (Dromaeosaundae); Apleryx
I)
auslralis(Aves).
Left
pesin
dorsal
view:
Euparkeria
J) capensis',K)
Lagosuchus talampayensis': L)
Herrerasaurus ischigualaslensis' (Herrerasaundae*); M)Leso-
ihosaurus diagnosucus (Omithischia);N)
Anchisaurus polyzelus'
(Sauropodomorpha); O)
based
on
Procompsognalhus Inassicus'(Theropodamceriae
sedis).
and
Scgisaurus halh'
and
Dilophosaurus welheriUi(Ceratosauna); P)
Archaeopler\'x hihographica'
(Avialae).
Medial
view
of
articulationbetweendigit
and
I metatarsal
II:Q)Compsognalhuslongipes(Coelurosauna).DrawingA,GafterMadsen(1976);BafterOstrom(19766),CafterOsmolskaetal.(1972);DafterMartinetal.
(1980);EafterGalton(1974);FafterWelles(1984);GafterGilmore(1920);HafterOstrom(19696);
after
J Cruickshank(1979);KafterBonaparte(1975a);Lafter
Reig(1963);MafterBakkerandGalton(1974);NafterMarsh(1896);PafterOstrom(1976t2).QafterTarsitanoandHecht(1980),
GAUTHIER:
SAURISCHIAN
MONOPHYLY 53

'

I
54 ORIGINOFBIRDSANDEVOLUTIONOFFLIGHT

9^" ^.

Ornithodira

una

Figure
Cladogram
7. depicting
phylogenetic
relationships
among
Archosauna.

rS- ,x*
X .<^
.V* .<^^ .<-^^
■&". .<^
^v-
<o^ -°^ .-^^" .^<^°
^' o G o^ o®

Maniraptora
oelurosauria
et anurae
da

Figure
Cladogram
8. depicting
phylogenetic
relationships
among
comparatively
well
known
theropods
andother
dinosaurs.
GAUTHIER:
SAURISCHIAN
MONOPHYLY 55

Maniraptora

Coelurosauria
Tetanurae

Theropoda

Figure
Cladogram
9. depicting
phylogenetic
relationships
among
all
theropods
considered
in
this
analysis.
Gauthier, Jacques. 1986. "Saurischian monophyly and the origin of birds."
Memoirs of the California Academy of Sciences 8, 1–55.

View This Item Online: https://www.biodiversitylibrary.org/item/53408

Permalink: https://www.biodiversitylibrary.org/partpdf/74988

Holding Institution
Smithsonian Libraries

Sponsored by
Smithsonian

Copyright & Reuse

Copyright Status: In copyright. Digitized with the permission of the rights holder.
License: http://creativecommons.org/licenses/by-nc-sa/3.0/
Rights: https://biodiversitylibrary.org/permissions

This document was created from content at the Biodiversity Heritage Library, the world's
largest open access digital library for biodiversity literature and archives. Visit BHL at
https://www.biodiversitylibrary.org.

This file was generated 16 April 2022 at 17:25 UTC