Palaeont. afr., 31, 1-38 (1994)
THE PHYLOGENY OF BASAL ARCHOSAURS
by
Lars Juul
Bernard Price Institute for Palaeontological Research,
University of the Witwatersrand, Johannesburg, Private Bag 3, Wits 2050, South Africa.
ABSTRACT
Archosaur systematics has received much attention from the mid 1970s and several influential
works on this topic have emerged. As discrepancy exists among proposed phylogenies, some of
the most important of the papers in question are assessed here. Characters used in cladistic
analysis have often been selected too uncritically or phrased too vaguely to be of diagnostic
value, and previously used, seemingly valid apomorphies have been disregarded by later
workers, sometimes for no apparent reason.
In the present paper a character matrix for the Archosauria was assembled by critically
incorporating, and often modifying, characters used in earlier works on archosaur systematics. A
phylogeny resulting from cladistic analysis of the matrix compiled here supports the monophyly
of Archosauria, Crown-group Archosauria Crurotarsi, Ornithodira, Dinosauromorpha and
Dinosauria, and disputes the existence of Suchia less Ornithosuchidae and of a monophyletic
Crocodylotarsi. A new taxon, Dromaeosuchia, is erected for a clade consisting of Ornithosuchidae,
Crocodylomorpha and Gracilisuchus plus Postosuchus.
KEYWORDS: phylogeny, archosaurs, assessment.
GENERAL INTRODUCTION
The purpose of this paper is to review and critically
assess what are here construed to be the most important
and influential, recent phylogenies (see below) proposed
for basal archosaurs. This is a justifiable task as these
phylogenies differ somewhat in their placement and
composition of the constituent taxa. Furthermore not
all characters listed in the earlier studies are discussed
in the later ones in which some previously used,
apparently valid apomorphies/characters were left out;
additionally a number of characters that have been
employed are too vaguely defined to have systematic
meaning. Thus, as a consequence, the objective here is
also to establish a comprehensive, better defined
character assemblage as basis for archosaur systematics.
It is the paraphyletic assemblage of basal archosaurs
previously known as the "Thecodontia" (e.g. Charig,
1976b; Bonaparte, 1982) that is the main concern of
the present paper; the interrelationship of these animals
and their relationships with other taxa is often less well
understood than in the case of other archosaurs. Thus
the group in question corresponds to the non
crocodyliform, non-dinosaurian, non-pterosaurian
archosaurs; note that basal crocodyliforms
+Protosuchidae, Orthosuchus, Gobiosuchus and
Sphenosuchus (Benton and Clark, 1988) are not
included above and will not be treated in detail in this
study as these have been and still are considered closely
affiliated with the taxa included in Mesoeucrocodilia
(see also Sereno and Wild, 1992).
BACKGROUND
At present two conflicting definitions of Archosauria
are in use: "traditionally" the clade consisting of
Proterosuchidae and all later evolved relatives has been
termed Archosauria (e.g. Charig, 1976a; Paul, 1984;
Benton, 1985, 1990a). However, recently the concept
has been redefined as consisting of the two extant
(archosaurian) taxa, Aves and Crocodylia, their most
recent common ancestor and all its fossil descendants
(Gauthier, 1986; crown group concept of Hennig, 1971)
and the term Archosauriformes has been applied to this
new Archosauria + the successive outgroups
Proterochampsidae (and more recently Euparkeria) ,
Erythrosuchidae and Proterosuchidae (Gauthier, 1986).
Here the traditional definition will be adhered to.
The notion that dinosaurs evolved from 'thecodont',
or 'thecodontian' (Charig, 1976b), ancestry is a long
held one, but there has been much disagreement among
authors about the exact relationship between
'thecodonts' and dinosaurs. This discrepancy has
existed partly because of an incomplete fossil record,
but also due to a lack of applied stringent cladistic
method in earlier archosaur systematics; resulting in,
among other things, that Dinosauria has been conceived
as a polyphyletic group (e.g. Charig, 1976a; Bonaparte,
1978; Cruickshank, 1979) because of the use of
primitive characters in phylogenetic reconstruction.
However from the mid 1980s, with the work of
Gauthier (1986) on saurischian monophyly and the
origin of birds, in connection with which he supplied a
diagnosis of a monophyletic Dinosauria, grew a
consensus among palaeontologists that dinosaurs had
but a single origin. This had been suggested earlier,
notably by Bakker <).Ild Galton (1974), but did not gain
broad acceptance at the time. At presellt only few
workers seriously question dinosaurian 'monophyly;
e.g. Charig (1991a).
2
MATERIAL AND METHODS
The phylogenies reviewed are: Gauthier (1986),
Gauthier et al. (1988), Benton and Clark (1988),
Benton (1990a), and Sereno (1991b); also Novas
(1992), Sereno and Novas (1992) and Parrish (1992;
1993) will be referred to below.
These papers are here seen as epitomising the
knowledge of the systematics of archosaurs and basal
dinosaurs resting on the foundation of othenecent and
older studies such as Romer 0956; 1972b), Bakker
and Galton (1974), Bonaparte (982), Chatterjee (1982),
Paul (1984) and Benton (1985).
Numbers (1-263) in parentheses below refer to
characters/character states in the relevant diagnoses in
appendix 1; references to nodes also pertain to the
cladograms in appendix 1.
Unless otherwise specified, definitions and diagnoses
by Benton are from Benton and Clark (1988), those of
Gauthier from Gauthier (1986) and those of Sereno
from Sereno 0991 b).
All characters, which are not contradicted by original
descriptions or fossil material examined by the present ·
author, are used to construct a cladogram of the
Archosauria. Some characters here deemed as being
phrased too vaguely were left out.
Prolacerta, Y oungina and Rhynchosauria were used
as archosaur outgroup taxa. Specimens of Pro lacerta
(BP/1/2675 and BP/l/2676), the type specimen of
Pricea (BP/1/471), the skull of which does not differ
morphologically from that of Prolacerta (Gow 1975),
and specimens of Youngina (BP/1/2871 and BP/1/
3859), augmented by descriptions by~Broom (1922)
and Gow (975), were used to determine the relevant
character states in these taxa. As representatives of
Rhynchosauria specimens of Meso such us (SAM 5882,
SAM 7416 and SAM 6536) and Howesia (SAM 5886
and SAM 5884) were examined because these genera
are the most likely to exhibit character states plesio
morphic for rhynchosaurs (pers. comm. David Dilkes);
in instances where this was not the case descriptive
work on other rhynchosaur genera was consulted
(Benton, 1990b; Benton and Kirkpatrick, 1989).
The 'branch-and-bound' search option of the
computer implemented algorithm PAUP Phylogenetic
Analysis using Parsimony (version 3.0s) developed by
David Swofford (Swofford, 1990) was used to derive
most parsimonious cladograrris. All characters were
treated as unordered ("Fitch characters") and scaled to
equal weight, i.e. a base weight of 1000 was assigned
to binary characters, a weight of 500 to 3-state
characters, a weight of 333 to 4-state characters etc.
This is recommended by Swofford (1990) whenever.
multi state characters are in operation so that these will
not count for more steps than binary ones during
search for shortest tree(s). Th~ DELTRAN or delayed
transformation option, which resolves ambiguities in
character configurations by preferring parallelisms over
reversals (Swofford, 1990), was chosen for optimising
the character states in the new diagnoses (apomorphy
list) (appendix 2).
The following abbreviations precede specimen
numbers referred to in the text and identify the place of
storage of the specimens: BP/1/, Bernard Price Institute
for Palaeontological Research (vertebrates); GHG,
Geological Survey, Pretoria; SAM, South African
Museum; SMNS, Staatliches Museum fur Naturkunde,
Stuttgart.
If not clearly stated otherwise then only those taxon
names followed by a specimen number refer to personal
observations by the present author.
CRITICAL REVIEW
Introduction
The four papers that are the main subjects of this
review (Gauthier 1986, Gauthier et al. 1988, Benton
and Clark 1988 and Sereno 1991b) differ in certain
aspects: no discussion of the characters listed was
provided by Gauthier et al. (1988) and only
synapomorphies of dinosaur taxa were discussed to
any extent by Gauthier (1986). Benton briefly examined
characteristics of different archosaurian taxa and stated
that his analysis was an initial attempt at discovering
the interrelationships of these, and in order not to
ignore potential synapomorphies rather imprecise
characters were included in the diagnoses, which will
become apparent below. In contrast Sereno gave good
account of the characters he employed and often
reviewed synapomorphies previously used by others,
frequently disregarding these. Most of the findings
offered by Sereno were anticipated or reached in an
earlier paper (Sereno and Arcucci, 1990).
All workers used PAUP to obtain their respective
cladograms and all appear to have coded their characters
as unordered with equal weight (unweighted). The
configuration of taxa presented by Benton is not the
most parsimonious resulting from his analysis;
according to the author himself Ornithosuchia was
placed as the sister group of Postosuchus plus
Crocodylomorpha in the cladogram entailing the fewest
steps, but because this arrangement involved numerous
reversals a 'slightly less parsimonious' cladogram was
preferred. This procedure amounts to aposteriori coding
of some characters as irreversible (,Camin-Sokal'
characters) .
For clarity the following discussion falls in four
parts pertaining to levels or clades in the archosaur
phylogenies; the first three of these divisions do of
course not correspond to any monophyletic groupings
or clades, but are only practical conventions: 0) non
crown group Archosauria, (2) non-crocodylomorph
crocodile-line Archosauria; basal crocodylomorphs
(Pseudohesperosuchus, Saltoposuchus, Dibothrosuchus
and Sphenosuchus) are not treated in this analysis as
these and Crocodyliformes are held here to,
unproblematically, constitute a monophyletic taxon,
(3) non-dinosaur bird-line archosaurs, and (4) basal
branching points within Dinosauria. Such a partitioning
is possible because there is a general consensus among
the authors, whose phylogenies are being discussed
here, about which taxa reside at the various levels.
Some overlap between the sections is of course
unavoidable; not least because of the fact that the
character distributions on the different phylogenies
sometimes conflict in listing the same synapomorphies
as diagnostic of different or incompatible clades.
Non-Crown Group Archosauria
As shown by the cladograms, the~is consensus
among the authors whose papers are reviewed here
that Erythrosuchidae and Proterosuchidae constitute
successive outgroups to the rest of Archosauria. Earlier,
the Proterosuchidae and Erythrosuchidae had been
included in a paraphyletic 'Proterosuchia' (Charig and
Sues, 1976), sometimes with Proterochampsidae
(Romer 1972b; Carroll, 1987) or Euparkeria
(Cruickshank, 1972). Bonaparte (1984) also
hypothesised an erythrosuchian ancestry for
Rauisuchidae and included the two taxa in a suborder,
' Erythrosuchia'; previously Hughes (1963) had
proposed similar ideas. These groupings, however,
were based on phenetic similarity (overall resemblance)
and did not reflect a 'natural' phylogenetic pattern.
The diagnoses of Archosauria by Gauthier et al.
(1988) and Benton share four characters: (1) postfrontal
reduced, (3) antorbital fenestra present, (6) presence of
an ossified laterosphenoid and (9) possession of a
fourth trochanter on femur. Presence of an antorbital
fenestra (3) is undoubtedly synapomorphic for this
clade as it is present in all its members except advanced
crocodilians, and not present in the archosaurian
outgroups Rhynchosauria (Mesosuchus; SAM 2871),
Prolacertiformes (Prolacerta; BP/1/2675, Priceal
Prolacerta; BP/1/471) and Youngina (BP/1/2871).
Character (1) has some credit to it as the postfrontals
in the just mentioned three taxa are larger than the
postfrontals in archosaurs, but the way the character is
phrased is too weak for it to be of diagnostic value.
Given a new specimen with postfrontals of a size
intermediate between that of Proterosuchus and the
immediate outgroup taxa of Archosauria there is no
way, other than a subjective evaluation, of deciding
whether the postfrontals of such a specimen would be
reduced enough to be diagnosed by synapomorphy (1).
It is necessary to redefine this type of character in a
quantified manner such that e.g. the length of the
feature in question is given as a proportion of another
osteological measurement. Even though the possibility
that such two measurements may not display isometry
throughout ontogeny remains, it is here believed
preferable to quantify characters instead of using terms
such as reduced, elongated, broad, etc. to specify
conditions as the result of the former procedure is
more open to objective testing.
In addition to Benton and Clark (1988) and Gauthier
et al. (1988), Parrish (1992) also holds the presence of
a laterosphenoid (6) to be an archosaurian
synapomorphy, although Cruickshank (1972) explicitly
stated that this feature is not possessed by
Proterosuchus. Benton quoted a personal observation
of J. M. Clark of this feature in Proterosuchus in
"
3
support of his use of the character; a recent study by
Clark et al. (1993) verifies the presence of a
laterosphenoid in Proterosuchus. Contra Gauthier et
al. (1988) and Benton who believed it to be a
characteristic of all archosaurs, Parrish (1992) claimed
that the possession of a fourth trochanter (9) only
diagnoses a clade consisting of Erythrosuchidae and
higher archosaurs, and it is indeed impossible to discern
a raised fourth trochanter on femora from published
drawings ofproterosuchids (Cruickshank, 1972; Parrish
1986), and Cruickshank (1972) did not mention any
such feature in his account of the femur in
Proterosuchus, nor did Hughes (1963).
Romer (1972d) claimed that there was no evidence
for the presence of a fourth trochanter in Gracilisuchus.
Parrish furthermore listed the possession of a lateral
mandibular fenestra (17) as synapomorphic for
Archosauria (Archosauriformes in his terms); Benton
and Gauthier et al. (1988) purported this to be
diagnostic of Erythrosuchidae + all higher archosaurs
only and Cruickshank (1972) did not figure a
mandi bular fenestra in his reconstruction of
Proterosuchus claiming the apparent presence of this
feature in some Proterosuchus specimens to be an
artefact. After personal examination of a
Proterosuchus specimen (GHG 231) with both lower
rami preserved, found after 1972, I all} convinced that
the mandibular fenestra is present in this taxon. In
addition see Clark et al. (1993) and WeIman and
Flemming (1993), however, the mandibular fenestrae
are relatively much bigger in specimen GHG 231 than
reconstructed by the latter two authors.
Character (12), 'marginal teeth laterally
compressed', in contrast is potentially synapomorphic
for Prolacertiformes + Archosauria as it is present in
Prolacerta (BPl/1/2675) and all non-ornithodiran
archosaurs (with the notable exceptions of
Stagonolepidae (Walker, 1961) and Doswellia
(Weems, 1980); in addition, only part of the dentition
in parasuchians (e.g. Chatterjee, 1978) show the
apomorphic state), it is however, not very pronounced
in Proterosuchus (GHG 231) and Erythrosuchus
(BP/1/5207). The character is absent in Youngina
(BP/1/2871) and rhynchosaurs (Mesosuchus SAM
6536; see also e.g. Benton, 1990b).
Among the more dubious apomorphies noted by
Gauthieretal. (1988) at node 1 (their Archosauriformes)
is 'upper temporal fenestra dorsally oriented'
(5): a dorsally oriented upper temporal fenestra is also
present in the archosaurian outgroups Prolacertiformes
(PricealProlacerta BP/1/471), Young ina (BP/1/2871)
and Rhynchosauria (Carroll, 1987), thus the character
must be considered invalid; note that Benton did not
list this character in his diagnosis of the same clade.
The latter author, however, listed 'posterior border of
infratemporal fenestra bowed' (11) as an archosaurian
synapomorphy; the meaning of this character is unclear
and the infratemporal fenestra posterior borders are
bowed, in at least some rhynchosaurs, as well as in
Pricea (BP/1/471), in different manners . Benton's
4
character (13) 'no ectepicondylar groove or foramen
on humerus' is almost identical to character (25) by
Gauthier et al. (1988) of node 2; the outgroup taxa
Prolacerta (BP/1/2675), Youngina (Gow, 1975) and
Mesosuchus (per. obs) all feature an ectepicondylar
groove albeit that in Mesosuchus is not very
pronounced. A specimen that is possibly a proterosuchid
features an ectepicondylar groove on the humerus
(Charig and Sues 1976: 17) and there is a deep
ectepicondylar groove on the humerus of Stagonolepis.
Thus this character will need further work before its
distribution can be ascertained with certainty.
The occurrence of thecodont dentition (Gauthier
et al., 1988); characters (8), 'premaxillary teeth
implanted in deep sockets', and (18), 'maxillary and
dentary teeth in deep sockets', node 1 and 2
respectively) character (50) of Benton who has listed
thecodont dentition as a single trait of node 3) has been
reported in a confusing manner. Hughes (1963) and
Charig and Sues (1976: 13) purported 'Proterosuchia'
(Proterosuchidae and Erythrosuchidae) as having
'subthecodont' dentition. However, Charig and Sues
later in the same paper when describing the marginal
dentition of Chasmatosaurus (Proterosuchus) in
general (not only that of the premaxillary; character (8)
by Gauthier et al., 1988) confusingly used the term
thecodont. The teeth in the premaxilla of Archosaurus
(onl y tooth bearing bone recovered of this genus) were
also referred to as being of thecodont implantation; so
was the dentition of Garjainia (Erythrosuchidae), and
Arizonasaurus, this genus, though, may be
rauisuchian (Galton, 1977; Chatterjee, 1985) rather
than erythrosuchid, has 'deep oval tooth sockets'.
Only Erythrosuchus, was described as having
'subthecodont' marginal dentition (Charig and Sues,
1976); this is indeed an ill-defined condition as noted
by Romer (1956). A large fragment of an
erythrosuchid maxilla (BP/l/4540) features very
deeply rooted teeth; at a point where the height of the
maxilla is 115 mm a tooth root is present which is
65 mm in length. The crown of this tooth is not
complete, but an approximate estimate is that the
length of the root is more than 45 percent of that of the
total tooth length. This also seems to be the case for the
premaxillary dentition of Proterosuchus, however, the
teeth of !he maxilla of the same animal are less deeply
rooted (pers. obs). Because it was not possible to
obtain a quantitative expression of the relative lengths
of root to crown for many archosaurs and thereby
make meaningful distinction between the terms
'subthecodont', and 'thecodont' characters (8) and
(18) are not considered further here. In addition
subthecodont is also used by some workers to refer to
the condition where interdental plates are present
rather than as a measure of root length. On the basis of
the material available to the present author it was not
possible to establish the exact distribution of
interdental plates within Archosauria; nevertheless
they are present in rauisuchids (,Kupferzell
rauisuchid'; SMNS unnumbered), Postosuchus (Wu
and Chatterjee, 1993), Herrerasaurus (Sereno and
Novas 1994) and a number of theropods, appear to be
so in Ornithosuchus (Walker, 1964) and in modified
form in the maxilla of Massospondylus (SAM K
1314). Interdental plates are not present in
rhynchosaurs or Prolacerta (pers obs.).
Serrated teeth (7) are present in Proterosuchus
(GHG 231),Erythrosuchus a!ricanus(BP/1/5207) and
other toothed archosaurs except stagonolepids, and
pterosaurs and some dinasaurs, but not in rhynchosaurs,
Prolacerta (BP/1/2675), Youngina (BP/1/2871) or
Milleretta (BP/l/2876 and BP/1/3318).
Synapomorphy (2) 'postparietal fused' is
corroborated by the presence of this condition in e.g.
Proterosuchus (Cruickshank, 1972), Shansisuchus
(Charig and Sues, 1976), Euparkeria (Ewer, 1965) and
the presence of discrete postparietals in Y oungina
(Romer, 1956) and Petrolacosaurus (Carroll: 1987);
however because postparietals are absent inProlacerta,
Trilophosaurus, rhynchosaurs and possibly some
eosuchians (Romer, 1956) the character is only
incorporated tentatively here.
There is some disagreement among authors about
the presence/absence of intercentra in different segments
of the axial column in basal archosaurs: Gauthier et al.
(1988) believe that the absence of intercentra 'from all
postcervical trunk vertebrae' (23) (note that their
character (22) is included in (23) making it redundant)
is diagnostic at their and Benton's node two, and that
the absence of intercentra from postaxial cervicals
diagnoses node three, while Benton holds that' no
intercentra are present behind the axis in the presacral
column at node two (29). As pointed out by Sereno it
is problematic for the phylogeny proposed by Gauthier
et al. (1988), and that of Benton, that intercentra occur
all along the spine of Euparkeria; thus the latter authors
are forced to recognise reversals of respectively
characters (23 and 41 'intercentra absent from postaxial
cervicals ') and (29) in this genus. Charig and Sues
(1976) reported that in Erythrosuchidae ' .. . presence of
intercentra not certain but probable' in Erythrosuchus,
'cervical centra .... without intercentra' in Garjainia,
'no intercentra between axis and tail' in Shansisuchus
and 'no precaudal intercentra' in the possible
erythrosuchid Cuyosuchus. According to Parrish (1992)
Fugusuchus is the 'only erythrosuchid for which an
intercentrum is preserved' and that 'anteroventral and
posteroventral bevelling of the cervical and trunk centra
in Vjushkovia triplicostata and of the cervical centra in
Garjainia prima (for which trunk centra are unknown)
indicates that these species possessed intercentra as
well' . Charig and Sues (1976) did not explicitly discuss
indirect evidence (shape of centra) for intercentra in
Erythrosuchidae, even though they suspected their
presence. Parrish's observations however, and the
osteology of Fugusuchus, seem to invalidate the
assignment of Gauthier et al.' s (1988) character (23)
and Benton's character (29) to node two. Sereno arguing
similarly chose to move Euparkeria to a basal position
in the archosaur phylogeny because of its full
complement of intercentra, which are not present in
Proterochampsidae and crown group archosaurs, and
on grounds of tarsal anatomy.
There is an apparent discrepancy between the
character lists of Gauthier et ai. (1988) and Benton at
node two; this is because Gauthier et ai. concerned
themselves with dental characters while Benton focused
on pelvic morphology. In addition to .!!!..e problem of
thecodont dentition, which has been discussed above,
the issue of palatal tooth characters is raised here.
'Teeth on transverse process of pterygoid absent' (21)
seems a valid apomorphy for node two, but again the
position of Euparkeria in Gauthier et ai.'s (1988)
phylogeny is problematic (as acknowledged by the
author himself (1986: 43»: Euparkeria exhibits teeth
on vomer, palatine and pterygoid; no crown group
archosaur does that. In Erythrosuchidae, palatal teeth
are absent (Sereno; Parrish 1992; contra Gauthier:
176) and this is hypothesised to constitute an apomorphy
for that family (i.e. to have been lost independently in
Erythrosuchidae and crown group archosaurs) by
Parrish (1992), as palatal teeth are present in
Proterosuchus (Cruickshank, 1972), Proterochampsidae
(Walker, 1968; Romer, 1971c; Sues, 1976) and
Doswellia (Weems, 1980). Gauthier, however, because
of his placement of Euparkeria, is forced to invoke
parallel loss of palatal teeth in his Pseudosuchia (77)
and Ornithosuchia (166) or, alternatively, a reversion
of this character in Euparkeria to the plesiomorphic
condition.
Character (18) refers to the question of thecodont
dentition which has already been discussed and character
(20) 'dentary teeth with an enlarged tooth within few
teeth from symphysis' is not readily supported by the
relevant literature; further because of the way it is
formulated it could too easily comprise more than a
single character and is therefore hardly valid.
Of the four pelvic characters listed in Benton's
diagnosis of node two, 'iliac blade has a small anterior
process'(34), 'pubis has a strongly downturned tuber
when seen in side'view' (35) and 'ischium has a large
posteroventral process (the ischium is longer than the
iliac blade)' (36) are well supported in literature e.g.
compare Cruickshank (1972, Figure 8a) and Carroll
(1987, Figure 13-2a,c,d) with Charig and Sues (1976,
Figure 8L) and Parrish (1986, Figure 4-11»; overlooked
by Gauthier et ai. (1988) who do not address pelvic
morphology at this node in the cladogram at all.
Although the conditions described by (34) and (35) are
somewhat difficult to delimit these characters are
applicable to taxa at this level; in contrast character
(33) "pelvis is markedly three-rayed, with a long down
turned pubis and ischium" is too loosely defined to be
of much systematic value and is partly covered by (34).
Both Gauthier et ai. (1988) and Benton noted
characters pertaining to the humerus and pectoral
girdle at node two, but without overlap of diagnoses.
Gauthier et ai. "scapula narrow above glenoid" (24),
although it describes a recognisable condition different
from that of Proterosuchus, is likely to characterise a
-, ..
5
more inclusive clade as it is found in, at least,
Protosaurus (Carroll 1987), Trilophosaurus (Romer,
1956) and the rhynchosaurs Scaphonyx (Benton and
Kirkpatrick, 1989), Rhynchosaurus (Benton 1990b)
and to a lesser extent Paradapedon (Carroll 1987).
Benton's 'deltopectoral crest extends at least one
quarter of the way down the shaft ofthe humerus' (31),
though it appears well defined, cannot be affirmed as it
is not possible to determine exactly where on the
humerus the deltopectoral crest ends; see below under
(235) for an alternative phrasing of this kind of
character. His use of the character 'loss of anterior
proximal 'hook' on the metatarsal V' (39) at node two
is invalidated by the presence of a "hooked" fifth
metatarsal, in at least, the erythrosuchid Shansisuchus
shansisuchus (Charig and Sues, 1976), Euparkeria
(Ewer, 1965), Stagonoiepis (Walker, 1961),
Parasuchus (Chatterjee, 1978) and Saurosuchus
(Rauisuchidae) (Sill, 1974); see below, synapomorphy
(241) , for further discussion of this feature.
At node three Benton placed Proterochampsidae
+ Doswellia, Gauthieret ai. (1988) Proterochampsidae,
and Sereno, who did not treat the first two nodes in his
analysis, Euparkeria as sister group to the rest of
Archosauria. As reason for his placement ofEuparkeria
in the cladogram Sereno lists two characters ((67)
'postaxial intercentra absent'; (68) 'contiguous crural
facets on astragalus ') shared by Proterochampsidae and
crown-group archosaurs, not shared by Euparkeria,
and considered most of the characters (40,41,42,44,
45, 47, 49, 50, 51, 52, 53) listed by Gauthier et ai.
(1988) and Benton, in support of their configuration of
taxa at node three, problematic. Indeed this appears to
a great extent to be true: character (40) 'parietal foramen
absent' also diagnoses node two as all but one of five
genera of erythrosuchids listed by Parrish (1992) share
the absence of a parietal or pineal foramen. This is the
most parsimonious interpretation since it is not the
most basal of the erythrosuchid genera in which a
parietal opening is present. However, this character has
a very peculiar distribution; thus a parietal foramen is
also present in some individuals of Piateosaurus
according to Galton (1990), but the present author was
unable to positively identify this feature in any BPI
specimen of the closely related Massospondyius.
Characters (41) 'intercentra absent from postaxial
cervicals' and (50) 'thecodont dentition' have already
been discussed above. Characters (42) "humerus with
reduced epicondyles", (47) "femoral condyles not
projecting markedly beyond shaft" and (49) "otic notch
well developed" will here be disregarded for reasons
similar to those given in the discussion of character (1);
in addition see Sereno (p. 9) for further critique of the
former two of these characters. The condition specified
by (44), 'inner two digits of hands and feet more robust
than outer two' , also seems to be present inShansisuchus
(Charig and Sues, 1976, Figure lOT), and furthermore
good manual material is lacking for a number of
archosaur taxa (see below) which makes the application
of (44) too speculative. An intertrochanteric fossa (45)
6
is present in the outgroup taxa used here (Pro lacerta,
Youngina and Rhynchosauria), Proterosuchus
(Cruickshank, 1972) and Erythrosuchidae; the members
of the latter family are unique in retaining an
intertrochanteric fossa and also possessing a fourth
trochanter (Parrish, 1992). An intertrochanteric fossa is
absent in Proterochampsidae, Euparkeria (Parrish,
1992), and crown-group archosaurs. In Proterosuchus
all ribs seem to be either one- or two-hea:ded and in
Erythrosuchidae two-headed ribs is the plesiomorphic
condition while three headed ribs is apomorphic for
Vjushkovia (Parrish, 1992) - all of which makes
character (51) 'ribs all one- or two-headed' problematic.
Semi-erect or erect gait (52) appears to be a characteristic
of node three, however this character may in fact
encompass a whole suit of characters (e.g. pelvic,
femoral and tarsal). Benton's character (53) 'possession
of 'crocodiloid' tarsus ... ' must be the character Sereno
claimed corresponds to his 'loss of a bony
astragalocalcaneal canal ' (57) even though he
disregarded parts of Benton's apomorphy later in the
same paper (Sereno: 10; note that the same author is
incorrect in stating that Benton and Gauthier et al.
(1988) listed respectively his character '3' and '5' (here
48 and 58) as synapomorphies for this group).
Characters pertaining to body armour were also
listed: (54) ' single paramedian osteoderm pair per
cervicodorsal vertebra' and (55) 'dorsal body
osteoderms' by Benton and Sereno respectively. The
dorsal body armour of Chanaresuchus (Protero
champsidae) consists of a single median row of
osteoderms, 2.85 per vertebra in the cervicodorsal
series (see Romer, 1972c); Doswellia, in contrast, was
heavily armoured with scutes arranged in paramedian
files (Weems, 1980), but what the exact number of
lateral scute rows per vertebra was in this animal, is
unclear. In Euparkeria there is an approximately one
to one correspondence of vertebrae to paramedian
scutes; this is also the case for Ornithosuchus (Walker,
1964) and many crocodile-line archosaurs including
parasuchians (Chatterjee, 1978), Stagonolepis (Walker,
1961) and Protosuchus (Colbert and Mook, 1951). In
rauisuchids, however, the situation is somewhat
different, e.g. in Ticinosuchus there are two paramedian
scutes per vertebra (Krebs 1965, plate 3); in
'Mandasuchus ' two and a half (Charig pers. comm.).
Rounded osteoderms has been found for Erythrosuchus
(BP/1/5207), however, the arrangement of these is
uncertain. Of the other synapomorphies listed at node
three by Sereno ' interclavicle with reduced, tablike
lateral processes' (56) is not well supported in literature:
the relevant part of the interclavicle is missing in
Euparkeria (Ewer, 1965), neither clavicles nor
interclavicles are present in Chanaresuchus (Romer,
1972c) and although cha:r:.acter (56) is present in
Doswellia (Weems, 1980) the systematic position of
this animal is uncertain (Sues, 1992) (Gauthier et al.
(1988) listed a similar character (72) as synapomorphic
for crown group archosaurs, but noted that it might
characterise a more inclusive taxon).
The two characters «67), (68) respectively 'postaxial
intercentra absent' and 'contiguous crural facets on
astragalus ') cited by Sereno in support of the monophyly
of Proterochampsidae + crown group archosaurs
(excluding Euparkeria) stand against the five (62-66)
characters listed by Benton in favour of a monophyletic
taxon consisting of Euparkeria and crown group
archosaurs, but exclusive of Proterochampsidae, and
the seven (41, 62, 69-73) characters listed by Gauthier
et al. (1988) being synapomorphic for crown group
archosaurs including Euparkeria.
Character (67) is valid at this level and (68) seems to
be so also. The most parsimonious distribution of (62),
the presence of antorbital fossa, on the archosaur
cladogram, depends on resolution of the inter
relationships of Proterochampsidae: if Gualosuchus
turns out to be the most basal member of this family
(62) will be diagnostic of a clade including Euparkeria,
Proterochampsidae and crown group archosaurs because
Gualosuchus is the only proterochampsid in which an
antorbital fossa or depression is present; furthermore
this feature is of similar construct as that of Euparkeria
where the dorsocaudal process of the maxilla has a
sharp prong which overlaps the lacrimal inside the
fossa. If Gualosuchus does occupy a basal position
within Proterochampsidae we would, in the phylogenies
of both Benton and Sereno, have to invoke a secondary
loss of (62) in proterochampsids other than Gualo
suchus and the character would require two-steps in .
both phylogenies, but if the latter genus is not the most
basal member of its family the configuration of character
(62) will require a total of at least three steps in
Sereno's cladogram, but only a minimum of two in
that of Benton (as well as in that of Gauthier et al.).
Benton did also specify in (62) that the antorbital
fenestra lying in the fossa be large; a large antorbital
fenestra may also have been the ancestral condition in
Proterochampsidae, but dorsoventral flattening of the
snout region could be the reason why this skull opening
is less pronounced in proterochampsids, and extreme
lengthening of e.g. the premaxilla and maxilla in these
animals also makes the antorbital fenestra look relatively
smaller. An ahtorbital fossa is also present in the
erythrosuchids Vjushkovza sinensis, Shansisuchus, and
Erythrosuchus (BP/l/5207; Parrish, 1992; character
16 of table 1) in the latter genus the fossa has a striking
resemblance to that of Euparkeria (Parrish 1992,
Figure 6).
The condition specified by character (63) 'nasals
run forwards between the nares' is present in at least
one of specimen Proterosuchus (GHG 363), some
erythrosuchids (Charig and Sues 1976, Figure 6), but
could not be verified for Prolacerta (BP/1/471) contra
Gow (1975; Figure 12), and (64) 'diapophysis is placed
fairly high on the neural arch of cervical vertebrae' in
at least Erythrosuchus and Shansisuchus. Characters
(65) 'parapophysis transfers to the neural arch in anterior
dorsal vertebrae' and (66) 'diapophysis and
parapophysis fuse in the posterior dorsal vertebrae and
the ribs become single-headed' may very well also be
valid for Chanaresuchus and Proterochampsidae
(Romer, 1972c).
Characters (41) and (72) have already been discussed;
character (69) 'exoccipital fused with opisthotic in
adult' does not seem to be a characteristic ofEuparkeria
(Ewer 1965, Figure 2a), but is pre~ent in
Proterochampsa (Sill, 1967; Figure 3) and
Chanaresuchus (Romer, 1972c), and (11) 'calcaneal
tuber posteriorly directed' seems to be featured by
Proterochampsidae (see Romer 1972c).
Euparkeria+crown group archosaurs, however, share
two characters: (54) 'possession of dermal armour with
one pair of osteoderms per vertebra' and (70) 'apex of
neural spine expanded in dorsal view'. The former
character was originally listed by Benton as being
synapomorphic for a more inclusive group, but does
only characterise Euparkeria and crocodile-line
archosaurs, and note that both characters are most
parsimoniously interpreted as defining a monophyletic
Euparkeria+crocodile line archosaurs, were it not for
the three characters (see below) shared by crown group
archosaurs (exclusive of Euparkeria) and for the two
characters, (67) 'postaxial intercentra absent' and (68)
'contiguous crural facets on astragalus', shared by
Proterochampsidae and crown group archosaurs.
The systematic status of Doswellia is problematic;
Benton placed this genus as the sister group of
Proterochampsidae (node 3.1) on account of two
apomorphies shared by these taxa: (60) 'loss of
postfrontal (parallelism with Crocodylomorpha)' and
(61) 'pelvis massive, and not three-rayed'. This is
obviously rather weak: evidence; the latter of these
characters is not a well defined one and the former not
an autapomorphy for the group. Sues (1992) claimed
that a detailed comparative study revealed no other
synapomorphies for Doswellia plus Proterochampsidae.
In addition to (60) Doswellia displays other features
which have been used to characterise crocodile-line
archosaur clades, such as broad, medial interpterygoidal
contact (Crocodylotarsi), 'spine tables' (see below),
lateral deflection of iliac crest (Protosuchus) and
additional vertebrae incorporated into sacrum
(Ornithosuchidae, some rauisuchians). However, some
aspects of this enigmatic archosaur's morphology are
quite primitive; e.g. pterygoid dentition, weakly
differentiated proximal femur and extensive
participation, of pubis in acetabulum.
Crown-Group Archosaurs
Before dealing with crocodile-line archosaurs the
characters uniting these and the bird-line archosaurs in
the crown group clade (node four in Gauthier's
cladogram; node five in Benton's and Sereno's
cladograms) will be examined. All the synapomorphies
listed at this node by Gauthier have been discussed
above and none seem valid unless Euparkeria is
included in crown group Archosauria. Of the characters
cited by Benton only (77)'palatal teeth absent' is clearly
apomorphic at this level; (74) 'parietals send posterior
processes into the occiput which meet the supraoccipital'
",
7
is also present in Proterochampsidae (Romer, 1971c;
Sill, 1967) and Euparkeria (Ewer, 1965) and it is
uncertain at which level (75) 'discrete postparietal and
exoccipitals absent beyond juvenile stages of
development' applies: According to Westphal a discrete
postparietal is present in some parasuchians (Westphal,
1976), and the exoccipital seems not to be fused with
the opisthotic in Nicrosaurus gregorii (same ref.,
Figure 3a); Chatterjee, however, reconstructs
Parasuchus hislopi with fused exoccipital and
opisthotic, but finds no postparietal (=interparietal) in
this species (probably juvenile specimens) (Chatterjee,
1978). The exoccipital and opisthotic bones are also
coossified inAngistorhinus (another parasuchian), but,
as in the case of Parasuchus, no postparietal appears to
have been present (Mehl, 1915a). Stagonolepis may,
like Typothorax, have had a separate postparietal,
nevertheless the exoccipitals and opisthotics are fused
in this animal (Walker, pers. comm.); in Gracilisuchus
a discrete postparietal appears to be present, but nothing
definite can be said about the condition of the
exoccipitals (Romer, 1972d) and the state of the
character is hard to discern from literature for many
other taxa as well as the ontogenetic stage of these.
Thus the use of part of character (75), as two characters:
'exoccipitals and opisthotics fused' and 'absence of
discrete postparietal ', here may be premature. If
character (76) "pterygoids meet medially in the palate"
is rephrased as: 'pterygoids with medial contact for at
least a third of their length' it may diagnose crocodile
line archosaurs as it is present in Parasuchus
(Chatterjee, 1978) , Rutiodon (Westphal, 1976)
Angistorhinus (Mehl, 1915a), Stagonolepis (Walker,
1961), Postosuchus (Chatterjee, 1985), Pseud
hesperosuchus and Hemiprotosuchus (Bonaparte,
1971). In contrast pterygoids in Coelophysis and
Syntarsus have no medial contact (Colbert, 1989) and
in Lesothosaurus these bones meet at their extreme
anterior end, and through a pair of processes posteriorly
(Sereno, 1991a); it is not certain what the state of this
character originally was in Sauropodomorpha.
In addition to 'palatal teeth absent' (77) Sereno lists
'calcaneal tuber directed more than 45 degrees
posterolaterally ' (78) and 'Calcaneum with contiguous
articular surfaces for fibula and distal tarsal 4' (79) as
crown group synapomorphies. The calcaneal tuber in
crown group archosaurs seems to be more posteriorly
directed than in basal archosaurs and character (79) is
not contradicted by original descriptions or fossil
material available to the present author.
Non-crocodylomorph crocodile-line Archosauria
The question of tarsal anatomy is raised in earnest
here as Gauthier, Benton and Sereno all cite several
synapomorphies, uniting crocodile line archosaurs,
pertaining to this anatomical region. This question has
received more than its fair share of attention by workers
in archosaur systematics and the structure of 'astragalus
and calcaneum' has preoccupied minds to a degree
which has sometimes led to an almost overshadowing
8
of other aspects of morphology. This subject will
therefore be treated at some length here again.
It has long been recognised that two rather different
tarsal constructs are present among archosaurs: Romer
(1971b) referred to Krebs (1963) as being the worker
who first clearly brought out the distinction between a
crurotarsal and mesotarsal ankle joint; previously also
Walker (1961) gave good account for the function of
the crurotarsal ankle joint. In the crurotarsal ankle,
rotation of the pes relative to the crus, in the
parasaggital plane, occurs through a joint between the
astragalus, which is functionally part of the lower
limb, and the calcaneum which moves with the foot. In
the mesotarsal type of ankle the axis of rotation of the
pes on the crus runs mediolaterally between the
proximal- and distal tarsals. Krebs later (1974)
grouped pseudosuchians (?sensu Benton) and
crocodylomorphs together in Suchia because these
archosaurs all have an ankle joint of the former type.
Further distinction was made by Bonaparte (1971: 97)
who chose to distinguish between two ankle types of
the crurotarsal configuration; these have later come to
be known as the 'crocodile-normal' (CN) and the
'crocodile reversed type' (CR) (Chatterjee, 1978). In
the CN ankle the pivotal articulation between the
astragalus and calcaneum occurs through a peg on the
former element and a receiving socket on the latter,
while in the CR type the peg is on the calcaneum and
the socket on the astragalus. The (CN) ankle is found
in crocodile-line archosaurs and the CR ankle is only
recognised, today, as being present III
Omithosuchidae.
Cruickshank (1979) put forward a phylogeny which
suggested a fundamental dichotomy of 'thecodonts'
into CN and CR lines, arising from a 'Proterosuchian'
ancestor with a mesotarsal ankle and leading to
respectively Pro sauropoda, Omithischia + Coeluro
sauria and Sauropoda + Camosauria. This scheme was
elaborated upon by Chatterjee (1982) who proposed a
phylogeny of 'thecodont' tarsi where, an ankle construct
of a 'primitive mesotarsal' type, found in
Proterosuchus, gave rise to CN and CR tarsal forms
which were further transformed into respectively
'advanced mesotarsal normal' and 'advanced mesotarsal
reversed' ankle forms; the latter two types occurring in
different dinosaurian taxa. Cruickshank and Benton
(1985) revising Cruickshank's and Chatterjee's earlier
proposals chose to recognise only one 'advanced
mesotarsal' ankle type supporting a monophyletic
Dinosauria, and introduced yet another term 'modified
primitive mesotarsal' referring to the ankle construct
found in Euparkeria, Chanaresuchus and
Erythrosuchus; the latter genus, however, is more likely
to possess a tarsus closely resembling that of
Proterosuchus (Chatterjee 1982; Parrish 1992). In the
primitive mesotarsal (PM) ankle a double articulation,
two pairs of pegs and sockets, exists between the two
proximal elements with an upper peg on the calcaneum
and a lower peg on the astragalus, separated by a
perforating foramen. In the 'modified primitive
mesotarsal' (MPM) tarsus the perforating foramen
between the astragalus and calcaneum has been lost
and the articulation surface between the two proximals
is relatively smaller and less sculptured than in PM,
CN and CR forms. In the AM ankle the astragalus is
very much enlarged at the expense of the calcaneum
and has an ascending process that locks the tibia; the
two proximal tarsi articulate with each other through
plain facets and are closely adpressed against each
other and against the tibia and fibula.
The different ankle constructs have been correlated
with various gait or posture 'grades' attributed to
archosaurian taxa: the PM ankle with a 'sprawling
posture', the CN and CR ankles with a 'semi-improved
posture' and AM ankles with a 'fully improved posture'
(Chatterjee, 1982).
Sereno disregarded the CN -CR nomenclature
claiming that the two ankle types are basically of the
same design, with a large ventral and a smaller dorsal
articulation between the two proximal elements, and
that they really only differ in the polarity of the ventral
astragalocalcanear articulation: in CN archosaurs this
articular surface of the astragalus is convex while it is
concave in CR archosaurs. But proximal tarsals in
many archosaurs are not readily comparable because
topographical details often differ widely and when
looking at the astragalocalcanear articulation, Sereno's
view does not seem to receive more support than the
'traditional' CN-CR one from the many excellent
drawings made of archosaur tarsi. Indeed it appears,
looking only at intratarsal articulation morphology,
one can choose to see the tarsus in omithosuchids as
having been derived from a 'CN' ancestor, as would be
a consequence of Sereno's phylogeny, or as having
evolved from a mesotarsal ankle (MPM) type, as hinted
at by Cruickshank and Benton (1985), equally well
were it not for the fact that both the CN and CR ankles
are both of the rotary, crurotarsal type where the
astragalus is functionally part of the crus. Sereno also
cited other tarsal characters in favour of derivation of
"CR" archosaurs from a 'CN' ancestor: (88)
'hemicylindrical calcaneal condyle for articulation with
fibula', (89) 'astragalus with flexed tibial articular
surface' (108 of Gauthier), (90) 'robust calcaneal tuber
with shaft wider than high', and (91) 'calcaneal tuber
with flared distal end'; all easily distinguishable
characters which lend support to a monophyletic
Crurotarsi (inclusive of Omithosuchidae), as defined
by Sereno.
In addition to tarsal morphology Gauthier and
Benton, but not Sereno, cited 'cervical ribs short and
stout' (82) as synapomorphic for crocodile line
archosaurs. This character is present in parasuchians
(Chatterjee 1978; Westphal 1976), Gracilisuchus
(Romer, 1972d), Stagonolepis (Walker, 1961),
Postosuchus (Chatterjee, 1985) and less extreme in
Ticinosuchus (Krebs, 1965) and Ornithosuchus
(Walker, 1964). In contrast cervical ribs are slender in
Euparkeria, Chanaresuchus, long and slender in
Lagerpeton, Herrerasaurus (Sereno, 1993; Figure 16)
Coelophysis (Colbert, 1989), Anchisaurus and in
Lesothosaurus diagnostic us the cervical ribs are short,
but slender (Weishampel and Witmer, 1990a). For
character (84) 'deltopectoral crest extends less than
one quarter of the way down the humerus shaft' in
Benton's list at node six I refer to the treatment of
character (31) in the previous section. Sereno cited
three characters with reference to the ap~ndages, (85)
' proximal humerus strongly arched under inner
tuberosity', (86) 'anterior trochanter of fibula robust
and knobshaped' and (87) 'distal end of fibula wider
than proximal end' in support of his Crurotarsi. These
claimed synapomorphies are hardly valid for reasons
pointed out by Parrish (993) and I refer to him for
critique of these characters.
Moving to the next node in the crocodile-line clade
Gauthier, Benton and Sereno (node six, seven and
seven respectively) all agree to leave Parasuchia
branching off as the basal-most taxon member. The
former two authors both believed that the absence of
the septomaxilla (94) is synapomorphic for the taxa
included at this node. Sereno, in contrast, saw this
condition as being plesiomorphic for crown group
archosaurs as Parasuchia is the only one of these taxa
possessing a septomaxilla; thus he considered the
presence of a septomaxilla a parasuchian apomorphy
and denied the possibility that the feature, because of
its singular morphology in parasuchians, is homologous
with the septomaxilla in primitive diapsids (Sereno
p. 16). Let it be noted here that no septomaxilla was
reported for Euparkeria (Ewer, 1965),Proterochampsa
(Sill, 1967), Gualosuchus and Chanaresuchus (Romer,
1971c) and that septomaxillae are possibly present in
Proterosuchus (Cruickshank, 1972); the data on the
distribution of this character are here considered too
sparse for them to be included in the analysis below.
Postparietal morphology (95) will be treated in detail
below. Where only adult specimens are available
character (96) 'fusion of second intercentrum and first
centrum in juvenile or earlier stages of ontogeny',
listed by Gauthier, of course cannot be assessed with
confidence, and 'triradiate pelvis' (97) is here
considered too imprecise to have diagnostic meaning;
furthermore the pelves of e.g. Chanaresuchus (Romer,
1972c) and Cuyosuchus (Erythrosuchidae) (Charig,
1976a; Figure 9c) may also very well be considered
' triradiate' . Character (98) "screw joint' tibio'astragalar
articulation', also by Gauthier, was considered
imprecise by Sereno and he rephrased it ' astragalus
with flexed tibial articular surface' (89), so that it also
was diagnostic of Parasuchia, considering the form of
the tibio-astragalar joint in this taxon to be basically
similar to that of higher crocodile line archosaurs; this
seems in concordance with the osteological evidence
(e.g. Parrish, 1986). Characters (99) 'fully developed
crocodile-normal crurotarsaljoint' and (107) 'advanced
crocodile-normal tarsus', by respectively Gauthier and
Benton, do not refer to any specific condition found in
the members of this clade not found in Parasuchia, but
rather mirror the fact that tarsal characters already
w,
9
present in Parasuchia are in some sense proliferated or
accentuated in higher crocodile-line archosaurs; e.g.
the calcaneal tuber is further posteriorly directed, the
astragalar peg for articulation with the calcaneum more
prominent and the proliferation of the proximal
articulation surfaces of the astragalus more pronounced.
Gauthier also 'noted osteoderms on ventral surface
of tail ' (100) for this clade; such dermal ossifications
seem to be present in only the members Stagonolepidae,
Saurosuchus (Bonaparte, 1981) and Ticinosuchus, as
noted by Sereno; the extent to which this character is
distributed in Rauisuchidae cannot be fully appreciated
as many members of this family are very poorly
preserved. In any case ventral dermal armour is also
present in some parasuchians (Gregory, 1962) and it
may be a primitive feature for crocodile-line archosaurs.
Benton's 'lower temporal fenestra is reduced in size
and triangular in shape, with a dorsal point' (101) is
also quoted by Sereno, but in a slightly rephrased form:
'postorbital-squamosal temporal bar anteroposteriorly
short with subtriangular laterotemporal fenestra' (10).
Even though 'reduced in size' is somewhat imprecise
Benton's phrasing of the character is preferable, if
applied to a clade of crocodile-line archosaurs exclusive
of Parasuchia and Omithosuchidae (in the latter taxon
the laterotemporal fenestra is neither reduced nor
particularly triangular in shape), as a (sub)triangular
lower temporal fenestra with a dorsal point is present in
Gracilisuchus (Romer, 1972d), Stagonolepidae
(Walker, 1961), Postosuchus (Chatterjee, 1985) and
Saurosuchus (Bonaparte, 1981), and because the
postorbital-squamosal temporal bar in the rauisuchians
Luperosuchus and possibly Heptasuchus (Bonaparte,
1984) does not seem to have been any shorter than that
of omithosuchids (Bonaparte, 1971; Walker, 1964). In
addition (01) and (10) show a lot of variation within
terminal taxa, e.g. Prestosuchus, Pseudhesperosuchus
andLotosaurus (Parrish, 1993) do not exhibit a reduced
laterotemporal fenestra and nor, as in Omithosuchidae,
is it triangular in shape in the latter two taxa, and can
only safely be said to characterise Stagonolepidae,
Gracilisuchus and Postosuchus. It is doubtful if
character (102) 'axial diapophysis is reduced or absent'
is diagnostic at this level as axial diapophyses are only
very weakly developed in Euparkeria (Ewer, 1965)
and Chanaresuchus (Romer, 1972c), and Sereno did
not use this character at any level in his cladogram.
Neither character (03) "no pubo-ischiadic plate, and
much reduced contact between pubis and ischium" will
here be regarded synapomorphic at node seven in
Benton's cladogram; broad puboischiadic contact, with
a plate-like development present, exists inStagonolepis
of approximately the same extent as that seen in the
parasuchids Parasuchus and Rutiodon (Westphal,
1976). In the rauisuchid Ticinosuchus (Krebs, 1965)
the puboischiadic symphysis, though not as prominent
as the one found in Stagonolepis, is at least of the same
length as those in Euparkeria and Chanaresuchus.
Benton acknowledged that several of the characters he
included in the diagnosis of this clade are paralleled in
10
his Ornithosuchia: (103), (104) 'pubis is long and
narrow and subvertically oriented', (lOSa) 'pubis is
longer than ischium' and (108) 'digit V of the foot is
reduced (shorter than I)'. These characters are
equivocally distributed on his cladogram as pointed
out by Sereno. This is not true for the parallelisms (97)
'triradiate pelvis', (98) 'screw joint..' between the two
clades listed by Gauthier because he included
Euparkeria as the most basal member of ' his
Ornithosuchia; this, however, causes other serious
. problems as acknowledged by Gauthier and described
above.
Character (104) 'pubis is long and narrow and
subvertically oriented' does describe a pubic structure
different from that found in Euparkeria, Chanare
suchus and parasuchians, and is valid as such, but, like
characters (33, 61 and 98), it may be too loosely
defined. In contrast (1 OSa) 'pubis is longer than ischium'
refers to a more exact pelvic condition which clearly
distinguishes itself from the pubo ishiadic allometry of
the just mentioned three taxa; character (10Sa) is valid
for all non-crocodylomorph crocodile-line archosaurs
(except for some rauisuchids and, of course, Parasuchia)
where the relevant elements are preserved. After
examining the prosauropod femurs in the collection of
the Bernard Price Institute I would refrain from using
character (106) 'proximal head of femur is turned
inwards at about 4So' as post-mortem distortion of the
femur seems to be common and obscures the true
inflection of the proximal head of this bone; thus
creating serious uncertainty with regard to (106) where
only a single or very few specimens are preserved.
'Digit V of the foot is reduced (shorter than I), (108) is
also present in Chanaresuchus, and ornithodirans, see
e.g. (Romer, 1971b) (Reig, 1963); in Euparkeria these
digits are about the same length. If rephrased 'digit V
shorter than I' the character will support the monophyly
of Proterochampsidae + crown group archosaurs. The
rephrasing is necessary because digit V of some higher
crocodile-line archosaurs is not reduced compared to
the same digit in Euparkeria, as pointed out by Sereno
(p. 12).
Sereno's definition of this clade differs from those
of Benton and Gauthier mainly by the inclusion of
Ornithosuchidae. The latter two authors believed this
family to be the sister taxon of Ornithodira. By using
parts of Benton's and Gauthier's characters (10Sa) and
(116) 'length of pubis exceeds three times width of
acetabulum' respectively Sereno arrived at a character,
'pubis is longer than ischium and at least three times
anteroposterior diameter of acetabulum' (10Sb), which
is not paralleled in basal ornithodirans, such as
Lagosuchus (Sereno; Figure 2S), and, hence, is
synapomorphic of his Suchia (except for Stagonoiepis,
some advanced crocodilians, as no!ed by the author
himself, and some rauisuchids) + Ornithosuchidae, and
also of higher dinosauriforms. In the latter group the
character is present in Dilophosaurus (Welles, 1984;
Figs. 30, 31), Coeiophysis (Colbert, 1989; Figure 77),
Lesothosaurus (Sereno, 1991a; Figure 10), in one
specimen of Sellosaurus (SMNS 17928) (Galton, 1984;
Figure Ie) and Herrerasaurus (Novas, 1993; Figure
S), but not inStaurikosaurus (Galton, 1977; Figure 2b)
nor quite in Lagosuchus (Sereno Figure 2S). Character
(10Sb) is thus synapomorphic for at least Eudinosauria
among bird-line archosaurs and character (lOSa) for at
least Lagosuchus and all other Dinosauriformes (the
terms Dinosauriformes and Eudinosauria are used here
in accordance with Novas, 1992; see appendix 1).
Character (109) 'posterior pubic acetabular margin
recessed' seems to be present only in Ornithosuchidae,
Postosuchus, possibly in some stagonolepids, and
among crocodylomorphs in at least Terrestrisuchus;
this character was concordantly only used by Sereno as
a tentative synapomorphy at node seven in his
cladogram.
Sereno did not try to resolve the relationships of his
Suchia and proposed only one synapomorphy (110)
"postorbital-squamosal bar short with subtriangular
laterotemporal fenestra', which has been discussed
above, for this taxon.
Benton, in contrast, had Gracilisuchus the next taxon
branching off from the crocodile line and quoted several
synapomorphies for the remaining taxa (node eight):
(96) and (100) have been treated above; the question of
the presence/absence of postparietals (9S) have also
been touched upon before, but will be dealt with in
more detail here: In Parasuchia taxa possessing a
postparietal do occur (Westphal, 1976) and, as noted
above, in Stagonolepidae a postparietal is present in
Typothorax, however, its presence could not be
established for Stagonolepis because of incomplete
preservation of the occiput (Walker, 1961).
Gracilisuchus retains the bone (Romer, 1972d), but it
appears to be absent inPostosuchus, (Chatterjee, 1985),
crocodylomorphs; e.g. Terrestrisuchus (Crush, 1984)
and Pseudhesperosuchus (Bonaparte, 1971), and in
ornithosuchids (Walker, 1964; Bonaparte, 1971).
Among rauisuchids Prestosuchus (Barberena, 1978)
possesses a postparietal, but for other members of the
taxon the state of character (9S) is difficult to ascertain
because of poorl y preserved or even missing hindskulls;
this is true for e.g. Teratosaurus (Benton, 1986),
Rauisuchus (Ruene, 1935-1942), Luperosuchus
(Romer, 1971a), Fasoiasuchus (Bonaparte, 1981),
'Mandasuchus' (fragmentary maxillae and a fragment
of the right dentary were the only skull material
recovered for this animal; Charig, 19S6) and also
Ticinosuchus; even though the type specimen is always
figured as a complete skeleton, only the maxilla and
dentary have been prepared and the rest of the skull is
_ reconstructed with Euparkeria as model! (Krebs, 1965).
Complete skull material exists for three taxa:
Heptasuchus clarki (Dawley et ai., 1979),Prestosuchus
chiniquensis (Barberena, 1978) and Saurosuchus gaiiiei
(Bonaparte, 1981). A satisfactory description of the
latter specimen has never been given; Sill (1974) gave
a detailed account of a fragmentary specimen of this
species and Dawley et ai. only supplied a preliminary
description of Heptasuchus.
According to Benton character (111), 'Pit between
basioccipital and basisphenoid', corresponds to
' foramen intertypanicum' of living crocodilians and
the author cites J. M. Clark's unpublished Ph.D. thesis
in support of the character. The term ' foramen
intertypanicum' does not appear to be otherwise used
in descriptions of archosaurs, nor is it used by Romer
(1946; 1956), Romer and Parsons (1.9.8,6) or Carroll
(1987). In Stagonolepis an excavation or a fossa is
present between the basioccipital and the basisphenoid
(Walker, 1961; Figure 5), this is also the case in
Postosuchus (Chatterjee, 1985; Figure 7). Whether or
not any of these features are the foramen intertypanicum
of modern crocodiles I am unable to say. In
Pseudhesperosuchus the basicranium is not preserved
(Bonaparte, 1971) and in Terrestrisuchus no
basisphenoid could be identified (Crush, 1984), and
Crush did not mention if a fossa or pit could be judged
present at the suture between the basioccipital and the
basisphenoid from the appearance of the former
element. In Ornithosuchus the preservation state of the
braincase does not allow comparison with respect to
basioccipital - basisphenoid morphology, nor does
Bonaparte ' s description of the basicranium of
Riojasuchus (Bonaparte, 1971). In rauisuchids studies
of the relevant skull parts were not available. No
foramen intertympanicum or fossa is present in
Euparkeria, Parasuchus (Chatterjee , 1978),
Dilophosaurus (Welles, 1984 ),Lesothosaurus (Sereno,
1991a) or Sellosaurus (Galton and Bakker, 1985).
Of the taxa included at node eight in Benton's
cladogram Stagonolepis is not diagnosed by ' accessory
neural spine on caudal vertebrae' (112); the character
was not noted by Chatterjee (1985) for Postosuchus,
but is present in Terrestrisuchus (Crush, 1984),
probably in P seudhesperosuchus (Bonaparte, 1971;
Figure 27d) and a number ofrauisuchians: Ticinosuchus
(Krebs, 1965), 'Mandasuchus' (pers. obs.),
Prestosuchus loricatus and Rauisuchus triradentes
(Huene, 1935-1942). The character may also be present
in other rauisuchians, but this was either not noted or
the relevant axial material has been too incompletely
preserved.
The first of the characters (113) 'first (atlantal)
intercentrum much longer than wide' listed by Gauthier
as synapomorphic for his node seven is inaccurate; it is
not present in e.g. Ticinosuchus , Postosuchus or
Terrestrisuchus. Character (114) 'axial diapophysis
reduced or absent' is phrased identically to (102), but
has been given a separate number here because of the
imprecise meaning of 'reduced', a relative term which,
when used at different nodes compares the characters
to two different sets of outgroup taxa. However, the
axial diapophyses are just as reduced in Gracilisuchus
(Romer, 1972d) and Parasuchus (Chatterjee, 1978) as
in e.g. Postosuchus (Chatterjee, 1985) and equally
reduced in Gracilisuchus (Romer, 1972d), Ticinosuchus
and Riojasuchus (Bonaparte, 1971). Information on
the character is not available for Stagonolepis (Walker,
1961) and many rauisuchids.
11
'Enlarged, pneumatic, basipterygoid processes' (115)
is also a very dubious synapomorphy at this level.
Basisphenoid elements are missing in Terrestrisuchus
(Crush, 1984), P seudhesperosuchus (Bonaparte, 1971)
and rauisuchids, except for possibly Prestosuchus; only
in Postosuchus are large basipterygoid processes
present, somewhat larger than in Parasuchus,
Stagonolepis and Gracilisuchus. To what extent these
processes are pneumatic in the different taxa is
impossible to determine from descriptive studies.
Because of the extent of the missing data on character
(115) it will not be considered further here. Character
(116) 'length of pubis exceeds three times width of
acetabulum' turns out to be identical to (105b) for all
practical purposes as the length of the pubis is never
exactly three times the width of the acetabulum.
Gauthier also listed 'fewer than four phalanges in
pedal digit five' (117) as being synapomorphic at node
seven in his cladogram; this character also diagnoses
Euparkeria (Ewer, 1965), Chanaresuchus (Romer,
1972c) and Riojasuchus (Bonaparte, 1971).
In contrast to Gauthier who placed Rauisuchidae
and Stagonolepidae (Aetosauria) as successive
outgroups to Crocodylomorpha, Benton saw the two
former taxa as constituting a monophyletic sister group
(his Pseudosuchia) to crocodylomorphs and listed five
synapomorphies (118-122) in support pf monophyletic
Rauisuchidae plus Stagonolepidae: Character (118)
'dorsal centra very constricted in ventral view' is
verified by literature for Rauisuchus, Prestosuchus
chiniquensis (Huene, 1935-1942), Saurosuchus and
Stagonolepis, but in addition diagnoses, at least,
Postosuchus and Ornithosuchus as well; for many
other taxa description does not permit secure
establishment of the presence/absence of the condition
specified by (118). Character (119) 'acetabulum is
subhorizontal and faces downwards, giving a 'pillar
like' erect posture of the hindlimb' can only be said to
be semi-present inStagonolepis (Walker, 1961, Figure
17) and Ticinosuchus as the tilt of the acetabulum
towards the horizontal is only slight in these animals;
in contrast the acetabular portion of the ilium faces
fully downwards in 'advanced' rauisuchids such as
Saurosuchus (Bonaparte, 1984). 'Iliac blade is oriented
subhorizontally' (120) is diagnostic at this node even
though the ilium of Stagonosuchus does not conform
to this specification; however, the vertical orientation
of the iliac blade in this animal could be interpreted as
a reversal because Stagonosuchus displays a number
of advanced rauisuchid characters (e.g. ventrally facing
acetabulum and three sacral vertebrae). Character (121),
'iliac blade is long and low', does not apply to
Stagonolepis any more than to Euparkeria or some
parasuchians (Westphal, 1976); however a number of
rauisuchids display a distinctive looking low iliac blade
in which the posterior process is very elongate. Hence
a version of (121) , exact, morphometrically defined,
may prove a rauisuchid synapomorphy. Character (122)
'pubis attaches to anteroventral face on the ilium' is
found in many archosaurs other than rauisuchids and
12
stagonolepids, e.g. Ornithosuchus, Euparkeria and
Chanaresuchus.
Benton also diagnosed a monophyletic Rauisuchidae
using Ticinosuchus and Luperosuchus as representatives
of this family. The rauisuchids were never a clearly
delimited group as e.g . the stagonolepids,
omithosuchids or parasuchians. This is in part due to
the fact that many rauisuchid genera are representedby
very fragmentary fossil material only, and partly due
to the use of primitive - or otherwise invalid characters
in the diagnoses of Rauisuchidae and Rauisuchia; e.g.
two of four diagnostics listed by Chatterj ee and
Majumdar (1987) for Rauisuchia 'teeth robust recurved,
laterally compressed with serrated edges' and 'pubis
and ischium elongate with long median symphysis'
are clearly also characteristic of other archosaurs; the
former trait is found in Riojasuchus, Herrerasaurus,
many theropods and erythrosuchids, and both characters
are present in Ornithosuchus. The same author also
listed 'maxillary teeth less than 12' , 'short cervicals' ,
'closed acetabulum', ' lack of foot in pubis' and
'quadrupedal pose' as synapomorphies of Rauisuchidae
within Rauisuchia; none of these are valid: the tooth
count in omithosuchids and Stagonolepis is also less
than 12; the cervicals in Parasuchia and omithosuchids
are of the same length as those of Ticinosuchus and
'Mandasuchus', and 'closed acetabulum' , 'lack of foot
in pubis' and 'quadrupedal pose' are found in numerous
other archosaurs.
Benton in comparison, gave a number of potential
synapomorphies for Rauisuchidae; these will here be
discussed in relation to the genera: Rauisuchus,
Saurosuchus, Ticinosuchus, Luperosuchus,
Prestosuchus, Stagonosu chus , Hepta suchus,
Tikisuchus, Fasolasuchus and Teratosuchus and the
poposaurids Poposaurus and Postosuchus . Other
nominal or potential rauisuchids and poposaurids are
Basutodon (Kitching and Raath, 1984), Lotosaurus
(Parrish, 1993), (the 'Kupferzell rauisuchid' ; SMNS
unnumbered) and (following Benton, 1986 and
Chatterjee, 1985; the '?' marks are the former author 's)
?Wangisuchus, ?Fenhosuchus, ?Arizonasaurus (listed
as an erythrosuchid by Charig and Sues, 1976),
Anisodontosaurus, ?Procerosuchus, and a number of
unnamed or undescribed forms from England, Germany,
America, Canada, India and South Africa. These latter
taxa are mostly omitted or only briefly mentioned in
the following discussion because the fossil material is
fragmentary and only displays characteristics also found
in most other lower archosaurs; in some cases the
descriptions of these forms were unavailable to me and
some of the taxa are as yet undescribed. In addition
three rauisuchids has been described from the Russian
region: Vjushkovisaursus (Ochev, 1982), Vytshegdo
such us and Dongusuchus (S ennikov, 1988) .
Vjushkovisaursus is represented by a small fragment
of a pterygoid, some vertebrae and a humerus. The
pterygoid fragment and vertebrae are not diagnostic;
the humerus is strongly arched under its inner tuberosity
(85) and could belong to a rauisuchid or a stagonolepid
equally well. Vytshegdosuchus is also represented by
very fragmentary remains only. The most interesting
of these is part of an ilium which has a small anterior
process, relatively low blade (incomplete posteriorly)
and a supraacetabular crest; this combination of features
makes it likely that Vytshegdosuchus is a rauisuchian.
The Dongusuchus material consists of miscellaneous
limb bones that could belong to a number different
archosaurian taxa.
'Mandasuchus' appears to be closely related to
Prestosuchus (Charig, 1956) and Ticinosuchus on
account of the similarity of the dorsal osteoderms,
which are of almost exactly the same shape, in these
genera; furthermore the ilium is slightly horizontally
inclined in both 'Mandasuchus' and Ticinosuchus
(Charig pers. comm.).
Character (123) 'extra slit-like fenestra between
maxilla and premaxilla' is present in Saurosuchus, '
Luperosuchus (Romer, 1971a), Heptasuchus,
Tikisuchus, Teratosaurus and Postosuchus, but not in
Prestosuchus; for many other taxa data regarding this
feature are missing. The character may also be present
in lushatyria (a taxon of uncertain affinity),
(Kalandadze and Sennikov, 1985) and possibly
Euparkeria (the material available to the present author
did not allow secure establishment of whether the
feature is present in this taxon); a fenestra between the
premaxilla and maxilla is also present in Shansisuchus
(Erythrosuchidae) (Parrish, 1992), Erythrosuchus
africanus (BP/1/5207; most clearly seen on the left
aspect of the specimen) contra Parrish, (1992) and
Chalishevia (a probable erythrosuchid) (Ochev, 1980).
Though the feature is figured as being subcircular
rather than slit-like for these taxa at least one specimen
d' Erythrosuchus (BP/1/5207) has a dorsoventrally
elongated fenestra between the premaxilla and maxilla.
A fenestra or foramen also occurs between the
premaxilla and maxilla of Herrerasaurus (Sereno et
al., 1993), many carnosaurs (Molnar et al. 1990),
Diplodocus (Mcintosh, 1990), and amongst
pro sauropods in at least Massospondylus (e.g. BP/1/
5241) andPlateosaurus (SMNS 13200) (Juul in prep.).
A subnarial foramen, of about the same relative size as
that found in H errerasaurus appears to be present in
Proterosuchus, and Mesosuchus, but not in other
rhynchosaurs (David Dilkes, pers. comm.).
Synapomorphy (124) 'movable joint between the
maxilla and premaxilla' appear to be positively present
in at least Tikisuchus (Chatterjee and Majumdar, 1987)
Teratosaurus, and (125) 'main antorbital fenestra is
low in front' is here considered a too vaguely phrased
~haracter. In contrast (126) 'tall orbit with a 'stepped'
postorbital/jugal bar behind' is better defined and
supported for Saurosuchus, H eptasuchus, Prestosuchus
and Tikisuchus, but, as (123), diagnose Postosuchus in
addition to rauisuchids. Character (127) 'lacrimal forms
a heavy ridge over the orbit' is clearly invalid as the
lacrimal is not part of the roof above the orbit in any of
taxa for which good skull material has been recovered.
The presence or non-presence of 'Proximal distance
between the ischia is less than that between the pubes'
(128) is extremely difficult to establish for most
rauisuchids and outgroup taxa from literature; however,
(1 28) is definitely present in Saurosuchus (Bonaparte,
1981; Figure 24), but also in Stagonolepis (Walker,
pers. comm). "Pubis is shorter than the ischium' (129)
is present in at least Saurosuchus and Ticinosuchus
(among rauisuchids), non-crown group-ru:chosaurs and
Parasuchia, but not in Gracilisuchus, Postosuchus,
Ornithosuchus or Lagosuchus.
Recently Parrish (1993) has published a phylogeny
of crocodylotarsan archosaurs in which he separates
Prestosuchus, Ticinosuchus and Saurosuchus from the
' traditional' Rauisuchidae into a family of their own,
the Prestosuchidae; a similar idea was proposed by
Charig (1956). It is not surprising if Rauisuchidae (as
generally perceived), being so vaguely characterised,
is a paraphyletic assemblage. However, the feeling
here is that it is a serious shortcoming of Parrish'
(1993) analysis that a number of relevant characters,
listed above, were not employed; e.g. (126) .. "stepped"
postorbital/jugal bar, (119) orientation of the
acetabulum, etc. Furthermore in some instances there
is no correspondence between his character list (table
1) and character state matrix (table 2), e.g. character 11
is given as a binary character in the character list, but 3
states are indicated in the matrix for the same character.
Thus the matrix is difficult to test and lacking important
characters. Nonetheless three characters (checked
against original descriptive work) from Parrish (1993)
are incorporated in the present analysis and in deference
to this author his Prestosuchidae (and not 'other'
rauisuchids) is employed as they incorporate the better
described, often more complete taxa.
Benton also saw Postosuchus, Terrestrisuchus,
Protosuchus and Crocodylia (the latter three taxa
forming a restricted Crocodylomorpha) constituting a
monophyletic group. Two of the characters, (115)
' enlarged, pneumatic, basipterygoid processes' and
(116) 'length of pubis exceeds three times width of
acetabulum', cited in support of this grouping have
been discussed previously; (115) is not substantiated
by literature for Postosuchus, Terrestrisuchus and
Protosuchus. In contrast (116) is present in the three
taxa just mentioned and in Saurosuchus (Bonaparte,
1981; Figure 24A), Ornithosuchus and some
dinosaurs; among rauisuchids at least Ticinosuchus
does not display the condition specified by this
character, but Saurosuchus does. Character (130)
'posterior border of lower temporal fenestra is not
bowed' is a reversal of (11); again the phrasing of this
character is considered too vague here . 'Short
descending process of squamosal and tall
quadratojugal that contacts the postorbital' (131) is
present in Postosuchus, Protosuchus (Crompton and
Smith, 1980), Hemiprotosuchus (Bonaparte, 1971)
all Orthosuchus (Walker, 1970), but no
quadratojugal-postorbital contact exists in
Terrestrisuchus (Crush, 1984), Sphenosuchus or
Crocodylus (Walker, 1970). Character (132)
",
13
'maxillary-vomer secondary palate' is dubious as
synapomorphy at this level: after examining
Postosuchus material Parrish (1991) stated that the
nature of the maxillary-vomer relationship in this
animal could not be determined. Otherwise (132) is
slightly developed in Terrestrisuchus (Crush, 1984)
and extensively in Eusuchia (Romer, 1956). However,
a secondary palate is not present in
Pseudhesperosuchus (Bonaparte, 1971; Figure 24),
and in Protosuchus this feature is rudimentary formed
by either vomers or palatines; no maxillary
participation was mentioned by Crompton and Smith
(1980). Character (132) is also present in
Chanaresuchus (as noted by Benton); in Protero
champsa a secondary palate is also developed, but
without participation of the vomers (Sill, 1967).
Clavicles (133), 'reduction or loss of clavicle', were
not described for Terrestrisuchus (Crush, 1984),
Protosuchus (Colbert and Mook, 1951),
Pseudhesperosuchus (Bonaparte, 1971) (all
crocodylomorphs) or Postosuchus (Chatterjee, 1985),
but are definitely present in Ticinosuchus (Krebs, 1965),
Stagonolepis (Walker 1961), Ornithosuchus (Walker,
1964), Euparkeria (Ewer, 1965) and Proterosuchus
(Cruickshank, 1972). Character (134) 'forelimb:
hindlimb ratio is about 0.5' is, like character (132), not
a very convincing synapomorphy at . this level: the
forelimb: hindlimb ratio of Postosuchus (Chatterjee,
1985) is approximately 0.4, in Gracilisuchus (Romer,
1972d) it is 0.54, in Euparkeria (Ewer, 1965) about
0.62, in Ticinosuchus 0.66 (Krebs, 1965) and ranges
from 0.63 to 0.67 in crocodylomorphs (all measure
ments are taken from skeletal reconstructions or
calculated from figures given in the respective
descriptions).
'Acetabulum perforated' (135) is, apart from being
paralleled in ornithosuchians, unproblematic ally
apomorphic at node nine in Benton's cladogram. A
'supra'acetabular crest on ilium' (136) appears to be
possessed by Saurosuchus, Poposaurids,
Terrestrisuchus, Protosuchus, Ornithosuchus (Walker,
pers. comm), Lagerpeton (Sereno and Arcucci, 1993),
Lagosuchus and dinosaurs. No ilium is preserved for
Pseudhesperosuchus (Bonaparte, 1971). In basal
pterosaurs the pelvis is too transformed for valid
comparison; here the upper border of the acetabulum is
formed by the extremely low iliac blade and it is
unclear if a supra-acetabular crest proper is present.
'Pedal digit V has no phalanges' (137) was also listed
by Benton in support of a monophyletic
Crocodylomorpha plus Postosuchus; this character is
present in Postosuchus, Protosuchus, higher
crocodylomorphs (e.g. Crocodylus and Stenosaurus),
Chanaresuchus (Proterochampsidae), Lagerpeton,
Lagosuchus (Romer, 1971b) and the early dinosaurs
Coelophysis (Colbert, 1989) and Dilophosaurus
(Welles, 1984), but not in Sauropodomorpha,
Herrerasaurus (Reig, 1963; Sereno and Novas 1992;
Figure 2, Novas, 1993) and pterosaurs where a fifth toe
is retained; the pes is missing in Staurikosaurus
14
(Colbert, 1970) and the phalangeal formula for the
fifth digit is unknown in basal ornithischians
(Weishampel and Witmer, 1990a). It is problematic to
attribute stances or gaits to fossils and character (138)
'stance is digitigrade' relies too heavily on conjectural
evidence for use in cladistic analysis; it would of
course be ridiculous not to assume that the elongate
and closely appressed metatarsals of e.g. Terrestri
suchus and Lagosuchus implies some sort of digital
stance in vivo, but it is less straightforward to interpret
taxa like Pseudhesperosuchus and Omithosuchidae in
this respect.
Non-dinosaur bird-line archosaurs
Here are included the basal members of the other
main lineage of archosaurs; the lineage giving rise to
dinosaurs and, eventually, birds.
Their evolutionary success has been contributed to,
among other things, the erect posture, parasaggital gait
and hindlimb morphometrics of early bird-line
archosaurs which enabled these to function as high
speed cursors; the key morphological acquisitions being
deep, perforated acetabulum, intumed, offset femoral
head, prominent fourth trochanter, elongate epipodials
and metatarsus, and an advanced mesotarsal ankle.
Naturally these and similar characters have been widely
used to diagnose bird-line archosaurs in phylogenetic
studies; not all equally well defined as will become
apparent below.
'Ornithosuchia'
As mentioned previously Gauthier, in contrast to
Benton, chose to include Euparkeria in Omithosuchia
(node eight), however, the characters cited in favour of
this arrangement are not impressive: ' squamosal
reduced and descending ramus gracile' (139); in
Shansisuchus, Vjushkovia (Erythrosuchidae) (Parrish,
1992; Figure 7 and 9) and Luperosuchus the squamosal
does not seem relatively larger nor does its descending
ramus appear less gracile than in Riojasuchus.
Furthermore the squamosal looks very dissimilar in
e.g. Euparkeria, Riojasuchus and Herrerasaurus, and
in the latter genus its descending ramus can hardly be
called gracile (Sereno and Novas, 1992, Figure Ib).
The meaning of character (140) 'centra steeply inclined
in at least the first four postatlantal cervicals' is probably
similar to that oJ (209), see below, and 'modifications
in the hindlimb and girdle correlated with semierect
gait' (141) clearly describes a character complex rather
than a single character and should have been listed
accordingly in order to be useful in cladistic analysis.
It is uncertain what the absent 'ventral flange' of the
astragalus (142) is, and character (143) has been
discussed above; a 'crocodile reversed ankle joint,
with peg on calcaneum and socket. on astragalus' is
only present in Omithosuchidae. Character (144) 'pedal
digit five with fewer than four phalanges' is also present
in Chanaresuchus as well as the taxa at node seven in
Gauthier's cladogram; thus this character requires three
steps when distributed on the cladogram whether it
originated once at node three or three times at nodes
three, seven and eight.
Benton listed numerous apomorphies in support of a
less inclusive Omithosuchia. (145, 146, 148-150) all
are paralleled in Suchia and have been evaluated above;
s9 have (139,142). Ofthe remaining characters 'manual
digit I is short and equipped with a divefging claw'
(147) is hard to test for Omithosuchidae andLagerpeton
as good manual material is lacking in these taxa,
however, ornithosuchids appear to possess a fIrst manual
digit that is relatively short compared to the three
middle digits, as do pterosaurs,Lesothosaurus (Sereno,
1991a), Herrerasaurus (Sereno and Novas, 1992),
Coelophysis (Colbert, 1989), Dilophosaurus (Welles,
1984) and prosauropods; only in the latter two taxa is a
diverging claw present on this digit. A lesser or anterior
trochanter (151) is only present in Omithosuchidae
(Walker, 1964; Bonaparte, 1971) Lagosuchus;
Pseudolagosuchus (Novas, 1992), dinosaurs, but not
in Lagerpeton (Sereno and Arcucci, 1994; Figure 2);
in pterosaurs no lesser trochanter was noted by
Wellnhofer (1978). Character (152) 'fourth trochanter
is a sharp flange' is distributed in the same manner as
(151) apart from also being present in Lagerpeton.
Character (153) 'shaft of femur is bowed dorsally' is
too imprecise and does not define a condition
characteristic of Omithosuchia only; the femurs of e.g.
Chanaresuchus and Ticinosuchus also conform to the
phrasing of (153). A prominent cnemial crest is present
on the tibiae (154) of Gracilisuchus, Omithosuchidae,
dinosaurs and probably Lagerpeton (Sereno; Figure
18C); pterosaurs also feature, as do many other
archosaurs, a cnemial crest on the tibia, but only a
weakly developed one.
Gauthier diagnosed a similar clade (his Omitho
suchia exclusive of Euparkeria), several of the synapo
morphies he listed in support of it are paralleled in
crocodile-line archosaurs or have been discussed already
in other connections: (155 , 156, 160 and 161). Of the
remaining characters (162) 'anterior trochanter on femur
appears early in post-hatching ontogeny' is clearly
dubious; the phrasing 'early post-hatching ontogeny'
is imprecise and ontogenetic series are not available
for most of the relevant taxa. Neither is 'coracoid
tubercle lies close to glenoid fossa and coracoid
foramen ' (157) accurate enough for the current purpose.
Also (164) 'fifth metatarsal gracile' is too vague;
furthermore the fifth metatarsal does not seem anymore
gracile in omithosuchids than in any other archosaurs,
e.g. Gracilisuchus and Terrestrisuchus. Character (158)
'first metacarpal with offset distal condyles, and pollex
- directed medially and bearing enlarged ungual' is two
rather than one character; the part of it which pertains
to ungual proportions has been dealt with above. The
other part of this character, first metacarpal with offset
distal condyles, is present in Omithosuchidae (Walker,
1964; Bonaparte, 1971), Herrerasaurus (Sereno and
Novas, 1992), Coelophysis (Colbert, 1989), many
pro sauropods and Lesothosaurus (Sereno, 1991a), but
does not seem to be so in pterosaurs or any other
archosaurs; manual material is missing in
Staurikosaurus (Colbert, 1970), Lagerpeton (see
Sereno; Figure 18C) and Gracilisuchus (considered an
omithosuchid by Romer, 1972d). Because digits are
very sparsely preserved in basal omithosuchians, even
more so than metacarpals, character (159), 'manus
more asymmetrical than in pseudosuchians, with inner
digits much larger than outer digits' , CaJl!!Qt be assessed
with confidence and (163) 'aliform fourth trochanter'
is here seen as just an alternative phrasing of
character (152).
Ornithodira
Gauthier, Benton and Sereno (nodes ten, eleven and
eight in respective cladograms) included nearly the
same taxa in their definition of the next avian clade,
Ornithodira. Within this clade Sereno placed pterosaurs
as the basal-most taxon member while Benton and
Gauthier remained inconclusive on the relative
systematic positions of Pterosauria and Lagosuchus
depicting the interrelationship of these taxa with
Dinosauria as an unresolved tricothomy on their
cladograms.
Gauthier, as the only one, listed two omithodiran
synapomorphies pertaining to the atlas-axis complex:
(166) 'atlantal intercentrum enlarged, completely
surrounding odontoid ventrally and laterally and fitting
into prominent recessed area below odontoid on axis'
and (167) 'axial intercentrum, and then odontoid, fuses
to axis at cessation of growth'. The former condition
seems to be present to the same extent in Riojasuchus
as it is in the pterosaur Rhamphorhynchus ; it is not
present in Dilophosaurus (Theropod a) and relevant
material is lacking in many taxa, e.g. Staurikosaurus
and basal omithischians. The latter character cannot be
properly tested in the fossil record; hence both (166)
and (167) are not included in the present analysis. The
postfrontal is absent (165) in proterochampsids,
Doswellia, crocodylomorphs, pterosaurs and dinosaurs
within Archosauria (again exact data were unobtainable
for basal dinosauromorphs). Although a bit vaguely
formulated (168) 'modification of cervical centra and
zygapophyses that combine to yield an S-shaped neck' ,
also listed by Gauthier, does describe a neck
morphology only found in ornithodirans among
archosaurs; notwithstanding that the neck curvature is
very weak in Lagerpeton, Lagosuchus and early
pterosaurs. Character (169) 'zygapophysial facets nearly
vertically disposed in all but proximal part of the tail'
by the same author, however, could not be verified and
in many theropods, e.g. Coelophysis and
Dilophosaurus, zygapophysial facets are almost
horizontally inclined. The absence of an interclavicle
(170) was regarded synapomorphic at this level by
Gauthier, Benton and Sereno; indeed this character is
well supported by descriptive literature (although the
character could not be verified for Lagerpeton,
Lagosuchus and Pseudolagosuchus by the present
author). The clavicle is purported to be 'reduced and
gracile' (171) and 'rudimentary or absent' (202) by
",
15
Gauthier and Sereno respectively; again this character
could not be tested here with regard to Lagerpeton,
Lagosuchus and Pseudolagosuchus, but clavicles are
not present in pterosaurs (Wellnhofer, 1978) or most
dinosaurs. These two workers also both listed characters
pertaining to the deltopectoral crest; (175) 'apex of
deltopectoral crest placed distally on humerus' and
(203) 'subrectangular deltopectoral crest'. The latter,
by Sereno, is preferable because it is more concise and
describes a condition easily recognisable in e .g.
Lagerpeton, Eudimorphodon, Herrerasaurus,
Dilophosaurus and Lesothosaurus. Two scapulocora
coid characters listed by Gauthier 'glenoid facet on
scapulacoracoid faces posteroventrally' (172) and
'coracoid small, with subcircular profile, and lying in
nearly the same plane as the scapula' (173) do not
seem to be unique to omithodirans. In contrast (176)
'less than five phalanges in manual digit four and less
than three phalanges in manual digit five' by the same
author are present in all ornithodirans where the relevant
digits are preserved, but it is uncertain whether it also
extends to Omithosuchidae. Character (174) 'Forelimbs
less than 55% of hindlimb length, and hindlimb very
long relative to length of trunk' (Gauthier) is difficult
to assess, the exact forelimb/hindlimb ratio cannot be
obtained for omithosuchids so it is uncertain if (174)
also characterises this family; in any case this ratio is
approximately .54 in Gracilisuchus, .55 inLagerpeton,
.52 in Lagosuchus, .51 in Scleromochlus, .48 in
Herrerasaurus, .41 in Coelophysis and plus .60 in
other archosaurs except pterosaurs in which the forelimb
is too transformed for comparison (most ratios were
calculated from measurements taken from skeletal
reconstructions and hence may be prove to contain
errors).
'At least three vertebrae involved in sacrum' (177)
is not a valid synapomorphy for Omithodira unless
one is willing to accept a reversal to the plesiomorphic
condition of two sacral vertebrae in Lagosuchus, and
probably Lagerpeton; the character is present in
Pterosauria, Scleromochlus (Krebs, 1976) and dinosaurs
(except maybe Staurikosaurus; see below). In contrast
to Gauthier who listed it synapomorphic at this level, a
more recent study (Novas, 1992) claims that a brevis
shelf on ventral surface of the postacetabular portion
of ilium (178) is present only in Saurischia (now also
including Herrerasauridae; see below (Sereno and
Novas, 1992)) and Omithischia; this is in concordance
with drawn reconstructions of omithodirans. A shelf
like projection is also present on the ilium of
P oposaurus (see Mehl, 1915b; Chatterjee, 1985; Figure
25), but it is hardly likely that this feature is homologous
with the brevis shelf in dinosaurs. Gauthier also cited
several characteristics of the hindlimb in support of
Ornithodira: (179) 'birdlike distal end of femur,
prominent anterior and posterior intercondylar grooves,
with the latter constricted by prominent external tibial
condyle ... ' is an extensive character complex of which
many of the sub-characters are found in omithosuchids
and rauisuchids. An anterior intercondylar groove is
16
absent in Lesothosaurus, basal 'thyreophorans', and
both anterior and posterior intercondylar grooves are
absent in heterodontosaurids; in some more advanced
ornithischians (e.g. psittacosaurids; Sereno, 1990)
where both these grooves are present it would appear
to be a case of parallelism. 'Tibia as long or longer
than femur' (180) is present inLagerpeton, Lagosuchus,
Pseudolagosuchus, Staurikosaurus (Colbert, 1970),
theropods, Lesothosaurus and Terrestrisuchus (Crush,
1984), but not in prosauropods and, peculiarly enough,
Herrerasaurus. Having cited (181) 'fibula thin and
strongly tapered distally and calcaneum reduced' it is
clearly redundant also to list (182) 'astragalus
transversely widened', and (183) astragalus and
calcaneum with smooth rollerlike articular surfaces
abutting against depressed distal tarsals' is a bit vague
and could be applied to lower archosaurs also, e.g.
Euparkeria. Further redundancy, as noted by Sereno
(p. 38), is present in listing both (184) 'metatarsals
elongate and closely appressed' and (185) 'pes
digitigrade'; characters similar to the former have been
listed by Benton: (200) 'metatarsals II-IV are closely
bunched as a unit' and Sereno 'compact metatarsus
with proximal third of metatarsals 1-4 shafts closely
appressed'. Benton's or Sereno's versions are superior
because they do not address the question of elongation
. of the metatarsus; this is done separately by Sereno in
character (208) 'metatarsal 2-4 elongate with metatarsal
3 more than 50 percent of tibial length'. Character
(200) and (208) seem to be autapomorphic for
Omithodira. In contrast there is much homoplasy in
Gauthier's 'pedal digit five reduced, does not exceed
length of metatarsal IV, and composed of no more than
two phalanges' (187). This character is present in
Chanaresuchus, Gracilisuchus, Postosuchus,
Crocodylomorpha and Omithodirans; in Pterosauria
the fifth pedal digit is highly variable in length and it is
uncertain what the original condition was in this group,
and because of incomplete preservation character (187)
cannot be securely assessed for Omithosuchidae either;
in addition the phrasing of the character or its absence
is not applicable to many taxa, e.g. Rhynchosauria and
Pro lacerta.
Absence of dermal armour has been construed, in
the light of other character evidence, to be the result of
a secondary loss in omithodirans (Gauthier; Sereno;
character 188); body armour, though, is present in
higher theropods and omithischians.
Benton listed some additional characters at this level:
(189) 'presacral vertebral column is divided into three
regions (cervical, cervical-thoracic, lumbar),; such
partitioning of the axial column could also be applied
to other archosaurs, e.g. ornithosuchids and crocodylo
morphs. 'Centra are steeply inclined in at least cervicals
3-6' (190) is as (140) not clear, but both probably refer
to the condition described by Sereno in synapomorphy
(209); see below. Character (191) 'zygapophyses of
the middle and posterior caudals are inclined pos
teroventrally' is neither supported nor contradicted by
descriptive work; . furthermore it is difficult to assess
its validity as there is often little more than a hyposphene
remaining posteriorly to the neural spine on the arches
ofthe relevant vertebrae. 'Fourth trochanter runs down
one third to one-half the length of the femur shaft'
(195) is not convincing: Benton noted himself that the
character is paralleled in Erythrosuchus and
Chanaresuchus, and. present in Omithosuchidae, but
not in pterosaurs. In addition this character seems to be
size related: the fourth trochanter is relatively more
distally positioned in large than in smaller specimens
of Euskelosaurus (luul in prep.). ' Knee articulates at
90° ' (196) is extremely difficult of assess as an
omithodiran synapomorphy. (197) 'Mesotarsal ankle
joint with astragalus and calcaneum fused to tibia'; in
Ceratosauria (Rowe and Gauthier, 1990) and
Heterodontosaurus (Omithischia) (Weishampel and
Witmer, 1990a) astragalus and calcaneum are fused
with the tibia, at least in adult specimens, however;
Romer noted that in Lagerpeton the proximal tarsals
are fused to each other, but not to the tibia. Benton's
notion that the omithodiran calcaneum has no tuber at
all (198) is contradicted by the presence of a rudimentary
tuber on the calcanae of Lagosuchus, Pseudolagosuchus
and Ischisaurus (Herrerasauridae) (Novas, 1989);
supporting Sereno's character (206) ' calcaneal tuber
rudimentary or absent'. 'Ascending process of
astragalus fits between tibia and fibula' (199) could
also describe the condition in Riojasuchus (compare
Sereno; Figure 7B,F,G with Novas, 1989; Figs. 2(5),
5(2,10)), furthermore the ascending process on the
astragalus of Lagerpeton is situated at the posterior
margin of the element (Sereno and Arcucci, 1993), not
the anterior as in other dinosauromorphs; thus a detailed
study is needed to assess whether the different ascending
processes of archosaur astragali are homologous. Indeed
in many other archosaurs there is a proximal astragalar
process with a tibial facet medially and fibular facet
laterally (see e.g. Sereno; Figure 6 and 8). Further
characters listed by Benton at his node nine have
already been reviewed above (e.g. parallelisms in
crocodile-line archosaurs).
Sereno also purported ' anterior cervical centrum
length longer than mid-dorsal length' (201), 'femoral
shaft bowed anteriorly along at least 80 percent of
femoral length' (204), 'posterior groove on astragalus
absent' (205) and 'distal tarsal 4 subequal in transverse
width to distal tarsal 3' (207) to be omithodiran
synapomorphies. Only the latter appear to be valid:
character (201) is present, to various degrees, in a lot
of archosaurs other than ornithodirans; e.g.
Ticinosuchus, Postosuchus, some crocodylomorphs and
, Gmcilisuchus. Furthermore it is not present in
Scleromochlus nor seems to be so in Lagerpeton. The
tarsal features of different archosaurs alleged to be
'posterior grooves' in character (205) are morpho
logically very different and it is questionable if they
are homologous (compare Sereno; Figs. 3-8) and
estimation of the presence of (204) is highly prone to
error as noted by Sereno himself. In contrast character
(207) appears to be autapomorphic for Omithodira.
D inosauromorpha
The long-limbed avian-clade archosaursLagerpeton,
Lagosuchus and Pseudolagosuchus are held to define
a monophyletic taxon together with dinosaurs by Novas
(1992) and Sereno. The latter worker gave a number of
characters diagnosing this Dinosauromorpha. At least
one autapomorphy was included: 'astragalus with acute
anteromedial comer' (211), present ~ Lagosuchus,
Pseudolagosuchus and dinosaurs (Novas, 1989); Sereno
also claims its presence in a specimen of Lagerpeton,
but no mention of Pterosauria was made in this
connection nor was I able determine the presence/non
presence of (211) in this taxon.
'Cervical column following strong sigmoid curve
with dorsal offset of the anterior face of the centrum
present as far posteriorly as the ninth or tenth presacral '
(209), 'forelimb 50 percent or less of hindlimb length '
(210) and 'distal tarsal 4 with articular surface for
metatarsal 5 limited to half of its lateral surface' (213)
are all disregarded here as dinosauromorph
synapomorphies: (209) is a very unreliable character
because of diffic