A new species of basal rhynchosaur (Diapsida:
Archosauromorpha) from the early Middle Triassic of
South Africa, and the early evolution of Rhynchosauria

RICHARD J. BUTLER1,2*, MARTÍN D. EZCURRA1,2, FELIPE C. MONTEFELTRO1,3,
ADUN SAMATHI2,4 and GABRIELA SOBRAL5

1School of Geography, Earth and Environmental Sciences, University of Birmingham, Edgbaston,
Birmingham B15 2TT, UK
2GeoBio-Center, Ludwig-Maximilians-Universität München, Richard-Wagner-Str. 10, 80333 Munich,
Germany
3Departamento de Biologia e Zootecnia, UNESP, Rua Monção 226, 15385-000, Ilha Solteira, Brazil
4Division of Paleontology, Steinmann Institute, University of Bonn, Nussallee 8, 53115, Bonn, Germany
5Museum für Naturkunde Berlin, Leibniz-Institut für Evolutions- und Biodiversitätsforschung,
Invalidenstraße 43, 10115 Berlin, Germany

Received 14 August 2014; revised 8 January 2015; accepted for publication 14 January 2015

Rhynchosauria was an important clade of herbivorous archosauromorph reptiles during the Triassic, with a world-
wide distribution. We describe a new genus and species of early rhynchosaur, Eohyosaurus wolvaardti gen. et
sp. nov., from the early Middle Triassic (early Anisian) Cynognathus Assemblage Zone (Subzone B) of the Karoo
Supergroup, South Africa. Eohyosaurus wolvaardti is known from a single skull, and is recovered as the sister
taxon of Rhynchosauridae in a new phylogenetic analysis. Cynognathus Subzone B has previously yielded the
stratigraphically oldest well-understood rhynchosaur species, Mesosuchus browni and Howesia browni. Eohyosaurus
wolvaardti increases the rhynchosaur diversity within this stratigraphical horizon to three species. Intriguingly,
all currently confirmed rhynchosaur occurrences from the Early Triassic to earliest Middle Triassic are from South
Africa. This may suggest a relatively restricted palaeogeographical distribution for early rhynchosaurs, followed
by a global dispersal of rhynchosaurids during the Middle Triassic.

© 2015 The Linnean Society of London, Zoological Journal of the Linnean Society, 2015, 174, 571–588.
doi: 10.1111/zoj.12246

ADDITIONAL KEYWORDS: Anisian – Burgersdorp Formation – Cynognathus assemblage zone – phylog-
eny – Rhynchosauridae.

INTRODUCTION

Rhynchosaurs were an important group of herbivo-
rous archosauromorph reptiles that played a key role
in many terrestrial ecosystems during the Triassic, par-
ticularly in Gondwana. The clade diversified during
the Middle Triassic and early Late Triassic, with abun-
dant remains known from Tanzania, South Africa,
Madagascar, India, Brazil, Argentina, the UK, and the
USA (Huene, 1938; Chatterjee, 1980; Benton, 1983a,

1990; Dilkes, 1995, 1998; Langer et al., 2000; Nesbitt
& Whatley, 2004; Montefeltro, Langer & Schultz, 2010;
Ezcurra et al., 2014; Montefeltro et al., 2013; Mukherjee
& Ray, 2014). The youngest stratigraphical occur-
rences of rhynchosaurs occur close to the Carnian–
Norian boundary (Langer et al., 2007; Martínez et al.,
2011), and the clade has been considered as a victim
of the possible Carnian–Norian extinction event (Benton,
1983b, 1990). Most rhynchosaurs (i.e. those belong-
ing to the clade Rhynchosauridae) were squat, pig-
like herbivores, with unusual skulls characterized by
bony beaks and complex tooth plates.*Corresponding author. E-mail: butler.richard.j@gmail.com

bs_bs_banner

Zoological Journal of the Linnean Society, 2015, 174, 571–588. With 6 figures

© 2015 The Linnean Society of London, Zoological Journal of the Linnean Society, 2015, 174, 571–588 571

D
ow

nloaded from
 https://academ

ic.oup.com
/zoolinnean/article-abstract/174/3/571/2453122 by U

niversidade Estadual Paulista J�
lio de M

esquita Filho user on 06 M
ay 2019

mailto:butler.richard.j@gmail.com


The earliest rhynchosaur, Noteosuchus colletti, is
known from the Lower Triassic Lystrosaurus Assem-
blage Zone of South Africa (Carroll, 1976; Dilkes, 1998;
Ezcurra, Scheyer & Butler, 2014), shortly after the
Permo-Triassic mass extinction. Two basal rhynchosaurs,
Howesia browni and Mesosuchus browni, are known
from the middle part (Subzone B) of the overlying
Cynognathus Assemblage Zone (AZ; Dilkes, 1995, 1998),
of early Middle Triassic (early Anisian) age. These taxa
are considered key for understanding the origin and
early evolution of rhynchosaurs, as many aspects
of their anatomy are intermediate between other
early archosauromorphs and more derived rhynchosaurs
(Rhynchosauridae). Mesosuchus browni, the
stratigraphically oldest well-understood rhynchosaur,
is recovered as the most basal rhynchosaur taxon in
many phylogenetic analyses (e.g. Dilkes, 1998; Hone
& Benton, 2008), and is represented by excellent cranial
and postcranial material. It lacks the well-developed
tooth plates that are typical of later rhynchosaurs. By
contrast, H. browni is known from relatively fragmen-
tary material and is considerably less well under-
stood, but does possess expanded and well-developed
tooth plates.

Here, we describe a third taxon of basal rhynchosaur
from Subzone B of the Cynognathus AZ of South
Africa (Fig. 1). This taxon provides additional
insights into the early evolutionary history of the
clade, including cranial evolution, diversity, and
biogeography.

METHODS
INSTITUTIONAL ABBREVIATIONS

BRSUG, University of Bristol, School of Earth Sci-
ences, Bristol, UK; FZB, Fundação Zoobotânica do Rio
Grande do Sul, Porto Alegre, Brazil; MACN-Pv, Museo
Argentino de Ciencias Naturales ‘Bernardino Rivadavia’,
Paleontología de Vertebrados, Buenos Aires, Argenti-
na; PVSJ, Museo de Ciencias Naturales, Universidad
de San Juan, San Juan, Argentina; NHMUK, Natural
History Museum, London, UK; SAM, Iziko South African
Museum, Cape Town, South Africa; SHYMS, Shrop-
shire Museums, Ludlow, UK; UFRGS, Instituto de
Geociências, Universidade Federal do Rio Grande do
Sul, Porto Alegre, Brazil; WARMS, Warwickshire
Museum, Warwick, UK.

COMPUTED TOMOGRAPHIC SCANNING

SAM-PK-K10159 was scanned in the Museum für
Naturkunde Berlin with a Phoenix|X-ray Nanotom (GE
Sensing and Inspection Technologies GmbH, Wunstorf,
Germany) using a tungsten target and a Cu filter
0.1 mm thick in modus 0, averaging 3, skip 2. The scan
comprised 1440 slices with an exposure time of 1 s,
85 kV, current of 400 μA, and a voxel size of 22.79 μm.
X-ray slices were reconstructed with the software
DATOS|X-RECONSTRUCTION v. 1.5.0.22 (GE Sensing
and Inspection Technologies GmbH, Phoenix|X-ray) and
analysed in VG STUDIO MAX 2.1 (Volume Graphics,

Figure 1. Geographical and stratigraphical setting of the type specimen of Eohyosaurus wolvaardti gen. et sp. nov.
A, map of South Africa and surrounding regions showing the location of the farm Lemoenfontein 44 (marked with a
star); B, simplified stratigraphy of the upper portion of the Beaufort Group (stratigraphical level of E. wolvaardti marked
with a star). Abbreviations: CHANGH., Changhsingian; P, Permian.

572 R. J. BUTLER ET AL.

© 2015 The Linnean Society of London, Zoological Journal of the Linnean Society, 2015, 174, 571–588

D
ow

nloaded from
 https://academ

ic.oup.com
/zoolinnean/article-abstract/174/3/571/2453122 by U

niversidade Estadual Paulista J�
lio de M

esquita Filho user on 06 M
ay 2019



Heidelberg, Germany). The specimen was too large to
be scanned in its entirety, so scans focused on areas
of particular interest (tooth rows, endocranium). The
endocranial anatomy will be explored in detail else-
where. A copy of the computed tomography (CT) data
has been provided to the SAM, to be archived along
with the specimen.

CLADISTIC ANALYSES

In order to investigate the phylogenetic position of
SAM-PK-K10159, a novel parsimony analysis was
conducted based upon a modified version of the data
matrix of Montefeltro et al. (2013). The matrix of
Montefeltro et al. (2013) was originally constructed
in order to investigate relationships within
Rhynchosauridae. For the present analysis we expand-
ed both taxonomic and character sampling in order
to allow us to address the relationships of both basal
(non-rhynchosaurid) members of Rhynchosauria and
Rhynchosauridae.

We added two non-rhynchosaurian outgroups
(Prolacerta broomi and Protorosaurus speneri) and the
two currently recognized species of Rhynchosaurus
(Rhynchosaurus articeps and Rhynchosaurus brodiei)
as discrete operational taxonomic units. By contrast,
the postcranial material EXEMS 79/1992 from the Otter
Sandstone of the UK, previously referred to Fodonyx
spenceri (Hone & Benton, 2008), was not included in
the analysis given its fragmentary nature and uncer-
tain taxonomic status and phylogenetic position (Langer
et al., 2010; Montefeltro et al., 2013).

We added characters from previously published
phylogenetic analyses relevant to the basal relation-
ships of Rhynchosauria and Rhynchosauridae, includ-
ing proposed synapomorphies of these two clades
(Benton, 1983a, 1985; Dilkes, 1998), as well as a number
of new characters. Scoring of taxa was based primari-
ly on firsthand examination of specimens, with the
exception of Isalorhynchus genovefae, which was scored
using Langer et al. (2000) and Whatley (2005),
Stenaulorhynchus stockleyi scored based on Huene
(1938), Dilkes (1998), Langer & Schultz (2000), and
Hone & Benton (2008), and Fodonyx spenceri scored
based on Benton (1990) and unpublished photo-
graphs provided by Max Langer. The final matrix (14
taxa and 93 characters) was analysed using TNT v.
1.1 (Goloboff, Farris & Nixon, 2008) using the implic-
it enumeration algorithm. Multistate characters 70,
73, and 75 were treated as additive. Zero branch length
nodes were collapsed following the search. Absolute
and GC bootstrap frequencies were calculated after
10 000 pseudoreplicates. The character list and
synapomorphy list are provided as Appendices 1 and
2. The phylogenetic data matrix is provided as online
Supporting Information.

SYSTEMATIC PALAEONTOLOGY
DIAPSIDA OSBORN, 1903 SENSU LAURIN, 1991

ARCHOSAUROMORPHA HUENE, 1946 SENSU

GAUTHIER, KLUGE, & ROWE, 1988

RHYNCHOSAURIA OSBORN, 1903 SENSU DILKES, 1998

EOHYOSAURUS WOLVAARDTI GEN. ET SP. NOV.
(FIGS 2–5)

Etymology
Genus name from the Greek eos, meaning ‘early’ or
‘dawn’, hyos, meaning ‘pig’ or ‘hog’, and sauros, meaning
‘lizard’ or ‘reptile’. Name refers to the common de-
scription of rhynchosaurs as ‘pig-reptiles’ and the early
stratigraphical occurrence of the new taxon. The species
is named after Frederik Petrus Wolvaardt, who dis-
covered the type and only known specimen in Decem-
ber 2000. The name is therefore intended to translate
as ‘Wolvaardt’s early pig-reptile’.

Holotype
SAM-PK-K10159, partial skull missing the anterior end,
with associated incomplete lower jaws (Figs 2–5).

Locality and horizon
Farm Lemoenfontein 44, Rouxville District, Free State
Province, South Africa. Burgersdorp Formation,
Cynognathus Assemblage Zone, Subzone B (early Middle
Triassic: early Anisian; Fig. 1).

The specimen was found as loose float on boulder-
strewn slopes at the base of a cliff (coordinates:
30°36′43.7′′S, 26°37′22.9′′E). The same locality has also
yielded remains of the archosauriform Erythrosuchus,
the cynodont Trirachodon, the kannemeyeriid dicynodont
Kannemeyeria, procolophonids, and the bauriid
therocephalian Microgomphodon oligocynus (S. Kaal,
pers. comm. 2013; D. Wolvaardt, pers. comm. 2013; Abdala
et al., 2014).

Diagnosis
Characterized by the following autapomorphy: jugal
with elongate dorsal process that forms the entire ante-
rior margin of the infratemporal fenestra and that ar-
ticulates anteriorly with the entire posterior margin of
an elongate ventral process of the postorbital. Distin-
guished from other rhynchosaurs on the basis of the
following unique combination of characters: (1) max-
illae and dentaries mediolaterally expanded (present
in Howesia browni and Rhynchosauridae sensu Dilkes,
1998; mediolaterally expanded maxillae/dentaries absent
in Mesosuchus browni); (2) teeth present on the occlusal
and lingual surfaces of the maxillae and dentaries (present
in H. browni and Rhynchosauridae; occlusal teeth only
present in M. browni); (3) maxilla lacks a longitudinal
occlusal groove and dentary lacks occlusal blade (groove
and blade present in Rhynchosauridae; groove and blade

NEW BASAL RHYNCHOSAUR FROM SOUTH AFRICA 573

© 2015 The Linnean Society of London, Zoological Journal of the Linnean Society, 2015, 174, 571–588

D
ow

nloaded from
 https://academ

ic.oup.com
/zoolinnean/article-abstract/174/3/571/2453122 by U

niversidade Estadual Paulista J�
lio de M

esquita Filho user on 06 M
ay 2019



also absent in M. browni and H. browni); (4) occlusal
margin of maxilla offset ventrally from the ventral
margin of the main body of the jugal (occlusal margin
not offset ventrally in M. browni; offset ventrally in
Rhynchosauridae, and possibly in H. browni); (5) pres-
ence of a short anguli oris crest on the lateral surface
of the maxilla (crest absent in M. browni; present in
Rhynchosauridae but placed on the jugal; condition un-
certain in H. browni); (6) posterior process of jugal is
short and terminates at approximately 50% of the
anteroposterior length of the infratemporal fenestra
(process proportionately longer in M. browni, and con-
tacts quadratojugal in most Rhynchosauridae; process

is also short in H. browni); (7) elongate posterior process
of the postorbital terminates above the anterior margin
of the ventral process of the squamosal (posterior process
shorter and does not terminate above the ventral process
of the squamosal in M. browni and H. browni; present
in Rhynchosauridae); (8) elongate ventral process of
the squamosal extends for more than 50% of the pos-
terior margin of the infratemporal fenestra (ventral
process does not extend for more than 50% of the pos-
terior margin in H. browni; present in M. browni and
Rhynchosauridae); (9) sagittal crest on parietal (sag-
ittal crest absent in M. browni; present in H. browni
and Rhynchosauridae).

Figure 2. Eohyosaurus wolvaardti gen. et sp. nov., SAM-PK-K10159. Photographs of holotype skull in right (A) and
left (B) lateral views; and interpretive line drawings of holotype skull in right (C) and left (D) lateral views. Hatched
areas represent broken surface. Dashed lines represent missing bones and uncertain sutural contacts. Arrows indicate
tooth impressions in rock matrix. Asterisks and associated letters indicate approximate planes of the computed tomog-
raphy cross-sections shown in Figure 5A, B and C. Abbreviations: a, angular; ao, anguli oris crest; ar, articular; d, dentary;
hy, hyoid bone; j, jugal; lb, long bone; lbm, long bone mould; m, maxilla; p, parietal; pf, postfrontal; po, postorbital; q,
quadrate; qj, quadratojugal; sa, surangular; scl, sclerotic plates; sf, surangular foramen; sq, squamosal; st, supratemporal.
Scale bars = 20 mm.

574 R. J. BUTLER ET AL.

© 2015 The Linnean Society of London, Zoological Journal of the Linnean Society, 2015, 174, 571–588

D
ow

nloaded from
 https://academ

ic.oup.com
/zoolinnean/article-abstract/174/3/571/2453122 by U

niversidade Estadual Paulista J�
lio de M

esquita Filho user on 06 M
ay 2019



Description

General
SAM-PK-K10159 is relatively complete posteriorly, but
the anterior end of the skull is incomplete (Figs 2, 3).
On the left side, the premaxilla and anterior half of
the maxilla are entirely missing, the external surface
of the posterior half of the maxilla has broken away,
the prefrontal and lacrimal are represented by small
fragments lacking their external surfaces, and the
external surface of the anterior process of the jugal
has broken away (Fig. 2B, D). On the right side, the
premaxilla and the tip of the maxilla are missing, much
of the anterior part of the preserved maxilla is rep-
resented by an impression of the medial surface of the
element, and only fragments of the lacrimal and
prefrontal are preserved (Fig. 2A, C). Dorsally, the

nasals, frontals, anterior end of the parietals, and much
of the postfrontals are missing, although a natural cast
of the endocranium and nasal capsule is exposed. In
the lower jaw, the anterior end of the dentary is missing
on both left and right sides, and much of the exter-
nal surface of the left mandible is missing, but some
details of the morphology of the latter are revealed by
a natural mould. The left side of the skull has addi-
tionally been displaced ventrally relative to the right
side, and also appears to have been compressed
anteroposteriorly. Posteriorly, the parietals appear to
have been displaced dorsally relative to the occipital
region.

As preserved, the maximum anteroposterior length
along the sagittal plane of the skull on the left side,
from the broken anterior margin of the maxilla to the

Figure 3. Eohyosaurus wolvaardti gen. et sp. nov., SAM-PK-K10159. Photographs of holotype skull in dorsal (A)
and occipital (B) views; and interpretive line drawings of holotype skull in dorsal (C) and occipital (D) views. Hatched
areas represent broken surface. Dashed lines represent missing bones and uncertain sutural contacts. Abbreviations: j,
jugal; op, opisthotic; p, parietal; pf, postfrontal; po, postorbital; q, quadrate; qf, quadrate foramen; qj, quadratojugal; sa,
surangular; scl, sclerotic plates; so, supraoccipital; sq, squamosal; st, supratemporal. Scale bars = 20 mm.

NEW BASAL RHYNCHOSAUR FROM SOUTH AFRICA 575

© 2015 The Linnean Society of London, Zoological Journal of the Linnean Society, 2015, 174, 571–588

D
ow

nloaded from
 https://academ

ic.oup.com
/zoolinnean/article-abstract/174/3/571/2453122 by U

niversidade Estadual Paulista J�
lio de M

esquita Filho user on 06 M
ay 2019



Figure 4. Eohyosaurus wolvaardti gen. et sp. nov., SAM-PK-K10159, close-up photographs. Anterior end of the skull
as preserved in left (A) and right (B) lateral views; temporal region of the skull in right lateral view (C); left squamosal,
quadrate, and quadratojugal in lateral view (D); and parietals and posterior skull roof in dorsal view (E). Abbreviations:
ao, anguli oris crest; ap.qj, anterior process of the quadratojugal; d, dentary; dp.j, dorsal process of the jugal; j, jugal;
lf.q, lateral flange of the quadrate; m, maxilla; p, parietal; pf, postfrontal; po, postorbital; pp.j, posterior process of the
jugal; pp.po, posterior process of the postorbital; qf, quadrate foramen; qj, quadratojugal; sc, sagittal crest; scl, sclerotic
ring; shm, impression of the horizontal occlusal shelf of the maxilla; t.po, thickened orbital margin of the postorbital;
vp.sq, ventral process of the squamosal.

576 R. J. BUTLER ET AL.

© 2015 The Linnean Society of London, Zoological Journal of the Linnean Society, 2015, 174, 571–588

D
ow

nloaded from
 https://academ

ic.oup.com
/zoolinnean/article-abstract/174/3/571/2453122 by U

niversidade Estadual Paulista J�
lio de M

esquita Filho user on 06 M
ay 2019



posterior margin of the quadrate condyles, is 68.2 mm.
Based on the dimensions of the orbit and infratemporal
fenestra (Table 1), SAM-PK-K10159 is similar in size
to a referred specimen of the early rhynchosaur
M. browni (SAM-PK-6536; Dilkes, 1998) that has a basal
skull length (premaxilla to quadrate condyles) of 89 mm,
and we suggest that the complete length of SAM-PK-
K10159 would have been similar. Our estimate of skull
length suggests that the skull was longer than broad,
similar to the condition in M. browni (Dilkes, 1998),
H. browni (Dilkes, 1995), and the rhynchosaurids
R. articeps (Benton, 1990), S. stockleyi (Huene, 1938),
and Bentonyx sidensis (Langer et al., 2010).

Although the dorsal rim of the orbit is not pre-
served on either side of SAM-PK-K10159, the orbit
appears to be slightly anteroposteriorly longer than
dorsoventrally tall on the right side (that of the left
side appears to be anteroposteriorly compressed). Despite
deformation, the orbits face mostly laterally, unlike the
condition in the hyperodapedontines I. genovefae
(Whatley, 2005), Teyumbaita sulcognathus (UFRGS-
PV-0232, UFRGS-PV-0298; Montefeltro et al., 2010) and

Hyperodapedon spp. (Hyperodapedon mariensis, FZB-
PV-1867; Hyperodapedon sanjuanensis, MACN-Pv 18185;
Hyperodapedon huenei, UFRGS-0132), in which the
orbits face mostly dorsally. Fragments of the right scle-
rotic ring appear to be present in the right orbit, and
some fragments of the left sclerotic ring may be present
on the left side (Figs 2–4: scl).

The infratemporal fenestra has a trapezoidal outline
on the right side (the left side is anteroposteriorly com-
pressed), being longer anteroposteriorly at its ventral
margin than at its dorsal margin. The supratemporal
fenestra has a broadly rounded anterior margin, expands
in transverse width posteriorly reaching a maximum
width just posterior to the midlength of the parietal,
and then tapers to a rounded posterolateral corner.

Maxilla
The maxilla is poorly preserved on both sides of
the skull, but its external morphology is best
preserved on the right side (Figs 2, 4: m). As in
all other rhynchosaurs and non-archosauriform
archosauromorphs, an antorbital fenestra and fossa are

Figure 5. Eohyosaurus wolvaardti gen. et sp. nov., SAM-PK-K10159, micro-computed tomography (micro-CT) cross-
sections through the tooth-bearing bones (A, B, C). Approximate positions of the CT sections are shown in Figure 2. Ab-
breviations: d, dentary; d.occ, occlusal surface of the dentary; d.t, dentary lingual tooth; l.m, left maxilla; m.occ, occlusal
surface of the maxilla; m.t, maxillary lingual tooth; ptg.t, pterygoid tooth; r.m, right maxilla; v, vomer; v.t, vomerine tooth.
Scale bars = 2 mm.

NEW BASAL RHYNCHOSAUR FROM SOUTH AFRICA 577

© 2015 The Linnean Society of London, Zoological Journal of the Linnean Society, 2015, 174, 571–588

D
ow

nloaded from
 https://academ

ic.oup.com
/zoolinnean/article-abstract/174/3/571/2453122 by U

niversidade Estadual Paulista J�
lio de M

esquita Filho user on 06 M
ay 2019



absent. The occlusal margin is nearly straight in lateral
view, but the preserved part of the anterior end of the
left maxilla and the impression of the anterior part
of the right maxilla suggests that it curved gently dor-
sally toward the contact with the premaxilla. This cur-
vature is not as marked as typically occurs in
rhynchosaurids (e.g. Benton, 1983a, 1984, 1990; Langer
& Schultz, 2000; Nesbitt & Whatley, 2004; Montefeltro
et al., 2010; Mukherjee & Ray, 2014), but does differ
from the completely straight occlusal margin present
in M. browni (SAM-PK-6536; Dilkes, 1998). The pos-
terior end of the occlusal margin of the maxilla is offset
ventral relative to the ventral margin of the jugal, con-
trasting with the condition in M. browni (SAM-PK-
6536; Dilkes, 1998), in which the ventral margin of
the jugal is in line with the occlusal margin of the
maxilla.

The posterior-most portion of the maxilla tapers
dorsoventrally, apparently forming an anterodorsally-
to-posteroventrally orientated suture with the jugal,
although the exact path of this suture is difficult to

establish owing to poor preservation. The contact of
the maxilla with the lacrimal is unclear on both sides
of the skull.

The lateral surface of the maxilla is mostly flat, and
is slightly expanded laterally at the occlusal margin.
Additionally, the lateral surface of the maxilla bears
a distinct ridge (Figs 2, 4: ao), here proposed to be ho-
mologous to the anguli oris crest of rhynchosaurids (Sill,
1970; Langer & Schultz, 2000). In Eohyosaurus
wolvaardti this ridge extends from the contact of the
maxilla with the ventral margin of the jugal, along the
lateral surface of the former bone, and terminates an-
teriorly below the anteroventral corner of the orbit.
Below this ridge the lateral surface of the maxilla is
slightly bevelled inwards. Although we propose that
this feature is homologous with the anguli oris crest,
its position is slightly different from that in most
rhynchosaurids, where it is positioned on the bounda-
ry between the jugal and the maxilla (Sill, 1970; Langer
& Schultz, 2000).

Where the anterior part of the right maxilla has
broken away, an impression of the bone demon-
strates that the ventral part of this element is strong-
ly transversely expanded, and that this transverse
expansion becomes stronger posteriorly. This trans-
verse expansion forms a nearly horizontal, ventrally
facing occlusal surface laterally. Medially the occlusal
surface forms a near right angle with the lingual surface
of the maxilla. This transverse expansion is very similar
to that present in H. browni (Dilkes, 1995) and
rhynchosaurids (Benton, 1984; Langer & Schultz, 2000;
Nesbitt & Whatley, 2004; Montefeltro et al., 2010), but
differs from the transversely narrower occlusal surface
present in M. browni (Dilkes, 1998).

Jugal
The jugal forms the ventral margin of the orbit and
much of the anterior and ventral margins of the
infratemporal fenestra (Figs 2–4: j). The lateral surface
of the main body of the jugal possesses several
subcircular pits that are irregularly distributed (Figs 2,
4C). Similar pits are present on the jugal of H. browni
(SAM-PK-5884). The orbital margin is marked by a
low thickening that is continuous with a similar thick-
ened margin on the orbital margin of the postorbital.
This thickened orbital margin is present to some degree
in all rhynchosaurs, but is particularly marked in the
basal rhynchosaurids R. articeps (SHYMS G-132/
1982), R. brodiei (WARMS Gz6097), S. stockleyi (Huene,
1938) and B. sidensis (BRSUG 27200). However, this
orbital thickening is less well developed in F. spenceri
(Benton, 1990) and hyperodapedontines (Benton, 1983a;
Whatley, 2005; Montefeltro et al., 2010).

The posterior process of the jugal is short, and ter-
minates at approximately the anteroposterior midlength
of the infratemporal fenestra (Fig. 4C: pp.j). This short

Table 1. Selected measurements of the holotype speci-
men of Eohyosaurus wolvaardti gen. et sp. nov. (SAM-
PK-K10159), in mm. Values in parentheses indicate
incomplete measurements, and the value given is the
maximum measurable. Maximum deviation of the digital
calliper equals 0.02 mm but measurements were rounded
to the nearest 0.1 mm

Length Height Width

Skull (68.2) c. 39 –
Jugal (36.2) 19.8 –
Jugal posterior process 9.8 – –
Jugal posterior process base – 3.9 –
Jugal ascending process – 13.9 –
Orbit 24.4 20.5 –
Postorbital posterior process 12.7 – –
Postorbital ventral process 11.4 – –
Squamosal 18.4 – –
Squamosal ventral process – 13.4 –
Squamosal ventral process

at base
– – 8.0

Infratemporal fenestra 25.1 21.4 –
Supratemporal fenestra 20.4 – 13.0
Quadratojugal height – 17.4 –
Quadratojugal ventral

margin
7.6 – –

Quadrate height – 24.6 –
Quadrate distal end – – 8.9
Parietals maximum – – 25.9
Parietals minimum – – 9.2
Lower jaw length (83.8) – –
Surangular height – 12.4 –
Hyoid length (52.4) – –

578 R. J. BUTLER ET AL.

© 2015 The Linnean Society of London, Zoological Journal of the Linnean Society, 2015, 174, 571–588

D
ow

nloaded from
 https://academ

ic.oup.com
/zoolinnean/article-abstract/174/3/571/2453122 by U

niversidade Estadual Paulista J�
lio de M

esquita Filho user on 06 M
ay 2019



process is similar to that of H. browni (Dilkes, 1995)
and R. articeps (NHMUK R1236), but differs from the
proportionately longer process of M. browni (Dilkes,
1998) and most rhynchosaurids (Benton, 1983a, 1990;
Langer & Schultz, 2000; Montefeltro et al., 2010), which
extends nearly as far as or beyond the anterior margin
of the ventral process of the squamosal (e.g. H. huenei:
UFRGS-PV-0132T). The posterior process of the jugal
of SAM-PK-K10159 is orientated almost directly
posteriorly, as also occurs in M. browni (SAM-PK-
6536) and H. browni (SAM-PK-5884). By contrast, in
rhynchosaurids the posterior process of the jugal is more
strongly directed lateroventrally as well as posteri-
orly and, as a result, the temporal region of the skull
is considerably broader than the orbital region (e.g.
R. articeps: NHMUK R1236, SHYMS G-132/1982).

Although the medial surface of the posterior process
of the jugal is covered with sediment, it seems highly
likely that the jugal−quadratojugal bar was incom-
plete, as in H. browni (Dilkes, 1995), M. browni (Dilkes,
1998), and R. articeps (NHMUK R1236). The dorsal
process of the jugal is elongate, and forms the entire
anterior margin of the infratemporal fenestra, taper-
ing to a very slender strip of bone at its dorsal tip (Fig. 4C:
dp.j). By contrast, the dorsal process of the jugal is
proportionately shorter in H. browni (Dilkes, 1995),
M. browni (Dilkes, 1998), R. articeps, and R. brodiei
(Benton, 1990). Moreover, in SAM-PK-K10159 the
postorbital bar is formed along its entire length by an
articulation between the dorsal process of the jugal and
the elongate ventral process of the postorbital. This
differs from the condition in B. sidensis (Langer et al.,
2010), S. stockleyi (Huene, 1938), and hyperodapedontines
(Chatterjee, 1974; Benton, 1983a; Langer & Schultz,
2000; Whatley, 2005; Montefeltro et al., 2010), in which
the bar is formed mostly by the dorsal process of the
jugal. The very elongate dorsal process combined with
the presence of a well-developed ventral process of the
postorbital represents an autapomorphy of E. wolvaardti.
The anterior margin of the dorsal process of the jugal
contacts the posterior margin of the ventral process of
the postorbital, differing from the condition in
T. sulcognathus and Hyperodapedon spp. in which the
ventral process of the postorbital fits into a slot in the
dorsal margin of the dorsal process of the jugal (Whatley,
2005; Montefeltro et al., 2010).

Prefrontal/lacrimal
The anterior border of the orbit is delimited by the
mould of the medial projections of the ventral pro-
cesses of the prefrontal and possibly lacrimal. This
mould shows that the anterior wall of the orbit was
well extended medially, as also occurs in R. brodiei
(WARMS Gz6097). There is a nearly horizontal groove
that is preserved as a natural mould situated
anteroventral to the orbit, and which would most likely

have been placed within the lacrimal. This groove
appears to have exited on the anterior margin of the
orbit, in the same position as the pair of foramina that
are present for the exit of the naso-lacrimal duct in
the lacrimal of R. brodiei (WARMS Gz6097).

Frontal
The frontal is completely missing, but the natural mould
of its ventral surface is preserved (Fig. 3A, C). This
mould shows the morphology of the olfactory tract and,
more anteriorly, the transversely expanded olfactory
bulbs. The morphology of the olfactory apparatus re-
sembles that of other basal archosauromorphs (e.g.
Tasmaniosaurus triassicus: Ezcurra, 2014).

Postorbital
The postorbital is a triradiate bone (Figs 2–4: po). The
posterior process is elongate and extends posteriorly,
reaching a point level with the posterior border of the
infratemporal fenestra (Fig. 4C: pp.po), as also occurs
in all rhynchosaurids (e.g. R. articeps: NHMUK R1237,
SHYMS G-132/1982; B. sidensis: Langer et al., 2010;
T. sulcognathus: Montefeltro et al., 2010; Hyperodapedon
spp.: Chatterjee, 1974; Benton, 1983a; Langer & Schultz,
2000). Although not complete on either side, its pos-
terior termination was gently rounded, as shown by
the articular surface preserved on each squamosal. By
contrast, in M. browni (SAM-PK-6536; Dilkes, 1998)
and H. browni (Dilkes, 1995) the posterior process of
the postorbital is proportionally shorter, ending ante-
rior to the posterior border of the infratemporal fenestra,
and tapers to a point rather than being broadly rounded.
The orbital margin of the postorbital of SAM-PK-
K10159 is thickened (Fig. 4C: t.po), and posterior to
this thickening the lateral surface of the bone is gently
concave.

The ascending process of the postorbital forms a broad
transverse suture with the postfrontal. The contacts
in the postorbital−postfrontal−parietal region are not
clear, but the ascending process appears to continue
medially as a broad process that forms the entire ante-
rior margin of the supratemporal fenestra, as in
M. browni (SAM-PK-6536; Dilkes, 1998), H. browni
(Dilkes, 1995), R. articeps (SHYMS G-132/1982, Benton,
1990), and S. stockleyi (Montefeltro et al., 2010).

Postfrontal
The lateral part of the postfrontal is preserved on the
left side of the skull, contributing to the posterodorsal
part of the orbital margin (Figs 2–4: pf). The postfrontal
appears to be excluded from the anterior border of the
supratemporal fenestra.

Parietal
The parietals are fused to one another, without any
trace of a median suture (Figs 2–4: p), resembling the

NEW BASAL RHYNCHOSAUR FROM SOUTH AFRICA 579

© 2015 The Linnean Society of London, Zoological Journal of the Linnean Society, 2015, 174, 571–588

D
ow

nloaded from
 https://academ

ic.oup.com
/zoolinnean/article-abstract/174/3/571/2453122 by U

niversidade Estadual Paulista J�
lio de M

esquita Filho user on 06 M
ay 2019



condition present in other rhynchosaurs (Dilkes, 1998).
The anterior ends of the parietals are damaged, and
the presence or absence of a pineal foramen cannot
be determined. A sagittal crest is present along the
full length of the midline of the parietals as pre-
served (Fig. 4E: sc). This differs from the condition in
M. browni (SAM-PK-6536; Dilkes, 1998), in which the
sagittal midline bears a concavity that is bounded lat-
erally by ridges that define the supratemporal fenestrae,
but resembles the condition in H. browni (Dilkes, 1995)
and rhynchosaurids (Benton, 1983a, 1990; Langer et al.,
2010; Montefeltro et al., 2010). As preserved, the
posterolateral processes of the parietals are directed
posterolaterally in dorsal view. The posterolateral ori-
entation is similar to the condition in M. browni (SAM-
PK-6536), H. browni (SAM-PK-5885), R. articeps
(SHYMS G-132/1982), and S. stockleyi (Montefeltro et al.,
2010), but differs from the laterally orientated pro-
cesses in B. sidensis (BRSUG 27200), I. genovefae
(Whatley, 2005), T. sulcognathus (UFRGS-PV-0232T,
UFRGS-PV-0298), and Hyperodapedon spp. (e.g.
H. huenei: UFRGS-PV-0132T; H. mariensis: FZB-PV-
1867, Montefeltro et al., 2010). In addition, the distal
tips of the posterolateral processes do not curve an-
teriorly as occurs in S. stockleyi (Montefeltro et al., 2010)
and the ‘Mariante rhynchosaur’ (UFRGS-PV-0168T).

In posterior view the posterolateral processes are ven-
trally deflected at an angle of nearly 40° from the hori-
zontal (Fig. 3B, D). This ventral deflection is a unique
condition amongst rhynchosaurs in which the
posterolateral processes are typically directed nearly
directly laterally in posterior view. This may repre-
sent an autapomorphy of E. wolvaardti, but this is dif-
ficult to confirm given the post-mortem deformation
that has affected the skull (including the dorsal dis-
placement of the parietals relative to the supraoccipital),
and this character is not included in the diagnosis. The
posterolateral processes of the parietals are
dorsoventrally deep, but taper at their distal ends where
they fit into a notch within the supratemporal. As in
all other rhynchosaurs, there is no postparietal bone.

Supratemporal
The supratemporal is well preserved on the left side
and lies between the posterolateral process of the pa-
rietal and the medial surface of the squamosal (Figs 2,
3: st). This condition is different from T. sulcognathus
and Hyperodapedon spp., in which this bone is absent
(Benton, 1983a; Langer & Schultz, 2000; Langer et al.,
2010; Montefeltro et al., 2010). The supratemporal in
SAM-PK-K10159 is transversely broad in occipital view
but contrasting with this species it also forms a small
part of the posterior margin of the supratemporal
fenestra as in M. browni (SAM-PK-6536). Posteriorly
the supratemporal contacts the dorsal edge of the tip
of the paroccipital process, and forms most of the dorsal

margin of the post-temporal fenestra. This condition
differs from that present in B. sidensis (BRSUG 27200;
Langer et al., 2010), in which this bone lacks a
ventromedial process and, as a result, is limited to a
more lateral portion of the margin of the fenestra.

Squamosal
The squamosal has four distinct processes (Figs 2, 3:
sq). The anterior process is broadly overlapped by the
extensive posterior process of the postorbital. The ante-
rior process forms only small portions of the lateral
border of the supratemporal fenestra and the dorsal
border of the infratemporal fenestra. The ventral process
is proportionally elongate, extending for more than 75%
of the height of the infratemporal fenestra (Fig. 4C:
vp.sq). This is similar to the condition in M. browni
(SAM-PK-6536; Dilkes, 1998), F. spenceri (Benton, 1990),
and T. sulcognathus (UFRGS-PV-0232T, Montefeltro
et al., 2010), but differs from the abbreviated process
that is present in H. browni (Dilkes, 1995) and
R. articeps (SHYMS G-132/1982). Additionally, the
narrow anteroposterior width of the ventral process
is similar to the condition in M. browni (Dilkes, 1998),
H. browni (Dilkes, 1995), R. articeps (SHYMS G-132/
1982), S. stockleyi (Montefeltro et al., 2010), and
F. spenceri (Benton, 1990), but differs from the wide
ventral process of the squamosal that is present in the
hyperodapedontines I. genovefae (Whatley, 2005),
T. sulcognathus (Montefeltro et al., 2010), and
Hyperodapedon spp. (Chatterjee, 1974; Benton, 1983a;
Langer & Schultz, 2000). The medial process of the
squamosal contacts the supratemporal, but it is unclear
whether it contacted the parietal. Finally, there is a
short posteroventral process that extends between the
supratemporal and the head of the quadrate, reach-
ing the paroccipital process. The lateral surface of the
squamosal possesses several small subcircular pits.

Quadratojugal
The quadratojugal is a strap-like, slightly posteriorly
bowed bone in lateral view (Figs 2–4: qj), resembling
that of M. browni (SAM-PK-6536; Dilkes, 1998). The
bone is best preserved on the left side, and is damaged
slightly at its very anteroventral margin. A short,
rounded anterior process of the quadratojugal appears
to have been present (Fig. 4: ap.qj), similar to the con-
dition in R. articeps (SHYMS G-132/1982), but differ-
ing from the condition in M. browni in which an anterior
process is absent (Dilkes, 1998). The quadratojugal ter-
minates posteriorly adjacent to the quadrate con-
dyles. It forms the lateral margin of an oval quadrate
foramen, shared with the quadrate. The size of the quad-
rate foramen is similar to that of M. browni (SAM-
PK-6536) and R. articeps (SHYMS 3), but differs from
the more reduced aperture that is present in F. spenceri
(Benton, 1990), T. sulcognathus (Montefeltro et al., 2010),

580 R. J. BUTLER ET AL.

© 2015 The Linnean Society of London, Zoological Journal of the Linnean Society, 2015, 174, 571–588

D
ow

nloaded from
 https://academ

ic.oup.com
/zoolinnean/article-abstract/174/3/571/2453122 by U

niversidade Estadual Paulista J�
lio de M

esquita Filho user on 06 M
ay 2019



and a number of South American specimens of
Hyperodapedon spp. (PVSJ 680, UFRGS-PV-0408T,
UFRGS-PV-0149T).

Quadrate
The quadrate is strongly bowed along its length, with
a concave posterior margin in lateral view (Figs 2–4:
q). Ventrally, the articular surface for the mandible is
divided into two strongly convex condyles. The medial
condyle is moderately better developed transversely and
extends further ventral than the lateral condyle, re-
sembling the condition present in M. browni (SAM-
PK-6536), but differing from the condition in
rhynchosaurids in which the condyles are equally well
developed (e.g. R. articeps, NHMUK R1236; B. sidensis,
BRSUG 27200; T. sulcognathus UFRGS-PV-0232). The
lateral flange of the quadrate forms the dorsal margin
of the quadrate foramen and overlaps the posterior
margin of the ventral process of the squamosal (Figs 3,
4: lf.q, qf).

Occiput
The supraoccipital is plate-like (Fig. 3: so) and similar
to that of M. browni (Dilkes, 1998) and H. browni
(Dilkes, 1995), but differs from the pillar-like
supraoccipital that is present in Rhynchosauridae (e.g.
R. articeps, Benton, 1990; B. sidensis, Langer et al., 2010;
T. sulcognathus, Montefeltro et al., 2010). The
supraoccipital has a very low midline ridge. The
paroccipital process is strap-like, and as preserved is
posterolaterally orientated, but appears to have been
displaced from its articulation with the supraoccipital.
Its ventral margin is straight as in M. browni,
T. sulcognathus (Montefeltro et al., 2010), and some speci-
mens of Hyperodapedon (e.g. UFRGS-PV-0149T), dif-
fering from the convex ventral margin that is present
in R. articeps (SHYMS G-132/1982), S. stockleyi (Huene,
1938), the ‘Mariante rhynchosaur ’ (UFRGS-PV-
0168T), B. sidensis (BRSUG 27200), and H. huenei
(UFRGS-0132). The paroccipital process forms the
ventral margin of the slit-like post-temporal fenestra.

Lower jaw, general
The dentary, surangular, angular, and articular are pre-
served (Fig. 2). The mandibular anatomy is generally
very similar to that of M. browni (SAM-PK-6536; Dilkes,
1998). There is no mandibular fenestra, as in all
rhynchosaurs. The maximum dorsoventral height of
the mandible is less than a quarter of the mandibu-
lar length, differing from hyperodapedontines in which
the mandible is higher than a quarter of the man-
dibular length (Benton, 1990; Dilkes, 1998; Whatley,
2005; Montefeltro et al., 2010).

Dentary
The dentary is incomplete anteriorly on both sides, but
a natural mould on the right side indicates that the

bone tapered in dorsoventral height towards its ante-
rior end, and would probably have ended approxi-
mately level with the anterior margin of the maxilla
(Fig. 2: d), as in M. browni (SAM-PK-6536; Dilkes, 1998).
Our estimation suggests that the dentary comprises
less than half of the total length of the mandible, similar
to the condition in M. browni and S. stockleyi (Benton,
1984, 1990; Dilkes, 1998). Natural moulds indicate that,
as in the maxilla, the dentary was transversely ex-
panded, and further show that multiple tooth rows were
present, and that teeth were present on the lingual
as well as occlusal surfaces. Posteriorly, the dentary
has an elongate posterodorsal process that overlaps the
lateral surface of the surangular.

Surangular
The surangular is a large element that dominates the
mandible in lateral view, extending for more than 70%
of the mandibular length, and extending posterior to
the glenoid fossa, forming the lateral surface of the
retroarticular process (Figs 2, 3: sa). The surangular
has strongly convex dorsal and ventral margins, and
a gently concave lateral surface. The lateral border of
the glenoid fossa (which is not exposed) forms a low
tuberosity. Immediately below this tuberosity there is
a small posterior surangular foramen (Fig. 2: sf).

Angular
The angular is exposed as a narrow strip of bone ventral
to the surangular (Fig. 2: a).

Articular
The articular is poorly preserved, but it can be deter-
mined that it contributed to a well posteriorly devel-
oped and upturned retroarticular process (Fig. 2: ar),
resembling the condition in H. browni (Dilkes, 1995).

Hyoid apparatus
The right ceratobranchial of the hyoid apparatus is pre-
served, and is a slender, rod-like element (Fig. 2: hy).
As preserved, it extends posteriorly beyond the level
of the craniomandibular joint, and is missing its pos-
terior end.

Dentition
Micro-CT data demonstrate that the maxillae and
dentaries are transversely expanded, forming dental
plates with multiple rows of conical teeth, with teeth
present both on occlusal and lingual surfaces of the
bones (Fig. 5: d.occ, m.occ). This condition resembles
the jaw apparatus of H. browni (Dilkes, 1995). The tooth-
bearing surfaces of the maxilla and dentary are divided
into a nearly horizontally orientated occlusal surface,
and a nearly vertical, medially facing surface, which
bears the lingual teeth (Fig. 5: d.t, m.t). This condi-
tion is different from the lingual margin of the tooth

NEW BASAL RHYNCHOSAUR FROM SOUTH AFRICA 581

© 2015 The Linnean Society of London, Zoological Journal of the Linnean Society, 2015, 174, 571–588

D
ow

nloaded from
 https://academ

ic.oup.com
/zoolinnean/article-abstract/174/3/571/2453122 by U

niversidade Estadual Paulista J�
lio de M

esquita Filho user on 06 M
ay 2019



plates of H. browni, in which the occlusal surface more
gradually merges into the lingual surface (Dilkes, 1995).
As in H. browni (Dilkes, 1995), there is no occlusal
groove, but in some places the occlusal surface of the
maxilla appears to be gently concave. The presence of
dental plates in both the maxilla and dentary in SAM-
PK-K10159 is different from the groove and blade
jaw apparatus of all rhynchosaurids, including the
fragmentary Middle Triassic form Ammorhynchus
navajoi and the ‘Chañares form’ (Benton, 1983a, 1984,
1990; Langer et al., 2000; Nesbitt & Whatley, 2004;
Montefeltro et al., 2010; Ezcurra et al., 2014).

The most clearly distinguishable and most com-
plete teeth are the most lingual. These teeth are clearly
set within shallow sockets and appear to be fused to
the surrounding bone, although they are still distin-
guishable. More labially positioned teeth, and teeth on
the occlusal surface, are more completely fused and
difficult to distinguish from the surrounding bone, with
those on the occlusal surface probably worn complete-
ly level with the bone.

As in M. browni there is clearly a single row of tightly
packed teeth on the vomer (Fig. 5: v.t), and slightly
smaller teeth are also present along the thickened
medial parts of the pterygoids (Fig. 5: ptg.t). It is unclear
whether or not palatine teeth are present. The palatal
teeth of the vomer and pterygoid are similar in size
to the lingual teeth present on the maxilla and dentary.

Possible tibia and fibula
There is a partial long bone preserved adjacent to the
posterior end of the right side of the skull that par-
tially obscures the craniomandibular joint (Fig. 2A, C:
lb). Next to it, there is a natural mould of a long bone
shaft that possesses the same longitudinal orienta-
tion (Fig. 2A, C: lbm). The size, curvature of the shaft,
and morphology of the preserved articular end suggest
that the preserved long bone is the distal half of a fibula
and the natural mould next to it may be of a tibia.
The probable fibula is bowed along its length and the
distal articular surface is subcircular.

PHYLOGENETIC ANALYSIS

The parsimony analysis yielded a single most parsi-
monious tree of 146 steps with a consistency index of
0.6781 and a retention index of 0.7565. Eohyosaurus
wolvaardti is recovered as the sister taxon of
Rhynchosauridae (sensu Dilkes, 1998) (Fig. 6). Under
suboptimal constrained topologies one additional step
is necessary to recover E. wolvaardti as the sister taxon
of H. browni (Templeton’s test P = 1.000), five addi-
tional steps for a position as the sister taxon of
M. browni (Templeton’s test P = 0.2891), and eight ad-

ditional steps for a position as a member of
Rhynchosauridae (Templeton’s test P = 0.0078).

DISCUSSION
PHYLOGENETIC IMPLICATIONS

Our phylogenetic results place E. wolvaardti as closer
to rhynchosaurids than are M. browni and H. browni
on the basis of one synapomorphy: the elongated pos-
terior process of the postorbital that reaches the
level of the posterior border of the infratemporal
fenestra (character number 29: 0→1). However, the
E. wolvaardti + Rhynchosauridae node is weakly sup-
ported (Table 2), and an alternative position as the sister
taxon to the morphologically similar H. browni is not
a significantly worse explanation of the data (Templeton
test results).

Rhynchosauridae is supported to the exclusion of
M. browni, H. browni, and E. wolvaardti on the basis
of eight synapomorphies: orbit with elevated rim along
the jugal, postorbital, frontal, prefrontal, and lacri-
mal (character number 5: 0→1); jugal without multi-

Figure 6. Time-calibrated single most parsimonious tree
showing the phylogenetic relationships of Eohyosaurus
wolvaardti gen. et sp. nov. and other rhynchosaurs re-
covered in the phylogenetic analysis presented here. The
length of the vertical bar representing each terminal taxon
represents chronostratigraphical uncertainty rather than
true stratigraphical range. Abbreviations: Ans, Anisian; Chx,
Changhsingian; Crn, Carnian; E., Early; Hynae,
Hyperodapedontinae; Ind, Induan; Lad, Ladinian; Nor,
Norian; Ole, Olenekian; PER., Permian; Rht, Rhaetian; Wuc,
Wuchiapingian.

582 R. J. BUTLER ET AL.

© 2015 The Linnean Society of London, Zoological Journal of the Linnean Society, 2015, 174, 571–588

D
ow

nloaded from
 https://academ

ic.oup.com
/zoolinnean/article-abstract/174/3/571/2453122 by U

niversidade Estadual Paulista J�
lio de M

esquita Filho user on 06 M
ay 2019



ple pits on the lateral surface of its main body (character
number 17: 1→0); jugal subtemporal (= posterior) process
with a distinct lateroventral orientation with respect
to the sagittal axis of the snout (character number 19:
0→1); squamosal with a very long medial process,
forming the entire or almost the entire posterior border
of the supratemporal fenestra (character number 40:
0→1); supraoccipital with an inverted V-shape (‘pillar-
like’; character number 48: 0→1); blade and groove jaw
apparatus, where dentary blade(s) fit precisely into max-
illary groove (character number 60: 1→2); and absence
of vomerine and pterygoid teeth (character numbers
61 and 63: 0→1). Several of the synapomorphies re-
covered here for Rhynchosauridae are recognized here
for the first time, and the clade is one of the best sup-
ported in our analysis (i.e. Bremer support=8, boot-
strap frequencies = 99%; Table 2).

The overall topology of the tree is consistent with
that recovered by Montefeltro et al. (2013) (Fig. 6), in-
cluding the sister grouping of the ‘stenaulorhynchines’
S. stockleyi and the ‘Mariante rhynchosaur’, and the
placement of F. spenceri as the sister group of a
monophyletic Hyperodapedontinae (I. genovefae,
T. sulcognathus, and Hyperodapedon spp.). In the present
analysis B. sidensis was unambiguously recovered as
closely related to hyperodapedontines, rather than in
a polytomy amongst derived rhynchosaurids. However,
the support for this clade is low (Table 2).

The addition of R. articeps and R. brodiei to the
phylogenetic analysis resulted in a paraphyletic genus
Rhynchosaurus, with the latter species being recov-
ered as more closely related to hyperodapedontines than
to R. articeps. The clade composed of R. brodiei and
other rhynchosaurids to the exclusion of R. articeps is
supported by three synapomorphies: jugal higher than
maxilla below the ventral border of the orbit (charac-
ter number 11: 0→1); prefrontal with a deeply concave
dorsomedial surface near the orbital rim (character

number 21: 0→1); and frontal without a groove on its
dorsal surface (character number 23: 1→0). This clade
possesses relatively high bootstrap support (> 75%,
Table 2). Rhynchosaurus brodiei is recovered in a
polytomy together with ‘stenaulorhynchines’ and more
derived rhynchosaurids. Although the results of our
phylogenetic analysis suggest that R. brodiei may need
to be transferred to a new genus, further anatomical
and taxonomic work is necessary on the hypodigm series
of Rhynchosaurus to confirm this, and is beyond the
scope of the present work.

EARLY RHYNCHOSAUR DIVERSITY AND BIOGEOGRAPHY

Our description of E. wolvaardti demonstrates the pres-
ence of three different rhynchosaur species in the tet-
rapod assemblage of Subzone B of the Cynognathus
AZ. This species richness of rhynchosaurs resembles
that present in some younger rock sequences such as
the Santa Maria Sequence 2 of southern Brazil (Langer
et al., 2007), and shows that rhynchosaurs were already
rather taxonomically diverse by the early Middle Tri-
assic, at least in South Africa (Fig. 6). The only
rhynchosaur known from rocks older than Middle Tri-
assic is N. colletti, from the earliest Triassic of South
Africa (Carroll, 1976), which has been supported as a
member of Rhynchosauria by a recent phylogenetic
analysis (Ezcurra et al., 2014).

Intriguingly, the stratigraphically oldest known
rhynchosaur (N. colletti), and the three most basal
rhynchosaur (non-rhynchosaurid) taxa in the present
phylogenetic analysis (which did not include N. colletti)
all come from South Africa. Although the impacts of
incomplete sampling must be considered, this is sug-
gestive that the early rhynchosaur radiation during
the Early and earliest Middle Triassic may have been
geographically restricted, and that rhynchosaur origins
may lie in southern Pangaea. An important test of this

Table 2. Bremer index and bootstrap frequency supports for the nodes of the single most parsimonious tree recovered
in the phylogenetic analysis

Node Bremer index
Absolute bootstrap
frequency (%)

GC bootstrap
frequency (%)

Rhynchosauria 5 93 92
Howesia + Rhynchosauridae 1 56 35
Eohyosaurus wolvaardti + Rhynchosauridae 1 64 58
Rhynchosauridae 8 99 99
Rhynchosaurus brodiei + Hyperodapedontinae 2 79 74
Stenaulorhynchinae 2 47 35
Bentonyx + Hyperodapedontinae 1 26 14
Fodonyx + Hyperodapedontinae 2 64 55
Hyperodapedontinae 8 99 99
Teyumbaita + Hyperodapedon 6 95 95

NEW BASAL RHYNCHOSAUR FROM SOUTH AFRICA 583

© 2015 The Linnean Society of London, Zoological Journal of the Linnean Society, 2015, 174, 571–588

D
ow

nloaded from
 https://academ

ic.oup.com
/zoolinnean/article-abstract/174/3/571/2453122 by U

niversidade Estadual Paulista J�
lio de M

esquita Filho user on 06 M
ay 2019



hypothesis would be the restudy of the material of
Eifelosaurus triadicus, an overlooked reptile from the
early Middle Triassic (early Anisian) of Germany (Jaekel,
1904), which has occasionally been identified as
rhynchosaurian (e.g. Huene, 1929).

Early rhynchosaurid rhynchosaurs achieved a global
distribution during the Anisian, being known from the
USA (Nesbitt & Whatley, 2004), Tanzania (Huene, 1938),
and the UK (Benton, 1990). This global distribution
was also obtained by hyperodapedontine rhynchosaurids
during the early Late Triassic, occurring in the UK,
Madagascar, Zimbabwe, India, Brazil, Argentina,
Canada, and the USA (Baird, 1963; Chatterjee, 1980;
Benton, 1983a; Langer et al., 2000; Montefeltro et al.,
2010, 2013; Mukherjee & Ray, 2014). The success of
hyperodapedontines was, however, relatively short-
lived, with the clade apparently suffering extinction
somewhere close to the Carnian–Norian boundary.

ACKNOWLEDGEMENTS

R. J. B. and M. D. E. are supported by an Emmy
Noether Programme Award from the Deutsche
Forschungsgemeinschaft (BU 2587/3-1 to R. J. B.) and
a Marie Curie Career Integration Grant (PCIG14-GA-
2013-630123 ARCHOSAUR RISE to R. J. B.). F. C. M.’s
travel to the University of Birmingham and partici-
pation in this project was partially supported by the
Brazil Visiting Fellows Scheme of the University of Bir-
mingham. We thank F. Wolvaardt for collecting the type
specimen of E. wolvaardti and providing contextual in-
formation on the type locality, R. Smith (SAM) for bring-
ing the specimen to our attention, and R. Smith and
S. Kaal (SAM) for facilitating its study and loan. We
thank the following curators, collections managers, and
researchers who provided access to specimens in their
care: D. Hone (BRSUG), A. M. Ribeiro and J. Ferigolo
(FZB), A. Kramarz (MACN-Pv), S. Chapman (NHMUK),
R. Martínez (PVSJ), S. Kaal and R. Smith (SAM), D.
Lockett (SHYMS), C. L. Schultz (UFRGS), J. Radley
(WARMS). David Dilkes, Max Langer, and an anony-
mous referee provided helpful comments that im-
proved the final version of this manuscript.

REFERENCES

Abdala F, Jashashvili T, Rubidge BS, van den Heever
J. 2014. New material of Microgomphodon oligocynus
(Eutherapsida, Therocephalia) and the taxonomy of south-
ern African Bauriidae. In: Kammerer CF, Angielczyk KD,
Fröbisch J, eds. Early evolutionary history of the Synapsida.
Vertebrate Paleobiology and Paleoanthropology Series,
Dordrecht: Springer, 209–231.

Baird D. 1963. Rhynchosaurs in the Late Triassic of Nova Scotia.
Special Papers of Geological Society of America 73: 107.

Benton MJ. 1983a. The Triassic reptile Hyperodapedon from
Elgin: functional morphology and relationships. Philosophi-
cal Transactions of the Royal Society of London, Series B 302:
605–717.

Benton MJ. 1983b. Dinosaur success in the Triassic: a non-
competitive ecological model. The Quarterly Review of Biology
58: 29–55.

Benton MJ. 1984. Tooth form, growth, and function in
Triassic rhynchosaurs (Reptilia, Diapsida). Palaeontology 27:
737–776.

Benton MJ. 1985. Classification and phylogeny of the diapsid
reptiles. Zoological Journal of the Linnean Society 84: 97–
164.

Benton MJ. 1987. The phylogeny of rhynchosaurs (Reptilia,
Diapsida), and two new species. In: Currie PM, Coster EH,
eds. Fourth symposium on Mesozoic terrestrial ecosystems,
Vol. 1. Drumheller: Tyrrell Museum of Paleontology,
12–17.

Benton MJ. 1990. The species of Rhynchosaurus, a rhynchosaur
(Reptilia, Diapsida) from the Middle Triassic of England. Philo-
sophical Transactions of the Royal Society of London, Series
B 328: 213–306.

Carroll RL. 1976. Noteosuchus – the oldest know rhynchosaur.
Annals of the South African Museum 72: 37–57.

Chatterjee S. 1974. A rhynchosaur from the Upper Triassic
Maleri Formation of India. Philosophical Transactions of the
Royal Society of London, Series B 267: 209–261.

Chatterjee S. 1980. The evolution of rhynchosaurs. Mémoires
de la Société Géologique de France, Nouvelle Série 139: 57–
65.

Dilkes DW. 1995. The rhynchosaur Howesia browni from
the Lower Triassic of South Africa. Palaeontology 38: 665–
685.

Dilkes DW. 1998. The Early Triassic rhynchosaur Mesosuchus
browni and the interrelationships of basal archosauromorph
reptiles. Philosophical Transactions of the Royal Society of
London, Series B 353: 501–541.

Evans SE. 1988. The early history and relationships of the
Diapsida. In: Benton MJ, ed. The phylogeny and classifica-
tion of the tetrapods, Vol. 1. Oxford: Clarendon Press, 221–
260.

Ezcurra MD. 2014. The osteology of the basal archosauromorph
Tasmaniosaurus triassicus from the Lower Triassic of Tas-
mania, Australia. PLoS ONE 9: e86864.

Ezcurra MD, Scheyer TM, Butler RJ. 2014. The origin and
early evolution of Sauria: reassessing the Permian saurian
fossil record and the timing of the crocodile-lizard diver-
gence. PLoS ONE 9: e89165.

Ezcurra MD, Trotteyn MJ, Fiorelli LE, von Baczko MB,
Taborda JRA, Iberlucea M, Desojo JB. 2014. The oldest
rhynchosaur from Argentina: a Middle Triassic rhynchosaurid
from the Chañares Formation (Ischigualasto–Villa Unión Basin,
La Rioja Province). Paläontologische Zeitschrift 88: 453–
460.

Gauthier JA, Kluge AG, Rowe T. 1988. Amniote phylog-
eny and the importance of fossils. Cladistics 4: 105–209.

Goloboff PA, Farris JS, Nixon KC. 2008. TNT, a free program
for phylogenetic analysis. Cladistics 24: 774–786.

584 R. J. BUTLER ET AL.

© 2015 The Linnean Society of London, Zoological Journal of the Linnean Society, 2015, 174, 571–588

D
ow

nloaded from
 https://academ

ic.oup.com
/zoolinnean/article-abstract/174/3/571/2453122 by U

niversidade Estadual Paulista J�
lio de M

esquita Filho user on 06 M
ay 2019



Hone DWE, Benton MJ. 2008. A new genus of rhynchosaur
from the Mid Triassic of SW England. Palaeontology 51: 95–
115.

Huene F. 1929. Über Rhynchosaurier und andere Reptilien
aus den Gondwana-Ablagerungen Südamerikas. Geologische
und paläontologische Abhandlungen, neue Folge 17: 1–62.

Huene F. 1938. Stenaulorhynchus, ein Rhynchosauride der
ostafrikanischer Obertrias. Nova Acta Leopoldina, neue Folge
6: 83–121.

Huene F. 1946. Die grossen Stämme der Tetrapoden in
den geologischen Zeiten. Biologisches Zentralblatt 65: 268–
275.

Jaekel O. 1904. Über ein neues Reptil aus dem Buntsandstein
der Eifel. Zeitschrift der Deutschen Geologischen Gesellschaft
56: 90–94.

Langer MC, Boniface M, Cuny G, Barbieri L. 2000. The
phylogenetic position of Isalorhynchus genovefae, a Late Tri-
assic rhynchosaur from Madagascar. Annales de Paléontologie
86: 101–127.

Langer MC, Montefeltro FC, Hone DWE, Whatley R,
Schultz CL. 2010. On Fodonyx spenceri and a new
rhynchosaur from the Middle Triassic of Devon. Journal of
Vertebrate Paleontology 30: 1884–1888.

Langer MC, Ribeiro AM, Schultz CL, Ferigolo J. 2007.
The continental tetrapod-bearing Triassic of South Brazil. In:
Lucas SG, Spielmann JA, eds. The global Triassic. Albu-
querque: New Mexico Museum of Natural History and Science
Bulletin 41, 201–218.

Langer MC, Schultz CL. 2000. A new species of the Late
Triassic rhynchosaur Hyperodapedon from the Santa
Maria Formation of South Brazil. Palaeontology 43: 633–
652.

Laurin M. 1991. The osteology of a Lower Permian eosuchian
from Texas and a review of diapsid phylogeny. Zoological
Journal of the Linnean Society 101: 59–95.

Martínez RN, Sereno PC, Alcober OA, Colombi CE, Renne
PR, Montañez IP, Currie BS. 2011. A basal dinosaur from
the dawn of the dinosaur era in southwestern Pangaea. Science
331: 206–210.

Montefeltro FC, Bittencourt JS, Langer MC, Schultz CL.
2013. Postcranial anatomy of the hyperodapedontine
rhynchosaur Teyumbaita sulcognathus (Azevedo and Schultz,
1987) from the Late Triassic of southern Brazil. Journal of
Vertebrate Paleontology 33: 67–84.

Montefeltro FC, Langer MC, Schultz CL. 2010. Cranial
anatomy of a new genus of hyperodapedontine rhynchosaur
(Diapsida, Archosauromorpha) from the Upper Triassic of
Southern Brazil. Earth and Environmental Sciences Trans-
actions of the Royal Society of Edinburgh 101: 27–52.

Mukherjee D, Ray S. 2014. A new Hyperodapedon
(Archosauromorpha, Rhynchosauria) from the Upper Trias-
sic of India: implications for rhynchosaur phylogeny. Palae-
ontology 57: 1241–1276.

Nesbitt S, Whatley R. 2004. The first discovery of a
rhynchosaur from the Upper Moenkopi Formation (Middle
Triassic) of northern Arizona. PaleoBios 24: 1–10.

Osborn HF. 1903. The reptilian subclasses Diapsida
and Synapsida and the early history of the Diaptosauria.

Memoirs of the American Museum of Natural History 1: 449–
507.

Sill WD. 1970. Schaphonyx sanjuanensis, nuevo rincosaurio
(Repitilia) de la Formacion Ischigualasto, Triasico de San Juan,
Argentina. Ameghiniana 7: 341–354.

Whatley R. 2005. Phylogenetic relationship of Isalorhynchus
genovefae, the rhynchosaur (Reptilia, Archosauromorpha) from
Madagascar. PhD Thesis, University of California, Santa
Barbara.

APPENDIX 1

Character list arranged in anatomical order for the
phylogenetic analysis conducted here. Characters
adapted from previously published analyses are ac-
knowledged accordingly. The numbering of charac-
ters used in the present analysis is shown between
square brackets and that of Montefeltro et al. (2013)
is shown in parentheses. Character 28 from the origi-
nal character list of Montefeltro et al. (2013) was not
included in the current analysis because of the poor
distinction between its character states.

[1] (1) Skull dimensions: longer than broad (0); broader
than long (1) (Benton, 1984).

[2] (2) Skull height: < 50% of the midline length (0);
> 50% of the midline length (1) (Hone & Benton,
2008).

[3] External nares: separate (0), single medial naris
(1) (Benton, 1985).

[4] (3) Orbit orientation: mostly lateral (0); mostly
dorsal (1) (Langer & Schultz, 2000).

[5] Orbit with elevated rim along the jugal, postorbital,
frontal, prefrontal and lacrimal: absent (0); present
(1) (new character).

[6] (4) Orbital medial margin: rounded (0);
forming a marked angle (1) (Montefeltro et al.,
2010).

[7] Lower temporal fenestra: open ventrally (0), closed
ventrally (1) (Dilkes, 1998).

[8] Premaxilla ventral margin: horizontal (0), down-
turned (1) (Benton, 1985).

[9] Premaxilla and prefrontal contact: absent (0),
present (1) (Dilkes, 1998).

[10] Shape of cranial margin of nasal at midline: strong-
ly convex with anterior process (0) or transverse
with little convexity (1) (Dilkes, 1998).

[11] (5) Jugal and maxillary heights below the ventral
border of the orbit: maxilla higher (0); jugal higher
(1) (Benton, 1984).

[12] (6) Jugal−lacrimal contact: minimal (0); exten-
sive contact of the jugal anterior process (1)
(Whatley, 2005).

[13] (7) Anguli oris crest: absent (0); present (1) (modi-
fied from Benton, 1984).

[14] Anterior extension of the anguli oris crest:
restricted to the main body of the jugal (0);

NEW BASAL RHYNCHOSAUR FROM SOUTH AFRICA 585

© 2015 The Linnean Society of London, Zoological Journal of the Linnean Society, 2015, 174, 571–588

D
ow

nloaded from
 https://academ

ic.oup.com
/zoolinnean/article-abstract/174/3/571/2453122 by U

niversidade Estadual Paulista J�
lio de M

esquita Filho user on 06 M
ay 2019



extending onto the anterior process of the jugal,
but not the maxilla (1); extending onto the maxilla,
but not the anterior process of the jugal (2) (modi-
fied from Benton, 1984).

[15] (8) Jugal surface dorsal to anguli oris crest: lacking
a secondary crest (0); with a secondary anguli oris
crest (1) (Langer & Schultz, 2000).

[16] (9) Lateral overlap of maxilla by jugal: absent or
minimally expanded (0); well developed (1)
(Whatley, 2005).

[17] Jugal with multiple pits on the lateral surface of
its main body: absent (0); present (1) (taken from
Dilkes, 1995; used here for the first time in
phylogenetic analysis).

[18] (10) Jugal subtemporal process: height > 50% of
the length (0); height < 50% of the length (1)
(Dilkes, 1995).

[19] Jugal subtemporal process with a distinct
lateroventral orientation with respect to the sag-
ittal axis of the snout: absent (0); present (1) (new
character).

[20] (11) Relative widths of postorbital bar and lower
temporal fenestra: < 0.4 (0); > 0.4 (1) (Langer &
Schultz, 2000).

[21] (12) Dorsomedial surface of prefrontal near the
orbital rim: flat or slightly concave (0); deeply
concave (1) (Whatley, 2005).

[22] (13) Procumbent lacrimal and prefrontal
anterolateral margin: absent (0); present (1)
(Whatley, 2005).

[23] (14) Groove on the dorsal surface of the frontal:
absent (0); present (1) (Dilkes, 1995).

[24] Dorsal groove on frontal: longitudinally extend-
ed along most of the surface of the frontal (0);
anterolaterally-to-posteromedially extended
along the posterior half of the frontal (1) (new
character).

[25] (15) Well-marked ‘V’-shaped crest along
frontal−postfrontal contact, anterior to the margin
of the supratemporal fossa: absent (0); present (1)
(Montefeltro et al., 2010).

[26] (16) Frontal and parietal midline lengths: frontal
longer (0); parietal longer (1) (Benton, 1987).

[27] (17) Postfrontal: excluded from upper temporal
fenestra border (0); forming the upper temporal
fenestra border (1) (Dilkes, 1998).

[28] (18) Postfrontal dorsal surface: flat (0); marked-
ly concave (1) (Dilkes, 1995).

[29] Posterior extension of the posterior process of the
postorbital: considerably anterior to the level of
the posterior border of the infratemporal fenestra
(0); level with the posterior border of the
infratemporal fenestra (1) (modified from Dilkes,
1998).

[30] (19) Postorbital anteroventral process: expand-
ing ventral to the level of the orbital midpoint

(0); expanding dorsally to orbital height mid-
point (1) (Dilkes, 1998).

[31] (20) Postorbital ventral process: expands anteri-
or to the jugal (0); fits dorsal to the jugal (1)
(Whatley, 2005).

[32] (21) Postorbital−parietal suture: visible in
dorsal view (0); hidden in dorsal view (1) (Dilkes,
1998).

[33] Parietals: separate (0), fused (1) (Benton, 1985).
[34] Parietal foramen: always, or sometimes, present

(0), always absent (1) (Benton, 1985).
[35] Parietal table: broad (0), constricted and with sag-

ittal crest (1) (Dilkes, 1998).
[36] (22) Parietal body: not expanded laterally at

midlength (0); expanded laterally at midlength (1)
(Montefeltro et al., 2010).

[37] (23) Parietal transverse process: posterolaterally
directed (0); laterally directed (1) (Montefeltro et al.,
2010).

[38] (24) Distal tip of parietal transverse process: not
anteriorly curved (0); anteriorly curved (1)
(Montefeltro et al., 2010).

[39] (25) Squamosal ventral process: thinner than 50%
of dorsoventral length (0); broader than 50% of
dorsoventral length (1) (Benton, 1990).

[40] Squamosal medial process: short, forming less than
half of the posterior border of the supratemporal
fenestra (0); long, forming entire or almost entire
posterior border of the supratemporal fenestra (1)
(new character).

[41] (26) Relative position of quadratojugal and
squamosal processes: squamosal ventral process
anterior to quadratojugal dorsal process (0);
squamosal ventral process overlapping the
quadratojugal dorsal process (1) (Whatley, 2005).

[42] Quadratojugal anterior process: completely absent
(0), present (1) (Dilkes, 1998).

[43] (27) Supratemporal: present (0); absent (1) (Benton,
1984).

[44] Supratemporal with a bifurcated medial border,
in which a ventromedial process extends under-
neath the posterolateral process of the parietal:
present (0); absent (1) (new character).

[45] (29) Pterygoid midline suture length: greater than
or equal to the distance between the posterior
margin of the suture and the basipterygoid ar-
ticulation (0); less than the distance between the
posterior margin of the suture and the
basipterygoid articulation (1) (Whatley, 2005).

[46] Ectopterygoid reaches lateral corner of trans-
verse flange of pterygoid: no (0), yes (1). (Dilkes,
1998)

[47] (30) Elements forming the border of the subor-
bital fenestra: ectopterygoid, palatine, and maxilla
(0); ectopterygoid and palatine only (1) (Dilkes,
1995).

586 R. J. BUTLER ET AL.

© 2015 The Linnean Society of London, Zoological Journal of the Linnean Society, 2015, 174, 571–588

D
ow

nloaded from
 https://academ

ic.oup.com
/zoolinnean/article-abstract/174/3/571/2453122 by U

niversidade Estadual Paulista J�
lio de M

esquita Filho user on 06 M
ay 2019



[48] Supraoccipital shape: plate-like (0); inverted
V-shape (1). (Dilkes, 1995)

[49] (31) Occipital condyle position: anterior to
craniomandibular articulation (0); aligned to
craniomandibular articulation (1) (Benton, 1984).

[50] (32) Basioccipital and basisphenoid/parasphenoid
lengths: basisphenoid/parasphenoid longer (0);
basioccipital longer (1) (Langer & Schultz, 2000).

[51] (33) Relative positions of the basipterygoid process
of the basisphenoid and the ectopterygoid process
of the pterygoid: at the same level (0), basipterygoid
process of the basisphenoid posterior to
ectopterygoid process of the pterygoid (1) (Dilkes,
1995).

[52] (34) Basipterygoid process dimensions (dorsoventral
length, anteroposterior width): longer than wide
(0); wider than long (1) (Langer & Schultz,
2000).

[53] Jaw symphysis: formed largely by dentary (0),
formed only by splenial (1) (Dilkes, 1998).

[54] Divergence of dentaries in front of symphysis:
absent (0), present (1) (Dilkes, 1998).

[55] (35) Mandible depth:< 0.25× the total length (0);
> 0.25× the total length (1) (Benton, 1984).

[56] (36) Dentary length: half, or less, than the total
mandibular length (0); greater than half of the
total mandibular length (1) (Benton, 1990).

[57] Premaxillary teeth: present (0), absent (1) (Benton,
1985).

[58] Tooth implantation: subthecodont or thecodont (0),
ankylothecodont (1) (Benton, 1985).

[59] Maxilla occlusal ventral margin: horizontal (0),
convex (1). (modified from Dilkes, 1998).

[60] Tooth occlusion: single-sided overlap (0), flat oc-
clusion (1), blade and groove jaw apparatus, with
dentary blade(s) fitting precisely into maxillary
groove(s) (2) (Benton, 1985).

[61] Vomerine teeth: present (0), absent (1) (Dilkes,
1998).

[62] Palatine teeth: present (0), absent (1) (Dilkes,
1998).

[63] Pterygoid teeth: present (0), absent (1). (Benton,
1983a)

[64] (37) Medial maxillary groove: absent (0); present
but not reaching the anterior half of the maxilla
(1); present and reaching the anterior half of the
maxilla (2) (Benton, 1984).

[65] (38) Maxillary area lateral to main groove: nar-
rower than the medial area (0); same width or
broader than the medial area (1) (Benton, 1990).

[66] (39) Maxillary cross-section lateral to main groove:
crest-shaped (0); cushion-shaped (1) (Langer et al.,
2000).

[67] (40) Tooth rows lateral to main maxillary groove:
a single clear row (0); more than one clear row
(1) (Langer & Schultz, 2000).

[68] (41) Number of tooth rows medial to main max-
illary groove: two rows and scattered teeth
(0); three or more tooth rows (1) (Langer et al.,
2000).

[69] (42) Occlusal tooth rows on the anterior half of
the maxilla: four or more tooth rows (0); fewer
than four tooth rows (1) (Whatley, 2005).

[70] (43) Maxillary lingual teeth: absent (0); scat-
tered teeth (1); large number of teeth on the medial
surface of the bone (2) (Benton, 1984), ORDERED.

[71] (44) Maxillary teeth: only conicals (0); conicals and
‘pyramidal’ (1) (Whatley, 2005).

[72] (45) Dentary teeth: only conicals (0); conical and
anteroposteriorly compressed (1) (Whatley,
2005).

[73] Number of rows of teeth on dentary: one (0), two
(1), more than two full rows (2). (Benton, 1983a),
ORDERED.

[74] (46) Posterior-most dentary teeth: on the anteri-
or half of lower jaw (0); on the posterior half of
lower jaw (1) (Langer & Schultz, 2000).

[75] (47) Lingual dentary teeth: absent (0); present and
forming one row (1); present and forming more
than one row (2) (Benton, 1984), ORDERED.

[76] (48) Dentary teeth on the dentary lingual surface:
scattered (0); crowded (1) (Benton, 1985).

[77] (60) Axis ventral keel: present (0); absent (1)
(Montefeltro et al., 2013).

[78] (61) Axial parapophysis: present (0); absent (1)
(Montefeltro et al., 2013).

[79] (62) Cervical postaxial vertebrae ventral
keel: absent (0); present (1) (Montefeltro et al.,
2013).

[80] (49) Truncal vertebrae with ossified intercentrum:
present (0); absent (1) (Evans, 1988).

[81] (50) Epipophyses on cervical postzygapophyses:
spine-shaped (0); crest-shaped (1) (Whatley,
2005).

[82] (63) Position of the transverse process of cranial
truncal vertebrae: at the level of prezygapophysis
(0); posteriorly located in the vertebra centrum
(1) (Montefeltro et al., 2013).

[83] (51) Second sacral vertebra: with a notch between
the iliac articular surface and the posterior process
(0); posterior process continuous to the iliac ar-
ticular surface (1) (Dilkes, 1998).

[84] (52) Caudal vertebrae neural spines: height twice
the length (0); height less than twice the length
(1) (Dilkes, 1998).

[85] (53) Interclavicle: posterior process longer than
twice the length of the lateral process (0); pos-
terior process shorter than twice the length of the
lateral process (1) (Dilkes, 1998).

[86] (64) Supinator process on the external surface of
humeral ectepicondyle: absent (0); present and
hook-shaped (1); present and formed by a low

NEW BASAL RHYNCHOSAUR FROM SOUTH AFRICA 587

© 2015 The Linnean Society of London, Zoological Journal of the Linnean Society, 2015, 174, 571–588

D
ow

nloaded from
 https://academ

ic.oup.com
/zoolinnean/article-abstract/174/3/571/2453122 by U

niversidade Estadual Paulista J�
lio de M

esquita Filho user on 06 M
ay 2019



supinator ridge and ligament groove (2)
(Montefeltro et al., 2013).

[87] (54) Posterior process of the coracoid: present (0);
absent (1) (Benton, 1984).

[88] (55) Dorsal margin of the ilium: anterior process
< 15% of the length of the posterior process (0);
anterior process > 15% of the length of the pos-
terior process (1) (Dilkes, 1995).

[89] (56) Pubic tubercle on the lateral surface of the
pubic shaft: present (0); absent (1) (Whatley,
2005).

[90] (57) Internal trochanter: continuous with the
femoral head (0); separated from femoral head (1)
(Whatley, 2005).

[91] (65) Crest on anteromedial region of tibial
shaft: absent (0); present (1) (Montefeltro et al.,
2013).

[92] (58) Relative size of astragalar articular facets:
tibial facet greater than centrale facet (0); centrale
facet greater than tibial facet (1) (Langer &
Schultz, 2000).

[93] (59) Metatarsal I: longer than broad (0); broader
than long (1) (Hone & Benton, 2008).

APPENDIX 2

Synapomorphies of the single most parsimonious tree
recovered in the present analysis.

Rhynchosauria: 3: 0→1; 9: 0→1; 23: 0→1; 60: 0→1;
73: 0→2.

Howesia + Rhynchosauridae: 10: 0→1; 34: 0→1; 59:
0→1; 70: 0→1.

Eohyosaurus wolvaardti + Rhynchosauridae: 13: 0→1;
29: 0→1.

Rhynchosauridae: 5: 0→1; 17: 1→0; 19: 0→1; 40: 0→1;
48: 0→1; 60: 1→2; 61: 0→1; 63: 0→1.

Rhynchosaurus brodiei + Hyperodapedontinae: 11:
0→1; 21: 0→1; 23: 1→0.

Stenaulorhynchinae: 25: 0→1; 38: 0→1; 64: 0→2; 68:
0→1; 70: 1→2; 76: 0→1.

Bentonyx + Hyperodapedontinae: 12: 0→1.
Fodonyx + Hyperodapedontinae: 1: 0→1; 2: 0→1; 5:

1→0; 66: 0→1.
Hyperodapedontinae: 14: 0→1; 16: 0→1; 39: 0→1;

41: 0→1; 55: 0→1; 71: 0→1; 72: 0→1; 74: 0→1.
Teyumbaita + Hyperodapedon spp.: 30: 0→1; 31: 0→1;

52: 0→1; 67: 0→1; 68: 0→1; 69: 1→0; 81: 0→1.

588 R. J. BUTLER ET AL.

© 2015 The Linnean Society of London, Zoological Journal of the Linnean Society, 2015, 174, 571–588

D
ow

nloaded from
 https://academ

ic.oup.com
/zoolinnean/article-abstract/174/3/571/2453122 by U

niversidade Estadual Paulista J�
lio de M

esquita Filho user on 06 M
ay 2019